Recurrent Fuzzy Neural Network Research Articles

Adaptive dynamic friction observer and recurrent fuzzy neural network estimator design with backstepping control

In this study, robust non-linear dynamic friction control is considered using a dynamic friction observer and intelligent control. An adaptive dynamic friction observer based on the LuGre friction model is proposed to estimate the friction parameters and a directly immeasurable friction state variable. A recurrent fuzzy neural network (RFNN) approximator and reconstructed error compensator are also designed to give additional robustness to the control system under friction model uncertainty. A proposed composite control scheme with a basic backstepping controller is applied to the position tracking control of the servo system. The performances of the proposed friction observer and the friction controller are demonstrated by some simulations and experiments.

Design of Dynamic Petri Recurrent Fuzzy Neural Network and Its Application to Path-Tracking Control of Nonholonomic Mobile Robot

This paper focuses on the design of a dynamic Petri recurrent fuzzy neural network (DPRFNN), and this network structure is applied to the path-tracking control of a nonholonomic mobile robot for verifying its validity. In the DPRFNN, the concept of a Petri net and the recurrent frame of internal-feedback loops are incorporated into a traditional FNN to alleviate the computation burden of parameter learning and to enhance the dynamic mapping of network ability. Moreover, the supervised gradient-descent method is used to develop the online-training algorithm for the DPRFNN control. In order to guarantee the convergence of path-tracking errors, analytical methods based on a discrete-type Lyapunov function are proposed to determine varied learning rates for DPRFNN. In addition, the effectiveness of the proposed DPRFNN control scheme under different moving paths is verified by experimental results, and its superiority is indicated in comparison with FNN, RFNN, Petri FNN, and PRFNN control systems.

Non-linear friction compensation using backstepping control and robust friction state observer with recurrent fuzzy neural networks

A robust precise tracking control for a servomechanical system with non-linear dynamic friction is presented. The LuGre friction that is adopted as a non-linear friction model contains both a directly immeasurable friction state variable and the uncertainty caused by incomplete parameter identification and change of the condition of contact surface. To provide an efficient solution to these problems, a composite robust control scheme is proposed, which consists of a robust friction state observer, a recurrent fuzzy neural network (RFNN) approximator, and an adaptive reconstructed error compensator with backstepping control. A robust friction state observer is designed to estimate the unknown internal state of the LuGre friction model. Next, a proposed RFNN scheme approximates the lumped friction torque uncertainty. Finally, an adaptive error compensator eliminates a reconstructed error arising from RFNN approximation. Some simulations and experiments for a ball-screw servosystem are carried out to demonstrate the performance of a proposed control scheme.

Precise friction control for the nonlinear friction system using the friction state observer and sliding mode control with recurrent fuzzy neural networks

This paper deals with a tracking control problem of a mechanical servo system with nonlinear dynamic friction which contains a directly immeasurable friction state variable and an uncertainty caused by incomplete parameter modeling and its variations. In order to provide an efficient solution to these control problems, we propose a composite control scheme, which consists of a friction state observer, a RFNN approximator and an approximation error compensator with sliding mode control. In first, a sliding mode controller and friction state observer are designed to estimate the unknown internal state of the LuGre friction model. Next, a RFNN is developed to approximate an unknown lumped friction uncertainty. Finally, an adaptive error compensator is designed to compensate an approximation error of RFNN. Some simulations and experiments on the mechanical servo system composed of ball-screw and DC servo motor are executed. Their results give a satisfactory performance of the proposed control scheme.

