Command Filtered Adaptive Fuzzy Backstepping Control Method of Uncertain Non‐linear Systems
Yongchao Wang,Lu Cao,Zhang Sheng-xiu,Xiaoxiang Hu,Fei Yu
DOI: https://doi.org/10.1049/iet-cta.2015.0946
2016-01-01
Abstract:IET Control Theory & ApplicationsVolume 10, Issue 10 p. 1134-1141 Regular PapersFree Access Command filtered adaptive fuzzy backstepping control method of uncertain non-linear systems Yongchao Wang, Corresponding Author Yongchao Wang wyc031566@163.com Department of automatic control, Xi'an Research Institution of Hi-Technology, Xi'an, 710025 People's Republic of ChinaSearch for more papers by this authorLijia Cao, Lijia Cao Department of automatic control, Xi'an Research Institution of Hi-Technology, Xi'an, 710025 People's Republic of China College of Automation and Electronic Information, Sichuan University of Science & Engineering, Zigong, 643000 People's Republic of ChinaSearch for more papers by this authorShengxiu Zhang, Shengxiu Zhang Department of automatic control, Xi'an Research Institution of Hi-Technology, Xi'an, 710025 People's Republic of ChinaSearch for more papers by this authorXiaoxiang Hu, Xiaoxiang Hu Department of automatic control, Xi'an Research Institution of Hi-Technology, Xi'an, 710025 People's Republic of ChinaSearch for more papers by this authorFeixiong Yu, Feixiong Yu Department of automatic control, Xi'an Research Institution of Hi-Technology, Xi'an, 710025 People's Republic of ChinaSearch for more papers by this author Yongchao Wang, Corresponding Author Yongchao Wang wyc031566@163.com Department of automatic control, Xi'an Research Institution of Hi-Technology, Xi'an, 710025 People's Republic of ChinaSearch for more papers by this authorLijia Cao, Lijia Cao Department of automatic control, Xi'an Research Institution of Hi-Technology, Xi'an, 710025 People's Republic of China College of Automation and Electronic Information, Sichuan University of Science & Engineering, Zigong, 643000 People's Republic of ChinaSearch for more papers by this authorShengxiu Zhang, Shengxiu Zhang Department of automatic control, Xi'an Research Institution of Hi-Technology, Xi'an, 710025 People's Republic of ChinaSearch for more papers by this authorXiaoxiang Hu, Xiaoxiang Hu Department of automatic control, Xi'an Research Institution of Hi-Technology, Xi'an, 710025 People's Republic of ChinaSearch for more papers by this authorFeixiong Yu, Feixiong Yu Department of automatic control, Xi'an Research Institution of Hi-Technology, Xi'an, 710025 People's Republic of ChinaSearch for more papers by this author First published: 01 June 2016 https://doi.org/10.1049/iet-cta.2015.0946Citations: 45AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract The command filtered adaptive fuzzy backstepping control method has been proposed in this study. By using the command filter, it is not only the states, the actuator constraints, and the problem of 'explosion of complexity' in conventional backstepping have been solved, but also the calculation of partial derivatives is unnecessary, so that control law and update law become succinct. Fuzzy logic systems, designed in this study, are utilised in real-time identification of the uncertain non-linear functions. Throughout the process of controller design, the transformed style and Butterworth low-pass filter are applied to solve the problem of unknown control directions. Using the Lyapunov theorem, it has been shown that the approach can guarantee all the signals of the resulting closed-loop system are semi-globally uniformly ultimately bounded. Finally, an application of the method has been made to the controller design of the attack angle of a hypersonic vehicle. The simulation results are given to prove the effectiveness of the method presented in this study. 