Showing papers by "Mojtaba Ahmadieh Khanesar published in 2016"

Direct Model Reference Adaptive Fuzzy Control of Networked SISO Nonlinear Systems

Abstract: This study presents a novel direct model reference fuzzy controller as applied to the control of a nonlinear system over a network subject to variable network induced time delay. The proposed method uses Pade approximation to cope with this condition. Unlike most approaches seen in the literature, which are mostly model based and necessitate the solution of a set of linear matrix inequalities, the proposed approach is online and can be applied to nonlinear systems with a little knowledge about the structure of the system and the values of its parameters. The stability of the proposed method is proved using an appropriate Lyapunov function. The approach is implemented and tested on a dc motor with nonlinear characteristics and nonlinear state-dependent disturbance. It is shown that it is capable of controlling the system over a network subject to variable network-induced time delay with bounded tracking error. In addition, the effect of packet losses is considered in the implementation part and it is seen that the system can be controlled under these conditions too.

Optimal design of adaptive type-2 neuro-fuzzy systems

Abstract: Graphical abstractDisplay Omitted HighlightsLearning algorithms of T2FLS are reviewed.Hybrid learning of parameters are reviewed particularly.The learning algorithms for T2FLS are divided into three categories.Comparison of the three categories is discussed at the end. Type-2 fuzzy logic systems have extensively been applied to various engineering problems, e.g. identification, prediction, control, pattern recognition, etc. in the past two decades, and the results were promising especially in the presence of significant uncertainties in the system. In the design of type-2 fuzzy logic systems, the early applications were realized in a way that both the antecedent and consequent parameters were chosen by the designer with perhaps some inputs from some experts. Since 2000s, a huge number of papers have been published which are based on the adaptation of the parameters of type-2 fuzzy logic systems using the training data either online or offline. Consequently, the major challenge was to design these systems in an optimal way in terms of their optimal structure and their corresponding optimal parameter update rules. In this review, the state of the art of the three major classes of optimization methods are investigated: derivative-based (computational approaches), derivative-free (heuristic methods) and hybrid methods which are the fusion of both the derivative-free and derivative-based methods.

COOA: Competitive optimization algorithm

Abstract: This paper presents a novel optimization algorithm based on competitive behavior of various creatures such as birds, cats, bees and ants to survive in nature. In the proposed method, a competition is designed among all aforementioned creatures according to their performances. Every optimization algorithm can be appropriate for some objective functions and may not be appropriate for another. Due to the interaction between different optimization algorithms proposed in this paper, the algorithms acting based on the behavior of these creatures can compete each other for the best. The rules of competition between the optimization methods are based on imperialist competitive algorithm. Imperialist competitive algorithm decides which of the algorithms can survive and which of them must be extinct. In order to have a comparison to well-known heuristic global optimization methods, some simulations are carried out on some benchmark test functions with different and high dimensions. The obtained results shows that the proposed competition based optimization algorithm is an efficient method in finding the solution of optimization problems.

Type-2 Fuzzy Neural Networks

Abstract: The two most common artificial intelligence techniques, FLSs and ANNs, can be used in the same structure simultaneously, namely as “fuzzy neural networks.” The advantages of ANNs such as learning capability from input-output data, generalization capability, and robustness and the advantages of fuzzy logic theory such as using expert knowledge are harmonized in FNNs. In this chapter, type-1 and type-2 TSK fuzzy logic models are introduced. Instead of using fuzzy sets in the consequent part (as in Mamdani models), the TSK model uses a function of the input variables. The order of the function determines the order of the model, e.g., zeroth-order TSK model, first-order TSK model, etc.

Fundamentals of Type-1 Fuzzy Logic Theory

Abstract: While Boolean logic results are restricted to 0 and 1, fuzzy logic results are between 0 and 1. In other words, fuzzy logic, as a super set of conventional Boolean logic, defines some intermediate values between sharp evaluations like absolute true and absolute false. That means fuzzy sets can handle concepts we commonly face in daily life, like very old , old , young and very young . In this chapter, we introduce the concepts of type-1 fuzzy sets and T1FLCs.