Adaptive output feedback control based on DRFNN for AUV

The tracking control problem of AUV in six degrees-of-freedom (DOF) is addressed in this paper. In general, the velocities of the vehicles are very difficult to be accurately measured, which causes full state feedback scheme to be not feasible. Hence, an adaptive output feedback controller based on dynamic recurrent fuzzy neural network (DRFNN) is proposed, in which the location information is only needed for controller design. The DRFNN is used to online estimate the dynamic uncertain nonlinear mapping. Compared to the conventional neural network, DRFNN can clearly improve the tracking performance of AUV due to its less inputs and stronger memory features. The restricting condition for the estimation of the external disturbances and network's approximation errors, which is often given in the existing literatures, is broken in this paper. The stability analysis is given by Lyapunov theorem. Simulations illustrate the effectiveness of the proposed control scheme.

FPGA Implementation of a Recurrent Neural Fuzzy Network with On-Chip Learning for Prediction and Identification Applications

In this paper, a hardware implementation of a recurrent neural fuzzy network (RNFN) used for identification and prediction is proposed. A recurrent network is embedded in the RNFN by adding feedback connections in the second layer, where the feedback units act as memory elements. Although the back propagation (BP) learning algorithm is widely used in the RNFN, BP is too complicated to be implemented using hardware. However, we use the simultaneous perturbation method as a learning scheme for hardware implementation to overcome the above-mentioned problems. The hardware implementation of the RNFN uses random access memory (RAM), which stores all the parameters of a network. This design method reduces the number of logic gates used. The major findings of the experiment show that field programmable gate arrays (FPGA) implementation of the RNFN retains good performance in identification and prediction problems.

Evolutionary algorithm-based learning of fuzzy neural networks. Part 2: Recurrent fuzzy neural networks

Fuzzy neural networks (FNN) as opposed to neuro-fuzzy systems, whose main task is to process numerical relationships, can process both numerical (measurement based) information and perception based information. In spite of great importance of fuzzy feed-forward and recurrent neural networks for solving wide range of real-world problems, today there are no effective training algorithm for them. Currently there are two approaches for training of FNN. First approach is based on application of the level-sets of fuzzy numbers and the back-propagation (BP) algorithm. The second approach involves using evolutionary algorithms to minimize error function and determine the fuzzy connection weights and biases. The method based on the second approach was proposed by the authors and published in Part 1 of this paper [R.A. Aliev, B. Fazlollahi, R. Vahidov, Genetic algorithm-based learning of fuzzy neural networks. Part 1: feed-forward fuzzy neural networks, Fuzzy Sets and Systems 118 (2001) 351–358]. In contrast to the BP and other supervised learning algorithms, evolutionary algorithms do not require nor use information about differentials, and hence, they are most effective in case where the derivative is very difficult to obtain or even unavailable. However, the main deficiency of the existing FNN based on the feed-forward architecture is its adherence to static problems. In case of dynamic or temporal problems there is a need for recurrent fuzzy neural networks (RFNN). Designing efficient training algorithms for RFNN has recently become an active research direction. In this paper we propose an effective differential evolution optimization (DEO) based learning algorithm for RFNN with fuzzy inputs, fuzzy weights and biases, and fuzzy outputs. The effectiveness of the proposed method is illustrated through simulation of benchmark forecasting and identification problems and comparisons with the existing methods. The suggested approach has also been used for real applications in an oil refinery plant for petrol production forecasting.

Recurrent fuzzy neural network by using feedback error learning approaches for LFC in interconnected power system

In this study, a new adaptive controller based on modified feedback error learning (FEL) approaches is proposed for load frequency control (LFC) problem. The FEL strategy consists of intelligent and conventional controllers in feedforward and feedback paths, respectively. In this strategy, a conventional feedback controller (CFC), i.e. proportional, integral and derivative (PID) controller, is essential to guarantee global asymptotic stability of the overall system; and an intelligent feedforward controller (INFC) is adopted to learn the inverse of the controlled system. Therefore, when the INFC learns the inverse of controlled system, the tracking of reference signal is done properly. Generally, the CFC is designed at nominal operating conditions of the system and, therefore, fails to provide the best control performance as well as global stability over a wide range of changes in the operating conditions of the system. So, in this study a supervised controller (SC), a lookup table based controller, is addressed for tuning of the CFC. During abrupt changes of the power system parameters, the SC adjusts the PID parameters according to these operating conditions. Moreover, for improving the performance of overall system, a recurrent fuzzy neural network (RFNN) is adopted in INFC instead of the conventional neural network, which was used in past studies. The proposed FEL controller has been compared with the conventional feedback error learning controller (CFEL) and the PID controller through some performance indices.

