See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/284010053

An improved grey neural network model for

predicting transportation disruptions

Article in Expert Systems with Applications · October 2015

DOI: 10.1016/j.eswa.2015.09.052

CITATIONS READS

0 62

6 authors, including:

Tong Shu Shou Chen

Hunan University Hunan University
28 PUBLICATIONS 26 CITATIONS 45 PUBLICATIONS 138 CITATIONS

SEE PROFILE SEE PROFILE

Shouyang Wang Kin Keung Lai

Chinese Academy of Sciences City University of Hong Kong
250 PUBLICATIONS 3,587 CITATIONS 612 PUBLICATIONS 6,424 CITATIONS

SEE PROFILE SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Kin Keung Lai
letting you access and read them immediately. Retrieved on: 21 October 2016
 Expert Systems With Applications 45 (2016) 331–340

Contents lists available at ScienceDirect

Expert Systems With Applications

journal homepage: www.elsevier.com/locate/eswa

An improved grey neural network model for predicting

transportation disruptions
Chunxia Liu a,b, Tong Shu a,∗, Shou Chen a, Shouyang Wang a,c, Kin Keung Lai d,e, Lu Gan f
a
Business School of Hunan University, Changsha, Hunan 410082, China
b
Department of Business Administration, Hunan University of Finance and Economics, Changsha, Hunan 410205, China
c
Academy of Mathematics and Systems Sciences, Chinese Academy of Sciences, Beijing 100080, China
d
International Business School, Shaanxi Normal University, Xi’an 710062, China
e
Department of Management Sciences, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong, China
f
Office of Humanities and Social Sciences, Hunan University, Changsha, Hunan 410082, China

a r t i c l e i n f o a b s t r a c t

Keywords: Transportation disruption is the direct result of various accidents in supply chains, which have multiple neg-
Transportation disruptions ative impacts on supply chains and member enterprises. After transportation disruption, market demand
GM(1,1) model
becomes highly unpredictable and thus it is necessary for enterprises to better predict market demand and
Neural network
optimize purchase, inventory and production. As such, this article endeavors to design an improved model
Prediction
of grey neural networks to help enterprises better predict market demand after transportation disruption
and then the empirical study tests its feasibility. This improved model of grey neural networks exceeds the
conventional grey model GM(1,1) with respect to the fact that the raw data tend to show exponential growth
and data variation is required to be moderate, demonstrating the good attribute of nonlinear approximation
in terms of neural networks, setting up standards for selecting the number of neurons in the input layer of
BP neural networks, increasing the fitting degree and prediction accuracy and enhancing the stability and
reliability of prediction. It can be applied to sequential data prediction in transportation disruption or muta-
tion, contributing to the prediction of transportation disruption. The forecasting results can provide scientific
evidence for demand prediction, inventory management and production of supply chain enterprises.
Crown Copyright © 2015 Published by Elsevier Ltd. All rights reserved.

1. Introduction In 1998, the computer breakdown at Hong Kong International Air-

The management of disruption risk in supply chains has been a
priority for enterprises and a focus of research in academia. In re-
cent years, with trends such as global purchases by enterprises, non-
core business outsourcing, single source supply and lean produc-
tion, supply chains have increasingly expanded in terms of space
and have significantly shorter supply cycles. The changes in space
and time increase the probability of occurrence of disruption. Also,
the moments of natural disasters, economic fluctuations, epidemics,
terrorism, wars and others are likely to occur, which renders sup-
ply chains fragile and the probability of supply chain disruption in-
creases. Transportation disruption is one aspect of the risk manage-
ment of supply chain disruption.
∗
Corresponding author. Tel.: +86 13087316938.
E-mail addresses: liucx1999@163.com (C. Liu), shutong@hnu.edu.cn (T. Shu),
chenshou@hnu.edu.cn (S. Chen), sywang@amss.ac.cn (S. Wang), mskklai@cityu.edu.hk
(K.K. Lai), 532339537@qq.com (L. Gan).
port delayed the transport of a multitude of goods and passengers.
The 911 incident resulted in movement of goods on the borders be-
tween the USA and Canada coming to a standstill and Ford Motor
Company was forced to stop production at 5 assembly plants. In 2002
strikes along the western coast of USA resulted in COSCO’s failure to
unload cargo which returned, causing a loss of $24,000,000 in two
weeks. The ice damage in China in 2008, floods in Thailand in 2011
and the Ya’an earthquake in China in 2013 led to paralysis of trans-
portation facilities on a large scale. The frequent occurrence of various
transportation disruptions directly disrupt movement of raw materi-
als of enterprises and production cost increases, affecting the steady
pace of production, health and safety of consumers, and the prosper-
ity and stability of society.
Transportation disruption in supply chains is the direct result of
various accidents such as natural disasters (e.g., ice damage, ocean
disasters, and earthquakes, etc.) and social accidents (e.g., strikes and
terrorist attacks), affecting supply chains and their member enter-
prises negatively. A number of scholars have investigated transporta-
tion disruption in terms of recognition and strategies. Sheffi (2001)

