Inter national J our nal of Electrical and Computer Engineering (IJECE) V ol. 11, No. 2, April 2020, pp. 1186 1199 ISSN: 2088-8708, DOI: 10.11591/ijece.v11i2.pp1186-1199 r 1186 Obser v er -based tracking contr ol f or single machine infinite b us system via flatness theory Mohammad P ourmahmood Aghababa 1 , Bogdan Marinescu 2 , Flor ent Xa vier 3 1,2 Ecole Centrale de Nantes, LS2N, France 3 R ´ eseau de T ransport d’Electricit ´ e (R TE), France Article Inf o Article history: Recei v ed Dec 11, 2019 Re vised Jul 31, 2020 Accepted Aug 17, 2020 K eyw ords: Critical clearing time Flat system Input constraints State observ er T ransient stability ABSTRA CT In this research, we aim to use the flatness control theory to de v elop a useful control scheme for a single machine connected to an infinite b us (SMIB) system taking into account input magnitude and rate saturation constraints. W e adopt a fourth-order non- linear SMIB model along an e xciter and a turbine go v ernor as actuators. According to the flatness-based control strate gy , first we sho w that the adopted nominal SMIB model is a flat system. Then, we de v elop a full linearizing state feedback as well as an outer inte gral-type loop to e nsure suitable tracking performances for the po wer and v oltage as well as the angular v elocity outputs. W e assume that only the angular v e- locity of the generator is a v ailable to be measured. So, we pro vide a linear Luenber ger observ er to estimate the remaining states of the system. Also, the saturation nonlinear - ities are transferred to the linear part of the system and the y are canceled out using their estimations. The ef ficienc y and usefulness of the proposed observ er -controll er ag ainst f aults are illustrated using simulation tests in Eurostag and Matlab . The results sho w that the clearing critical time of the introduced methodology is lar ger than the classical control approaches and the proposed observ er -based flatness controller e xhibits o v er much less control ener gy compared to the classic IEEE controllers. This is an open access article under the CC BY -SA license . Corresponding A uthor: Bogdan Marinescu Ecole Centrale de Nantes LS2N, Nantes, France Email: Bogdan.Marinescu@ec-nantes.fr 1. INTR ODUCTION No w adays, the electricity has become as an important and vit al component of the life and industry . So, the electrical po wer netw orks should be in a secure operation with a reas onable stability mar gin to produce the demanded electricity . The ability of a po wer system in maintaining in the machines synchronous operation point after occurrence of a disturbance and/or f ault is usually interpreted as its transient stability concept. T o retain the po wer system stability in a suitable limit in the e v ent of uncertainties, f aults and disturbances, it is necessary to add control actions, such as e xciters and go v ernors, to the system to impro v e dynamics till the circuit break er opening and reclosing times [1]. The critical clearing time (CCT) is one useful and applied f actor to measure the transient stability mar gin of a po wer machines. The CCT stands for the maximum time during which a f ault can be applied without missing the system’ s stability . Such a stability mar gin depends on the design of the controls of generators connected to the grid [2]. T o enhance the CCT of a grid, some control de vices should be designed and implemented in the netw ork. T o synthesis and analyze the performance of the controllers on the CCT of a po wer grid, a single J ournal homepage: http://ijece .iaescor e .com Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Elec & Comp Eng ISSN: 2088-8708 r 1187 machine connected to an infinite b us (SMIB) po wer syst em model is usually adopted to a v oid the unnecessary comple xity of the po wer system in the control design phase. Generally speaking, there are tw o main controller classes for this case: (i) standard controllers and (ii) adv anced control techniques. The first class belongs to the well-kno wn standard IEEE controllers. More details about the classic IEEE controllers can be found in [3, 4]. Although the classic IEEE controllers are