Sliding Mode Control of a Class of Uncertain Nonlinear Time-Delay Systems Using LMI and TS Recurrent Fuzzy Neural Network

This paper proposes the sliding mode control using LMI techniques and adaptive recurrent fuzzy neural network (RFNN) for a class of uncertain nonlinear time-delay systems. First, a novel TS recurrent fuzzy neural network (TS-RFNN) is developed to provide more flexible and powerful compensation of system uncertainty. Then, the TS-RFNN based sliding model control is proposed for uncertain time-delay systems. In detail, sliding surface design is derived to cope with the non-Isidori-Bynes canonical form of dynamics, unknown delay time, and mismatched uncertainties. Based on the Lyapunov-Krasoviskii method, the asymptotic stability condition of the sliding motion is formulated into solving a Linear Matrix Inequality (LMI) problem which is independent on the time-varying delay. Furthermore, the input coupling uncertainty is also taken into our consideration. The overall controlled system achieves asymptotic stability even if considering poor modeling. The contributions include: i) asymptotic sliding surface is designed from solving a simple and legible delay-independent LMI; and ii) the TS-RFNN is more realizable (due to fewer fuzzy rules being used). Finally, simulation results demonstrate the validity of the proposed control scheme.

The Modeling and the Sensor Fault Diagnosis of a Continuous Stirred Tank Reactor with a Takagi-Sugeno Recurrent Fuzzy Neural Network

In this paper, a novel Takagi-Sugeno recurrent fuzzy neural network (TSRFNN) is constructed for modeling and sensor fault diagnosis of a Continuous Stirred Tank Reactor (CSTR), a nonlinear dynamic system. The TSRFNN is composed of 9 layers, including premise network and consequence network. The temporal information is embedded in the TSRFNN by adding the feedback connections between the output layer and the input layer of the fuzzy neural network (FNN). It is assumed that the inputs are Gaussian membership functions; the product operation is utilized for the premise and implication, and the weighted center-average method is adopted for defuzzification. The network is a Fuzzy Basis Function(FBF). The general approximation characteristic of the network was proven by the theory reasoning. The identification of the TSRFNN consists of two steps: structure identification and parameter identification. Unsupervised clustering is used to determine the structure of the fuzzy system, the number of fuzzy rules, and the membership functions of the premise using the input-output data of a system. Then in the parameter identification, the Dynamic Backpropagation (DBP) is adopted to determine the membership functions of the conclusion of the fuzzy system. Then the network is applied to the modeling and diagnosis of a CSTR system. The network is applied to set up the sensors models of a CSTR system, including the models of the temperature sensor faults and the models of concentration sensor faults. To set up a high performance diagnosis model, a persistent signal was chosen to sufficiently activate the plant. Finally an Adaptive Threshold Algorithm based on statistics is used to fault diagnose. If there are sensor faults in the CSTR, the bias between the actual output of the plant and the output of the network surpasses the threshold and the faults are detected on-line according to it. The effectiveness of the modeling and diagnosis approach proposed was verified by the simulation results with Matlab. Therefore, the proposed network and the fault diagnosis approach can be employed to a CSTR system successfully and be extended to the modeling and fault detection of other nonlinear systems.

