Browsing by Author "Wang, Chi-Hsu"

Now showing 1 - 20 of 37

A Dynamic Neural Network Model for Nonlinear System Identification
(2008)
Wang, Chi-Hsu
;
Chen, Pin-Cheng
;
Lin, Ping-Zong
;
Lee, Tsu-Tian
;
電控工程研究所
In this paper, a new dynamic neural network based on the Hopfield neural network is proposed to perform the nonlinear system identification. Convergent analysis is performed by the Lyapunov-like criterion to guarantee the error convergence during identification. Simulation results demonstrate that the proposed dynamic neural network trained by the Lyapunov approach can obtain good identifted performance.
A New Intelligent Fuzzy Controller for Nonlinear Hysteretic Electronic Throttle in Modern Intelligent Automobiles
(2013-06-01)
Wang, Chi-Hsu
;
Huang, De-Yu
;
電機工程學系
;
電控工程研究所
;
Department of Electrical and Computer Engineering
In order to control the nonlinear hysteretic electronic throttle adopted in modern automobiles, a new result for the tracking control of nonlinear hysteretic system is first proposed in this paper. Therefore we can realize the intelligent fuzzy logic controller in a well-behaved and systematic manner. A new closed-loop Back-propagation tuning is also proposed for the tuning of the fuzzy output membership functions to yield better tracking result. Finally, the controller synthesis is performed in a real-time environment using dSpace MicroAutobox and advanced microcontroller development board to yield excellent tracking results with cost-effective implementation.
A New Learning Algorithm for a Fully Connected Neuro-Fuzzy Inference System
(2014-10-01)
Chen, C. L. Philip
;
Wang, Jing
;
Wang, Chi-Hsu
;
Chen, Long
;
電機資訊學士班
A traditional neuro-fuzzy system is transformed into an equivalent fully connected three layer neural network (NN), namely, the fully connected neuro-fuzzy inference systems (F-CONFIS). The F-CONFIS differs from traditional NNs by its dependent and repeated weights between input and hidden layers and can be considered as the variation of a kind of multilayer NN. Therefore, an efficient learning algorithm for the F-CONFIS to cope these repeated weights is derived. Furthermore, a dynamic learning rate is proposed for neuro-fuzzy systems via F-CONFIS where both premise (hidden) and consequent portions are considered. Several simulation results indicate that the proposed approach achieves much better accuracy and fast convergence.
Adaptive asymmetric fuzzy neural network controller design via network structuring adaptation
(2008-10-16)
Hsu, Chun-Fei
;
Lin, Ping-Zong
;
Lee, Tsu-Tian
;
Wang, Chi-Hsu
;
電控工程研究所
This paper proposes a self-structuring fuzzy neural network (SFNN) using asymmetric Gaussian membership functions in the structure and parameter learning phases. An adaptive self-structuring asymmetric fuzzy neural-network control (ASAFNC) system which consists of an SFNN controller and a robust controller is proposed. The SFNN controller uses an SFNN with structure and parameter learning phases to online mimic an ideal controller, simultaneously. The structure learning phase consists of the growing-and-pruning algorithms of fuzzy rules to achieve an optimal network structure, and the parameter learning phase adjusts the interconnection weights of neural network to achieve favorable approximation performance. The robust controller is designed to compensate for the modeling error between the SFNN controller and the ideal controller. An online training methodology is developed in the Lyapunov sense, and thus the stability of the closed-loop control system can be guaranteed. Finally, the proposed ASAFNC system is applied to a second-order chaotic dynamics system. The simulation results show that the proposed ASAFNC can achieve favorable tracking performance. (C) 2008 Elsevier B.V. All rights reserved.
Adaptive SOM-Based Fuzzy Neural Network Controller Design for Multi-Agent System Dispatching and Path Planning
(2012)
Wang, Chi-Hsu
;
Hung, Kun-Neng
;
電機工程學系
;
Department of Electrical and Computer Engineering
In this paper, a new adaptive SOM-based Fuzzy Neural Network controller is proposed for dynamic dispatching and path planning of multi-agent systems. The traditional self-organizing map (SOM) is to exclusively search the real-time shortest paths for all agents to go to their targets. After this traditional dispatching, the weighting factors of our new SOM-based fuzzy neural network (FNN) controller are activated to force the agents toward their corresponding targets. The FNN controller is the main controller which combines the fuzzy rules with the neural network. A monitoring controller is also designed to reduce the error between FNN controller and ideal controller. By the Lyapunov constraints, the weighting factors for the proposed SOM-based FNN controller are updated to guarantee the stability of the path planning system. From the simulation results, it can be obviously seen that the proposed new method is capable of effectively dispatching the agents and controlling their trajectories to the corresponding targets.