simple, their parameters are needed to be appropriately adjusted and their stabil ity re gions are limited. In the w orks [5–7] some intelligent heuristic optimization methods ha v e been proposed for finding the suitable parameters of the re gulators. Ho we v er , since their approaches requires implementation of some iterati v e numerical algorithms , their practi cal implementations will be dif ficult in online and f ast response needed situations. On the other hand, the approaches in the second class use some so-called adv anced control strate gies to enhance the transient stability of the po wer machines. There are se v eral e xamples in the literature for this cate gory which include sliding mode control [8], fuzzy control [9], nonlinear control [10], dynamic in v ersion control [11] and optimal control [12], etc. In [13], a po wer system stabilizer has been proposed for synchronous machines based on con v en- tional fuzzy-PID and type-1 fuzzy controller combined with a sliding mode control strate gy . The w ork [14] has proposed an adapti v e w a v elet netw ork-based nonlinear e xcitation control for po wer systems without consider - ing the go v ernor dynamics. T o impro v e the stability of the v oltage re gulation and to enhance the damping of lo w frequenc y po wer system oscillations of SMIB systems, an e xtended reduced-order observ er along with an automatic v oltage re gulator has been de v eloped in [15]. In [16], the relationship between transient stability = instability and conca vity = con v e xity of the phase-plane tr ajectory has been found and a transient instability cri- terion has been deri v ed for real-time instability detection and the SMIB system has been stabilized. Rout et al [17] ha v e sho wed that the SMIB system can possess chaotic and oscillatory dynamics when the system parame- ters f all into a certain area. Accordingly , the y ha v e designed an adapti v e controller based on LaSalle’ s in v ariant principle to mak e the system oscillations damped. The paper [18] has in v estig ated the problem of transient sta- bility and v oltage re gulation for a SMIB system via a modified backstepping control design method. Ho we v er , most of the pre vious w orks either ha v e not considered the ef fects of input saturations, the y ha v e been designed for some simplified linear and/or nonlinear models of the SMIB, there are usually steady state errors on the outputs of their methods or the y ha v e assumed that all the states of the system are a v ailable to be measured. The concept of dif ferentially flat nonlinea r systems w as first introduced by Fliess et al [19, 20]. The scheme is an e xtension from the input-output scheme with zero internal dynamics. A system is con- sidered to be dif ferentially flat if all its state v ariables and its control inputs can be e xpressed as functions of one single algebraic v ariable which is the so-ca lled flat output, and also as functions of the flat-output’ s deri v a- ti v es. The dif ferential flatness property enables the transformation of the nonlinear system’ s dynamics into the linear canonical form and the design of a state feedback controller through the application of pole placement techniques in the linearized equi v alent model of the system. The construction of the feedback la w is done by a simple in v ersion of system equations with respect to the system input. Although this technique has been applied to se v eral nonlinear and linear mechanical systems [21–23], its application to the control of po wer systems has been limited to a fe w w orks [24–26] and [27]. Ho we v er , the pre vious w orks ha v e not focused on the transient stability mar gin and the y either ha v e not carried out the ef fects of the actuator saturations or the y ha v e assumed that all the states of the synchronous machines are a v ailable to be measured. In this research, i nspired by the flatness control theory , we propose a full linearizing state feedback for the system to cancel out the nonlinearities of the system and to obtain a linear canonical (Bruno vsk y) form for it. Then, we transfer the input saturation nonlinearities of the system to the adopted