http://dx.doi.org/10.1016/j.eswa.2015.09.052
0957-4174/Crown Copyright © 2015 Published by Elsevier Ltd. All rights reserved.
332 C. Liu et al. / Expert Systems With Applications 45 (2016) 331–340
found that transportation disruption is one of the factors generat-
ing supply chain risks. He stated that risk can be mitigated by in-
creasing the visualization of the transportation process and optimiz-
ing the inventory decisions and partnership. Wilson (2007) examined
the effects of transportation disruption on the attributes of supply
chains by systematic dynamics and found that transportation dis-
ruption affects the first suppliers and their inventory most. Zhang
and Lam (2015) investigate the transportation disruption caused by
typhoons at ports and the loss they cause and finds that loss con-
sists of 4 parts: reputation loss, shippers’ loss, transporters’ loss
and port operators’ loss. Taking China’s railways and airline com-
panies as examples, Ouyang, Pan, Liu, and He (2015) discuss mea-
sures to mitigate the fragility of various transportation systems in
terms of the complementary relationship of the infrastructure sys-
tem. Hishamuddin, Sarker, and Essam, (2013), (2015) consider the
two-stage recovery model of production and inventory in transporta-
tion disruption, and find that the optimal order quantity and pro-
duction can be determined and the total cost can be minimized
during the recovery period. Khademi et al. (2015) take Tehran with
frequent earthquakes as an example, analyze the fragility of trans-
portation networks, and point out that after disasters the demand
for tourism becomes highly uncertain and the traditional methods
cannot be applied to predictions. They recommend some methods
to evaluate the loss caused by earthquakes and recover the trans-
portation. Bravo and Vidal (2013) develop an optimal model for sup-
ply chain goods by mathematical planning and provide an exam-
ple and build a model in terms of abnormal transportation. The
literature discussed above has shown that many researchers of-
fer strategies for transportation disruption from multiple perspec-
tives and note that the market demand is highly unpredictable af-
ter transportation disruption, affecting the production and inventory
significantly.
The uncertainty of market demand after transportation disrup-
tion can be optimized from many aspects such as production, inven-
tory and distribution in supply chain enterprises. Chiu, Lee, Chiu, and
Cheng (2013) explore the supply chain in multinational companies to
ensure that products can be sold in multiple places; quality is guar-
anteed; the optimal quantity of products and means of transportation
can be determined, thus minimizing the cost of production, inventory
and transportation of supply chain members. Based on Generic Bill of
Materials (GBOM), Shu, Chen, Wang, and Lai (2014) investigate the
risk control of supply chain enterprises and tackle the unpredictabil-
ity of production after disruption by combining random simulation
and neural networks. Costantino, Gravio, Shaban, and Tronci (2015)
design the real time statistics process control (SPC) system to coun-
teract the whip effect through control charts, hence achieving the sta-
bility of competitive inventory. The results show that SPC system can
satisfy the principle of smooth complements. Pasandideh, Niaki, and
Asadi (2015) consider a three-tier supply chain consisting of manu-
facturers, distribution centers with uncertain service and customers
to achieve optimization of multiple products and multiple stages in
supply chains. They deploy the model of mixed integral linear plan-
ning and the method of multiple-target decisions, to achieve mini-
mization of total cost and maximization of average quantity of goods
for customers.
Market demand is unpredictable and highly uncertain particularly
after transportation disruption and thus enterprises have to manage
their inventory and production accordingly. Now, if the market de-
mand is well predicted, the cost of inventory and production as well
as risk can be reduced and sales can be optimized. Little research so
far has involved prediction of market demand for supply chain en-
terprises after transportation disruption. For this reason, this article
designs an improved model of grey neural networks to help supply
chain enterprises better forecast market demand after transportation
disruption. Also, an empirical study is conducted to test the feasibility
of the model.
2. An improved prediction model of grey neural networks
Deng (1982) proposed the grey system theory which addresses
samples of some known and some unknown information and is an
uncertain system of inadequate information. The advantages are a
simple modelling process and concise expressions, which are exten-
sively employed in multiple disciplines. By contrast, this prediction
method is suitable for sequences with stable rates of variation. In par-
ticular, when the sequence of data changes dramatically, where devi-
ation varies in terms of prediction results, many scholars have made
improvements accordingly (Chang, Li, Huang, & Chen, 2015; Cristóbal
et al., 2015; Zeng & Liu, 2011).
Neural networks refer to concurrent nonlinear dynamic systems
on a large scale, with strength to handle nonlinear problems. Partic-
ularly, back propagation is most widely employed, in e-commerce,
industry, environmental science and many other fields (Chi & Zhao,
2012; Ray, Mukhopadhyay, Datta, & Pal, 2013; Wang, Zeng, Feng, &
Xia, 2013). In contrast, this method is likely to show flaws such as
local minimizers in longer operations.
There are merits and demerits of each algorithm and thus it is
common to integrate two or more methods and build a combined
model (Bates & Granger, 1969). Some scholars have endeavored to in-
corporate the grey model with other methods to build a combined
model, including particle swarm optimization (Hodzic & Tai, 2016),
DEMATEL method (Rajesh & Ravi, 2015), uncertainty theory (Memon,
Lee, & Mari, 2015), multivariate model (Intharathirat, Salam, Kumar,
& Untong, 2015), fuzzy mathematics method (Dewangan, Gangopad-
hyay, & Biswas, 2015) and Markov method (Xie, Yuan, & Yang, 2015),
among many others. A number of researchers have attempted to inte-
grate grey prediction model and neural network to handle nonlinear
problems, tackle unpredictable demand and increase prediction pre-
cision (Alvisi & Franchini, 2012; Hsu, 2011; Pai et al., 2015; Zhang, Xu,
Cui, He, & Tian, 2012).
It can be observed that in the combination of grey model and neu-
ral network model, the number of neurons in the input layers of BP
neural networks can be either 1 or random, which can control the in-
trinsic errors of the prediction model of neural networks, whereas
the deviation of grey model and neural network model cannot be
controlled at the same time. This leads to failure to optimally con-
trol the total deviation. As such, based on the grey model and the
neural network model, this article designs an improved prediction
model of grey neural networks, setting up the standards for choos-
ing the number of neurons in the input layer of BP neural networks,
namely, the optimal number of dimensions of the improved grey pre-
diction model is the figure for neurons in the input layer of BP neural
networks. This improved prediction model for determining the num-
ber of neurons in the input layer of BP neural networks exceeds the
conventional grey GM(1,1) which has strict requirements for the se-
quence of raw data, demonstrating the good attribute of nonlinear
approximation in terms of neural network, increasing the fitting de-
gree and prediction accuracy, and enhancing the stability and relia-
bility of prediction. It can be applied to sequential data prediction in
transportation disruption or mutation.
2.1. A grey prediction model
2.1.1. Traditional GM(1,1) model
GM(1,1) is a fundamental prediction model based on the grey the-
ory. GM is formed by the initial letters of the grey model; regard-
ing (1.1), the first 1 means that the prediction model only consists
of a 1-step equation and the second 1 means that there is only one
dependent variable in the model. As such, GM(1.1) is based on the
premise that 1-step equation and 1 dependent variable are satisfied.
In the Grey prediction model GM(1,1), the original sequence is cumu-
latively produced as the time sequence increases progressively. The
corresponding approximate differential equations are constructed to
 C. Liu et al. / Expert Systems With Applications 45 (2016) 331–340 333

display the development of data sequence. What follows are the pro- n − 1 equations are insolvable in terms of two unknown co-
cesses of model construction and solutions. efficients (a and b), and then the method of minimal squares
is used to achieve  = (BT B)−1 BT Y , and the approximate solu-
(1) If the original time sequence is given as x(0) = (x(0) (1),
tions â and b̂ can be achieved.
x(0) (2), . . . , x(0) (n)), in which n is the figure for the origi-
(5) Put â and b̂ into the grey differential equation, and the discrete
nal data and x(1) is cumulatively generated data in relation
corresponding functions are achieved:
to x(0) , i.e., 1-AGO, then x(1) = (x(1) (1), x(1) (2), . . . , x(1) (n)) is 