Analysis and Design of Indirect Adaptive Fuzzy Controller for Nonlinear Hysteretic Systems
(2015-03-01)
Wang, Chi-Hsu
;
Wang, Jyun-Hong
;
Chen, Chun-Yao
;
電機資訊學士班
;
Undergraduate Honors Program of Electrical Engineering and Computer Science
In order to handle the nonlinear properties of hysteretic systems, an indirect adaptive fuzzy controller (IAFC) is proposed in this paper. However, it is hard to directly identify the unknown hysteretic effects. Therefore, to overcome this problem, a dynamic hysteretic equation is employed and modified to construct the nonlinear properties of backlash-like hysteretic systems. Besides, the existence of an IAFC can be derived in this paper. Compared with the existing fuzzy control methods, our proposed IAFC is simpler and can handle the more general hysteretic problems with our new learning algorithm. Based on the learning algorithm, the adaptive and the control laws not only can be derived but the stability of the closed-loop system can also be guaranteed by the Lyapunov stability criterion. Finally, the proposed IAFC is compared to the adaptive backstepping control method, and the results show that our proposed IAFC can effectively handle the nonlinear properties in the backlash-like hysteretic systems.
Design and Implementation of Intelligent Surge Controller for Modern Turbo Charged Automobiles
(2014-06-01)
Wang, Chi-Hsu
;
Wang, Chih-Cheng
;
電機工程學系
;
電控工程研究所
;
Department of Electrical and Computer Engineering
The destructive surge transient response is an expected phenomenon on a turbo charged automobile. Therefore a new intelligent controller to prevent the surge phenomenon to prolong the life cycle of modern turbo charged automobiles is proposed in this paper. In order to do so, the surge boundary for a particular turbo charger in a specific automobile must be determined first. A new systematic way without performing a real experiment to find the surge boundary is therefore proposed in this paper. The modeling of the turbo charged system is first discussed with Moore-Greitzer model with the new finding of determining the Helmholtz frequencies on the compressor load curves (CLCs) on the compressor map diagram. By substituting the real physical parameters of SAAB NG 900 into the modeling equation, the manufacturer\'s surge boundary and real surge boundary for all turbo charged automobiles can be accurately verified. These new results are immediately adopted in this paper to design an intelligent surge controller to minimize the risk of surge phenomenon in a turbo charger automobile. This new intelligent surge controller has the capability to maintain a certain pressure to bypass the surge boundary with minimum pressure release. Excellent real-time experimental results have been obtained to validate that our new approach is very effective. This will have decent applications in modern turbo charged intelligent automobiles.
Design of Indirect Adaptive Fuzzy Control (IAFC) for Nonlinear Hysteretic Systems
(2014-01-01)
Wang, Chi-Hsu
;
Wang, Jyun-Hong
;
Chen, Chun-Yao
;
電機資訊學士班
;
Undergraduate Honors Program of Electrical Engineering and Computer Science
Nonlinear hysteretic phenomena occur in many physical systems, such as electronic throttle and solenoid valves in automobiles, piezoelectric sensors, and many other mechanical actuators. In order to handle the nonlinear properties of hysteretic systems, an indirect adaptive fuzzy controller (IAFC) is proposed in this paper. However, it is a hard task to directly identify unknown hysteretic effects. Firstly, to overcome this problem, a hysteretic function is employed to construct the nonlinear properties of backlashlike hysteretic systems. Then the existence of an indirect adaptive controller (IAFC) is derived in this paper. Unlike the existing fuzzy control methods, our proposed IAFC can deal with different kinds of hysteretic problems with adaptive and control laws. Based on the learning algorithm, the adaptive and control laws not only can be derived but the stability of the closed- loop system can also be guaranteed by the Lyapunov stability criterion. Finally, MATLAB software is used for simulations, and the results show that our proposed IAFC can effectively handle the nonlinear properties in some unknown hysteretic systems.
