Professional Documents
Culture Documents
Abstract
The Kalman filter and its nonlinear variants have been widely used for filtering and state estimation. However, models with severe
nonlinearities are not handled well by Kalman filters. Such a case is presented in this paper: the estimation of the overflow losses in a
hopper dredger. The overflow mixture density and flow-rate have to be estimated based on noisy measurements of the total hopper
volume, mass, incoming mixture density and flow-rate. In order to reduce complexity and make the tuning easier, a decomposition of the
nonlinear process model into two simpler subsystems is proposed. A different type of observer is considered for each subsystem—a
particle filter and an unscented Kalman filter. The performance is evaluated for simulated and real-world data and compared with the
centralized setting for four combinations of the particle filter and the unscented Kalman filter. The results indicate that the distributed
observer achieves the same performance as the centralized one, while leading to increased modularity, reduced complexity, lower
computational costs and easier tuning.
r 2007 Elsevier Ltd. All rights reserved.
Keywords: State estimation; Nonlinear systems; Nonlinear observers; Kalman filter; Particle filter
0967-0661/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.conengprac.2007.05.004
ARTICLE IN PRESS
Zs. Lendek et al. / Control Engineering Practice 16 (2008) 392–406 393
which may render the filter unusable for on-line or real-time PF would be required. However, the model can be
estimation. represented as two cascaded subsystems, which allows the
Since the above-mentioned methods are suboptimal, use of two observers. For these observers the combinations
their performance varies, depending on the application of UKF and PF are considered and the four possible
considered. While for a highly nonlinear and non-Gaussian combinations in the distributed setting are compared with
model, a PF is likely the best option, UKF may also yield the performance of a centralized PF for the whole system,
good performance with considerably lower computational both on simulated and experimental data.
costs. However, the design of an observer for a complex The structure of the paper is as follows. Section 2
nonlinear system for on-line estimation is problematic due reviews the UKF and the PF methodology. In Section 3,
to tuning difficulties and large computational costs. the proposed cascaded observer setting is given, while
Decentralized estimation has been studied in the context Section 4 presents the dynamic sedimentation model and
of large-scale processes and distributed systems. The the models used for estimation purposes. Sections 5 and 6
architecture in general takes the form of a network of give the results for the simulated and experimental data,
sensor nodes, each with its own processing facility. In case respectively. Finally, Section 7 concludes the paper.
of a fully decentralized system, computation is performed
locally and communication occurs between any two nodes. 2. Estimation methods
Each node shares information with other nodes and
computes a local estimate. Computation and communica- In this section, two methods for estimating the states and
tion is distributed over the network so that a global parameters of a nonlinear system are presented. Consider
estimate can be computed. Several topologies have the following discrete-time, possibly time-varying, non-
been proposed, depending on the particular application. linear system:
In case of large-scale processes (Vadigepalli & Doyle,
xk ¼ f ðxk1 ; vk1 Þ, (1)
2003a, 2003b), the network is in general in a hierarchical
form, with several intermediate and one final fusion node.
yk ¼ hðxk ; Zk Þ, (2)
For distributed systems, such as multiagent societies
(López-Orozco, de la Cruz, Besada, & Ruipérez, 2000; where k is the current time step, x the state variables, v; Z
Roumeliotis & Bekey, 2002; Schmitt, Hanek, Beetz, the noises of known distributions, y the measurements, f
Buck, & Radig, 2002), several fusion nodes exist, which the state transition model, h the measurement model.
process the data and send the information to the rest Note that the functions f and h may also depend on other
of the nodes. Observers for distributed estimation include, known inputs or parameters. However, for the ease of
but are not limited to decentralized KF and EKF filter notation, these variables are omitted. It is assumed that
(Durrant-Whyte, Rao, & Hu, 1990), information filter and system (1)–(2) is observable, in order to be able to estimate
PFs (Bolic, Djuric, & Hong, 2004; Coates, 2004). the states.