linear part of the model. T o mak e the obtained linear system to be controllable and observ able, some modifications are done on the nonlinear feedback control to modify the linear matrix of the system. Subsequently , a full order state linear observ er is designed for the system to est imate the states of the model using the only a v ailable output, i.e. the angular v elocity . T o pro vide tracking of v oltage and po wer reference commands, i nte gral outer loops are added to the linear part of the system . The Kalman filter and the linear quadratic re gulator approaches are adopted to deri v e the linear observ er and controller , respecti v ely , and to construct the frame w ork of the final augmented system. Finally , the ef ficienc y and applicability of the proposed observ er -based state feedback controller are re v ealed using some illustrati v e e xamples si mulated in the presence of short circuit f aults and magnitude and rate saturations of input signals, where the superiority of the introduced controller is re v ealed for its rob ustness ag ainst unw anted f aults increasing CCT of the machines and its lo w implementation costs reducing the control ener gy required for the stabilization of the system compared to the classic IEEE control schemes. Observer -based tr ac king contr ol for ... (Mohammad P ourmahmood Aghababa) Evaluation Warning : The document was created with Spire.PDF for Python.
1188 r ISSN: 2088-8708 2. SYSTEM MODELING AND PR OBLEM FORMULA TION Figure 1 sho ws a simple representation of a po wer system consisting of a single machine connected to an infinite b us through a transmission line with impedance Z = r e + j x ep . In the sequel, we will present the one-axis model, named also flux decay model, of the SMIB po wer system where the reader may refer to [1] for more details. T r an s m is s io n   lin e   = +    ~ ( , )   G   I n f in ite  b u s   ~ ( , )     Figure 1. A SMIB po wer system 2.1. One-axis model of a SMIB The mechanical dynamics of the generator , which correspond to the rotor’ s relati v e angle ( ) and the angular v elocity ( ! ), are described by: _ = ( ! ! s ) ! 0 (1) _ ! = ! s 2 H ( T m T f w T e ) (2) where ! s is the rated synchronous speed in p.u, ! 0 is the rated synchronous speed in rad/s, H is the machine’ s inertia, T m represents the mechanical torque applied to the shaft, T f w = D ( ! ! s ) is a friction windage torque with a damping coef ficient D and T e is the electrical torque defined by the follo wing e xpression T e = i q e 0 q + ( x q x 0 d ) i d i q (3) where x q is the synchrounous reactance of the generator in quadratic axis and x 0 d is its transient reactance in direct axis. The mechanical po wer , denoted P m , is defined as: P m = T m ! (4) Inserting in 2) T e gi v en by (3) and T m gi v en by (4) and assuming D = 0 , one gets _ ! = 1 2 H ( T m i q e 0 q ( x q x 0 d ) i d i q ) (5) The dynamic of the induced v oltage on the q -axis, denoted e 0 q , is gi v en by: _ e 0 q = 1 T 0 d 0 e 0 q ( x d x 0 d ) i d + E f d (6) where E f d represents the v oltage across the rotor field coil , which is considered as an input for the SMIB model to be designed. The mathematical equation of the go v ernor dynamics is gi v en belo w . _ P sv = 1 T g ( T cc P sv + K g ( ! ! s )) (7) in which T cc is a constant to be designed as a reference for mechanical torque and P sv is the steam v alv e position and it represents the second input for the SMIB model. The turbine dynamical equation is also gi v en by: _ T m = 1 T ch ( P sv T m ) (8) Int J Elec & Comp Eng, V ol. 11, No. 2, April 2020 : 1186 1199 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Elec & Comp Eng ISSN: 2088-8708 r 1189 The currents in dq -coordinates ( i d , i q ) are obtained by solving a set of algebraic equations, whic h represents the interaction of the machine with the po wer grid: i d = c 1 + c 2 e 0 q i q = c 3 + c 4 e 0 q (9) with c 1 = ( r s + r e ) sin ( ) + ( x q x ep ) cos ( ) ( r s + r e ) 2 + ( x 0 d + x ep )( x q + x ep ) V s c 2 = x q + x ep ( r s + r e ) 2 + ( x 0 d + x ep )( x q + x ep ) c 3 = ( x 0 d + x ep ) sin ( ) ( r s + r e ) cos ( ) ( r s + r e ) 2 + ( x 0 d + x ep )( x q + x ep ) V s c 4 = r e + r s ( r s + r e ) 2 + ( x 0 d + x ep )( x q + x ep ) Assuming r s = r e = 0 (i.e., o v erhead lines with ne glectable resistances compared to reactances), the currents i d , i q are simplified as follo ws: i q = V s x ep + x d sin ( ) i d = 1 x ep + x 0 d e 0 q V s x ep + x 0 d cos ( ) (10) The abo v e dynamics (1)-(10) can be re grouped in the follo wing fifth-order state form: _ x ( t ) = f ( x ( t ) ; u ( t )) ; x ( t 0 ) = x 0 (11) where x 1 = , x 2 = ! , x 3 = e 0 q , x 4 = T m , u 1 = E f d and u 2 = P sv . It is noted that in this w ork, we will include the go v ernor dynamics (7) in an outer control loop. It should be also noted that due to ph ysical limi tations, the control inputs ha v e to v erify the follo wing constraints: Magnitude limitation: The e xciter may generate positi v e and ne g ati v e v alues where symmetric bounds are generally considered: j u 1 j u 1 (12) Ho we v er , the steam v alv e position is limited between 0 (completely closed) and 1 (completely open): 0 u 2 u 2 (13) Rate limitation: Beside the magnitude saturation, the v ariation of steam v alv e position has to respect some limitations: 1 T c _ u 2 1 T o (14) where T o ( T c ) is the necessary time to pass from a completely closed (open) position to a completely open (closed) position. Remark 1: One can use the first-order deri v ati v e approximation _ u = u ( t ) u ( t ) (where is a small constant) to con v ert the rate saturation to a magnitude saturation as follo ws: 1 T c _ u 2 1 T o ) 1 T c u 2 ( t ) u 2 ( t ) 1 T o ) T c + u 2 ( t ) u 2 ( t ) T o + u 2 ( t ) (15) Observer -based tr ac king contr ol for ... (Mohammad P ourmahmood Aghababa) Evaluation Warning : The document was created with Spire.PDF for Python.
1190 r ISSN: 2088-8708 No w , defining u 2 M in = T c + u 2 ( t ) and u 2 M ax = T c + u 2 ( t ) and noting to 0 u 2 u 2 , we will ha v e u 2 L u 2 ( t ) u 2 H (16) with u 2 L = max f 0 ; u 2 M in g and u 2 H = min f u 2 ; u 2 M ax g . 2.2. SMIB Model in faulty mode operation The appearance of a temporary short circuit in the transmission line af fects considerably the system’ s stability . Let us c o ns ider the case of tw o parallel transmission lines with the same impedance Z 0 = 2( r e + j x ep ) . This allo ws us to preserv e the same impedance of the pre vious t ransmission line Z . The short circuit will happen at ti me t c at the middle of the second transmission line. This f aulty mode af fects the SMIB model by changing its parameters ( r e , x ep ) by 2 3 ( r e , x ep ) and V s by 1 3 V s as reported in [28]. Notice that in practice the short circuit may happen at an y point of the line. It has been e xamined here in the middle for the sak e of simplicity . F or e xample, if the short circuit appears at the terminal of the generator , V s will be replaced by zero in the dynamics (1)-(10). In this case, the machine’ s speed increases, with respect to the dynamic (17), as long as the short circuit is present and depending on its durat ion (denoted t ) the machine may lose synchronism. _ ! = ! s 2 H T m (17) 2.3. Contr ol objecti v es The main control objecti v e for the synchronous machine is to operate at synchronous speed ! s , main- tain a constant modulus of the terminal v oltage V r ef and achie v e a desired mechanical po wer P r ef under control inputs constraints. Accordingly , we will try to mimic the standard classic IEEE go v ernor model (7) in the closed-loop system. The modulus of the terminal v oltage of the synchronous generator , denoted v t , is gi v en by: v t = q v 2 d + v 2 q (18) where: v d = r e i d x ep i q + V s sin ( x 1 ) v q = r e i q + x ep i d + V s cos ( x 1 ) (19) 3. OBSER VER-B ASED NONLINEAR CONTR OLLER DESIGN In this sect ion, first we sho w that the nonlinear model SMIB system is flat. Then, we will c ompensate the ef fects of the input saturations. Finally , a linear state observ er is designed for the system to b uilt the final flatness-based controller . 