1 Introduction In the past few years, adaptive fuzzy backstepping method has made much progress and became one of the most popular robust and practical control methods for uncertain non-linear systems (see [1–22] and references therein). Yang and Zhou [1] and Mohammad and Hamid [2] presented adaptive fuzzy backstepping methods for single input single output (SISO) systems. Sui et al. [3] and Xu et al. [4] proposed the fault-tolerant controllers by using adaptive fuzzy backstepping methods. To overcome unknown backlash-like hysteresis properties of the actuators and the time delay which are common in the control systems, Boulkroune et al. [5] and Shahnazi et al. [6] presented the stable adaptive fuzzy controllers for the interconnected uncertain non-linear systems. However, the approaches proposed in [5, 6] are limited to be suitable for the controlled plants which must satisfy matching conditions. Li et al. [7], motivated by the aforementioned literatures, presented an output-feedback control approach for time-delay systems. It is not only the unknown backlash-like hysteresis properties of the actuators has been taken into account, but also can be applied for the complicated systems whose states are not measured and that do not satisfy the matching conditions. It is known that the phenomenon of the physical saturation of the actuators, which often degrades the performance of the systems or even severely damages the stability, is inevitable in the controllers design process. The analysis and designations of the systems with input saturations have been well studied in [8, 23] and references therein. Li et al. [8] presented a direct adaptive fuzzy backstepping approach. The hyperbolic tangent and Nussbaum gain functions are applied effectively to handle the non-linearities aroused by the input saturations. Chen et al. [23] proposed the scheme to handle the unknown control directions. However, it requires some details of the systems. For example, the control coefficient matrices and the scaling of whose spectral radius must be known. The filtered signals passing through the Butterworth low-pass filter (LPF) are introduced by Zou et al. [9] to circumvent algebraic loop problem. The assumptions that priori knowledge of the control directions can be foreknown are no longer required. The controller design for the systems with unmeasured states has been studied in [10–12, 24–27]. In [10, 11], a fuzzy disturbance observer based robust controller had been developed to achieve good tracking performance. The prescribed performance technique [13] and the barrier Lyapunov function [14] were combined with backstepping recursive to keep tracking errors be remained as a desirable neighbourhood of the origin [15]. According to the above-mentioned works, it can be concluded that: (i) the functions can be uncertain or even unknown, which implies the robustness of the method; (ii) they can be applied to the systems with time delay, unmeasured states, input saturation, unknown backlash-like hysteresis properties of the actuators, or even the output tracking error constraint, but not with states constraints. The existing results cannot limit the states of the systems to a specified scope which is common in the industrial control process. A backstepping control method based on a command filter of non-linear systems was developed by Farrell et al. [16, 17]. The paper [16] first introduced a command filter with magnitude, rate, and bandwidth constraints. A command filtered robust controller was designed for the aircraft. The work [17] has investigated the online approximation based on backstepping control method with the command filter introduced in [16]. To reinforce the robustness of the controller, Farrell et al. [18] presented a command filtered backstepping approach including the online adaptive identification of the normally uncertain parameters. The resulting closed-loop system maintains the stability that has been analysed with the intermediate states limited by the magnitude, rate, and bandwidth constraints. The literature [19] extended the method proposed in [18] to a vehicle controller to accommodate battle damage types of events. On the basis of a nested saturation function combined with immersion and invariance methodology, Hu and Zhang [20] designed a command filtered control law for vertical take-off and landing (VTOL) vehicles to deal with the uncertainty and thrust saturation. In [20], the maximum contribution is the influence of the frequency of the command filter on the closed-loop system was rigorously analysed. In [21], the effect of the command filter on closed-loop stability and performance were rigorously analysed. By using Tikhonov's theorem, the proof of stability was presented. The compensated error retained the standard stability properties of the backstepping recursive. To simplify the adaptive backstepping implementation considerably, Dong et al. [22] presented a command filtered implementation approach. It can be