In this paper, it is proposed to decompose a nonlinear The goal is to estimate the states of interest. Two methods
system model into cascaded subsystems, with the possibi- are considered: the UKF and the PF. Both methods use
lity of using different estimation methods for the sub- notions from probability theory, however, the UKF is a
systems. Many nonlinear systems can be represented as deterministic method, while PFs are stochastic. Both filters
cascaded, observable subsystems, which alone are less are recursive algorithms, that use all the provided informa-
complex than the original system. Separate observers can tion (model and observations) to estimate the current state of
be designed for the individual subsystems. Moreover, the system. The filters work in two steps: prediction and
different types of observers can be combined, depending update. The prediction step uses the system model and the
on the complexity and nonlinearity of the subsystems. This information incorporated so far in order to predict the
setting can be regarded as a cooperative multiagent system. process’ states. This stage is also known as the time update
Each agent has the task of observing one of the subsystems, step, as it projects the current state forward in time. The
possibly using different methods and relying on its own update stage uses the latest measurement to modify (correct)
measurement and the information gathered from other the projected state. This stage is also known as measurement
agents. In turn, each agent communicates its own results to update, since it incorporates the information brought by the
other agents. new measurement.
The proposed distributed observer design is applied to
the estimation of the overflow losses in a hopper dredger. 2.1. Unscented Kalman filter
The estimation of overflow losses is an essential step
toward the optimization of the separation process in the For linear systems corrupted by white Gaussian noise,
hopper, which is of vital importance for future improve- the KF is proven to be an optimal filter in the least mean
ment in dredging efficiency, accuracy and from the square sense. For nonlinear systems, several extensions
viewpoint of labor saving. In the considered process, exist: the EKF (based on linearizing the models around the
the measured variables are heavily corrupted by noise. The current states), and the family of sigma point KFs (Julier &
system is highly nonlinear, and for global state estimation a Uhlmann, 2002b; van der Merwe, 2004).
ARTICLE IN PRESS
394 Zs. Lendek et al. / Control Engineering Practice 16 (2008) 392–406
The fundamental problem of the EKF is that the can be non-positive semidefinite, in which case the filter
approximations will only be valid if all the higher order collapses.
derivatives of the nonlinear functions are effectively zero
around the current estimate. Another variant of the KF for 2.1.2. UKF algorithm
nonlinear systems, the UKF was developed by Julier and In order to apply the KF to system (1)–(2), the state
Uhlmann (1997). Since the original formulation of the variables are augmented with the state transition and
UKF, a number of sigma-point filters were described, measurement noise, and the state covariance with the state
which are based on approximating the distribution of the transition and measurement covariance. Assuming that the
states by deterministically chosen samples (sigma points) noises have means v and Z, and covariances Q and R,
and differ mostly on how these samples are generated. respectively, the augmented variables can be expressed as
These filters preserve the normal distribution and are valid
up to the second order of the Taylor expansion of the xak1 ¼ ½xTk1 vTk1 ZTk1 T ; na ¼ dimðxak1 Þ,
nonlinear functions.
The UKF is based on the unscented transformation, Pak1 ¼ diagð½Pxk1 Qk1 Rk1 Þ. (6)
which computes the statistics of a random variable The sigma points are computed based on the augmented
undergoing a nonlinear transformation. To compute the state and covariance.
statistics, a number of weighted samples called ‘‘sigma The prediction is extended with respect to the KF, since
points’’ are chosen deterministically, so that they comple- the sigma points have to be computed and propagated
tely capture the mean and covariance of the random through the state transition model to predict the new states.
variable. The predicted states also have to be propagated through
the measurement model in order to predict the measure-
2.1.1. Unscented transformation ment. The equations are described below.
Assume that an nx -dimensional random variable x has to Generate the sigma points for (6), according to (3). The
be propagated through the nonlinear function g in order to parameter k provides an extra degree of freedom to ‘‘fine
generate y: tune’’ the higher order moments of the approximation, and
can be used to reduce the overall prediction error. When
y ¼ gðxÞ. the states are assumed Gaussian, a useful heuristic is to
Assume also that x has a known mean xm and a known select na þ k ¼ 3 (Julier & Uhlmann, 2002a). For other
covariance Px . In this case, 2nx þ 1 sigma points are distributions, a different choice of k might be more
generated deterministically, so that they capture the mean appropriate. In this paper, it is assumed that the states
and variance. One of the selection procedures (Julier & are Gaussian and the above guideline is used.