achieved, in which x(1) (i) = ni=1 x(0) (i), i = 1, 2, . . . , n. If there x
(1) (1) = x(0) (1)
 i≥1 (5)
x
is no special notification, value i will remain the same in the (1) (i + 1) = b̂
â
+ e−âi x(1) (0) − b̂
â
following. Here, GM(1.1) model is shown as follows:
x(0) (i) + ax(1) (i) = b(a, b areunderminedcoefficients) (1) (6) x
(1) (i + 1) is recovered by cumulative reduction and the pre-
diction value of the original data is achieved.
(2) Test the quasi-smoothness of x(0) and the quasi-index pat- 
tern of x(1) . The sequence smoothness can be achieved by x
(0) (1) = x(0) (1)
i≥1 (6)
continuous functions while if the time sequence is com- x (i + 1) = x
( 0) (1) (i + 1) − x
(1) (i)
posed of discrete individual points, it is difficult to obtain
smoothness by this derivation. Here, the time sequence is (7) Precision test
supposed to show the attribute of being smooth, and the
x(0) (i+1) p(i+1)
• Pointwise test
quasi-smooth coefficient is p(i + 1) = . As p(i)
< 1 Absolute error: ε (0) (i) = x(0) (i) − x
(0) (i)
x(1) (i)
ε(t )
and p(i) < 0.5 are satisfied, the original time sequence x(0) is Relative error: e(i) = (0) × 100%
x (i)
considered a quasi-smooth sequence (Liu & Guo, 1991). The • Posterior difference ratio
normal non-negative quasi-smooth is derived by accumulat- 
1   (0) 2 1  (0)
n n
ing it once and the randomness is reduced, showing a gen-
S1 = x (i) − x̄(0) , in which x̄(0) = x (i)
eral trend of exponential growth. p(i + 1) < 0.5 together with n n
(1) (0) (i+1)+x(1) (i) i=1 i=1
σ (1) (i + 1)= x x(1()i+1
(i)
) x
=
x(1) (i)
= 1+ p(i + 1) < 1.5 are 
given; here x(1) can be derived by accumulating the original 1 
n  2 1 
n
S2 = ε(i) − ε̄ (0) in which ε̄ (0) = ε (0) (i)
time sequence x(0) once, which satisfies the quasi-index pat- n n
i=1 i=1
x(1) (i+1)
tern. The quasi-index pattern target is σ (1) (i + 1) = . If
x(1) (i)
S2
σ (1) (i) ∈ 1, 1.5 is given, then the quasi-index pattern is gen- C=
S1
erally satisfied (Liu & Guo, 1991).
(3) If p(i + 1) < 0.5, σ (1) (i + 1) ∈ 1, 1.5 is given, then the albe- Accuracy range is acceptable if C ∈ [0.35, 0.65] (Wa, 2010). When
faction differential equation of GM(1,1) is: C becomes smaller, the range of prediction errors decreases and
therefore the accuracy increases.
dx(1)
+ ax(1) = b (2) • Frequency of minor errors
dt
P = {P |ε(i) − ε̄| < 0.6745S1 }
where, a is the coefficient of x(1) and dx(1) is the differential of x(1) ,
and thus a can show the tendency of x(0) and x(1) . If a is grey develop- Accuracy range is acceptable if P ∈ [0.7, 0.95] (Wa, 2010). When
ment scale and b is the data derived from the background value, in- P becomes larger, the probability of minor errors increases and
dicating the changing patterns of the data, this can be viewed as the therefore the accuracy increases.
significant evidence to differentiate the grey model and the model
based on the general input and output. If b is the grey action quan- (8) Test of fitting degrees
tity; t indicates time; and x(1) (i) − x(1) (i − 1) = x(0) (i), x(0) (i) is the
(1) Grey correlation analysis involves judging degrees of closeness by
grey derivative. (dxdt ) + ax(1) = b is the simple transformation from determining approximation degrees of geometrical shapes related to
x(0) (i) + ax(1) (i) = b in Eq. (1). The background value x(1) (i) is the grey the original and forecast sequences. Then, the tendency of dynamic
number; x(1) (i) and grey derivatives do not satisfy the relationship development of the original sequence is quantified and analyzed by
of flat projection. As the time sequence consists of discreet individ- comparison. The procedure is to make the prediction and original se-
ual points, there are boundary points rather than inner points and quence non-dimensional, and to calculate the correlation coefficients
thus it is vacant between points. For this reason, the inner points are and degrees. Next, the closeness degree is determined by the correla-
derived from the mean value of the right and left boundary points. tion degree (Table 1).
Here, x(1) (i) and the grey derivative x(0) (i) fail to satisfy the rela- In the first step, variables become non-dimensional. Then calcu-
tionship of flat projection. The traditional method is that the se- late the initial image of the original and forecast sequence. Here, the
quence x(1) uses the adjacent mean value to derive the sequence w(1) , initial image is selected.
that is, w(1) = (w(1) (1), w(1) (2), . . . , w(1) (n)), in which w(1) (i + 1) =
0.5x(1) (i) + 0.5x(1) (i + 1). The albefaction differential equation in (2)  x(0)  x
(0)
x(0) = (0) , x
(0) =
can be transformed into the following grey differential equation: x ( 1) 
x(0) (1)
x(0) (i) + aw(1) (i) = b (3) In the second step, calculate the related number ξ i :

(4) Construct min min |y(k) − xi (k)| + ρ max max |y(k) − xi (k)|
⎡ (1matrix B, ⎤constant matrix Y and matrix A, in which
⎡ ⎤
−w ) (2) 1
ξi (k) = i k i k
x(0) (2) |y(k) − xi (k)| + ρ max max |y(k) − xi (k)|
⎢ . .⎥
.. ⎦, Y = ⎣ . . . ⎦, A =
a i k
B=⎣ .. , and therefore
b
( ) x (0) (n ) where ρ is the resolution ratio. If ρ becomes smaller, the resolution
−w (n)
1 1
ratio increases. If ρ ≤ 0.5463, the resolution ratio is the best. Gener-
Eq. (2) can be equally changed into:
ally, if ρ = 0.5 (Liu, Dang, & Fang, 2010), the calculation remains the
Y = BA (4) same here.
334 C. Liu et al. / Expert Systems With Applications 45 (2016) 331–340
Table 1
Accuracy test.
Levels Results Average relative error (ARE)
1 Very good ≤ 0.01
2 Good ≤ 0.05
3 Ordinary ≤ 0.1
4 Disqualified ≤ 0.20
In the third step, calculate the correlation degree ς i :
1 n
ςi = ξi (k), k = 1, 2, . . . , n
n k=1
In the fourth step, consider the fitting degree. If ς i ≥ 0.6, the fitting
degree is good and satisfies the fitting requirements. If ς i < 0.6, the
fitting degree is worse and cannot satisfy the fitting requirement (Li
& Hu, 2005).
2.1.2. An improved GM(1,1) model
In building of the actual prediction model, the forecasting data are
more likely to be affected by short-term data rather than the long-
term data. As a result, if all data in the original sequence are used
to build the model, forecasting results are excessively affected and
interfered by long-term data, particularly when the long-term data
are mutated and disrupted, and thus the recent changeable pattern
does not show. In order to improve the traditional Model GM(1,1),
the index pattern should be satisfied or approximately satisfied; the
speed of change of data sequence should not be too fast; the fitting
degree and forecasting accuracy should be improved; stability and
reliability should be enhanced; and the isochronal sequence within
x(0) (n) should be included in data for building the models. As such,
two methods are used to build the models. Firstly, change of number
of dimensions shows that x(0) (n) is the termination of the value se-
lected for building the models and then the values selected forward
can form a set of forecasting models. The optimal model can be se-
lected from the prediction models for future prediction, which can
be called the changeable starting point model. Another point to note
is the fixed number of dimensions. That is to say, the oldest informa-
tion is deleted and the latest information is embedded for continuous
metabolism, which is called the metabolism model.
The model building and solutions of the changeable models are
shown as follows.
(1) The original time sequence is x(0) = (x(0) (1), x(0) (2), . . . ,
x(0) (n)), in which n is the number of the original data;
(2) Model GM(1,1) built on x(0) = (x(0) (1), x(0) (2), . . . , x(0) (n)) is
the complete data model; and Model GM(1,1) is built on x(0) =
(x(0) (1), x(0) (2), . . . , x(0) (n)), i ≥ 2 is the partial data model;
(3) The number of the least time sequence is 4 (Deng,
2005), and thus a four-dimensional sequence is built:
(x(0) (n − 3), x(0) (n − 2), x(0) (n − 1), x(0) (n)). This continues
one by one, and finally n-dimensional sequence can be built:
(x(0) (1), x(0) (2), . . . , x(0) (n)), the original time sequence, and
n − 3 sub-sequences can be built in total;
(4) Make predictions by each sub-sequence in the traditional
Model GM(1,1) and then test the fitting degree and accuracy,
and select the optimal models to make predictions from the
set of models.
The model building and solutions of the metabolism models are
shown as follows.
(1) The original time sequence is x(0) = (x(0) (1), x(0) (2), . . . ,
x(0) (n)), in which n is the figure for the original data;
(2) n − 3 dimensions in the n-dimensional original time sequence
are fixed, such as four dimensions, five dimensions, till n-
dimension;
Posterior difference ratio C Frequency of minor errors P
<0.35 >0.95
<0.50 >0.80
<0.65 >0.70
≥ 0.65 ≤ 0.70
(3) With fixed dimensional numbers, such as 5-dimension, the
model set with n − 4 sub-sequences can be built.
 (0) 
x (1), x(0) (2), x(0) (3), x(0) (4), x(0) (5)
 (0) 
x (2), x(0) (3), x(0) (4), x(0) (5), x(0) (6)
···
 (0) 
x (n − 4), x(0) (n − 3), x(0) (n − 2), x(0) (n − 1), x(0) (n)
The treatment methods of other dimension numbers are ex-
actly the same.
(4) Make predictions by sub-models of the fixed dimensional
numbers in the traditional models GM(1,1) and test the fit-
ting degree and accuracy, determine the corresponding mean
value and then compare each mean value, to select the optimal
model to make predictions from the model set.
2.2. The improved neural network prediction model
2.2.1. Back-propagation neural network
At the moment, in actual use, most neural network models adopt
either feed-forward back-propagation network (BP network) or oth-
ers, including the multi-layer feed-forward error back-propagation
network, containing input, output and other implicit layers. Each
layer is made up of a number of neurons and the super-stratum and
substratum neurons are connected via weight; complete connection
is used to combine the neurons between layers; and there is no con-
nection between neurons within layers. When the learning samples
are provided for the neural network and are then spread forward
through advanced line in the network, if the error between the out-
put results and target output is beyond expectations, this forward
spreading process reverses to the backward process. The mistaken
signals return through the original route and the mistaken signals are
decreased by modifying the weight of neurons in each layer. As the
modification of the error of backward spreading continues, the corre-
sponding accuracy of the network to the input mode increases until
it meets the requirements.
BP network consists of the input, output and several explicit lay-
ers. Cybenko (1989) and Ito (1991) showed that the three-layer BP
neural networks can satisfy the omnipotent approximation theorem,
i.e., in the three-layer neural networks containing one explicit layer,
if the number of the explicit points is sufficient, any continuous func-
tion can be approximated within the boundaries, which, however,
does not mean that the three-layer neural networks are the most rea-
sonable. In terms of the same target function, Tamura and Tateishi
(1997) showed that less neurons are used in the four-tier structure
than in the three-layer neural structure. The implicit layer is more
than one layer and can improve the network training in terms of
speed. However, it takes longer for training in actual applications. The
training speed can be achieved by increasing the number of implicit
layers which simulates nonlinear continuous projections. Therefore,
three-layer BP neural network with one implicit layer is selected for
prediction here. The input layer of the BP neural networks is the sam-
ple data, while the output layer is the prediction results.
(1) The design of the model structure and functions selected
What BP model actually completes is the projection of input sam-
ple space on the output expectation space: the output sample of
ai → bj is i-dimension; the output expectation is j-dimension; and
 C. Liu et al. / Expert Systems With Applications 45 (2016) 331–340 335

