Professional Documents
Culture Documents
School of Control & Computer Engineering, North China Electric Power University, Changping District, Beijing, China.
Abstract
This work proposes a neural-fuzzy sliding mode control scheme for a hydro-turbine speed governor system.
Considering the assumption of elastic water hammer, a nonlinear mode of the hydro-turbine governor system
is established. By linearizing this mode, a sliding mode controller is designed. The linearized mode is subject
to uncertainties. The uncertainties are generated in the process of linearization. A radial basis function (RBF)
neural network is introduced to compensate for the uncertainties. The update formulas for the neural
networks are derived from the Lyapunov direct method. For the chattering phenomenon of the sliding mode
control, a fuzzy logic inference system is adopted. In the sense of Lyapunov, the asymptotical stability of the
system can be guaranteed. Compared with the internal mode control and the conventional PID control
method, some numerical simulations verify the feasibility and robustness of the proposed scheme.
Keywords: Hydropower Plant, Speed Governor, Water Hammer, Neural Network, Fuzzy Logic, Sliding
Mode Control, Chattering.
1. Introduction
With the increasing acute global energy crisis, non-linear dynamical behaviors of the
hydropower, an ideal renewable and clean energy, hydro-turbine system.
has been given a widely social concern. It is According to many different mathematical
generally known that the hydropower generation models, various control methods have been
has a great development potential and a broad proposed. Tan [4] has designed a PID tuning of
market prospect. A hydro-turbine governor system load frequency controller based on a two-degree-
is an important part of the hydropower plants [1], of-freedom internal model control. Using a
and the speed governor plays a vital role in high-gain observer to replace the coordinate
controlling the hydro-turbine speed. Therefore, transformation, Liu [5] has presented a nonlinear
research work on the hydro-turbine governor robust control strategy, which is only required to
system has becomes significant. measure the rotor speed. Kishor [6] has used the
A hydro-turbine governor system is a complex H-infinity method to approximate the non-linear
nonlinear object, which features time- variant and part of the penstock-turbine transfer function, and
non-minimum phases. For the non- linear has studied the hydraulic transient characteristics.
governor system, various mathematical models Qian [7] has presented a GA- based fuzzy sliding
and control methods have been presented. mode control approach for speed governing of a
Many different mathematical models have been hydro-turbine. Most of the methods mentioned
proposed in terms of the hydropower plants. For above only focus on the accurate linear model and
example, Paolo [2] has presented a detailed the non-linear dynamical analysis. However, few
numerical model for the dynamic behavior of the papers have paid attention to the uncertainties
Francis turbine, and has considered the effects of generated in the process of linearization of the
water hammer on the turbine. Chen [3] has non-linear mode, and the influence of elastic
introduced a novel model of a hydro-turbine water hammer upon the security and stability of a
system with a surge tank, and has studied the hydro-turbine governor system.
Yu & Qian/ Journal of AI and Data Mining, Vol 4, No 1, 2016.
In order to further investigate the hydro -turbine of the flowing fluid in the pipelines. The
governor system, a novel non-linear mathematical emergence of elastic water hammer can cause
model with elastic water hammer was developed. serious problems for the hydraulic turbine, and
By linearizing the nonlinear mode, the sliding then affects the whole power system. Therefore,
mode controller was designed. The advantage of the stability study of the hydroelectric power plant
the sliding mode control method is that in the with elastic water hammer is of paramount
sliding motion, the designed controller shows importance. In this paper, the mathematic models
strong robustness and high control accuracy in the of the water tunnel and penstock components are
presence of uncertainties and external presented under the assumption of elastic water
disturbances. However, the designed sliding mode hammer. The elastic water hammer is represented
controller based on the linearized model cannot by a delay block of e-Trs. Three components
stabilize the non-linear governor system. interconnected with each other constitute the
Furthermore, a chattering phenomenon [8] exists nonlinear governing system, as shown in figure 1.
in the conventional sliding mode control method. ey
Therefore, in this work, we presented a eqh mg0
neural-fuzzy sliding mode control scheme, which u 1 q h mt 1 x
eqy 2 hw eh
combines with the merits of the fuzzy set theory Ty s +1 y Tas eg
[9-10], neural networks [11-12], and sliding mode ex
control [13-14]. A radial basis function (RBF) 2 eTr s
neural network was introduced to compensate for eqx
1
the uncertainties generated in the process of s
linearization of the non-linear mode [15]. The
Figure 1. Transfer function block diagram of a
update formula of the network was deduced from hydro-turbine governor system.