Uhlmann, 2002a) computes the sigma points Xi as follows: Since the augmented state from which the sigma points
are generated has the form (6), the sigma points also have
X0 ¼ x m ; w0 ¼ k=ðnx þ kÞ,
the form:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Xi ¼ xm þ ½ ðnx þ kÞPx Ti , Xai;k1 ¼ ½Xxk1 Xvk1 XZk1 i ; i ¼ 0; . . . ; 2na , (7)
wi ¼ 1=½2ðnx þ kÞ; i ¼ 1; 2; . . . ; nx ,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi where Xxk1 denote the part of the sigma points that
Xi ¼ xm ½ ðnx þ kÞPx Ti , corresponds to the state, Xvk1 the part that corresponds to
wi ¼ 1=½2ðnx þ kÞ; i ¼ nx þ 1; . . . ; 2nx , ð3Þ the state transition noise and XZk1 the part that
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi corresponds to the measurement noise.
where k is a scaling parameter, ½ ðnx þ kÞPx i is the ith Propagate the sigma points corresponding to the state
row of the matrix square root of ðnx þ kÞPx , and wi is the and state transition noise through the state transition
weight associated with the ith sample. model:
The sigma points are now propagated though the
Xxi;kjk1 ¼ f ðXxi;k1 ; Xvi;k1 Þ; i ¼ 0; . . . ; 2na . (8)
nonlinear function g : Yi ¼ gðXi Þ, i ¼ 0; 1; . . . ; 2nx . The
mean and covariance of y are estimated as Predict the next state:
X
2nx X
2na
ym ¼ wi Yi , (4) x^ kjk1 ¼ wi Xxi;kjk1 (9)
i¼0 i¼0
and covariance:
X
2nx
T
Py ¼ wi ðYi ym ÞðYi ym Þ . (5) X
2na
i¼0 Pxkjk1 ¼ wi ðXxi;kjk1 x^ kjk1 ÞðXxi;kjk1 x^ kjk1 ÞT . (10)
i¼0
These estimates are accurate to the second order of the
Taylor series expansion of gðxÞ, for any nonlinear function. Propagate the transformed sigma points Xxi;kjk1 and the
However, in certain cases the computed covariance matrix part corresponding to the measurement noise XZi;k1
ARTICLE IN PRESS
Zs. Lendek et al. / Control Engineering Practice 16 (2008) 392–406 395
Update:
K k ¼ Pxy y
kjk1 ðPkjk1 Þ
1 . Kalman gain
x^ k ¼ x^ kjk1 þ K k ðyk y^ kjk1 Þ . correct the state
Pxk ¼ Pxkjk1 þ K k Pykjk1 K Tk . correct the covariance
end for
ARTICLE IN PRESS
396 Zs. Lendek et al. / Control Engineering Practice 16 (2008) 392–406
generated. Samples xik are drawn from an (chosen) Resample using Algorithm 3.
importance density function qðxik jxik1 ; yk Þ, and the weights end if
are updated, using the current measurement yk
pðyk jxik Þ pðxik jxik1 Þ Algorithm 3. Resampling
w~ ik ¼ wik1 (18)
qðxik jxik1 ; yk Þ
Input: fðxi ; wi ÞgN
i¼1
and normalized
Output: fðxinew ; winew ÞgNi¼1
w~ i for i ¼ 1; 2; . . . ; N do
P
wik ¼ PN k .
~ jk Compute cumulative sum of weights: wic ¼ ij¼1 wjk
j¼1 w
end for
If the importance density qðxk jxk1 ; yk Þ is chosen equal to Draw u1 from Uð0; N1 Þ
the state-transition PDF pðxk jxk1 Þ, the weight update
for i ¼ 1; 2; . . . ; N do
equation (18) becomes
Find xþi , the first sample for which wic Xui .
w~ ik ¼ wik1 pðyk jxik Þ. Replace particle i: xinew ¼ xþi , winew ¼ N1
The use of the transition prior as the importance density is uiþ1 ¼ ui þ N1
a common choice (Arulampalam et al., 2002) and it has the end for
advantage that it can be easily sampled and the weights are
easily evaluated.