allowed to be suitable for a broader category of systems. From the above discussions, we can learn that the command filter methodology have achieved much progress. However, the articles do not take the singular into the consideration. Our work aims to investigate the command filtered robust control of the uncertain systems with states and actuator physical constraints. The system style has transformed because the control directions are unknown, which may arise from the singular phenomenon that is ignored in [16, 17]. The fuzzy logic systems (FLSs) were utilised to identify the uncertain functions. The Butterworth LPF is used for circumventing the algebraic loop problem, then that all of the control directions of the transformed system are accurately acknowledged. Combining the command filter with the backstepping design technique, a robust control method has been proposed. It is provided that the proposed control method can guarantee all the signals are semi-globally uniformly ultimately bounded (SGUUB), states and actuator can be limited in the specify scopes. Moreover, the tracking errors converge to a small neighbourhood of the origin. Compared with the analogous literature, the main contributions are summarised as follows: (i) based adaptive fuzzy backstepping (CFAFB) control method can solve the states and actuator physical constraints. The hard nut is that the states constraints cannot be handled by the existing adaptive fuzzy backstepping control method. The second-order non-linear command filter contained rate and magnitude restrained functions. The command filter ensures that the commanded trajectory and its derivative satisfy these same constraints. (ii) By using the command filter, it is not only the states and the actuator constraints and the obvious obstacle of 'explosion of complexity' in the controller design process of the conventional backstepping have been averted, but also the complicated calculation of partial derivatives is unnecessary, so that the control and the update law are more concise. As we know, dynamic surface control (DSC) [24–27] is also used to solve the problem of 'explosion of complexity'. In essence, DSC method and the command filter make full use of the filters to obtain the derivatives of the command virtual control signal, so much so that the 'explosion of complexity' is avoided. In comparison, the command filter can ensure that the commanded trajectory and its derivative meet the magnitude and rate constraints to meet the states and actuator physical constraints. However, the first-order linear filter of the DSC method does not have this feature. (iii) The control method takes the unknown control direction, which may arise from the singular of the controller, into the consideration. From a practical perspective, most of the control direction of the controlled plant can be unknown. In this paper, the unknown control direction gain function can be transformed into 1 after introducing the Butterworth LPF. Hence, in the next controller design process, we can ignore the unknown control direction. The considered systems of the most literatures are of strict-feedback form whose control direction gain functions are 1. 2 Problem formulation and preliminaries 2.1 System descriptions and basic assumptions Consider the following controlled plant (1)where and are the state vector and the output of the system, respectively, and are unknown smooth functions, , and are all uncertain function due to parameter uncertainties. is the unknown external disturbance. Remark.Note that the uncertain non-linear function can be expressed as (2)Some reasonable assumptions listed as follows can be applied throughout the paper. Assumption 1.There exist constants , the inequality holds. Assumption 2 (Sui et al. [28]).For , , the inequality holds (3)where expresses the 2-norm of vector . Remark.The constants and mentioned in the above assumptions are all used for the stability analysis. They may be unknown. They are only for the analytical purposes, so we do not need them exactly or even the scope. Be equipped with the FLSs, Butterworth LPF, and command filter, our objective is to develop a robust control scheme for the controlled plant with unknown control direction so that all the signals are SGUUB, states and actuator physical constraints are satisfied. FLSs will be employed to identify unknown functions online in the controlled plant (1). According to [29], a FLS can be expressed (4)where , . The optimal parameter vector defined by the following inequality is only for analytical purposes (5) Lemma 1 (Wang [29]).For , which is continuous function and defined over a compact and , there exist a FLS and an ideal parameter such that (6) Lemma 2 (Ji and Xi [30]) (Young's inequality).For , the following inequality holds (7)where , , and . 