Dynamic optimal training of a three layer neural network with sigmoid function
(2006)
Wang, Chi-Hsu
;
電控工程研究所
;
Institute of Electrical and Control Engineering
This paper proposes a dynamical optimal training algorithm for a three layer neural network (NN) with sigmoid activation functions in the hidden and output layers. This three layer neural network can be used for classification problems, such as the classification of Iris data. The mathematical formulation of this three layer NN is rigorously derived first in this paper, so that the dynamical optimal training of it can be performed. The dynamical optimal training process for this three layer NN is therefore presented which guarantees the convergence of the training in a minimum number of epochs. This dynamical optimal training does not use fixed learning rate for training. Instead, the learning rates are updated for next iteration to guarantee the optimal convergence of the training result. Excellent results have been obtained for XOR and Iris data set.
DYNAMIC SYSTEM IDENTIFICATION USING HIGH-ORDER HOPFIELD-BASED NEURAL NETWORK (HOHNN)
(2012-11-01)
Wang, Chi-Hsu
;
Hung, Kun-Neng
;
電機工程學系
;
Department of Electrical and Computer Engineering
The high-order Hopfield-based neural network (HOHNN) with functional link net (FLN) has been developed in this paper for the purpose of uncertain dynamic system identification. In comparison with the traditional Hopfield neural network (HNN) and the high-order neural network (HONN), the compact structure of FLN, with a systematic order mathematical representation combined into the proposed HOHNN, has additional inputs for each neuron for faster convergence rate and less computational load. The weighting factors in HOHNN are tuned via the Lyapunov stability theorem to guarantee the convergence performance of real-time system identification. The robust learning analysis of HOHNN to improve the convergence in the performance is also discussed. The simulation results and computation analysis for different Hopfield-based neural networks are conducted to show the effectiveness of HOHNN in uncertain dynamic system identification.
Finding the Capacity of Fuzzy Neural Networks (FNNs) via Its Equivalent Fully Connected Neural Networks (FFNNs)
(2011)
Wang, Jing
;
Wang, Chi-Hsu
;
Chen, C. L. Philip
;
電子工程學系及電子研究所
;
Department of Electronics Engineering and Institute of Electronics
The capacity of Fuzzy Neural Network (FNN) is explored in this paper. The FNN is first transformed into an equivalent fully connected three layer neural network, or FFNN, via a new approach proposed in this paper. Then the lower and upper bounds of FNN can be found. To check the validity of the theoretical bounds, an example is illustrated with the trainings to yield excellent capacity matching with the theoretical bounds. This new finding has its emerging values in all engineering applications using FNN, such as intelligent adaptive control, pattern recognition, and signal processing, ... , etc.
Finding the Real Surge Boundaries of Turbo-charged Automobiles Using Intelligent Fuzzy Reasoning Technique
(2015-06-01)
Wang, Chi-Hsu
;
Wang, Chih-Cheng
;
資訊工程學系
;
Department of Computer Science
Engine with turbo charger needs to be aware of the potential damage caused by surge phenomena. The focus of this paper is to propose a new intelligent reasoning to find the real surge boundary for a turbo-charged automobile without performing real complicated automobile experiment. Four physical fuzzy variables of a turbo-charged automobile, i.e., throttle size, pressure ratio, engine displacement, and total effective pipe length between the throttle and turbo charger, are chosen as the input to our intelligent reasoning algorithm system to produce the mass air flow for correction. The mass air flow for correction can then be applied on the compressor map of turbo charger to determine the real surge boundary for a certain turbo-charged automobile. In comparison with the previous approach using quadratic interpolation formula for exact real surge boundary, our new approach can indeed produce nearly the same answer with simplified computation load. Excellent benchmark tests are conducted on Saab NG 900 and 9000 with turbo chargers, Garrett T25-60.