The posterior PDF is represented by the set of weighted The state estimate is computed as the weighted mean of
samples, conventionally denoted by the particles:
X
N X
N
arbitrary distributions. In such case, however, the mean so that the subsystem
of the posterior cannot be considered a correct estimate
x_ 1 ¼ f 1 ðx1 ; uÞ,
of the state, since no guarantees exist that the posterior
is unimodal.
y1 ¼ h1 ðx1 ; uÞ (21)
Due to the approximation of the posterior with weighted
samples, a large number of samples are necessary for is observable.
good performance. Hence, the algorithm is computa- Since both systems (19) and (21) are observable, this also
tionally involved, and not suitable for fast processes. means that the subsystem
While in theory PFs are the best of the above estimators,
for nonlinear systems, they cannot be considered a x_ 2 ¼ f 2 ðx1 ; x2 ; uÞ,
universal solution. Mild nonlinearities or fast processes
can be better handled by a KF. y2 ¼ h2 ðx1 ; x2 ; uÞ (22)
Incoming mixture
Overflow pipe
Qi , ρi
Overflow
Q o , ρo
Hopper
Pump ht ρm
ho
Sand bed ρs hs
optimization of the dredging performance is the minimiza- overflow pipe is full (e.g., because the valve inside the pipe
tion of the overflow losses. In the literature, a number of is engaged), the following model must be used:
sedimentation models have been proposed (Camp, 1946; pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
van Rhee, 2002), however, these models cannot be used as Qo ¼ k0o 2g maxðht ho ; 0Þ. (27)
a basis for control or optimization of the dredging process.
Clearly, there is some uncertainty in the modeling of the
The reason is that they are based on detailed (often PDE)
overflow rate. Moreover, due to the model’s switching
modeling of the physical phenomena and contain too many
nature, it is not straightforward to estimate its parameters.
uncertain parameters. Therefore, simplified models are
To model the overflow density ro , the density profile in
used, along with advanced signal processing and estimation
the mixture above the sand bed must be described.
techniques.
Generally, this profile can be approximated as a decreasing
function of the height above the sand, but the exact form of
4.1. Dynamic sedimentation model this function is highly uncertain and time varying. In this
paper, the following saturated affine approximation is
The sedimentation process in a hopper dredger can be used:
described by a model with three state variables: the total
mass in the hopper mt , the total volume V t of the mixture ro ¼ maxðrs kr ðho hs Þ; rw Þ. (28)
in the hopper and the mass of the sand bed ms (see Fig. 3).
While the first two variables can be derived from on-line The slope kr must be determined at every time instant such
measurements (the ship draught and the total level ht , that the average mixture density rm , computed from the
respectively), the mass of the sand bed is not measurable. mass balance relations, equals to the average of the density
The flow-rate Qi of the incoming mixture and the overflow profile:
height ho are the manipulated inputs and the incoming Z
1 ht
mixture density ri is in this context regarded as a measured rm ¼ maxðrs kr ðh hs Þ; rw Þ dh
disturbance. The volume and mass balance equations are hm h s
given by with hm ¼ ht hs . Solving this constraint for the model
V_ t ¼ Qi Qo , (23) (28) yields the following equation for the slope:
8
> 2ðrs rm Þ 1
_ t ¼ Qi ri Qo ro .
m (24) >
> for rm 4 ðrw þ rs Þ;
< hm 2
The rate of material sedimentation is a function of the kr ¼ 2
>
> ðrs rw Þ
settling velocity ðfs Þ and the scouring (erosion) effects ðfe Þ >
: otherwise;
2hm ðrm rw Þ
_ s ¼ fs ðrm Þfe ðQo ; hm Þ.
m (25)
The overflow rate Qo , the density ro and the functions in where the average mixture density is given by
(25) are modeled by using static relationships as detailed mt ms r ðmt ms Þ
below. rm ¼ ¼ s .