A novel compensation-based recurrent fuzzy neural network and its learning algorithm

Based on detailed study on several kinds of fuzzy neural networks, we propose a novel compensation-based recurrent fuzzy neural network (CRFNN) by adding recurrent element and compensatory element to the conventional fuzzy neural network. Then, we propose a sequential learning method for the structure identification of the CRFNN in order to confirm the fuzzy rules and their correlative parameters effectively. Furthermore, we improve the BP algorithm based on the characteristics of the proposed CRFNN to train the network. By modeling the typical nonlinear systems, we draw the conclusion that the proposed CRFNN has excellent dynamic response and strong learning ability.

Recurrent neuro fuzzy control design for tracking of mobile robots via hybrid algorithm

This paper proposes a TSK-type recurrent neuro fuzzy system (TRNFS) and hybrid algorithm- GA_BPPSO to develop a direct adaptive control scheme for stable path tracking of mobile robots. The TRNFS is a modified model of the recurrent fuzzy neural network (RFNN) to obtain generalization and fast convergence. The TRNFS is designed using hybridization of genetic algorithm (GA), back-propagation (BP), and particle swarm optimization (PSO), called GA_BPPSO. For the tracking control of mobile robot, two TRNFSs are designed to generate the control inputs by direct adaptive control scheme and hybrid algorithm GA_BPPSO. Through simulation results, we demonstrate the effectiveness of our proposed controller.

A block-diagonal recurrent fuzzy neural network for system identification

A recurrent fuzzy neural network with internal feedback is suggested in this paper. The network is entitled dynamic block-diagonal fuzzy neural network (DBD-FNN), and constitutes a generalized Takagi-Sugeno-Kang fuzzy system, where the consequent parts of the fuzzy rules are small Block-Diagonal Recurrent Neural Networks. The proposed model is applied to a benchmark identification problem, where a dynamic system is to be identified. Additionally, an application of the proposed model to the problem of the analysis of lung sounds is presented. Particularly, a filter based on the DBD-FNN is developed, trained with the RENNCOM method. Extensive experimental and simulation results are given and performance comparisons with a series of other models are conducted, highlighting the modeling characteristics of DBD-FNN as an identification tool and the effectiveness of the proposed separation filter.

Dynamic programming prediction errors of recurrent neural fuzzy networks for speech recognition

This paper proposes Mandarin phrase recognition using dynamic programming (DP) prediction errors of singleton-type recurrent neural fuzzy networks (SRNFNs). This method is called DP-SRNFN. The recurrent property of SRNFN makes it suitable for processing temporal speech patterns. A Mandarin phrase comprises monosyllabic words. SRNFN training is based on the word unit. There are N w SRNFNs for modeling N w words, and each SRNFN receives the current frame feature and predicts the next one of its modeling word. In recognizing N P phrases, the prediction error of each trained SRNFN is computed, and DP is used to find the optimal path that maps the input frames to the best matched SRNFNs (words) for each of the N P phrases. The accumulated error of each phrase model is computed from its optimal path and the one with the minimum error is the recognition result. To verify DP-SRNFN performance, this study conducted experiments on recognizing 30 Mandarin phrases. SRNFN training with noisy features for phrase recognition under different noisy environments was also conducted. DP-SRNFN performance is compared with the hidden Markov models (HMMs). Results show that DP-SRNFN achieves higher recognition rates than HMM in both clean and noisy environments.

Pattern recognition using neural-fuzzy networks based on improved particle swam optimization

This paper introduces a recurrent neural-fuzzy network (RNFN) based on improved particle swarm optimization (IPSO) for pattern recognition applications. The proposed IPSO method consists of the modified evolutionary direction operator (MEDO) and the traditional PSO. A novel MEDO combining the evolutionary direction operator (EDO) and the migration operation is also proposed. Hence, the proposed IPSO method can improve the ability of searching global solution. Experimental results have shown that the proposed IPSO method has a better performance than the traditional PSO in the human body classification and the skin color detection.