From fuzzy center average defuzzifier (CAD) to fuzzy lookup table controller (FLTC) with an efficient Heaviside search algorithm (HSA)
(2019-09-01)
Wang, Chi-Hsu
;
Hor, Kar-Chun
;
交大名義發表
;
電機學院
;
National Chiao Tung University
This paper proposes an alternate interpretation of the traditional lookup table controller via the center average defuzzifier (CAD) in fuzzy set theory with a set of disjoint block pulse membership functions (DBPMFs). The DBPMFs with CAD can be proved to be equivalent to the traditional fuzzy lookup table controller (FLTC). Further, a new Heaviside search algorithm (HSA) is also proposed to implement the FLTC in a more efficient way. The HSA is to find the indices for every fuzzy input variable independently, in order to speed up the table-searching process. The computational complexity of our new HSA is far less than the complexity of the traditional linear search algorithm (LSA), where HSA is the sum of R-i, where R-i is the number of membership functions for ith input variable in FLTC. The average computational complexity in LSA is the product of R-i divided by 2. The HSA also reduces the complexity of coding significantly in the same order as the comparison of computation complexities between HSA and LSA. The balancing of the inverted pendulum system is adopted as the benchmark to show the feasibility and efficiency of this new FLTC and HSA.
Fuzzy Control Using Intuitive Image Analysis
(2008)
Chen, Guan-Ming
;
Lin, Ping-Zong
;
Wang, Wei-Yen
;
Lee, Tsu-Tian
;
Wang, Chi-Hsu
In this paper, a novel fuzzy control scheme using intuitive image analysis is developed to imitate the intuitive human control behavior determined through human eyes. A CCD camera is used to gather the images of the controlled plant, and a simple algorithm is proposed to analyze the images. Unlike that in the visual servo control systems, the image information is utilized in a more intuitive way via the proposed image analysis algorithm. The difference between a reference image and the current image is numerically expressed and directly used by a fuzzy control system using a human-like control law. To investigate the effectiveness of the proposed fuzzy control scheme, it is applied to an inverted pendulum system. Simulation results show that the proposed scheme can achieve favorable tracking performance without prior knowledge of the controlled plant.
Fuzzy Lookup Table Controller (FLTC) via Center Average De-fuzzifier (CAD) with a New Efficient Heaviside Search Algorithm (HSA) for Lookup Table Implementation
(2018-01-01)
Wang, Chi-Hsu
;
Hor, Kar-Chun
;
Chen, Wei-Chuan
;
電機學院
;
電機工程學系
This paper proposes an alternate interpretation of the traditional lookup table controller via the Center Average De- fuzzier (CAD) in Fuzzy Set Theory with a set of Disjoint Block Pulse Membership Functions (DBPMFs). Further, a new Heaviside Search Algorithm (HSA) is also proposed to implement the Fuzzy Lookup Table Controller (FLTC) in a more efficient way. The computation complexity of our new HSA is O(Sigma R-i), where R-i is the number of membership functions for i th input variable in FLTC. Whereas the average computational complexity in traditional Linear Search Algorithm (LSA) is O((Sigma R-i)/2). In comparison with LSA, the HSA also reduces the complexity of coding significantly in the same order as the comparison of computation complexities between HSA and LSA. The balancing of Inverted Pendulum System (IPS) is adopted as the benchmark to show the feasibility and efficiency of this new FLTC and HSA.
Fuzzy-identification-based adaptive backstepping control using a self-organizing fuzzy system
(2009-05-01)
Chen, Pin-Cheng
;
Hsu, Chun-Fei
;
Lee, Tsu-Tian
;
Wang, Chi-Hsu
;
電控工程研究所
In this paper, a fuzzy-identification-based adaptive backstepping control (FABC) scheme is proposed. The FABC system is composed of a backstepping controller and a robust controller. The backstepping controller, which uses a self-organizing fuzzy system (SFS) with the structure and parameter learning phases to on-line estimate the controlled system dynamics, is the principal controller, and the robust controller is designed to dispel the effect of approximation error introduced by the SFS. The developed SFS automatically generates and prunes the fuzzy rules by the proposed structure adaptation algorithm and the parameters of the fuzzy rules and membership functions tunes on-line in the Lyapunov sense. Thus, the overall closed-loop FABC system can guarantee that the tracking error and parameter estimation error are uniformly ultimately bounded; and the tracking error converges to a desired small neighborhood around zero. Finally, the proposed FABC system is applied to a chaotic dynamic system to show its effectiveness. The simulation results verify that the proposed FABC system can achieve favorable tracking performance even with unknown controlled system dynamics.