V t ms =rs V t rs ms
If the outgoing mixture freely flows through the overflow
pipe, the flow-rate Qo is given by Validation based on measured data has shown that this
model is not very accurate, but it suffices for the tuning and
Qo ¼ ko maxðht ho ; 0Þ3=2 , (26)
first evaluation of the PF.
where ko is an uncertain parameter depending on the The settling function fs describes how the rate of
overflow pipe shape and circumference. However, if the sedimentation depends on the undisturbed settling velocity
ARTICLE IN PRESS
Zs. Lendek et al. / Control Engineering Practice 16 (2008) 392–406 399
vs and the mixture density: where vk and Zk are zero mean Gaussian noises, with
!b covariances Q and R, respectively. The equivalent prob-
rm rw rq rm abilistic model is expressed as
fs ðrm Þ ¼ Ars vs . (29)
rs rm rq rw
pðxk jxk1 Þ ¼ Nðxk ; f ðxk1 Þ; QÞ,
The scouring function fe describes the effect of erosion on
the sand bed due to the flow in the mixture (which is pðyk jxk Þ ¼ Nðyk ; hðxk Þ; RÞ.
considered to be equal to the overflow rate in steady state):
!
Using the above estimation model, a centralized PF has
Q2o
fe ðQo ; hm Þ ¼ max 1 ;0 . (30) already been successfully implemented in the data-acquisi-
kc h2m tion and monitoring system of a hopper dredger (Babuška,
The parameters of the entire model have been determined Lendek, Braaksma, & de Keizer, 2006). Here, different
by fitting the outputs of the simulation model to real data observers are compared in terms of their performance. The
from a ship, by using nonlinear least-squares optimization. following cases are considered:
4.2. The estimation problem (1) A centralized observer to simultaneously estimate the
values of both Qo and ro , see Fig. 4.
In order to estimate at each time step the overflow (2) Cascaded observers: the first observer estimates Qo
density and flow-rate, the volume and mass balance based on the volume balance (31), and the second
equations were discretized by using the Euler method: estimates ro based on the mass balance (32) and the
values obtained for Qo by the first observer, see Fig. 5.
V t;k ¼ V t;k1 þ TðQi;k1 Qo;k1 Þ, (31)
mt;k ¼ mt;k1 þ TðQi;k1 ri;k1 Qo;k1 ro;k1 Þ, (32) Note, that models (31)–(34) only approximate the
underlying true process. If the data were generated based
where the sampling period is T ¼ 5 s, which is also the on these models, a KF could be used. For the data
sampling period of the on-board data-acquisition system. generated from the sedimentation model, however, the
These state equations are augmented with a random-walk results obtained by the KF are too noisy and the KF
model for Qo and ro : becomes unstable for the experimental data.
Qo;k ¼ Qo;k1 þ Q;k1 , (33) It was found that the UKF cannot simultaneously
estimate both Qo and ro . Therefore, for the centralized
ro;k ¼ ro;k1 þ r;k1 . (34) observer, only PF is considered, while in the cascaded
setting both PF and UKF are used. The observers are first
The motivation for this choice results from the process applied to the simulated data and then, with the same
description in Section 4. The sedimentation models are parameters, to real measured data.
based on empirical modeling of the physical phenomena There is one more setting of observers that was
and contain too many uncertain parameters. By using a considered for this specific application: independent
random walk model, the use of the uncertain overflow observers. In this setting, one uses two observers: the first
model (26)–(27), the settling and scouring functions estimates Qo based on the volume balance (31). The second
(29)–(30), and the uncertain parameters is circumvented. estimates Qo ro based on the mass balance (32). The value
The augmented state, input and output vectors are defined of ro can be computed afterwards by dividing the estimate
as
0 1
Vt
B C ! !
B mt C Qi Vt
x¼B B C Vt
C; u ¼ ri ; y ¼ mt . Qo
@ Qo A Qi
Observer
ro i
o
mt
Measurements are available for the inputs Qi and ri and
the outputs V t and mt . The objective is to estimate Qo and
Fig. 4. Centralized observer.
ro on-line. The corrupting noises are considered zero-mean
Gaussians ðxi;k Nð0; nxi ÞÞ, and their standard deviations
are determined experimentally. Qo
Vt
Note, that in this particular case, the estimation model is Qi
Observer 1
described as Observer 2 o
i
xk ¼ f ðxk1 Þ þ vk1 , mt
11.5 1450
11 1400
10.5 1350
10
1300
9.5
ρo [kg/m3]
Qo [m3/s]
1250
9
1200
8.5
1150
8
7.5 1100
7 1050
6.5 1000
1 1.1 1.2 1.3 1.4 1.5 1.6 1 1.1 1.2 1.3 1.4 1.5 1.6
Time [h] Time [h]
Fig. 6. Centralized observer: results for Qo (a) and ro (b) using the particle filter (solid line—simulated data, dotted line—estimate).