3 Controller design and stability analysis 3.1 Controller design Underlying command filter, a robust controller is designed in this section for plant (1). The controller design process is composed of steps. Step 1: Consider the first subsystem (8)Equation (8) can be transformed as following: . The function contains variable while the virtual control is . This will arise from the algebraic loop problem. The Butterworth LPF is adopted, and (8) can be rewritten as (9)where is the filtered signal defined in [9, 10] which is used for circumventing the algebraic loop problem, is a Butterworth LPF, . Assumption 3., where is a bounded positive constant. Remark.According to [11], the replacements and are reasonable because most actuators have low-pass properties. We can define that , where is unknown function bounded by positive constant such that . Accordingly, the assumption is also reasonable. It is worth pointing that is not employed in the control design and can be unknown, and it is only used for the stability analysis. The uncertainty function can be approximated by the designed FLS (10)where is the optimal parameter vector, and denotes the approximate error. The virtual control law is designed as follows (11)where denotes the output tracking error, denotes the estimated value of the optimal parameter vector , is the estimated value of the approximate error of the FLS and whose adaptive law will be given later, , are the designed parameters of the system. and will be obtained from the following introduced command filter (14) by passing through and the expected state is constrained. is the compensated tracking error signal and defined as (12)where can be obtained from the following expressed filter (13)where variable will be obtained after passed through the command filter, will be given in the next step. Remark.Equation (13) is used to achieve the filtering value . Note that is computed without differentiation. can be obtained through integrating processes although the filter is expressed by differential equation. Assuming that the state constrained is , so the magnitude, rate, and bandwidth limited states and can all be obtained from the following designed filter [24] (14)In this second-order non-linear command filter, and are the intermediate variables of the filter, ; is the damping ratio of the system; denotes natural frequency. and are the rate and magnitude restrained function separately and can be described as follows (15) Remark.The outputs of the command filter and are all bounded and continuous when the input is bounded. The calculations of partial derivatives in the process of the traditional designation of the control law are unnecessary for the process of the calculation of the variable which is the output of the filter mentioned above. will track the variable exactly by choosing the appropriate parameters. As there are rate, magnitude, or bandwidth limitations on the signal that is intended to track the output of the command filter, so the command filter will impose some constraints on the command while also outputting the derivative of the command. When the states must remain in some operating envelope defined by magnitude limit and rate limit , so the command filter ensures that the commanded trajectory and its derivative satisfy these same constraints. Choosing the adaptation law for and as (16) (17)where , , , and are all the positive constants. Step i: We will design the control law . The th subsystem is considered as follows (18)Considering as the virtual input, Butterworth LPF is introduced, and (18) can be expressed as (19)where , and is the filtered signal. The non-linear unknown function in (19) can be real-time approximated by the designed FLS (20)The defined intermediate tracking errors and are similarly expressed as follows (21)where is the output of the filter designed as follows (22)where . Similarly, choosing the virtual control law as follows (23)where is the