This paper suggests novel hybrid learning algorithm with stable learning laws for adaptive network based fuzzy inference system
(ANFIS) as a system identifier and studies the stability of this algorithm. The new hybrid learning algorithm is based on
particle swarm optimization (PSO) for training the antecedent part and gradient descent (GD) for training the conclusion part.
Lyapunov stability theory is used to study the stability of the proposed algorithm. This paper, studies the stability of PSO
as an optimizer in training the identifier, for the first time. Stable learning algorithms for the antecedent and consequent
parts of fuzzy rules are proposed. Some constraints are obtained and simulation results are given to validate the results.
It is shown that instability will not occur for the leaning rate and PSO factors in the presence of constraints. The learning
rate can be calculated on-line and will provide an adaptive learning rate for the ANFIS structure. This new learning scheme
employs adaptive learning rate that is determined by input–output data. 相似文献
Stability and convergence analysis have been previously accomplished for some population-based search and swarm intelligence algorithms like Particle Swarm Optimization and Gravitational Search Algorithm. However, there is no adequate theoretical analysis for Bat Algorithm (BA) in the literature. The BA is a type of optimization algorithms which is inspired by the motion of small bats searching for hunting their preys. In this study, stability and convergence of the particle dynamics in the standard version BA are analyzed, and some restrictions are described. Then, new updating relations have been proposed. Also the dynamics of the algorithm have been investigated, and sufficient conditions for stability have been derived using Lyapunov stability analysis. Extensive simulation is used to examine the findings. The results confirm the theoretical predictions and indicate the stability and convergence of the proposed updating relations.
Type-1 fuzzy sets cannot fully handle the uncertainties. To overcome the problem, type-2 fuzzy sets have been proposed. The novelty of this paper is using interval type-2 fuzzy logic controller (IT2FLC) to control a flexible-joint robot with voltage control strategy. In order to take into account the whole robotic system including the dynamics of actuators and the robot manipulator, the voltages of motors are used as inputs of the system. To highlight the capabilities of the control system, a flexible joint robot which is highly nonlinear, heavily coupled and uncertain is used. In addition, to improve the control performance, the parameters of the primary membership functions of IT2FLC are optimized using particle swarm optimization (PSO). A comparative study between the proposed IT2FLC and type-1 fuzzy logic controller (T1FLC) is presented to better assess their respective performance in presence of external disturbance and unmodelled dynamics. Stability analysis is presented and the effectiveness of the proposed control approach is demonstrated by simulations using a two-link flexible-joint robot driven by permanent magnet direct current motors. Simulation results show the superiority of the IT2FLC over the T1FLC in terms of accuracy, robustness and interpretability. 相似文献
This paper proposes the generalized projective synchronization (GPS) of uncertain chaotic systems with external disturbance via Gaussian radial basis adaptive sliding mode control (GRBASMC). A sliding surface is adopted to ensure the stability of the error dynamics in sliding mode control. In the neural sliding mode controller, a Gaussian radial basis function is utilized to online estimate the system dynamic function. The adaptation law of the control system is derived in the sense of Lyapunov function, thus the system can be guaranteed to be asymptotically stable.The proposed method allows us to arbitrarily adjust the desired scaling by controlling the slave system. It is not necessary to calculate the Lyapunov exponents and the eigen values of the Jacobian matrix, which makes it simple and convenient. Also, it is a systematic procedure for GPS of chaotic systems and it can be applied to a variety of chaotic systems no matter whether it contains external excitation or not. Note that it needs only one controller to realize GPS no matter how much dimensions the chaotic system contains and the controller is easy to be implemented.The proposed method is applied to three chaotic systems: Genesio system, Lur’e like system and Duffing system. 相似文献
In this paper, we present new adaptive algorithms for the computation of the square root of the inverse covariance matrix. In contrast to the current similar methods, these new algorithms are obtained from an explicit cost function that is introduced for the first time. The new adaptive algorithms are used in a cascade form with a well-known adaptive principal component analysis to construct linear discriminant features. The adaptive nature and fast convergence rate of the new adaptive linear discriminant analysis algorithms make them appropriate for online pattern recognition applications. All adaptive algorithms discussed in this paper are trained simultaneously using a sequence of random data. Experimental results using the synthetic and real multiclass, multidimensional input data demonstrate the effectiveness of the new adaptive algorithms to extract the optimal features for the purpose of classification. 相似文献