Sliding mode fuzzy rule base bilateral teleoperation control of 2-DOF SCARA system

Abstract: In the present study, the control of a bilateral teleoperation system using a fuzzy logic system which operates based on the sliding mode control theory has been considered. Because of intrinsic time delay and uncertainties of this system we choose sliding mode control theory as a robust controller to avoid mentioned side effects. Furthermore, the utilization of some fuzzy rules on the sliding manifold helps to overcome chattering problems which may appear in sliding mode control signals. The rule base controller is derived which results in an admissible outcome in the tracking of master by slave, precisely. The proposed approach is simulated on one of the most commonly used types of robots in industry namely SCARA. Moreover, in the free and contact motion, the stability and transparency of bilateral teleoperation system which is of a great significance is guaranteed in the presence of time delay, parameter uncertainties and system disturbances with a high synchronization performance.

Maclaurin series expansion complexity-reduced center of sets type-reduction + defuzzification for interval type-2 fuzzy systems

Abstract: This paper provides a mathematical analysis that shows how the crisp output of an IT2 FLS that is obtained by using the Begian-Melek-Mendel (BMM) formula compares to the one obtained by using center-of-sets type-reduction followed by defuzzification (COS TR + D). This is made possible by reformulating the structural solutions of the two optimization problems that are associated with COS TR, and then expanding each of them using a Maclaurin series expansion. As a result of doing this, we show that BMM is the zero-order approximation to COS TR + D. Additionally, by retaining the zero-order and first-order terms from the Maclaurin series expansions, we provide a new Enhanced BMM, one that is non-iterative, has a closed form and is much faster than using the EKM algorithms for COS TR. Although the Enhanced BMM formula is slower than BMM, we demonstrate, by means of extensive simulations, that it is from 5% to 50% more accurate than is BMM for achieving the same numerical solution that is obtained from COS TR + D; and, it is at least 94% faster than when EKM is used for COS TR +D, which makes the Extended BMM a very strong candidate for use in real time applications of IT2 FLSs.

Recurrent interval type-2 neuro-fuzzy control of an electro hydraulic servo system

Abstract: This paper presents a recurrent interval type-2 neuro-fuzzy controller which benefits from a sliding mode theory-based training algorithm. The recurrent interval type-2 neuro-fuzzy benefits from recurrent type-2 membership functions with interval variances which are trained by a novel training method. Furthermore, the adaptation laws considered for the parameters of the controller benefit from an adaptive learning rate. The stability of the proposed training method is considered using an appropriate Lyapunov function. The proposed method is simulated on an electro hydraulic servo system. The results of simulations show that the proposed method can control the system with a satisfactory performance.

Sliding Mode Control Theory-Based Parameter Adaptation Rules for Fuzzy Neural Networks

Abstract: In this chapter, in order to deal with nonlinearities, lack of modeling several uncertainties, and noise in both identification and control problems, SMC theory-based learning algorithms are designed to tune both the premise and consequent parts of T2FNNs. Furthermore, the stability of the learning algorithms for control and identification purposes are proved by using appropriate Lyapunov functions. In addition to its well-known feature which is robustness, the most significant advantage of the proposed learning algorithm for the identification purposes is that the algorithm has a closed form, and thus it is easier to implement in real-time compared to other existing methods.

Model of car wing active control in order to increase stability of the car on corners of roads

Abstract: Active wing has the potential to increase the stability of racing cars by down-force as a result of aerodynamics pressure energy. In this paper a novel way of wing modelling and control is introduced. The proposed model is driven by using the second and the third Newton law and is also theoretically calculated by some simple formulation. Moreover, a proof to show how the performance increases is given. There are ways to work on other aspects of the subject because it will be the first step to intelligently control wings in order to increase road-holding without any special features on the road corners. Numerical and analytical arguments are considered to show step by step modeling spoiler. First the history of wing and spoiler is described then some research and improvement in the subject are given. Next the model of the wing plus its legs are discussed in details. Eventually suspension modelling according to wing model are given. The numerical results show the efficacy of the proposed method.

Artificial bee colony optimization of interval type-2 fuzzy extreme learning system for chaotic data

Abstract: The major challenge in the design of Interval type-2 fuzzy logic system (IT2FLS) is to determine the optimal parameters for their antecedent and consequent parts. This paper propose a novel hybrid learning algorithm for the design of IT2FLS. The proposed hybrid learning algorithm utilizes the combination of extreme learning machine (ELM) and artificial bee colony optimization (ABC) to tune the parameters of the consequent and antecedent parts of the IT2FLS, respectively. The effective forecasting performance of the proposed hybrid learning algorithm is analyzed by modeling a chaotic data set. It is found that the forecasted errors gradually decrease with decrease in the level of noise in data and vise versa.