estimated value of the approximated error of the FLS, is the filtered signal which is obtained by passing the virtual control law through the introduced command filter (14). is a positive constant. Choosing the adaptation law for and as (24) (25)where , , , and are the positive constants. Remark.It can be learned that designed in the previous step is passed through the command filter to produce and it is the first-order derivative . They are needed for calculating the virtual control law . Step n: In this crucial step, the control law will be given. Consider the th subsystem (26)Equation (26) can be transformed as follows after the Butterworth LPF introduced (27)where , and is the filtered signal. The non-linear function in (27) can be real-time approximated by the designed FLS (28)The expected control law input can be designed as follows (29)where is defined as the tracking error, is the estimated value of the optimal parameter of the FLS, is the estimated value of the approximate error of the FLS, is the filtered signal which is obtained by passing the virtual control law through the introduced command filter (14). is the compensated tracking error signal and defined as (30)where can be obtained from the following filter (31)Considering the actuator is constrained, the real control law which can maintain the physical constraints can be produced by passing the signal through the command filter. Select the adaptive update law for and (32) (33)where , , , and are the positive constants. Now, we have completed the task. It is obvious that there are many signals in the control process. The block diagram of the controller structure depicts the control process and signal flow, as shown in Fig. 1, and can offer convenience to understand each signal and the relationship among them. Fig. 1Open in figure viewerPowerPoint Block diagram of the controller structure (number inside the brackets denotes the equation number) 3.2 Stability analysis We will discuss the stability of closed-loop system. The dynamics of and , which are the tracking errors and the compensated tracking errors, will be given first (34) (35) (36)Similar to (34–36), the dynamics of compensated tracking errors can also be obtained from the following equations (37) (38) (39)Now, considering the control Lyapunov function candidate (40)The time derivative of Lyapunov function is given as follows (41)Substituting (37) into (41) yields (42) Lemma 3 (Polycarpou and Ioannou [31]).For , is satisfied. Applying Lemma 3, one has (43)By applying Young's inequality and Lemma 2, we can obtain the inequalities (44)The following inequalities can be obtained by applying adaptive update laws (45) (46)Substituting (44–46) into (43) yields (47)Similarly to the previous analysis and error dynamics, (43) can be rewritten as (48)Let , , then (48) can be rewritten as (49)According to Lyapunov theory and (49), all the signals are SGUUB. The errors can converge to a small neighbourhood of the origin. 4 Simulation study In this section, a simulation example was provided to illustrate the effectiveness and robustness of the developed CFAFB control scheme. Consider the longitudinal flight control of a hypersonic aircraft [32, 33]. The dynamic equations are (50)where , , , , and denote attack angle, pitch rate, flight-pitch angle, moment of inertia, and the speed of the vehicle. denotes radial distance from Earth's centre (51)where , , and are aerodynamic coefficients which have been simplified around cruising flight, whose errors are existed and inevitable. Let , . The parameters uncertainties and the disturbances of the hypersonic vehicle are always existed along the flight process. Hence, the dynamics can be written as (52)where , is the gravity acceleration, , , is the disturbance of the first subsystem, , , , is the external disturbance torque acting on the pitch axis. In the simulation model, is the sine wave jamming to the pitch axis and whose numerical value is . and are all uncertain function due to parameter uncertainties. Particularly, the aerodynamic coefficient errors are existed and inevitable. Hence, the functions are unknown and we should construct FLSs to approximate them in real time. The parameter uncertainties and disturbances are all applied for the course of simulation. The aerodynamic coefficients and the additive uncertainties of the parameters used for the simulation