60 50
45
50
40
40 35
30
# samples
# samples
30 25
20
20
15
10
10
5
0 0
−2 −1.5 −1 −0.5 0 0.5 1 1.5 2 −100 −50 0 50 100
Qo [m3/s] ρo [kg/m3]
For this simulation, only the inputs Qi and ri are fed (see also Fig. 5). Two types of filters are compared: the
with experimental data, corrupted by noise. The remaining UKF and the PF.
variables are computed in simulation without adding noise. The dynamic model is decomposed into two subsystems.
The results obtained with the different configurations of The first observer uses the model
observers are compared to the simulated values of ro and
V t;k ¼ V t;k1 þ TðQi;k1 Qo;k1 Þ þ V ;k1 ,
Qo . The standard deviations of the state transition and
measurement noise are given in Table 1. The PF used 1000
Qo;k ¼ Qo;k1 þ Q;k1 , (36)
samples, with resampling at a threshold of N T ¼ 900. The
presented results are the average of 30 simulations. where V t is the measured output. The second observer uses
the model
5.1. Centralized observer
mt;k ¼ mt;k1 þ TðQi;k1 ri;k1 Qo;k1 ro;k1 Þ þ m;k1 ,
The results obtained with a PF, based on model (35) are
ro;k ¼ ro;k1 þ r;k1 , (37)
presented in Fig. 6. The maximum standard deviation
computed point-wise for 30 trajectories of the state where mt is the measured output.
estimated by the PF is 0.1 for Qo and 5.1 for ro . The results obtained for Qo are presented in Fig. 8.
The residuals are computed as the difference between the The maximum standard deviation computed point-wise
simulated and estimated values of Qo and ro , respectively. for the 30 Monte Carlo simulations is 0.095. The statistics
The distribution of the residuals is presented in Fig. 7, of the residuals’ distributions are given in Table 3. These
while their statistics are given in Table 2. statistics are comparable with those obtained with the
centralized observer.
5.2. Cascaded observers For the cascaded observer the following combinations
are considered: UKF for both Qo and ro , PF for both Qo
This setting involves two observers in a cascade. The first and ro , UKF for Qo and PF for ro , and PF for Qo and
one estimates Qo using the volume balance (31) and a UKF for ro . In what follows, the observer setting is
random walk model for Qo . The second observer estimates denoted as observer 1–observer 2, i.e., PF–PF denotes that
ro based on the mass balance (32), a random walk model PF is used for both Qo and ro , UKF–PF means that UKF
for ro and the result obtained for Qo by the first observer is used for Qo and PF for ro , etc.
Table 2 Table 3
Statistics of residuals for the centralized observer Statistics of residuals of Qo ðm3 =sÞ
11.5 11.5
11 11
10.5 10.5
10 10
9.5 9.5
Qo [m3/s]
Qo [m3/s]
9 9
8.5 8.5
8 8
7.5 7.5
7 7
6.5 6.5
1 1.1 1.2 1.3 1.4 1.5 1.6 1 1.1 1.2 1.3 1.4 1.5 1.6
Time [h] Time [h]
Fig. 8. Cascaded observers: results for Qo using PF (a) and UKF (b) (solid line—simulated data, dotted line—estimate).
