Font Size: a A A

Study On Adaptive Fuzzy Control Via State Space Method

Posted on:2007-08-01Degree:DoctorType:Dissertation
Country:ChinaCandidate:Z N MiaoFull Text:PDF
GTID:1118360182495897Subject:Traffic Information Engineering and Control
Abstract/Summary:PDF Full Text Request
Fuzzy control is an important field of intelligent control. It has accessed great success through its developing history in the past decades. So do the fuzzy control theory. For the reason that fuzzy control imitates the inference and decision process of human and controls the plant based on the human's operation experience instead of the accurate math model, fuzzy control can not describes the system precisely just as the traditional control theory and modern control theory do. Further more;because the design process of fuzzy control system relies on the control experience of the expert or experienced operator, there is not a systemic method to choose the fuzzy variables, to make the fuzzy division for the variables, to choose proper membership functions and to construct the rule base. These limitations obstruct the further development and application of fuzzy control. This paper begins with the basic principle of fuzzy control. It proposes a new method for fuzzy system describing, analyzing and systemic design based on the successful modern control theory method and the pioneer works of fuzzy control. In this paper, the following research results are achieved:1. Presented the linguistic model for the controlled object definitely. In the traditional fuzzy control theory, the characteristic of the fuzzy control needs not the accurate math model is overmuch emphasized while the importance of controlled object model is ignored. This ideal makes the fuzzy system an open-loop system. The model of controlled object is presented in this paper. The linguistic model inherits the advantages of utilization of human's intuited knowledge while the precise math model absent. Meanwhile, the linguistic model of controlled object makes the fuzzy system a close-loop system and constructs a base for further study on system design and analyses.2. Utilize state space method in fuzzy control system. State space method is the base of system design and analyses in modern control theory and it has achieved great success in the field. In this paper, a new method named fuzzy state space is proposed for fuzzy control system. The description and implementation of fuzzy inference is discussed via the new method. It also provides a tool for representation of linguistic model of the controlled object. The fuzzy state space method provides a base for system representation and analyses.3. The effect on system performance caused by the fuzzy set definition isdiscussed. This paper also suggests a similarity criterion for fuzzy sets definition in the discourse of universe of fuzzy variables. Another criterion for rules consistency based on multi-dimension similarity is proposed.4. Utilize the ideal of adaptive inverse control in fuzzy system design and propose a systemic design method named adaptive inverse fuzzy control. Adaptive inverse is a success method in adaptive control theory and it accesses an outstanding performance especially in the system suffering disturbance or system parameter changing. This paper proposes a new structure of fuzzy system based on the idea of adaptive inverse control.5. A new genetic algorithm named multi-population genetic algorithm (MPGA) is suggested. An important task for construct adaptive inverse fuzzy control system is to construct the linguistic model of controlled object. Learning from data is an effective way to acquire the fuzzy model. MPGA simulates the basic principle of populations evolve respectively and adapt itself to the environment in the nature. It takes the rule base and rule parameters as two different populations and utilize genetic algorithm to learn fuzzy model from experiment data. The algorithm also involve real-coded genetic algorithm to access a better accuracy and convergent speed.6. The inverse fuzzy model is discussed. Based on the preexisted research results in such field, this work investigates the reversibility of a fuzzy model and different kind of inverse model for a given fuzzy model.7. A typical non-linear hysteretic system (domestic gas burning heater) is investigated as a controlled object to prove the correctness and effectiveness of the proposed method. The analyses and design process is discussed in detail. Some simulation result is presented in this paper.
Keywords/Search Tags:fuzzy control, fuzzy state space, adaptive inverse fuzzy control, multi population genetic algorithm, domestic gas burning heater
PDF Full Text Request
Related items