the dimensional number is the figure for the corresponding neu- 2.2.2. The improved back-propagation network
rons. The dimensional numbers of the input, output space show the • Attached momentum method

characteristics of vectors, and the number of neurons in each layer
is determined by features of the historic data and prediction in the
training.
If there are fewer neurons in the implicit layer, it is very difficult
for the network to identify the sample and the training is hard to com-
Rumelhart et al. (1986) put forward an approach to improve the
BP training time to ensure process stability. This method adds the
previous weighting variation of a positive proportion to each weight
modulating value. The specific representation is as follows.
∂ ET
plete. At the same time, the fault tolerance declines. If the number is ω(t + 1) = ω(t ) − (1 − α)η + α ω(t )
bigger, then the iterations increase and it takes longer to train the
∂ω(t )
network. Meanwhile, the universality of the network decreases, low-
ering the prediction capacity. In the specific design, the first step is
to determine the number of neurons in the implicit layer in terms of
the experiential formula. Next, training and contrast are followed for
different numbers of neurons. Finally, the number of neurons is de-
termined. The number of neurons in the implicit layer is determined
by Kolmogorov Theory (Qian & Gong, 2000). If i is the number of neu-
rons in the implicit layer and n is the number of neurons in the input
layer, the equation is as follows.
i = 2n + 1
The function of the neuron transmission in the network implicit
layer is Sigmoid (Rumelhart, Hintont, & Williams, 1986); the neu-
ron transmission in the input layer adopts the linear function Pure-
lin (Tian, 2009); and the training function uses the Trainbr approach
based on the Bayes algorithm (Tian, 2009), which can increase net-
work universality.
(2) The pre-treatment of sample data
In reality, most data are incomplete, noisy and inconsistent and
need to be packed by accumulating, deleting redundant features or
clustering; it is necessary to normalize the input and output sample
data in the network. That is to say, the quantitative value is limited in
the range of [0.1, 0.9]; the normalizing equation is shown as follows.
 w(0) (k) − w(0) min
 (0) (k) = 0.1 + (0.9 − 0.1)
w(0) max − w(0) min
(3) The selection of initial weighting and learning rate
In BP neural networks, the initial weight of the network is distinct
and thus the training results differ, which is caused by numerous local
minima related to errors in the curved surface. Generally speaking,
the initial weight of the network is a small random value, ensuring
each neuron works in the maximal range of the motivated function
slope variations. Also, it can prevent infinite changes of some unrea-
sonable weight after adequate continuous learning. Thus, the random
number (−1, +1) is commonly selected.
The learning rate decides the weight variation generated in each
cycle of training. A larger learning rate contributes to system instabil-
ity whereas a smaller learning rate costs longer training time, which
means very slow convergence. Thus, the smaller learning rate is se-
lected to ensure system stability and the range (0.01, 1) is generally
selected.
(4) Model evaluation index
The performance assessment indices of the model prediction in-
clude absolute errors of the predicted and actual values and relative
errors of the predicted and actual values and r coefficient.
where ω(t ) = ω(t ) − ω(t − 1); α is the momentum factor with
0.95 as the normal value. Here, the modification of weight value is
added to the memory of the modification direction for the previous
moment of weight value.
When the attached momentum method is used to modify the
weight of the network, the function of error on slopes is also consid-
ered and the influence of the changing tendency of the error curved
surface is also taken into account. The function is similar to a low
filtering device which ignores minor changes. Without the attached
momentum, the network might fall into local minima. The function
of the attached momentum is to skip the minimal values. The nature
of the attached momentum is to pass the effect of the final weight
variation through the momentum factor. In the training process, the
conditions to determine the attached momentum method are as
follows.
0, SSE (t ) > 1.04SSE (t − 1)
α= 0.95, SSE (t ) < SSE (t − 1)
α , SSE (t − 1) ≤ SSE (t ) ≤ 1.04SSE (t − 1)
where SSE is the square sum of the network output errors.
• Self-adaption learning ratio
Generally, the principles of modulating learning rate involve
checking whether the corrected weight reduces the error function ac-
tually. If so, the learning rate is small and some amount can be added.
If not, then the result is redundancy, which means the learning rate
must be reduced. This method can ensure that the network can use
the maximal learning rate.
The modulation equation of the self-adaptation ratio is as follows
(Tian, 2009).