Biped locomotion has attracted much attention in recent years. The most successful implemented methods in this area are based on two approaches, central pattern generator (CPG) and zero moment point (ZMP). Unfortunately, neither of these concepts can solely solve the movement challenge completely. In this study, we introduce a hybrid controller to combine the advantages of these methods. The proposed controller is based on two major approaches, CPG and ZMP. This hybrid controller is composed of a trajectory control system and a trajectory generator system. The trajectory control system applied to keep the robot stable uses ZMP as a real time control feedback. The trajectory generator system, which is composed of nonlinear oscillators, generates stable motions. The parameters of CPG are tuned by a new two-stage approach using differential evolution (DE) and bees algorithm (BA). Furthermore, performance of the proposed controller is verified using the robotic simulation software Webots. 相似文献
In a cement factory, a rotary kiln is the most complex component and it plays a key role in the quality and quantity of the final product. This system involves complex nonlinear dynamic equations that have not been completely worked out yet. In conventional modeling procedures, a large number of the involved parameters are crossed out and an approximation model is presented instead. Therefore, the performance of the obtained model is very important and an inaccurate model may cause many problems in the design of a controller. This study presents a Takagi-Sugeno (TS)-type fuzzy system called a wavelet projection fuzzy inference system (WPFIS) in which a dimension reduction section is used at the input stage of the fuzzy system. In order to clarify the structure of the extracted features, structural learning with forgetting (SLF) based on Minkowski norms is proposed. In addition, gradient descent (GD) was used as a training algorithm. The results show that the proposed method has higher performance in comparison with conventional models. The data collected from Saveh White Cement Company were used in our simulations. 相似文献
In this study, a novel robust fault diagnosis scheme is developed for a class of nonlinear systems when both fault and disturbance are considered. The proposed scheme includes both component and sensor fault with nonlinear system that transferred to nonlinear Takagi-Sugeno (T-S) model. It considers a larger category of nonlinear system when fuzzification is used for only nonlinear distribution matrices. In fact the proposed method covers nonlinear systems could not transform to linear T-S model. This paper studies the problem of robust fault diagnosis based on two fuzzy nonlinear observers, the first one is a fuzzy nonlinear unknown input observer (FNUIO) and the other is a fuzzy nonlinear Luenberger observer (FNLO). This approach decouples the faulty subsystem from the rest of the system through a series of transformations. Then, the objective is to design FNUIO to guarantee the asymptotic stability of the error dynamic using the Lyapunov method; meanwhile, FNLO is designed for faulty subsystem to generate fuzzy residual signal based on a quadratic Lyapunov function and some matrices inequality convexification techniques. FNUIO affects only the fault free subsystem and completely removes any unknown inputs such as disturbances when residual signal is generated by FNLO is affected by component or sensor fault. This novelty and using nonlinear system in T-S model make the proposed method extremely effective from last decade literature. Sufficient conditions are established in order to guarantee the convergence of the state estimation error. Thus, a residual generator is determined on the basis of LMI conditions such that the estimation error is completely sensitive to fault vector and insensitive to the unknown inputs. Finally, an numerical example is given to show the highly effectiveness of the proposed fault diagnosis scheme. 相似文献
A method of using particle swarm optimization (PSO) algorithm to design electromagnetic absorber is presented. To demonstrate effectiveness of the PSO algorithm three different design cases are optimized. To reduce the local minimum traps, a modified local search strategy is employed. Each design problem is optimized using genetic algorithm (GA) and four variants of PSO algorithms, namely global PSO (gbest), local PSO (lbest), comprehensive learning PSO (CLPSO), and modified local PSO (MLPSO). The results clearly show that the MLPSO is a robust, fast, and useful optimization tool for designing absorbers. A seven-layer absorber achieved by this method has reflection coefficient below 18.7 dB from VHF to 20 GHz. 相似文献
In the conventional CMAC-based adaptive controller design, a switching compensator is designed to guarantee system stability
in the Lyapunov stability sense but the undesirable chattering phenomenon occurs. This paper proposes a CMAC-based smooth
adaptive neural control (CSANC) system that is composed of a neural controller and a saturation compensator. The neural controller
uses a CMAC neural network to online mimic an ideal controller and the saturation compensator is designed to dispel the approximation
error between the ideal controller and neural controller without any chattering phenomena. The parameter adaptive algorithms
of the CSANC system are derived in the sense of Lyapunov stability, so the system stability can be guaranteed. Finally, the
proposed CSANC system is applied to a Chua’s chaotic circuit and a DC motor driver. Simulation and experimental results show
the CSANC system can achieve a favorable tracking performance. It should be emphasized that the development of the proposed
CSANC system doesn’t need the knowledge of the system dynamics. 相似文献