3.1. Checking out flatness of SMIB Consider a general case of the nonlinear system (11) with x ( t ) 2 R n and u ( t ) 2 R m . It is said to be dif ferentially flat if and only if there e xists a flat output z ( t ) 2 R m of the form : z = h ( x; u; _ u; :::; u ( r ) ) (20) such that x = ( z ; _ z ; :::; z ( q ) ) u =   ( z ; _ z ; ::: ; z ( q +1) ) (21) The components of the flat output z are dif ferentially independent. These equations yield, that for e v ery gi v en trajectory of the flat output t ! z ( t ) , the e v olution of all other v ariables of the system t ! x ( t ) and t ! u ( t ) are also gi v en without inte gration of the system of dif ferential equations Proposition 1 : The one-axis model of the single machine infinite b us po wer system gi v en by (11) is dif ferentially flat with the output z = [ x 1 x 4 ] T . Int J Elec & Comp Eng, V ol. 11, No. 2, April 2020 : 1186 1199 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Elec & Comp Eng ISSN: 2088-8708 r 1191 Proof: W e should pro v e that the remaining states x 2 = ! and x 3 = e 0 q as well as the control inputs u 1 = E f d and u 2 = P sv can be written as a function of the flat outputs and their successi v e deri v ati v es. First, from (2), one gets ! = _ w 0 + ! s (22) So, x 2 = ! is stated by the flat output . From (2), one obtains the follo wing relation for e 0 q . e 0 q = T m 2 H w 0 w AAV s i q (1 + AA ) i q (23) where AA = x q x 0 d x ep + x 0 d . The ne xt step is to e xpress the control inputs as functions of the outputs and their deri v ati v es. T o this end, in what follo ws, we replace the control inputs u 1 ( t ) and u 2 ( t ) with S at ( u 1 ) and S at ( u 2 ) , respecti v ely to include the corresponding saturation nonlinearities. As a result, using (6) and (23), one can obtain S at ( u 1 ) = T d 0 _ e 0 q + e 0 q + ( x d x 0 d ) i d (24) where _ e 0 q = h _ T m 2 H w 0 w s + AAV s ( _ iq cos ( ) sin ( ) i q ) i ( i q (1+ AA )) F ( X ) ( i q (1+ AA )) 2 with F ( X ) = _ iq (1 + AA )( T m 2 H w 0 w _ + AAV s cos ( ) i q ) and _ iq = V scos ( ) x ep + xq . And, according to the (8) one can easily obtain S at ( u 2 ) = T ch _ T m + T m (25) So, the proof is completed and the selected v ariables z = [ x 1 x 4 ] T are the flat outputs of the SMIB system. 3.2. Flatness-based linearizing state feedback Here, we consider the ne w state v ector = [ 1 2 3 4 5 ] T with 1 = x 1 = , 2 = _ 1 , 3 = 1 , 4 = x 4 = P m and 5 = _ P m . In what follo ws, we conclude a proposition to highlight the equi v alence between the nonlinear model of a SMIB po wer system and a linear controllable one. Before proceeding to the proposition, we first re write the saturation function as follo ws: S at ( u i ) = u i + u i ; i = 1 ; 2 : (26) where u i is defined as follo ws: u i = 8 < : u H i u i ; u i u H i 0 ; u L i u i u H i u L i u i ; u i u L i (27) Therefore, based on (26) and defining v 1 = and v 2 = _ T m , the control inputs u 1 (24) and u 2 (25) are reformed as (28) with v 1 = w s w 0 u 1 2 H and v 2 = u 2 T ch . u 1 = T d 0 [ _ T m 2 H w 0 w s ( v 1 + v 1 )+ AAV s ( _ iq cos ( ) sin ( ) i q )]( i q (1+ AA )) T d 0 F ( X ) ( i q (1+ AA )) 2 + T m 2 H w 0 w AAV s i q (1+ AA ) i q + ( x d x 0 d ) i d u 2 = T ch ( v 2 + v 2 ) + T m (28) Proposition 2 : Under the mapping = ( x ) gi v en by: ( x ) = 0 B B @ x 1 x 2 ! s ! 0 ( x 3 i q (1+ AA )+ x 4 + AAV s i q ) w 0 w s 2 H x 4 1 C C A (29) Observer -based tr ac king contr ol for ... (Mohammad P ourmahmood Aghababa) Evaluation Warning : The document was created with Spire.PDF for Python.
1192 r ISSN: 2088-8708 and the state feedback gi v en in (28), the nonlinear model of the SMIB is equi v alent to the follo wing semi-linear (i.e. linear system with nonlinear uncertainties) controllable system presented in (30). _ = A + B ( v + v ) (30) where: v = v 1 v 2 ; v = v 1 v 2 ; A = 2 6 6 4 0 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 3 7 7 5 ; B = 2 6 6 4 0 0 0 0 1 0 0 1 3 7 7 5 (31) Proof: By considering the flat v ariables x 1 and x 5 , the corresponding Bruno vsk y form has a full rank controllability matrix, which means that one can propose a state feedback that can linearize the SMIB model (11) [20, 29]. Using the abo v e dif feomorphism and replacing the control e xpression (28) in the dynamics of the ne w coordinates yields the abo v e linear controllable system (30). Remark 2: It