Fundamentals of Type-2 Fuzzy Logic Theory

Abstract: There are two different approaches to FLS design: T1FLSs and T2FLSs. The latter is proposed as an extension of the former with the intention of being able to model the uncertainties that invariably exist in the rule base of the system. In type-1 fuzzy sets, MFs are totally certain, whereas in type-2 fuzzy sets, MFs are themselves fuzzy. The latter results in a case where the antecedents and consequents of the rules are uncertain. In this chapter, we will introduce the basics of type-2 fuzzy sets, type-2 fuzzy MFs and T2FLCs.

Proposing Interval Activation Functions in Radial Basis Function Neural Network to Predict Nonlinear dynamic systems

Abstract: A Radial Basis Function Neural Network (RBFNN) is a general approximator. In this paper a granular activation function is proposed to improve its learning under the noisy conditions. The granular activation function is also named the interval activation function and it is typicaly the Gaussian function which benefits from having a fixed mean and an uncertain standard deviation. The hidden layer of the proposed network has a total of three parameters to train that it consists the means, the lower bounds of the standard deviations and the higher bounds of the standard deviations of the Gaussian functions. The output layer parameters for training are the means of the interval weights and the intervals of the weights. “K-Means clustering algorithm” method is used to train 1 Radial Basis Function Neural Network (RBFNN) 2 Granular activation functions 3 Bottom-up granulation 4 Nonlinear dynamic system with multiple time delays 5 Mackey glass chaotic time series 6 Granular Radial Basis Function Neural Network (GRBFNN) D ow nl oa de d fr om jo c. kn tu .a c. ir at 1 :0 6 + 03 30 o n F rid ay D ec em be r 21 st 2 01 8 2 اوت داهنشيپ لاعف عب زاس هزاب شيپ روظنم هب یا هداد ینيب عباوت هياپ رب یبصع ۀكبش رد زيون هب هتشغآ یاه یعاعش للها ،هنگنز یروهظ راي هنشت دمحم هناخ هيدمحا یبتجم ،بل رس Journal of Control, Vol. 9, No. 4, Winter 2016 دلج ،لرتنک هلجم 9 هرامش ، 4 ، ناتسمز 1394 these parameters. The purpose of the above learning method is regarded as one of the granular method presenting the bottom-up granulation which causes the input vectors clustered in the larger granules in the hidden layer. Gradient descend method is also used to train these parameters to compare with this novel method. The structure is tested with or without noisy data to identify a nonlinear dynamic system with multiple time delays and to predict a chaotic model, Mackey-Glass. It has been shown that the sensibility related to input alterations reduces because of using the granular activation function in RBF Neural Network structure and the response of Granular RBF Neural Network with noisy data is better than RBF Neural Network.

Gradient Descent Methods for Type-2 Fuzzy Neural Networks

Abstract: Given an initial point, if an algorithm tries to follow the negative of the gradient of the function at the current point to reach a local minimum, we face the most common iterative method to optimize a nonlinear function: the GD method. The main goal of this chapter is to briefly discuss a multivariate optimization technique, namely the GD algorithm, to optimize a nonlinear unconstrained problem. The referred optimization problem is a cost function of a FNN, either type-1 or type-2, in this chapter. The main features, drawbacks and stability conditions of these algorithms are discussed.

Fuzzy reference model for adaptive indirect Takagi-Sugeno model reference control

Abstract: Present study presents an indirect model reference fuzzy controller which benefits from a reference model which is itself fuzzy rather than a linear reference model. The use of a fuzzy reference model makes it possible to have more degrees of freedom in the choice of the reference model and to design a more appropriate reference trajectory for the system. The Stability analysis of the resulting system including the adaptation laws are proved via selecting an appropriate Lyapunov function. Designed for SISO nonlinear systems subjected to variable dynamics and also disturbances, the method is tested on chaotic systems in aforementioned conditions. Considering the facts that the use of fuzzy reference model adds more degrees of freedom to the system and is more flexible, the proposed control structure outperforms existing indirect adaptive model reference fuzzy controllers which use linear reference models in their structures.