the Lyapunov direct method so that the weight
convergence and system stability could be Considering a small variation around an operating
simultaneously guaranteed. Considering the point, the linearized equation of the Francis
chattering phenomenon of the sliding mode hydro-turbine can be given as
control, a fuzzy logic inference system was mt ex x ey y eh h
adopted to regulate the gain of the switching (1)
q eqx x eqy y eqh h
control law according to the system performance
[16]. In the sense of Lyapunov, the asymptotically The six hydro-turbine constants ex, ey, eh, eqx, eqy,
stability of the whole system could be guaranteed. and eqh can be defined as:
The simulation results obtained demonstrated the mt m m q q q
ex , ey t , eh t , eqx , eqy , eqh
effectiveness of the proposed scheme to improve x y h x y h
the dynamic performance of the system and The effects of surge tank are ignored, and the
attenuate the chattering phenomenon. transfer function of penstock with the incremental
This paper is organized as follows. Under the head and flow can be written as
assumption of elastic water hammer, the detailed h( s )
2hwth(0.5Tr s ) (2)
nonlinear mathematical model of a hydro-turbine q( s )
governor system is given in Section 2. The design
of a new neural-fuzzy sliding mode controller is Compared with the traditional simplified method
reported in Section 3. The simulation results for of the Taylor series expansion, this paper presents
the different states in the hydro-turbine governor a new simplified nonlinear method, as shown in
system are illustrated in Section 4. Finally, function (3).
conclusions are given in Section 5. When the hyperbolic tangent function is
converted to an exponential function, the transfer
2. Mode of hydro-turbine governor system function of penstock with the incremental head
2.1 Penstock and hydro-turbine and flow can be given as
When penstocks are short or medium in length, h( s )
Gh ( s ) 2hwth(0.5Tr s )
the hydraulic turbine model is adequate under the q( s )
(3)
assumption of the inelasticity of penstocks and the eTr s
2hw (1 2 )
incompressibility of fluid. However, considering 1 eTr s
the hydropower plants with long conduits, it has The new method takes into accounts the higher
to take into account the effects of elasticity of order terms of the system model. Therefore, the
penstock and compressibility of fluid. Water accuracies of the hydraulic tunnel and the
hammer is produced by the rapid velocity changes penstock are greatly improved. In figure 1, the red
86
Yu & Qian/ Journal of AI and Data Mining, Vol 4, No 1, 2016.
section means the hydraulic turbine part taking deviation x1=x. To eliminate the speed deviation,
elastic water hammer; this relation only depends the integral of x1 is utilized as an additional state
on the length of penstock. Considering the elastic x4, which is defined as
water hammer, the hydro-turbine transfer function
x4 x1dt (8)
relating to the mechanical power and the wicket 0
gate opening can be derived and written as The state space expression for the hydro-turbine
mt (s) 1 2hwe tanh(0.5Tr s) governor system with model uncertainties and
ey (4)
y(s) 1 2hweqh tanh(0.5Tr s) external disturbances is described as follows.
where, e=eqyeh/ey-eq, and hw is the normalized x ( A+A)x ( B B)u F (d (t ) g (t ))
hydraulic impedance of penstock, the water inertia Ax Bu f (9)
time is Tw=Trhw, the penstock reflection time is y CT x
Tr=2L/a, penstock length is L, and the water wave
where, x=[x1 x2 x3 x4]T is the state vector; A and
velocity is a. B are unknown or uncertain; d(t) values are
disturbances including the parameter variations
2.2 Wicket gate and servomechanism and load disturbances; g(t) is the external
The servomotor is used to amplify the control indefinite disturbance; f is the uncertainties
signal, and to provide power to operate the guide consisting of all unknown parts and disturbances;
vane. Neglecting small time constants, the transfer and CT=[1 0 0 0] is the output matrix.
function relating to the control signal u and the eg ex 1
wicket gate servomotor stroke y can be written as: 0 0
Ta Ta
y(s) 1 eqy eh eqh ey eqy eh eqh e y
1 ey
u (s) Ty s 1 (5) 0
A eqhTw Ty eqh eqhTw Ty eqh
where, Ty is the response time of the wicket gate 0 1 1
0
servomotor Ty Ty
1 0 0 0
2.3 Generator and network eqy eh eqh ey 1 T
B [0 0]
If the generator unit supplies for an isolated load, Ty eqh Ty
the dynamic process of the synchronous generator 1
unit only relating to the moment of inertia can be F [ 0 0 0]T
Ta
described as f Ax Bu F (d (t ) g (t ))
x(s) 1
(6)
[(mt mg 0 )(s)] Ta s eg The sliding mode controller consists of two stages.