should be noted that although, we ha v e not used the deri v ati v es of the saturation func- tions of the control inputs in proposition 1 to pro v e the flatness property of the SMIB system, the a v ailable discontinuities in the u may mak e the proposed dif feomorphism mapping to be local from the equi v alenc y point of vie w . Ho we v er , transferring the discontinuous term u to the semi-linear side as the term v and noting to this f act that the discontinuities of v are e xactl y equal to those of u (due t o this f act that v includes the e xact u ), will a v oid this problem and, therefore, the proposed dif feomorphism will be global. Remark 3: It should be noted that although, we ha v e not used the deri v ati v es of the saturation func- tions of the control inputs in Proposition 1 to pro v e the flatness property of the SMIB sys tem, the a v ailable discontinuities in the u mak e the proposed dif feomorphism mapping to be local from t he equi v alenc y point of vie w . In f act, when the control inputs hit the saturation limits, there will not be a one-by-one mapping between the semi-linear Bruno vsk y system and the corresponding nonlinear SMIB model an ymore. Ho we v er , e v en in this case , one can design proper linear control inputs v ( t ) to stabili ze the system lik e an output feedback linearization scheme. 3.3. Linear state obser v er design From flatness-control point of vie w , one of the main practical significance of the flat outputs of a dynamical system is that if the flat outputs are measurable, then all the system v ariables required for feedback can be directly computed without inte grating an y dif ferential equations. In our case, the rotor’ s relati v e angle and the mechanical po wer are selected as the system flat outputs. Ho we v er , in practice, these v alues cannot be easily measured. And, in most practical situations, the angular v elocity of the machine is the only a v ailable output to be measured. Accordingly , if we w ant to implement the abo v e-designed controller , we should fist de v elop an observ er for the system to estimate the states of the system using the angular v elocity output. In what follo ws, the observ er design procedure for the SMIB system is presented. Instead of directly de v eloping an observ er for the SMIB system gi v en in (1)-(10), we use the equi v alent semi-linear system (31) to propose an ef ficient linear Luenber ger observ er . T o this end, we first select the output matrix of the system as y = [0 1 0 0] T and check the observ ability property of the system (31). After checking the rank of the observ ability matrix, it is re v ealed that it is singular (it is rank ef ficient) and, therefore, the system (31) is not observ able. T o solv e this problem, we propose an alternati v e: modify the linear matrix A in (31) using modifying the linear control inputs v 1 and v 2 to ob t ain an observ able system. In this line, we add T m and to the control inputs u 1 and u 2 in (28), respecti v ely as formulated in (32). u 1 = T d 0 [ _ T m 2 H w 0 w s ( v 1 T m | {z } v 1 n + v 1 + T m | {z } v 1 n )+ AAV s ( _ iq cos ( ) sin ( ) i q )]( i q (1+ AA )) T d 0 F ( X ) ( i q (1+ AA )) 2 + T m 2 H w 0 w AAV s i q (1+ AA ) i q + ( x d x 0 d ) i d u 2 = T ch ( v 2 | {z } v 2 n + v 2 + | {z } v 2 n ) + (32) Int J Elec & Comp Eng, V ol. 11, No. 2, April 2020 : 1186 1199 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Elec & Comp Eng ISSN: 2088-8708 r 1193 The abo v e modifications are equi v alent to put a 1 in the third ro w-forth column of the matrix A and another 1 in the firth ro w-fifth column of the matrix A as follo ws: A = 2 6 6 4 0 1 0 0 0 0 1 0 0 0 0 1 1 0 0 0 3 7 7 5 (33) No w , the pair ( A; C ) is observ able and the pair ( A; B ) is controllable (their corresponding observ abil- ity and controllability matrices are full rank). The linear Luenber ger observ er is designed using the follo wing dynamics: _ b = A b + B V + L ( y C ) (34) where V = [ v 1 n v 2 n ] T is the linear control input to be designed later . The observ er g ain matrix L can be easily obtained using the Kalman command of the Matlab softw are. So, the estimated states b will be