Hybrid Training Method for Type-2 Fuzzy Neural Networks Using Particle Swarm Optimization

Abstract: Even if we have highly capable computers today, nature still has the best engineering designs. For example, it is not coincidence that the nose of an airplane is very similar to that of a dolphin. Not only the physical appearance but also the behavior of animals in nature are opening the doors for new theories. This chapter shows that nature is still helping humans make the most efficient and brilliant engineering designs.

Adaptive direct fuzzy control of SISO nonlinear systems using a fuzzy reference model

Abstract: This study presents a novel fuzzy adaptive controller comprising a fuzzy direct model reference mechanism to control uncertain nonlinear SISO systems. The proposed method benefits from a reference model which is itself fuzzy. Since the reference model is fuzzy it has more degrees of freedom to define a more appropriate dynamic behavior for the system to be controlled. The flexibility caused by the fuzzy reference model makes it possible for the system to outperform the case when the reference signal is linear in terms of rise time and settling time. In addition, using the proposed approach, tracking error of the reference system reduces significantly. The direct method of Lyapunov is used to prove the stability of the system. The proposed method is tested on a Duffing oscillator subject to disturbances.

Noise Reduction Property of Type-2 Fuzzy Neural Networks

Abstract: In this chapter, an attempt is made to show the effect of input noise in the rule base in a general way. There exist number of papers in literature claiming that the performance of T2FLSs is better than its type-1 counterparts under noisy conditions. We try to justify this claim by simulation studies only for some specific systems. However, in this chapter, such an analysis is done independent of the system to be controlled. For such an analysis, a novel type-2 fuzzy MF (elliptic MF) is proposed. This type-2 MF has certain values on both ends of the support and the kernel, and some uncertain values for other values of the support. The findings of the general analysis in this chapter and the aforementioned studies published in literature are coherent.

A multi-objective genetic type-2 fuzzy extreme learning system for the identification of nonlinear dynamic systems

Abstract: The major challenge in the design of Interval type-2 fuzzy logic system (IT2FLS) is to determine the optimal parameters for their antecedent and consequent parts. The most frequently used objective function for the design of IT2FLSs is root mean squared error (RMSE). However, other than RMSE, the maximum absolute error (MAE) for each of identification samples is very important. This paper propose a novel hybrid learning algorithm for the design of IT2FLS. The proposed algorithm benefits from the combination of extreme learning machine (ELM) and non-dominated sorting genetic algorithm (NSGAII) to tune the parameters of the consequent and antecedent parts of the IT2FLS, respectively. The proposed method is used for forecasting of nonlinear dynamic systems. It is shown that not only the proposed method results in low RMSE, MAE achieved is also satisfactory.

Case Studies: Control Examples

Abstract: In this chapter, three real-world control problems, namely anesthesia, magnetic rigid spacecraft and tractor-implement system are studied by using SMC theory-based learning algorithms for T2FNNs. For all the systems, the FEL scheme is preferred in which a conventional controller (PD, etc.) works in parallel with an intelligent structure (T1FNN, T2FNN, etc.). The proposed learning algorithms have been shown to be able to control these real-world example problems with satisfactory performance. Note that the proposed control algorithms do not need a priori knowledge of the system to be controlled.

Case Studies: Identification Examples

Abstract: In this chapter, the learning algorithms proposed in the previous chapters (GD-based, SMC theory-based, EKF and hybrid PSO-based learning algorithms) are used to identify and predict two nonlinear systems, namely Mackey-Glass and a second-order nonlinear time-varying plant. Several comparisons are made, and it has been shown that the proposed SMC theory-based algorithm has faster convergence than existing methods such as GD-based and swarm intelligence-based methods. Moreover, the proposed learning algorithm has an explicit form, and it is easier to implement than other existing methods. However, for offline algorithms for which computation time is not an issue, the hybrid training method based on PSO and SMC theory may be a preferable choice.

Extended Kalman Filter Algorithm for the Tuning of Type-2 Fuzzy Neural Networks

Abstract: In this chapter, the EKF algorithm is used to optimize the parameters of T2FNNs. The basic version of KF is an optimal linear estimator where system is linear and is subject to white uncorrelated noise. However, it is possible to use Taylor expansion to extend its applications to nonlinear cases. Finally, the decoupled version of the EKF is also discussed, which is computationally more efficient than EKF to tune the parameters of T2FNNs.