0.75ηt, SSE (t ) > 1.04SSE (t − 1)
η(t + 1) = 1.05ηt, SSE (t ) < SSE (t − 1)
ηt, SSE (t − 1) ≤ SSE (t ) ≤ 1.04SSE (t − 1)
The initial learning rate is highly random, implying that the num-
ber of networks trained by the self-adaptive learning rate is only a
fraction of networks trained at the fixed learning rate. Therefore, net-
work training at self-adaptive learning rate is extremely valid.
2.3. An improved prediction model of grey neural networks in
transportation disruption
The number of neurons in the input layer of BP neural networks
can be either 1 or random, which is not vigorous or scientific. An im-
proved prediction model of grey neural networks provides a scientific
method to select the number of neurons in the input layer of the BP
neural networks. In other words, high precision and fitting degrees in
Absolute error: ε (0) i = x(0) i − x
(0) i
the optimal prediction model is selected from grey prediction model,
Relative error: e(i) = |ε(0t)| × 100% determining the optimal number of dimensions, which is the figure
x i


  for neurons in the input layer of BP neural networks.
x(0) i − x(0) 1 x
(0) i − x
(0) 1
r=  2 2 3. An improved prediction model of grey neural networks in
x(0) i − x(0) 1 x
(0) i − x
(0) 1 transportation disruption and its empirical study

3.1. Background and data source

In which, x
(0) i is the predicted value of the BP neural network;

x i is the actual value; x(0) 1 d x(0) i is the mean value; and x

(0) (0) 1 is At the beginning of 2008, South China experienced the coldest
the mean value of x i.(0) weather in the preceding 50 years. Hunan province was most affected
336 C. Liu et al. / Expert Systems With Applications 45 (2016) 331–340

Table 2 self-equipped electricity generators can be used to resume produc-

Weekly effective sales (units) between 1st January and
tion. The raw materials inventory of the factory was sufficient and,
30th March in HX factory.
therefore, production was not greatly affected. As the road surface
Time Sales Time Sales Time Sales was frozen, effective sales (units) were still influenced significantly.
1 5285 6 4489 11 4900
Table 2 shows the weekly effective sales (units) of chairs in the Hu-
2 5313 7 4512 12 5215 nan HX Office and School Chair Furniture Factory from 1 January to
3 4492 8 4583 13 5340 30 March in 2008.
4 4480 9 4680 – –
5 4486 10 4795 – – 3.2. Modelling and an empirical study

3.2.1. The construction of traditional model GM(1,1) in transportation

disruption
in this snow disaster. It was estimated that 25,220,000 people in 14
(1) The sales sequence value is established as x(0) (Table 3); the
cities and towns and 112 counties were affected. Seven people died;
cumulative sequence 1-AGO is established as x(1) (Table 3);
167,000 people were relocated urgently; and 1,286,000 people could
(2) Test the quasi-smoothness of x(0) and the quasi-index pattern
not get clean water. An estimated 21,000 rooms in 31,000 houses col-
of x(1) . If i > 2, then p(i) < 0.5, σ (1) i ∈ 1, 1.5, and the require-
lapsed, affecting 8275 households. 131,000 rooms were damaged. The
ments of the quasi-smoothness and the quasi-index pattern
affected agricultural area was 1,293,000 square meters and the di-
are met, shown as p(i) and σ (1) i, respectively, in Table 3;
rect economic loss was 151.65 billion RMB. This snow disaster led
(3) Construct the matrix B and the constant matrix Y and the con-
to extensive paralysis of traffic facilities, particularly highways and
struction approach is the same as above and −w(1) (i) can be
superhighways in provinces. Some highways experienced traffic dis-
found in Table 3;
ruptions for a long period of time and the traffic ministry and traffic
(4) Apply LINEST in Excel and achieve â = −0.0092,
department in Hunan Province re-distributed traffic. Also, this snow
b̂ = 4467.5298;
disaster caused breakdowns, disconnections, lightening and tripping
(5) Put â and b̂ into the grey differential equation and achieve the
in many power transmission networks. There were several develop-
corresponding discrete function:
ment stages in this snow disaster. In the development stage (from 13 
to 15 January), the freezing areas in South China expanded to Hu- x
(1) (1) = 5285
 i≥1
nan Province where temperatures dropped to approximately zero. ( )
x i + 1 = 493008.6660e0.0092i − 487723.6660
1
In the reducing stage (from 16 to 17 January), snow remained in
the southeast of Hunan Province and the overall degree of freezing (6) Recover x (1) i + 1 cumulatively, and achieve the prediction
decreased. In the further development stage (from 18 January to 2 value of the original data:

x
February), snow spread from the west of Hunan and Guizhou to the (0) (1) = 5285
east and south of China, and the intensity increased gradually; snow 
 , i≥1
x( ) i + 1 = 493008.6660e0.0092i−1 e0.00492 − 1
0
was widespread throughout Hunan Province. In the declining stage
(from 3 to 6 February), the freezing area declined, snow in the south- (7) Accuracy test
west of Hunan decreased and the intensity fell obviously. From 5 to 6
February, the remaining snow disappeared gradually in the southeast
• Pointwise test
of Hunan. The relative error e(i) and the absolute error ɛ(0) i can be 
seen in
e
(i)
Hunan HX Office and School Chair Furniture Factory is engaged Table 3. By calculation, the mean relative error is e(i) = n =
in research and manufacture of modern office and school furniture, 4.2079%, which indicates that the prediction error is mainly
including various school desks and chairs, canteen tables and chairs, 4.2079% (approximately). The prediction error is generally low,
apartment beds, computer desks, soft and hard benches, step ladder and the accuracy is high and satisfies the requirements of Level
chair, office and school furniture for large and small colleges across all 2.
provinces of China. HX school furniture adheres to the international • Posterior difference value C
quality standard ISO9001, the environment standard ISO14001 and By calculation, the results are S1 = 84.0765, S2 = 70.6851, C =
S2
professional health safety management system T28001. The quality S1 = 0.8407; and C ≥ 0.65 shows that the prediction error range
of products is guaranteed and supply, transportation and installation is large, with low accuracy, and thus they fail to pass the test.
are integrated; the corporate spirit of “truth, innovation, inspiration • Frequency of minor errors P
and perfect product structure” has been implemented. During the By calculation, the results are ε̄ = 4812.85, p = 0.3077, indicating
snow period, in response to the disruption in electricity supply, the that prediction errors increase as the probability rises, with low

Table 3
Calculation processes and results of traditional model GM(1,1).

Time x(0) x(1) p(i) σ (1) (i) w(1) (i) ɛ(0) (i) e(i) x
(0)

1 5285 5285 – – – 0.0000 0.0000 5285

2 5313 10,598 1.0053 2.0053 −7941.50 776.2942 0.1461 4536.7058
3 449,244 15,090 0.4239 1.4239 −12844.00 −86.4529 0.0192 4578.4529
4 80 19,570 0.2969 1.2969 −17330.00 −140.5842 0.0314 4620.5842
5 4486 24,056 0.2292 1.2292 −21813.00 −177.1032 0.3948 4663.1032
6 4489 28,545 0.1866 1.1866 −26300.50 −217.0134 0.0483 4706.0134
7 4512 33,057 0.1581 1.1581 −30801.00 −237.3186 0.0526 4749.3186
8 4583 37,640 0.1386 1.1386 −35348.50 −210.0222 0.0458 4793.0222
9 4680 42,320 0.1243 1.1243 −39980.00 −157.1280 0.0336 4837.1280
10 4795 47,115 0.1133 1.1133 −44717.50 −86.6396 0.0181 4881.6396
11 4900 52,015 0.1040 1.1040 −49565.00 −26.5608 0.0054 4926.5608
12 5215 57,230 0.1003 1.1003 −54622.50 243.1045 0.0466 4971.8955
13 5340 62,570 0.0933 1.0933 −59900.00 322.3528 0.0604 5017.6472
 C. Liu et al. / Expert Systems With Applications 45 (2016) 331–340 337

Table 4 Table 6
Calculation processes and results of testing fitting degrees. Calculation results of the 5-dimension metabolism model GM(1,1).