studies can be taken from [32]. The design parameters were chosen as follows: , , , , , , , , , and . To accomplish the target of the controller, the command filter must be imposed by some constraints on the command while also outputting the derivative of the command for that there are rate, magnitude, and bandwidth limitations on the signals. We also take the dynamics of the actuator into consideration in this work. The command filter parameters can be learned from Table 1. Table 1. Command filter parameters Command variable , rad/s Magnitude limit Rate limit Damping factor 4 – 1.0 20 – 1.0 120 0.7 After the style of the controlled plant is transformed and the Buttterworth LPF is introduced (see (9), (19), and (27)), the method proposed in [8] can be applied to control the plant to track the reference signal. As shown in figures, the simulation results by using the proposed approach in this paper are compared with those by using the method in [8], where the states constraints and the physical constraints (elevator executive rate) were not taken into the consideration. In [8], the hyperbolic tangent function and Nussbaum function are applied to handle the input magnitude saturation. The simulation results in [8] are included to illustrate the effectiveness of the proposed method to handle the input magnitude saturation. Figs. 2, 3–4 show that the controlled plant can steadily track the reference trajectory in a satisfactory way under the action of the two controllers. Figs. 2 and 3 show the reference signal as solid lines, the command signal which is produced by passing the reference signal through the command filter as dashed lines, the attack angle as the dotted lines, the attack angle as the dash-dotted lines. As expected, the proposed command filtered fuzzy adaptive backstepping controller and the controller proposed in [8] provide stable tracking of the reference trajectories and make the tracking errors finally kept in the neighbourhood of the origin. Fig. 5 shows the state curves of the controlled plant under the two controllers. The pitch rates and are represented by solid line and dash-dotted line, respectively. Fig. 5 shows that throughout keeps in the saturated scope that can be seen in Table 1 while does not. The maximum value of the absolute value of the variable is . Fig. 6 shows the elevator deflection curves of the two controllers. In Fig. 6, the solid line represents the input of the elevator of the controller proposed in [8]. The elevator deflections and are represented by dashed and dotted line, respectively. Clearly, the elevator deflection curves under the action of the two controllers all satisfy the magnitude constraints as expected. Fig. 7 shows the curves of the elevator executive rates. Clearly, represented by solid line is far beyond the elevator processing ability. By comparison, represented by dash-dotted line satisfies the rate constraints that can be seen in Table 1 from beginning to end due to the rate restrained function. We can draw a conclusion that simulation results revealed the controller designed herein has the perfect control performance. The proposed method can be a good solution to the states and actuator physical property constraints. Fig. 2Open in figure viewerPowerPoint Curves of (solid line), (dashed line), (dotted line), and (dash-dotted line) Fig. 3Open in figure viewerPowerPoint Curves of (solid line), (dashed line), (dotted line), and (dash-dotted line) in specific areas Fig. 4Open in figure viewerPowerPoint Curves of (dotted line) and (solid line) Fig. 5Open in figure viewerPowerPoint Curves of (solid line) and (dash-dotted line) Fig. 6Open in figure viewerPowerPoint Curves of (solid line), (dashed line), and (dotted line) Fig. 7Open in figure viewerPowerPoint Curves of (solid line) and (dash-dotted line), where and are elevator executive rates of two methods, respectively Hereinafter, denotes the variable of the controlled plant under the action of the controller proposed in [8], denotes the variable of the controlled plant under the action of the controller proposed in this paper. 