ARTICLE IN PRESS
402 Zs. Lendek et al. / Control Engineering Practice 16 (2008) 392–406
The results obtained for ro using Qo estimated by PF much better than the centralized observer or the PF.
and UKF are presented in Fig. 9. The maximum point- However, considering the standard deviations (Table 4
wise standard deviation of 30 Monte Carlo simulations and Fig. 10), one can see that the estimates of ro obtained
for ro , based on the results of Qo given also by PF is by UKF are much noisier than those obtained by the
5.8, while for ro based on the results of Qo given by PF, both centralized and cascaded. Comparing the
UKF is 2.96. The distribution of the residuals is given results obtained by PF for ro with the centralized observer
in Fig. 10. The statistics of the residuals can be found in (Table 2), it can be seen that using a combination of two
Table 4. PFs leads to approximately the same results as the
centralized filter. However, the best result, based on the
5.3. Discussion statistics presented in Tables 3 and 4 was obtained with
cascaded observers, the combination UKF for Qo and PF
For simulated data, a good estimate is obtained with the for ro .
centralized observer (PF) for Qo , but the estimate of ro is
delayed relative to the one simulated (Fig. 6). 6. Results for experimental data
For the cascaded observers, considering only the
mean of the residuals (Table 4), one can conclude For the experimental data, the same model and
that the UKF performs the best for the estimation of ro , parameter values are considered as for the simulated
1450 1450
1400 1400
1350 1350
1300 1300
ρo [kg/m3]
ρo [kg/m3]
1250 1250
1200 1200
1150 1150
1100 1100
1050 1050
1000 1000
1 1.1 1.2 1.3 1.4 1.5 1.6 1 1.1 1.2 1.3 1.4 1.5 1.6
Time [h] Time [h]
1450 1450
1400 1400
1350 1350
1300 1300
ρo [kg/m3]
ρo [kg/m3]
1250 1250
1200 1200
1150 1150
1100 1100
1050 1050
1000 1000
1 1.1 1.2 1.3 1.4 1.5 1.6 1 1.1 1.2 1.3 1.4 1.5 1.6
Time [h] Time [h]
Fig. 9. Estimates of ro with the four possible filter combinations (solid line—simulated data, dotted line—estimate). (a) PF–PF. (b) PF–UKF.
(c) UKF–PF. (d) UKF–UKF.
ARTICLE IN PRESS
Zs. Lendek et al. / Control Engineering Practice 16 (2008) 392–406 403
50 50
45 45
40 40
35 35
30 30
# samples
# samples
25 25
20 20
15 15
10 10
5 5
0 0
−100 −50 0 50 100 −100 −50 0 50 100
ρo [kg/m3] ρo [kg/m3]
50 50
45 45
40 40
35 35
30 30
# samples
# samples
25 25
20 20
15 15
10 10
5 5
0 0
−100 −50 0 50 100 −100 −50 0 50 100
ρo [kg/m3] ρo [kg/m3]
Fig. 10. Cascaded observers: residuals for the estimates of ro . (a) PF–PF. (b) PF–UKF. (c) UKF–PF. (d) UKF–UKF.
Table 4 computed as
Statistics of residuals of ro ðkg=m3 Þ
1
Qo ro Mean Standard deviation
Qo;k ¼ Qi;k ðV t;kþ1 V t;k Þ,
T
PF PF 12.1986 25.1174
PF UKF 0.3354 54.3970 Qi;k ri;k ð1=TÞðmt;kþ1 mt;k Þ
ro;k ¼ . (38)
UKF PF 10.6790 21.5436 Qo;k
UKF UKF 1.2536 54.2936
As the result of this computation is very noisy, a first order
anti-causal low-pass filter was applied to the measured data
before computing Qo and ro , with an experimentally
data and the same combinations of observers. The chosen cut-off frequency of 0.05 Hz.
presented results for the PF are the average of 30
simulations.
Since no measurements of Qo and ro are available, 6.1. Centralized observer
the results are compared to the values of ro and Qo
directly computed from the volume and mass balance, i.e., The results obtained with the centralized observer (PF)
Eqs. (31)–(32). The overflow rate and the density are are presented in Fig. 11. The maximum standard deviation
ARTICLE IN PRESS
404 Zs. Lendek et al. / Control Engineering Practice 16 (2008) 392–406
computed point-wise over the 30 Monte Carlo simulations The sample-wise maximum standard deviation of the 30
is 1.76 for Qo and 42.87 for ro . Monte Carlo simulations for ro , based on the results of Qo
given also by PF is 36.91, while for ro based on the results
6.2. Cascaded observers of Qo given by UKF is 36.05.
The results obtained by the UKF and PF for Qo are 6.3. Discussion
presented in Fig. 12. The estimate of Qo obtained by the
UKF is noisier than that obtained by the PF, but For experimental data, a centralized observer (Fig. 11)
comparable to the result obtained by the centralized obtains a reasonably good estimate for ro , but the estimate
observer. The maximum sample-wise standard deviation of Qo is noisier than the one computed by (38).