Time x(0) x
(0) x(0)  x
(0) ξi Location of original e(i) C P ς
sequential data selected
1 5285 5285 1.0000 1.0000 1.0000
2 5313 4536.7058 1.0053 0.8584 0.3578 1–5 0.0340 0.4906 0.8 0.5806
3 4492 4578.4529 0.8455 1.0092 0.3333 2–6 0.3956 2.6850 0 0.2540
4 4480 4620.5842 0.9973 1.0092 0.8733 3–7 0.0021 0.5935 0.8 0.4686
5 4486 4663.1032 1.0013 1.0092 0.9124 4–8 0.0033 0.4481 1 0.7136
6 4489 4706.0134 1.0007 1.0092 0.9056 5–9 0.0033 0.2381 1 0.6724
7 4512 4749.3186 1.0051 1.0092 0.9525 6–10 0.0017 0.0805 1 0.5431
8 4583 4793.0222 1.0157 1.0092 0.9261 7–11 0.0005 0.0207 1 0.6232
9 4680 4837.128 1.0212 1.0092 0.8725 8–12 0.0076 0.2175 1 0.6036
10 4795 4881.6396 1.0246 1.0092 0.8419 9–13 0.0063 0.1597 1 0.5942
11 4900 4926.5608 1.0219 1.0092 0.8657 Mean 0.0505 0.5482 0.8444 0.5615
12 5215 4971.8955 1.0643 1.0092 0.5977
13 5340 5017.6472 1.0240 1.0092 0.8472

is met, but other sub-sequences do not meet the require-

accuracy. It fails to achieve the minimum of 0.7, and thus it does
not pass the test.
(8) Test of fitting degrees
Calculate the initial value (or mean value) of the original and pre-
diction sequence, as shown in Table 4. Calculate the correlation coeffi-

cient, as shown in Table 4. The results are ςi = 1n nk=1 ξi (k) = 0.7912,
suggesting that the fitting degree is good and satisfies the fitting re-
quirement.
The calculation process and results show that the required accu-
racy cannot be met due to the influence and interference of long-term
data mutation and disruption, although the test of fitting degree is
passed. This cannot demonstrate the recent change pattern.
ment. Comparison of the 7-dimension and 4-dimension sub-
sequences shows that indices (e(i) and ς ) of the 4-dimension
sub-sequence are better than 7-dimension sub-sequence. 4-
dimension sub-sequence is worse than 7-dimension sub-
sequence in index (C), but they are the same in index (P). The
4-dimension sub-sequence should be selected based on the
results. However, results of prediction of transportation dis-
ruption and mutation emphasize the stability and sequence of
the data. Thus, the fluctuation range of errors should be min-
imal and the posterior difference value should be the focus
of transportation disruption and mutation. Here, 7-dimension
and 4-dimension sub-sequences can be taken as the prediction
models.
• Metabolism model

(1) There are 10 fixed dimensions in the 13-dimension original

3.2.2. Constructing an improved GM(1,1) model in transportation
disruption
• Changeable starting point model
(1) With n = 13, 10 sub-sequences can be constructed.
4-dimension sequence x(0) = (x(0) (10), x(0) (11), x(0) (12),
x(0) (13));
5-dimension sequence x(0) = (x(0) (9), x(0) (10), x(0) (11), x(0) (12),
x(0) (13));
In turn, 10 sub-sequences can be constructed.
(2) Make predictions of each sequence on the basis of the tradi-
tional model GM(1,1) and then each model can be built, as
shown in Table 5. Calculate the mean relative error e(i) and
posterior difference value C and the minor error frequency P
can be found in Table 5. Calculate the grey correlation value ς ,
as shown in Table 5.
(3) Compare the accuracy and fitting degree of each model. 7-
dimension and 4-dimension sub-sequences can satisfy the re-
quirements of Level 1. The test requirement of fitting degree
time sequence.
(2) The fixed dimensions, such as the 5-dimension, can be calcu-
lated on the basis of the traditional model GM(1,1), as shown in
Table 6, and the corresponding mean value can also be calcu-
lated. The treatment approach of other dimension numbers is
exactly the same and hence only the final results are presented
here (Table 7).
(3) Compare each average, and 7-dimension is constantly consid-
ered as optimal. The 7-dimension sequence is selected for pre-
diction, and the results are exactly the same as the changeable
starting point model.
It is important to note that the seven-dimension sub-sequence
derived by the model of changing starting points is x(0) = (x(0) (7),
x(0) (8), x(0) (9), x(0) (10), x(0) (11), x(0) (12), x(0) (13)), with an average
relative error of 0.7521%. There are 6 sets of the seven-dimension
sub-sequence derived by the metabolism model, whereas the seven-
dimension sub-sequence derived by the model of changing starting
points is one of them, whose average relative error is the relative error
of the six corresponding seven-dimension sub-sequences, and their

Table 5
Calculation results of transferable starting point model in disruptions.

Numbers of sequential dimensions Model e(i) C P ς

13 493008.6660e 0.0092i
− 487723.6660 0.0421 0.8165 0.3077 0.7912
12 23568.5028e0.0181i − 23555.5028 0.0215 0.3567 1 0.5246
11 207259.2003e 0.0206i
− 202767.2003 0.0186 0.3453 1 0.5145
10 183916.9932e0.0233i − 179436.9932 0.0160 0.3081 1 0.4865
9 163827.0572e0.0264i − 159341.0572 0.0122 0.2329 1 0.5323
8 148746.9916e0.0294i − 144257.9916 0.0097 0.1871 1 0.5609
7 138779.0351e0.0321i − 134267.0351 0.0075 0.1662 1 0.6056
6 254940.5447e0.0187i − 250357.5447 0.0205 0.4137 1 0.4907
5 121492.6285e0.0395i − 116812.6285 0.0063 0.1597 1 0.5942
4 113815.0285e0.0424i − 109020.0285 0.0062 0.1831 1 0.6162
338 C. Liu et al. / Expert Systems With Applications 45 (2016) 331–340

Table 7 and the corresponding 7 neurons in the second set are (x(0) (2), x(0) (3),
Calculation results of metabolism model GM(1,1).
x(0) (4), x(0) (5), x(0) (6), x(0) (7), x(0) (8)). In turn, the corresponding 7
Numbers of sequential dimensions e(i) C P ς neurons in the final set are (x(0) (7), x(0) (8), x(0) (9), x(0) (10), x(0) (11),
x(0) (12), x(0) (13)). The corresponding neuron in the output layer is
13 0.0421 0.8165 0.3077 0.7912
12 0.0319 0.5990 0.9167 0.5800
the prediction value, and x(0) (8) can be predicted according to the
11 0.0327 0.3634 0.7273 0.5619 corresponding 7 neurons in the first group. In turn, x(0) (9), x(0) (10),
10 0.0211 0.4261 0.9000 0.5692 x(0) (11), x(0) (12) and x(0) (13) could be predicted. The neuron trans-
9 0.0169 0.3673 0.9556 0.6059 mission of the network explicit layer uses the Sigmoid function; the
8 0.0131 0.3167 0.9583 0.6067
neuron transmission of the network output layer uses the linear Pure-
7 0.0107 0.2927 0.9796 0.6166
6 0.0108 0.3115 0.8958 0.5705 lin function; and the training uses Trainbr on the basis of the Bayes
5 0.0505 0.5482 0.8444 0.5615 algorithm. The precision of the training target is 0.005; the terminal
4 0.0085 0.3342 0.9400 0.5652 training steps are 6000; and by 2705 times of training, the prediction

results are (x (0) ); the normalization results are ( (0) (k)); and the
index of model evaluation can be found in Table 8.
Table 8
Results and processes of the improved grey neural network
prediction model. 3.3. A comparison of prediction results