First the sliding surface is defined as
where, mt is the turbine torque (in per unit), mg0 is
s cT x c1 x1 c2 x2 c3 x3 c4 x4 (10)
the load torque (in per unit), Ta is the generator
unit mechanical time (in seconds), and eg is the where, ci (i=1,2,3,4) is the sliding surface
load self-regulation fact. coefficient vector, which is positive constant.
Secondly, the sliding mode control law is
3. Control designed and analysis designed, which is composed of the equivalent
3.1 Design of RBF network-based sliding mode control law and the switching control law
controller u ueq usw (11)
Taking into account the linearized small-signal
model of an ideal Francis turbine, the expression where, ueq forces the system trajectory to the
of penstock with inelastic water hammer and sliding mode stage, and is defined as the
non-friction can be obtained by the Taylor series equivalent control law. usw keeps the system
expansion. trajectory on the sliding surface and is defined as
h( s ) the switching control law.
2hwth(0.5Tr s ) Tw s (7)
q ( s ) Furthermore, ueq can be obtained by
Thus a 3-order mode of state space is obtained s cT x 0 (12)
by linearization of the simplified model, where Substituting (8) into (12) yields
the system state variables are independent and ueq (cT B)1 cT Ax (13)
measurable. The wicket gate servomotor stroke
relative deviation x3=y, the turbine torque relative Take into account the following positive definite
deviation x2=mt and the turbine speed relative function as a Lyapunov candidate function:
87
Yu & Qian/ Journal of AI and Data Mining, Vol 4, No 1, 2016.
88
Yu & Qian/ Journal of AI and Data Mining, Vol 4, No 1, 2016.
Degree of membership
and non-zero,and the sliding surface variable s is
0.8
the input of the fuzzy controller. When the s value
0.6
is zero, then the control law of the fuzzy sliding
0.4
mode controller is only determined by the
0.2
equivalent control. Similarly, when the s value is
0
non-zero, then that is composed of the equivalent
-30 -20 -10 0 10 20 30
control and the switching control. s
However, in practice application, only two control NB NM ZO PM PB
1
Degree of membership
rules cannot be enough to completely describe the
0.8
operation of a non-linear hydro-turbine governor
0.6
system. In order to eliminate this weakness and
attenuate the chattering phenomena 0.4
89
Yu & Qian / Journal of AI and Data Mining, Vol 4, No 1, 2016.
regulation time was greatly shortened. Compared Figure 4(c). Water gate deviation x3.
with the conventional PID control, the dynamic
performance of the state variables improved 0.25
0.02
0.1
0.01
Control input u
Fuzzy SMC
0
0.05
-0.01
-0.02 0
0 10 20 30 40 50 60
Time(s)
90
Yu & Qian / Journal of AI and Data Mining, Vol 4, No 1, 2016.
-0.04
Additional state x4
91
Yu & Qian / Journal of AI and Data Mining, Vol 4, No 1, 2016.
92
نشرهی هوش مصنوعی و داده کاوی
طراحی مدار فرمان برای نیروگاه برق آبی توسط کنترل ساختار متغیر حالت لغزان هوشمند
دانشکده کنترل و مهندسی کامپیوتر ،دانشگاه اترژی برق شمال چین ،منطقه چانگ پینگ ،پکن ،چین.
چکیده:
آ چکه این تحقیق یک روش کنترل عصبی فازی با حالت لغزان را برای یک سیستم مدیریت سرعت توربین آبی پیشنهاد میکند .با توجه بهه فهر
االستیک ،یک حالت غیر خطی سیستم فرماندار توربین آبی ایجاد شده است .با خطی سازی این حالت ،یک کنترل حالت کشهویی طراحهی شهده اسهت.
حالت خطی منوط به عدم قطعیت میباشد .عدم قطعیت در روند خطی تولید میشود .یک شبکه عصبی تهاب پایهه شهعاعی ( )RBFبهرای جبهران عهدم
قطعیت معرفی شده است .فرمول به روز رسانی شبکه های عصبی از روش مستقیم لیاپانوف استخراج شده است .برای این پدیده چترینگ از کنتهرل مهد
لغزشی ،یک سیستم استنتاج منطق فازی اتخاذ شده است .در مفهوم لیاپانوف ،ثبات مجانبی سیستم میتواند تضمین شود .در مقایسه بها کنتهرل حالهت
داخلی و روش کنترل PIDمعمولی ،برخی از شبیهسازیهای عددی امکانسنجی و استحکام طرح پیشنهادی ارائه شده را تأیید میکنند.
کلمات کلیدی. :نیروگاه برق آبی ،مدیریت سرعت ،آ االستیک ،شبکه عصبی ،منطق فازی ،کنترل حالت لغزان.