of the Monte Carlo simulations for Qo is 0.55. From Fig. 13 it can be seen that the UKF cannot handle
For the cascaded setting, the previous four combinations the estimation of ro , probably due to the high noise level of
are considered. The results obtained for ro with the four the measured variables. By using the combinations of
combinations are presented in Fig. 13. PF–PF and UKF–PF, approximately the same results are
15
1500
14
13 1400
12
1300
11
Qo [m3/s]
ρo [kg/m3]
10 1200
9
1100
8
7
1000
6
5 900
1 1.1 1.2 1.3 1.4 1.5 1.6 1 1.1 1.2 1.3 1.4 1.5 1.6
Time [h] Time [h]
Fig. 11. Centralized observer: results for Qo (a) and ro (b) using the particle filter (solid line—computed by (38), dotted line—estimate).
15 15
14 14
13 13
12 12
11 11
Qo [m3/s]
Qo [m3/s]
10 10
9 9
8 8
7 7
6 6
5 5
1 1.1 1.2 1.3 1.4 1.5 1.6 1 1.1 1.2 1.3 1.4 1.5 1.6
Time [h] Time [h]
Fig. 12. Cascaded observers: results for Qo using PF (a) and UKF (b) (solid line—computed by (38), dotted line—estimate).
ARTICLE IN PRESS
Zs. Lendek et al. / Control Engineering Practice 16 (2008) 392–406 405
1500 1500
1400 1400
1300 1300
ρo [kg/m3]
ρo [kg/m3]
1200 1200
1100 1100
1000 1000
900 900
1 1.1 1.2 1.3 1.4 1.5 1.6 1 1.1 1.2 1.3 1.4 1.5 1.6
Time [h] Time [h]
1500 1500
1400 1400
1300 1300
ρo [kg/m3]
ρo [kg/m3]
1200 1200
1100 1100
1000 1000
900 900
1 1.1 1.2 1.3 1.4 1.5 1.6 1 1.1 1.2 1.3 1.4 1.5 1.6
Time [h] Time [h]
Fig. 13. Cascaded observers: estimates of ro using the four possible combinations (solid line—computed by (38), dotted line—estimate). (a) PF–PF.
(b) PF–UKF. (c) UKF–PF. (d) UKF–UKF.
obtained for ro . These results are also comparable to those centralized observer. The overflow losses, represented by
obtained by a PF as the centralized observer. the overflow mixture density are estimated on the basis of
the measured total hopper volume, hopper mass, incoming
7. Conclusions mixture density and flow-rate.
The proposed approach uses straightforward nonlinear
This paper proposed a distributed setting for nonlinear mass balance equations and does not rely on complex
state-estimation. In many real-life applications, a complex overflow and sedimentation models which contain uncer-
process model can be decomposed into simpler subsystems, tain parameters and empirical functional relationships.
and observers can be designed for these individual The performance was evaluated in simulations and with
subsystems. This partitioning leads to increased modularity real measurements. The simulation results for this applica-
and reduced complexity of the problem, with reduced tion clearly indicate the best combination: cascaded
computational costs. Moreover, since the design and observers, using UKF or PF for the simpler subsystem
tuning of the observers becomes simpler, the overall (flow-rate) and PF for the more complex one (density).
performance can also be improved. In the future research, the theoretical conditions under
The distributed observer setting was applied to the which such a distribution of the process and the estimation
estimation of overflow losses in a hopper dredger. The is possible while maintaining the same performance as a
results were compared with those obtained by the centralized observer will be investigated.
ARTICLE IN PRESS
406 Zs. Lendek et al. / Control Engineering Practice 16 (2008) 392–406