Time x (0)
x
(0) ϖ (0) 
(k) ɛ (i)
(0)
e(i)
Comparison of the improved grey neural network model, the
1 5285 – 0.8488 – – traditional GM(1,1) and the improved GM(1,1) model (the model
2 5313 – 0.8749 – – of changing starting points is used instead for comparative pur-
3 4492 – 0.1112 – – poses), prediction values of the traditional GM(1,1), improved GM(1,1)
and the improved grey neural network model are x (0) (1), x
4 4480 – 0.1000 – –
(0) (2)
5 4486 – 0.1056 – –
6 4489 – 0.1084 – –

( )
dx (3), respectively and the corresponding relative errors are e(1),
0
7 4512 – 0.1298 – – e(2) and e(3), respectively. Each prediction value and the correspond-
8 4583 4584.8 0.1958 −1.8 0.0004
ing relative error can be found in Table 9.
9 4680 4680.2 0.2860 −0.2 0.0000
10 4795 4825.5 0.3930 −30.5 0.0064 The comparisons show that prediction accuracy of the improved
11 4900 4955.7 0.4907 −55.7 0.0114 model of grey neural networks is far greater than that of the tradi-
12 5215 5211.4 0.7837 102.6 0.0007 tional GM(1,1) model and that it is also superior to prediction accu-
13 5340 5354.4 0.9000 −14.4 0.0027 racy of the improved GM(1,1) model. As such, the improved model of
grey neural networks is a feasible prediction method for transporta-
tion disruption.
average relative error is 1.07%. Hence, there is a difference between
the average relative error in the model of changing starting points 4. Conclusions
and the model of metabolism.
Transportation disruption is the direct result of varied accidents
3.2.3. An improved prediction model of grey neural networks in such as natural disasters (e.g., ice damage, ocean disasters and earth-
transportation disruptions quakes, etc.) and social accidents (e.g., strikes and terrorist attacks),
The BP neural network consists of input, output and several ex- exerting multiple negative effects on supply chains and their member
plicit layers. Comparing the traditional GM(1,1) model, the model of enterprises, even leading to paralysis of supply chains. Market de-
changing starting points and the model of metabolism, it is optimal mand after transportation disruption is highly unpredictable while
to choose the seven-dimension sequence for the neural network pre- production and inventories of enterprises are affected to a large
diction. Therefore, the input layer consists of seven neurons: 4512, extent.
4583, 4680, 4795, 4900, 5215 and 5340. The implicit layer contains This article proposes an improved model of grey neural networks
15 neurons; the output layer contains 1 neuron, and the output is the for prediction of demand after transportation disruption. Firstly, the
ultimate prediction results. So the network typology constructed is conventional GM(1,1) model is deployed to predict the sequence of
7-15-1. It is important to note that 7 neurons are input into the out- transportation disruption; the inaccurate prediction is due to the mu-
put layer once, and there are 6 sets. The corresponding 7 neurons in tation and instability of the data. Secondly, based on this, GM(1,1)
the first set are (x(0) (1), x(0) (2), x(0) (3), x(0) (4), x(0) (5), x(0) (6), x(0) (7)) model of varying starting points and GM(1,1) model of metabolism

Table 9
A comparison of prediction results.

Time x(0) x
(0) (1) e(1) x
(0) (2) e(2) x
(0) (3) e(3)