A pipelined recurrent fuzzy model for real-time analysis of lung sounds

This paper presents a recurrent fuzzy-neural filter that performs the task of separation of lung sounds, obtained from patients with pulmonary pathology. The filter is a pipelined Takagi–Sugeno–Kang recurrent fuzzy network, consisting of a number of modules interconnected in a cascaded form. The participating modules are implemented through recurrent fuzzy neural networks with internal dynamics. The structure of the modules is evolved sequentially from input–output data. Extensive experimental results, regarding the lung sound category of crackles, are given, and a performance comparison with a series of other fuzzy and neural filters is conducted, underlining the separation capabilities of the proposed filter.

Compensation for secondary uncertainty in electro-hydraulic servo system by gain adaptive sliding mode variable structure control

Based on consideration of the differential relations between the immeasurable variables and measurable variables in electro-hydraulic servo system, adaptive dynamic recurrent fuzzy neural networks(ADRFNNs) were employed to identify the primary uncertainty and the mathematic model of the system was turned into an equivalent linear model with terms of secondary uncertainty. At the same time, gain adaptive sliding mode variable structure control(GASMVSC) was employed to synthesize the control effort. The results show that the unrealization problem caused by some system’s immeasurable state variables in traditional fuzzy neural networks(TFNN) taking all state variables as its inputs is overcome. On the other hand, the identification by the ADRFNNs online with high accuracy and the adaptive function of the correction term’s gain in the GASMVSC make the system possess strong robustness and improved steady accuracy, and the chattering phenomenon of the control effort is also suppressed effectively.

Recurrent Fuzzy Neural Network Control for Mimo Nonlinear Systems

Abstract This paper develops a design method of recurrent fuzzy neural network (RFNN) control system for multi-input multi-output (MIMO) nonlinear dynamic systerns. This control system consists of a state feedback controller and an RFNN controller. The state feedback controller is a basic stabilizing controller to stabilize the system, and the RFNN controller presents a robust controller to deal with uncertain parts of system dynamics and external disturbances. The adaptive laws of the RFNN parameters are derived based on the Lyapunov synthesis approach and a projection algorithm, so that the stability of the system and convergence of the parameters can be guaranteed. The simulation results for a robotic system and an ecological system confirm the effectiveness of the proposed design method.

INTELLIGENT SENSORLESS CONTROL OF TWO-PHASE LINEAR BRUSHLESS DC MOTOR BASED ON A RECURRENT FUZZY NEURAL NETWORK

This paper presents a recurrent fuzzy neural network (RFNN) controller to provide sensorless speed control for the two-phase linear brushless DC motor. It is not easy to arrange current and speed sensors along the linear motor's required railway. The RFNN controller has the capability of tracing the system dynamics due to the inherent learning function. With the help of RFNN controller, the conventional PID control can be made to provide better performance by tracing the system dynamics. The paper will demonstrate dynamic performance improvement scenarios. The recurrent structure also enhances the conventional fuzzy neural network (FNN) controller with embedded memory function. The results from experimental evaluation indicate that the PID with RFNN controller can provide better performance than the conventional PID controller.

Nonlinear active noise control using EKF-based recurrent fuzzy neural networks

Active Noise Control (ANC) system is commonly designed and implemented using adaptation algorithm and adaptive control structure. In this paper we present theoretical and experimental result of active noise control system using Recurrent Fuzzy Neural Network (RFNN). RFNN is developed by combining fuzzy logic and neural networks, aimed at producing better control system performance than if we use neural network or fuzzy logic separately. Using a control structure with two multilayer feedforward RFNNs (one RFNN serves as a nonlinear controller while the other one operates as a nonlinear plant model), a recursive least-squares algorithm based on Adjoint Extended Kalman Filter approach is employed for the training of the controller network. Extended Kalman Filter (EKF) algorithm is introduced to develop a new algorithm with faster convergence speed by using nonlinear recursive-least square method. Experimental result using DSP demonstrates effectiveness of the proposed RFNN structure and algorithm to attenuate unwanted noise.