used to construct the final controller (including linear controls v 1 n ; v 2 n and nonlinear ones u 1 ; u 2 ). T able 1. T ypical parameters of a SMIB po wer system in p.u P arameter ! s V s T 0 d 0 x d x q x 0 d H T ch r s r ep D x ep V alue 1 1 9.6 2.38 1.21 0.336 5 5 0 0 0 0.4 3.4. Output tracking loop The ne w v ector of control V has to guarantee the stability of the SMIB system and achie v e the control requirements discussed before. Thus, the state v ector has to be dri v en to a desired state r ef , which corre- sponds to a desired relati v e angle, a zero relati v e speed, a zero acceleration and a desired mechanical torque. Ho we v er , in addition to the dri ving the states of the semi-linear system to the desired the reference v alues, for better tracking purposes and good rob ustness properties, we add an output inte grator loop to the system to mak e it for ef ficient satisf action of the control objecti v es. On the other hand, one needs to cancel out the ef fects of the semi-linear ter ms v in the system. So, the output tracking loop includes thre e parts: i) an acti v e control for cancelling the semi-linear parts, ii) a feedforw ard tracking scheme for pro viding the tracking conditions of linear state space systems, and iii) an inte grator loop for including the output errors in the obtained equv alent linear system. These items are e xplained in the ne xt subsections. 3.4.1. Acti v e contr ol Here, we aim to cancel out the ef fects of the semi-linear parts v 1 n and v 2 n . Noting to the defi- nitions v 1 n = w s w 0 u 1 2 H + T m and v 2 n = u 2 T ch + , it is re v ealed that these terms are kno wn. So, we easily use the so-called acti v e control method [30] to directly remo v e them by a feedback control. Ho we v er , it should be noted that we cannot use t he states of the nonlinear system; ins tead, we use the outputs of the deri v ed observ er to b uild estimations for v 1 n and v 2 n . Accordingly , the estimated semi-linear parts will be added to the control input v n to cancel out the ef fects of the corresponding terms. In this line, one can define a ne w linear control input as follo ws: v n = v f v (35) where v f is the ne w linear control input and v = [ v 1 n v 2 n ] T . Based on the abo v e formulation, the equi v alent linear system for the nonlinear model of the SMIB becomes: _ = A + B v f y = C (36) Observer -based tr ac king contr ol for ... (Mohammad P ourmahmood Aghababa) Evaluation Warning : The document was created with Spire.PDF for Python.
1194 r ISSN: 2088-8708 3.4.2. F eedf orward tracking scheme Here, we incorporate the reference input into the observ er/controller state feedback. This method is useful for tracking of slo wly changing references. Consider the state space system (36). 1. W e first design a state feedback g ain K such that A B K is stable (with poles at nice locations); 2. Suppose that r ef is the reference input. W e w ould associate a steady state v ector ss = N r ef for an y constant reference input r ef . No w we define the control to be: V f = K ( ss ) + v ss (37) where V ss = N v f r ef is the steady state control input to maintain at ss . 3. T o define N v f and N , we consider the desired steady state relationships: _ ss = A ss + B v f ss = AN + B N v f r ef = 0 y ss = C ss = C N r ef = r ef (38) T o mak e this w ork for all r ef , we need to solv e: A B C 0 N N v f = 0 1 (39) 4. Once N and N v f ha v e been found, we can re write the control la w to be: V f = K ( N r ef ) + N v f r ef V f = K + N r ef (40) where N = N v f + K N . 