1 5285 5285 0.0000 – – – –

2 5313 4536.7058 0.1461 – – – –
3 4492 4578.4529 0.0192 – – – –
4 4480 4620.5842 0.0314 – – – –
5 4486 4663.1032 0.0395 – – – –
6 4489 4706.0134 0.0483 – – – –
7 4512 4749.3186 0.0526 4512.0000 0.0000 – –
8 4583 4793.0222 0.0458 4531.7979 0.0113 4584.8 0.0004
9 4680 4837.128 0.0336 4679.7828 0.0000 4680.2 0.0000
10 4795 4881.6396 0.0181 4832.6000 0.0078 4825.5 0.0064
11 4900 4926.5608 0.0054 4990.4075 0.0181 4955.7 0.0114
12 5215 4971.8955 0.0466 5153.3681 0.0120 5211.4 0.0007
13 5340 5017.6472 0.0604 5321.6502 0.0034 5354.4 0.0027
Average relative error (%) – 4.2079 – 0.7521 – 0.3592
 C. Liu et al. / Expert Systems With Applications 45 (2016) 331–340
are used for improvement of the prediction. The optimal models with
high precision and high fitting degrees are selected to determine the
optimal number of dimensions. The optimal number of dimensions is
the number of neurons in the input layer of neural networks. Finally,
the improved neural networks are employed for optimal prediction.
A good example is Hunan HX Office and School Furniture Fac-
tory, whose sales were partially disrupted by the snow storm in
South China in 2008. First, the traditional model is built. The cal-
culation results show that the requirement of prediction accuracy
cannot be satisfied due to the influence and interference of the long-
term data mutation and disruption, although the test of fitting de-
gree is passed. As such, the changeable starting points model and
metabolism model improve the traditional model GM(1,1). The cal-
culation results of the changeable starting points model show 7-
dimension and 4-dimension can be used as the prediction models.
Results of the metabolism model suggest that it is optimal when the
dimension number is fixed at 7. By comparing and integrating these
two models, the 7-dimension is selected as the prediction model, and
the optimal number of dimensions is 7. Ultimately, the BP neural net-
work is applied here and the network typology 7-15-1 is built. The
neuron transmission of the network explicit layer uses the Sigmoid
function; the neuron transmission of the network output layer uses
the linear Purelin function; and the training function uses Trainbr
based on the Bayes algorithm. The accuracy of the training target
is 0.005; the terminal training steps are 6000. By MATLAB 2011, in-
put 6 sets of data, and the accuracy requirement of 0.005 is satis-
fied through 2705 times of training. This shows the feasibility of the
method.
This approach to determining the number of neurons in the input
layer of BP neural network through the optimal number of dimen-
sions in the improved grey prediction model has exceeded the tradi-
tional grey model GM(1,1), which limits the sequence of the original
data. It also shows the characteristics of a better nonlinear approxi-
mation in neural networks so that the fitting degree and forecasting
precision are increased and the stability and reliability of prediction
are enhanced. This can contribute to the prediction of the market de-
mand after transportation disruption and the forecasting results may
provide information for production and inventory management deci-
sions, as well as demand prediction in supply chain enterprises. This
method can also be based on big data to construct similar model or
expert system, predicting the consequence of transportation disrup-
tion caused by accidents, and then some measures can be taken in
advance.
The empirical study in this article is grounded in the single supply
chain enterprise, which has not explored the upstream and down-
stream supply chain enterprises as well as other enterprises in other
regions affected. However, it is likely that results derived from the
data of single enterprises will not reflect the performance of corre-
sponding regions, markets and other relevant enterprises after dis-
ruption. In the future research, it is important to analyze the big data
concerning the upstream and downstream enterprises and the cor-
responding regions and markets after transportation disruption. As
such, it is possible that the findings might be more universal.
Acknowledgments
This article is financially supported by the Natural Science Foun-
dation of China [Grant numbers 71172194, 71390330, 71390331 and
71221001].
References
Alvisi, S., & Franchini, M. (2012). Grey neural networks for river stage forecasting with
uncertainty. Physics and Chemistry of the Earth, Parts A/B/C, 42-44, 108–118.
Bates, J. M., & Granger, C. W. J. (1969). Combination of forecasts. Operations Research
Quarterly, 20(4), 451–468.
339
Bravo, J. J., & Vidal, C. J. (2013). Freight transportation function in supply chain opti-
mization models: a critical review of recent trends. Expert Systems with Applica-
tions, 40(17), 6742–6757.
Chang, C. J., Li, D. C., Huang, Y. H., & Chen, C. C. (2015). A novel gray forecasting model
based on the box plot for small manufacturing data sets. Applied Mathematics and
Computation, 265(8), 400–408.
Chi, M., & Zhao, J. (2012). Application of BP neural network in evaluating e-business
performance for service industry. International Journal of Networking and Virtual
Organisations, 10(3/4), 374–385.
Chiu, S. W., Lee, C. H., Chiu, Y. S. P., & Cheng, F. T. (2013). Intra-supply chain system with
multiple sales locations and quality assurance. Expert Systems with Applications,
40(7), 2669–2676.
Costantino, F., Gravio, G. D., Shaban, A., & Tronci, M. (2015). A real-time SPC inventory
replenishment system to improve supply chain performances. Expert Systems with
Applications, 42(3), 1665–1683.
Cristóbal, J. R. S., Correa, F., González, M. A., Navamuel, E. D. R. D., Madariaga, E.,
Ortega, A., López, S., & Trueba, M. (2015). A residual grey prediction model for pre-
dicting s-curves in projects. Procedia Computer Science, 64(11), 586–593.
Cybenko, G. (1989). Approximation by superpositions of a sigmoidal function. Mathe-
matics of Control, Signals and Systems, 2(4), 303–314.
Deng, J. L. (2005). Basic methods of grey system. Wuhan: Central-China Science and
Technology Press.
Deng, J. L. (1982). Control problems of grey systems. Systems & Control Letters, 5(1),
288–294.
Dewangan, S., Gangopadhyay, S., & Biswas, C. K. (2015). Multi-response optimization
of surface integrity characteristics of EDM process using grey-fuzzy logic-based
hybrid approach. Engineering Science and Technology, 18(3), 361–368.
Hishamuddin, H., Sarker, R., & Essam, D. (2013). A recovery model for a two-echelon
serial supply chain with consideration of transportation disruption. Computers &
Industrial Engineering, 64(2), 552–561.
Hishamuddin, H., Sarker, R., & Essam, D. (2015). A simulation model of a three echelon
supply chain system with multiple suppliers subject to supply and transportation
disruptions. IFAC-PapersOnLine, 48(3), 2036–2040.
Hodzic, M., & Tai, L. C. (2016). Grey predictor reference model for assisting particle
swarm optimization for wind turbine control. Renewable Energy, 86(2), 251–256.
Hsu, K. T. (2011). Using a back propagation network combined with grey clustering
to forecast policyholder decision to purchase investment-inked insurance. Expert
Systems with Applications, 38(6), 6736–6747.
Intharathirat, R., Salam, P. A., Kumar, S., & Untong, A. (2015). Forecasting of municipal
solid waste quantity in a developing country using multivariate grey models. Waste
Management, 39(5), 3–14.
Ito, Y. (1991). Representation of functions by superpositions of a step or sigmoid func-
tion and their applications to neural network theory. Neural Networks, 4(3), 385–
394.
Khademi, N., Balaei, B., Shahri, M., Mirzaei, M., Sarrafi, B., Zahabiun, M., &
Mohaymany, A. S. (2015). Transportation network vulnerability analysis for the
case of a catastrophic earthquake. International Journal of Disaster Risk Reduction,
12(7), 234–254.
Li, H., & Hu, Q.Y. (2005). Forecasting and decision-making. Xi’an: Xi’an Electronic
Science.
Liu, S. F., & Guo, T. B. (1991). Grey system theory and its applications. Kaifeng: Henan
University Press.
Liu, S. F., Dang, Y. G., & Fang, Z. G. (2010). Grey system theory and its applications. Beijing:
Science Press.
Memon, M. S., Lee, Y. H., & Mari, S. I. (2015). Group multi-criteria supplier selection
using combined grey systems theory and uncertainty theory. Expert Systems with
Applications, 42(21), 7951–7959.
Ouyang, M., Pan, Z. Z., Liu, H., & He, Y. (2015). Vulnerability analysis of complementary
transportation systems with applications to railway and airline systems in China.
Reliability Engineering & System Safety, 142(11), 248–257.
Pai, T. Y., Lo, H. M., Wan, T. J., Chen, L., Hung, P. S., Lo, H. H., Lai, W. J., & Lee, H. Y. (2015).
Predicting air pollutant emissions from a medical incinerator using grey model and
neural network. Applied Mathematical Modelling, 39(5/6), 1513–1525.
Pasandideh, S. H. R., Niaki, S. T. A., & Asadi, K. (2015). Optimizing a bi-objective multi-
product multi-period three echelon supply chain network with warehouse relia-
bility. Expert Systems with Applications, 42(5), 2615–2623.
Qian, M. P., & Gong, G. L. (2000). Random process theory. Beijing: Beijing University
Press.
Rajesh, R., & Ravi, V. (2015). Modeling enablers of supply chain risk mitigation in elec-
tronic supply chains: a Grey-DEMATEL approach. Computers & Industrial Engineer-
ing, 87(9), 126–139.
Ray, A., Mukhopadhyay, S., Datta, B., & Pal, S. (2013). Prediction of atmospheric pressure
at ground level using artificial neural network. International Journal of Research in
Computer Science, 3(1), 11–18.
Rumelhart, D. E., Hintont, G. E., & Williams, R. J. (1986). Learning representations by
back-propagating errors. Nature, 323(9), 533–536.
Sheffi, Y. (2001). Supply chain management under the threat of international terrorism.
International Journal of Logistics Management, 12(2), 1–11.
Shu, T., Chen, S., Wang, S. Y., & Lai, K. K. (2014). GBOM-oriented management of produc-
tion disruption risk and optimization of supply chain construction. Expert Systems
with Applications, 41(1), 59–68.
Tamura, S., & Tateishi, M. (1997). Capabilities of a four-layered feed forward neural net-
work: four layers versus three. IEEE Transactions on Neural Networks, 8(2), 251–255.
Tian, Y. B. (2009). Mixed neural network technology. Beijing: Science Press.
Wa, D. (2010). The GM(1,1) forecast model of the Tibet rural consumption expenditure
trends and effect factors analysis. Statistical Research, 27(4), 79–82.
340 C. Liu et al. / Expert Systems With Applications 45 (2016) 331–340
Wang, R. J., Zeng, J. L., Feng, X. X., & Xia, Y. J. (2013). Evaluation of effect of plastic in-
jection molding process parameters on shrinkage based on neural network simu-
lation. Journal of Macromolecular Science, Part B: Physics, 52(1), 206–221.
Wilson, M. C. (2007). The impact of transportation disruptions on supply chain per-
formance. Transportation Research Part E: Logistics and Transportation Review, 43(4),
295–320.
Xie, N. M., Yuan, C. Q., & Yang, Y. J. (2015). Forecasting China’s energy demand and self-
sufficiency rate by grey forecasting model and Markov model. International Journal
of Electrical Power & Energy Systems, 66(3), 1–8.
Zeng, B., & Liu, S. F. (2011). Direct modeling approach of DGM(1,1) with approximate
non-homogeneous exponential sequence. Systems Engineering Theory and Practice,
31(2), 297–301.
Zhang, H. N., Xu, A. J., Cui, J., He, D. F., & Tian, N. Y. (2012). Establishment of neural
network prediction model for terminative temperature based on grey theory in
hot metal pretreatment. Journal of Iron and Steel Research, 19(6), 25–29.
Zhang, Y., & Lam, J. S. L. (2015). Estimating the economic losses of port disruption due
to extreme wind events. Ocean & Coastal Management, 116(11), 300–310.