5 Conclusion A CFAFB control scheme has been presented in this paper. The command filter with magnitude, rate, and bandwidth constraints is introduced to reach the target that the states and actuator satisfy the constraints and the physical property, respectively. The FLSs were used to approximate the unknown functions. The Butterworth LPF was introduced to effectively handle the unknown control direction which may arise from the phenomenon of singular. The proposed method does not require the assumptions with respect to the prior knowledge of the control direction functions and their derivatives. The stability properties were proved using Lyapunov methods. Simulation results show that satisfactory tracking performance can be obtained by applying the proposed control scheme. Future work will extend the proposed CFAFB approach to stochastic non-linear systems. The filtering error compensation should be designed to guarantee the approximation precision between the filtered signal and the command virtual control. 6 Acknowledgments The authors are grateful to editors and the anonymous reviewers for their critical and constructive comments, which help to improve the quality of this paper. This work was supported in part by the National Natural Science Foundation of China (nos. 61304001 and 61304239, 61203007). The authors thank Zhang J.J., Sun Q., and Qi N.X. for their suggestions for this work. 7 References 1Yang Y., and Zhou C.: 'Adaptive fuzzy stabilization for strict-feedback canonical nonlinear systems via backstepping and small-gain approach', IEEE Trans. Fuzzy Syst., 2005, 13, (1), pp. 104– 114 (doi: https://doi.org/10.1109/TFUZZ.2004.839663) 2Mohammad M., and Hamid R.: 'Adaptive fuzzy backstepping controller design for uncertain under-actuated robotics systems', Nonlinear Dyn., 2015, 79, (2), pp. 1457– 1468 (doi: https://doi.org/10.1007/s11071-014-1753-y) 3Sui S. Tong S.C., and Li Y.M.: 'Fuzzy adaptive fault-tolerant tracking control of MIMO stochastic pure-feedback nonlinear systems with actuator failures', J. Franklin Inst., 2014, 351, (6), pp. 3424– 3444 (doi: https://doi.org/10.1016/j.jfranklin.2014.03.016) 4Xu Y.Y. Tong S.C., and Li Y.M.: 'Prescribed performance fuzzy adaptive fault-tolerant control of non-linear systems with actuator faults', IET Control Theory Appl., 2014, 8, (6), pp. 420– 431 (doi: https://doi.org/10.1049/iet-cta.2013.0676) 5Boulkroune A. Saadand M.M., and Chekireb H.: 'Design of a fuzzy adaptive controller for MIMO nonlinear time-delay systems with unknown actuator nonlinearities and unknown control direction', Inf. Sci., 2010, 180, (24), pp. 5041– 5059 (doi: https://doi.org/10.1016/j.ins.2010.08.034) 6Shahnazi R. Pariz N., and Kamyad A.V.: 'Adaptive fuzzy output feedback control for a class of uncertain nonlinear systems with unknown backlash-like hysteresis', Commun. Nonlinear Sci. Numer. Simul., 2010, 15, (8), pp. 2206– 2221 (doi: https://doi.org/10.1016/j.cnsns.2009.09.004) 7Li Y.M. Tong S.C., and Li T.S.: 'Adaptive fuzzy output feedback control of MIMO nonlinear uncertain systems with time-varying delays and unknown backlash-like hysteresis', Neurocomputing, 2012, 93, (1), pp. 56– 66 (doi: https://doi.org/10.1016/j.neucom.2012.04.004) 8Li Y.M. Tong S.C., and Li T.S.: 'Direct adaptive fuzzy backstepping control of uncertain nonlinear systems in the presence of input saturation', Neural Comput. Appl., 2013, 23, (5), pp. 1207– 1216 (doi: https://doi.org/10.1007/s00521-012-0993-3) 9Zou A.M. Hou Z.G., and Tan M.: 'Adaptive control of a class of nonlinear pure-feedback systems using fuzzy back- stepping approach', IEEE Trans. Fuzzy Syst., 2008, 16, (4), pp. 886– 897 (doi: https://doi.org/10.1109/TFUZZ.2008.917301) 10Kim E., and Sungryul L.: 'Output feedback tracking control of MIMO systems using a fuzzy disturbance observer and its applications to the speed control of a PM synchromous motor', IEEE Trans. Fuzzy Syst., 2005, 13, (6), pp. 725– 741 (doi: https://doi.org/10.1109/TFUZZ.2005.859306) 11Tong S.C., and Li Y.M.: 'Observer-based adaptive fuzzy backstepping control of uncertain nonlinear pure-feedback systems', Sci. China (Inf. Sci.), 2014, 57, (2), pp. 1– 14 (doi: https://doi.org/10.1007/s11432-013-5043-y) 12Tong S.C. Li Y.M., and Shi P.: 'Observer-based adaptive fuzzy backstepping control of uncertain MIMO pure-feedback nonlinear systems', IEEE Trans. Fuzzy Syst., 2012, 20, (4), pp. 771– 785 (doi: https://doi.org/10.1109/TFUZZ.2012.2183604) 13Benchlious C.P., and Rovithakis G.A.: 'Adaptive control with guaranteed transient and steady state tracking error bounds for strict feedback systems', Automatica, 2009, 45, (2), pp. 532– 538 (doi: https://doi.org/10.1016/j.automatica.2008.08.012) 