3.4.3. Integrator loop T o t rack constant references for v t and T m without steady-state error , system (30) should be aug- mented by intermediate states X = [ _ T e v e w ] T where e v = K v ( v t V r ef ) with K v 0 as a constant g ain and e w = w w r ef are gi v en by e v = g 1 ( ) e p = g 2 ( ) ; (41) where g 1 ( ) and g 2 ( ) are nonlinear functions that can be found using equations (18) and (4). In our case, these functions are linearized around an operating point to get a local approximation in the form e v = C 1 ( ) + O ( 2 ) e T = C 2 ( ) + O ( 2 ) : (42) Ho we v er , to inspire the standard classic IEEE go v ernor model (7) in the closed-loop syst em, we modify the second inte grator as follo ws: e w = 1 T g ( T r ef P sv + K g ( ! ! s )) (43) Ne glecting high order terms in (42), the linear augmented system is _ X = A a X + B a _ v ; (44) where A a = 2 4 A 0 4 2 C 1 0 1 2 C 2 0 1 2 3 5 ; B a = B 0 2 2 ; (45) The pair ( A a , B a ) is pro v ed controllable in our case where the follo wing rank condition is v erified: r ank  B a A a B a A 2 a B a A 3 a B a A 4 a B a A 5 a B a  = 6 (46) Int J Elec & Comp Eng, V ol. 11, No. 2, April 2020 : 1186 1199 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Elec & Comp Eng ISSN: 2088-8708 r 1195 Hence, there e xists a stabilizing state feedback: _ v = K a X ; with K a = [ K K v K T ] (47) where K 2 R 2 4 , K v 2 R 2 1 and K T 2 R 2 1 . Inte grating the abo v e equation, the outer -loop has the follo wing form: v ( t ) = K ( t ) K v Z t 0 e v ( ) d K T Z t 0 e p ( ) d (48) It includes proportional and inte gral terms, which achie v e zero steady-state error in po wer and v olt age responses. One of the interesting approaches to design the g ain K a is the LQR (Linear Quadratic Re gulator) method. The g ain K a is computed by solving a Riccati equation and minimizes the follo wing inde x J = Z 1 0 X T ( ) QX ( ) + _ v T ( ) R _ v ( ) d (49) where Q and R are the weight matrices. 4. SIMULA TION RESUL TS In order to illustrate the performance of the proposed flatness controller in impro ving the transient sta - bility of the SMIB system, some computer simulations with Matlab and Eurostag are pro vided here. F or comparison purposes, classic controllers [3, 4] ha v e also been implemented and tested, which are described by the follo wing dif ferential equations: _ u 1 = 1 T a ( u 1 + K a ( v t V r ef )) (50) _ u 2 = 1 T g ( u 2 + K g ( ! s ! ) + T r ef ) (51) T o enhance the performance of the classic controller , we add a PID controller for the A VR. This PID controller will ensure an e xact tracking scheme for the term inal v oltage and will speed up the response time of it. Also, the classic go v ernor controller , generating u 2 , is equipped with a F ast V alving Scheme (FVS), which will enhance the transient stability [31]. Once a se v er disturbance is detected, the steam v alv e is closed with a f ast dynamic respecting the rate limitation ( _ u 2 = 1 =T c ). This situation is maintained for a time T 2 and after that the v alv e position is released to meet the dynamic (51) and respect the rate constraints. The parameters of the controllers are gi v en in T able 2 . T able 2. P arameters of the adopted controllers P arameter u 1 K p K I K d K a T a K g T g K v V r ef w r ef T c T o V alue 3 10 1 0 1 0.1 25 0.3 5 1 1 0.3 3 W e consider a short circuit in the middle of the second transmission line. Figure 2 illustrates the machine speed, mechanical torque and terminal v oltage obtained from Eurostag softw are. The CCT in this case is 585 ms. It is clear that the proposed approach sho ws appropriate rob ustness ag ainst hard f aults occurred in the netw ork. On the other hand, the classic controllers re v eal a maximum CCT equal to 528 ms. This means that the suggested flatness control algorithm has o v er 10% impro v ement in CCT compared to the IEEE common controllers. Also, Figure 3 to Figure 5 illustrate dif ferent state e v olutions for the proposed controller ag ainst the classic IEEE controllers which ha v e been obtained from Matlab . One can see that the proposed controller w orks well. In f a ct, the CCT obtained by the proposed controller is lar ger than that of the classic IEEE controllers. The applied control inputs Efd and Psv are illustrated in Figures 6 and 7, respecti v ely . One can see that the control inputs of the proposed scheme are feasible in practice. Also, the proposed controller has the ef fect of f ast v alving as it can be observ ed in Figure 7, which is an interresting feature that enhances the transient stability of the po wer system. Finally , Figure 8 depicts the dedicated control ener gies for both Observer -based tr ac king contr ol for ... (Mohammad P ourmahmood Aghababa) Evaluation Warning : The document was created with Spire.PDF for Python.