High-Order Hopfield-based Neural Network for Nonlinear System Identification
(2009)
Wang, Chi-Hsu
;
Hung, Kun-Neng
;
電控工程研究所
;
Institute of Electrical and Control Engineering
The high-order Hopfield neural network (HOHNN) with functional link net has been developed in this paper for the purpose of system identification of nonlinear dynamical system. The weighting factors in HOHNN will be tuned via the Lyapunov stability criterion to guarantee the convergence performance of real-time system identification. In comparison with the traditional Hopfield neural network (HNN), the proposed architecture of HOHNN has additional inputs for each neuron which has the advantages of faster convergence rate and less computational load. The simulation results for both HNN and HOHNN are finally conducted to show the effectiveness of HOHNN in system identification of uncertain dynamical systems. It is obvious from the simulation results that the performance of system identification for HOHNN is better than that of HNN.
Identifying an Unknown Flying Target in Missile Defense Systems Using Intelligent Fuzzy Neural Networks
(2017-01-01)
Wang, Chi-Hsu
;
Hor, Kar-Chun
;
Wang, Bin
;
電機工程學系
;
Department of Electrical and Computer Engineering
For radar automatic target recognition (RATR), this paper aims at identifying the incoming unknown flying missiles in the Missile Defense Systems (MDS), using intelligent fuzzy neural networks (FNNs) with intelligent feature extraction. The training data for FNNs is obtained by sampling the Radar Power Signal Envelop (RPSE) of the radar echo signal (from High Resolution Range Profile (HRRP) radar) for the incoming flying missiles under different azimuth and elevation angles. The RPSEs under different irradiation angles of incoming flying targets can be generated by a high frequency structural simulator (HFSS) package, which is close to real RPSE and has been adopted by academic researchers in this area. The premise part in FNN is composed of a set of Uniform Distributed Gaussian Membership Functions (UDGMFs) and the consequent part is a two layer Neural Network (NN) which can be trained by Dynamic Optimal Training Algorithm (DOTA). By using this approach, the identification of five different flying missiles is performed in this paper. From the bench mark test, this intelligent FNN configuration can identify the unknown flying missiles very accurately due to the fact that the number of training patterns is well below the capacity of the proposed FNN configuration under a certain noise intensity.
Indirect Adaptive Control Using Hopfield-Based Dynamic Neural Network for SISO Nonlinear Systems
(2009)
Chen, Ping-Cheng
;
Wang, Chi-Hsu
;
Lee, Tsu-Tian
;
電子工程學系及電子研究所
;
Department of Electronics Engineering and Institute of Electronics
In this paper, we propose an indirect adaptive control scheme using Hopfield-based dynamic neural network for SISO nonlinear systems with external disturbances. Hopfield-based dynamic neural networks are used to obtain uncertain function estimations in an indirect adaptive controller, and a compensation controller is used to suppress the effect of approximation error and disturbance. The weights of Hopfield-based dynamic neural network are on-line tuned by the adaptive laws derived in the sense of Lyapunov, so that the stability of the closed-loop system can be guaranteed. In addition. the tracking error can be attenuated to a desired level by selecting some parameters adequately. Simulation results illustrate the applicability of the proposed control scheme. The designed parsimonious structure of the Hopfield-based dynamic neural network makes the practical implementation of the work in this paper much easier.
Intelligent Adaptive Law for Missile Guidance Using Fuzzy Neural Networks
(2013-06-01)
Wang, Chi-Hsu
;
Hung, Kun-Neng
;
電機工程學系
;
Department of Electrical and Computer Engineering
In this paper, an adaptive fuzzy neural network (FNN) controller is proposed for missile guidance. The objective is for one defending missile (DM) to intercept an attacking missile (AM) in air battle scenario. The adaptive FNN controller is adopted to force the DM toward the AM under the existence of disturbance, and a monitoring controller is also designed to reduce the error between FNN controller and ideal controller. In comparison with the other multi-layered neural network controller, the proposed adaptive FNN controller which combines the fuzzy rules with the neural network can be easily designed. The weighting factors of our new FNN controller are activated to dispatch the DM toward the AM. Using the Lyapunov constraints, the weighting factors for the proposed FNN controller are updated to guarantee the stability of the path planning system. In our illustrated examples, the systematic battle environment is constructed. From the simulation results, the proposed adaptive FNN controller is capable of performing missile guidance, and it also shows that the computation load of our proposed approach is much less than that of using cerebellar model articulation controller (CMAC).