14Ren B.B. Ge S.S., and Tee K.P., et al.: 'Adaptive neural control for output feedback nonlinear systems using a barrier Lyapunov function', IEEE Trans. Neural Netw., 2010, 21, (8), pp. 1339– 1345 (doi: https://doi.org/10.1109/TNN.2010.2047115) 15Li Y.M., and Tong S.C.: 'Prescribed performance adaptive fuzzy output-feedback dynamic surface control for nonlinear large-scale systems with time delays', Inf. Sci., 2015, 292, (1), pp. 125– 142 (doi: https://doi.org/10.1016/j.ins.2014.08.060) 16Farrell J.A. Polycarpou M., and Sharma M.: 'Adaptive backstepping with magnitude, rate, and bandwidth constraints: aircraft longitude control'. Proc. of American Control Conf., 2003, pp. 3898– 3904 17Farrell J.A. Polycarpou M., and Sharma M.: 'On-line approximation based control of uncertain nonlinear systems with magnitude, rate and bandwidth constraints on the states and actuators'. Proc. of American Control Conf., 2004, pp. 2557– 2562 18Farrell J.A. Sharma M., and Polycarpou M.: 'Backstepping-based flight control adaptive function approximateon', J. Guid. Control Dyn., 2005, 28, (6), pp. 1089– 1102 (doi: https://doi.org/10.2514/1.13030) 19Farrell J.A. Polycarpou M., and Sharma M.: ' Adaptive backstepping with magnitude, rate, and bandwidth constraints: aircraft longitude control' ( Air Force Research Laboratory, Cincinnati, OH, 2006) 20Hu J.C., and Zhang H.H.: 'Immersion and invariance based command-filtered adaptive backstepping control of VTOL vehicles', Automatica, 2013, 49, (7), pp. 2160– 2167 (doi: https://doi.org/10.1016/j.automatica.2013.03.019) 21Farrell J.A. Polycarpou M., and Sharma M., et al.: 'Command filtered backstepping', IEEE Trans. Autom. Control, 2009, 54, (6), pp. 1391– 1395 (doi: https://doi.org/10.1109/TAC.2009.2015562) 22Dong W.J. Farrell J.A., and Polycarpou M., et al.: 'Command filtered adaptive backstepping', IEEE Trans. Control Syst. Tech., 2012, 20, (3), pp. 566– 580 (doi: https://doi.org/10.1109/TCST.2011.2121907) 23Chen M. Ge S.S., and Ren B.B.: 'Adaptive tracking control of uncertain MIMO nonlinear systems with input constraints', Automatica, 2011, 47, (3), pp. 452– 465 (doi: https://doi.org/10.1016/j.automatica.2011.01.025) 24Tong S.C. Li Y.M., and Feng G., et al.: 'Observer-based adaptive fuzzy backstepping dynamic surface control for a class of MIMO nonlinear systems', IEEE Trans. Syst. Man Cybern. B, Cybern., 2011, 41, (4), pp. 1124– 1135 (doi: https://doi.org/10.1109/TSMCB.2011.2108283) 25Tong S.C. Li Y.M., and Zhang H.: 'Adaptive neural network decentralized backstepping output-feedback control for nonlinear large-scale systems with time delays', IEEE Trans. Neural Netw., 2011, 22, (7), pp. 1073– 1086 (doi: https://doi.org/10.1109/TNN.2011.2146274) 26Li Y.M. Tong S.C., and Li T.S.: 'Composite adaptive fuzzy output feedback control design for uncertain nonlinear strict-feedback systems with input saturation', IEEE Trans. Cybern., 2015, 45, (10), pp. 1073– 1086 27Li Y.M. Tong S.C., and Li T.S.: 'Adaptive fuzzy output feedback dynamic surface control of interconnected nonlinear pure-feedback systems', IEEE Trans. Cybern., 2015, 45, (1), pp. 138– 149 (doi: https://doi.org/10.1109/TCYB.2014.2333738) 28Sui S. Tong S.C., and Li Y.M.: 'Adaptive fuzzy backstepping output feedback tracking control of MIMO stochastic pure-feedback nonlinear systems with input saturation', Fuzzy Sets Syst., 2014, 254, (1), pp. 26– 46 (doi: https://doi.org/10.1016/j.fss.2014.03.013) 29Wang L.X.: ' Adaptive fuzzy systems and control' ( Prentice-Hall, Englewood Cliffs, NJ, 1994) 30Ji H.B., and Xi H.S.: 'Adaptive output-feedback tracking of stochastic nonlinear systems', IEEE Trans. Autom. Control, 2006, 51, (2), pp. 355– 360 (doi: https://doi.org/10.1109/TAC.2005.863501) 31Polycarpou M.M., and Ioannou P.A.: 'A robust adaptive nonlinear control design', Automatica, 1996, 32, (3), pp. 423– 427 (doi: https://doi.org/10.1016/0005-1098(95)00147-6) 32Xu H.J. Mirmirani M.D., and Ioannou P.A.: 'Adaptive sliding mode control design for a hypersonic flight vehicle', J. Guid. Control Dyn., 2004, 27, (7), pp. 829– 838 (doi: https://doi.org/10.2514/1.12596) 33Wang Q., and Stengel R.F.: 'Robust nonlinear control of a hypersonic aircraft', J. Guid. Control Dyn., 2000, 23, (4), pp. 577– 585 (doi: https://doi.org/10.2514/2.4580) Citing Literature Volume10, Issue10June 2016Pages 1134-1141 FiguresReferencesRelatedInformation