應用T-S型式模糊類神經網路於系統鑑別、控制及影像壓縮

T-S type Fuzzy Neural Network System and Its Applications in System Identification, Control, and Image Compression.

賴威諭(Wei-Yu Lai)

指導教授 ： 李 慶 鴻

元智大學/工程學院/電機工程學系/碩士(2004年)

摘要

本論文提出一個Takagi-Sugeno (T-S)型態模糊類神經網路(稱為TSFNN)，並將其應用在非線性系統鑑別、控制及影像壓縮。T-S型態之模糊類神經網路之建立，主要以模糊類神經網路與T-S模糊邏輯理論為基礎，針對一非線性系統經由模糊類神經網路學習，並產生合適的模糊規則及相對應的系統狀態空間矩陣，所學習而成的網路將以T-S型式模糊邏輯系統表示。本論文中提出的TSFNN系統在學習上擁有收斂快速、規則數少和減少修改參數等優點，且在非線性系統鑑別過程中不需要了解系統的精確動態方程式，即可精確的將非線性系統建構出來。在控制器設計應用上，本文利用所學習而成的網路資訊藉由李亞普諾夫穩定理論推導，透過線性矩陣不等式(linear matrices inequality, LMI)運算，設計出最佳控制器及T-S型式模糊可變結構控制器。最後將T-S型式模糊類神經網路應用在數位影像壓縮中的預測編碼法，利用TSFNN系統建構一智慧型預測器，在模擬結果看出有非常好的成效。

關鍵字

模糊類神經網路 ； T-S模糊邏輯系統 ； 最佳化控制理論 ； 可變結構控制 ； 預測編碼法

並列摘要

In this thesis, the Takagi-Sugeno (T-S) fuzzy logic system and fuzzy neural network are used to implement the T-S type fuzzy neural network system (TSFNN). As previous literature, a nonlinear system can be approximated by a T-S fuzzy logic system. Herein, the TSFNN system is applied for identifying unknown dynamic systems, and the system has the following advantages- fast convergence, minimum rules and parameters, high accuracy approximation. From linguistic information, numeric data of dynamics system, the TSFNN system can be trained and then obtain the corresponding membership functions (MFs), T-S fuzzy rules. For system controller design, we adopt the learned T-S type fuzzy model to design the fuzzy optimal controller based on Lyapunov stability theorem and linear matrices inequalities (LMI). In addition, T-S type fuzzy variable structure compensator is developed to treat the external disturbance. Finally, we apply the TSFNN system in the image data compression technology. It is applied to be an intelligent predictor for predictive coding. Simulation results are shown to demonstrate the effectiveness of the TSFNN system.

並列關鍵字

Fuzzy neural network ； T-S fuzzy logic system ； optimal control theory ； variable structure control ； predictive coding

參考文獻

[1] Aiazzi, B., L. Alparone, and S. Baronti, “Fuzzy logic-based matching pursuits for lossless predictive coding of still images,” IEEE Trans. on Fuzzy Syst., vol. 10, pp. 473-483, 2002.

[2] Boulgouris, N. V., D. Tzovaras, and M. G. Strintzis, “Lossless image compression based on optimal prediction, adaptive lifting, and conditional arithmetic coding,” IEEE Trans. on Image Processing, vol. 10, no. 1, pp. 1-14, 2001.

[4] Buckley, J. J., “Theory of fuzzy controller: an introduction,” Fuzzy Sets Syst., vol. 51, pp.249-258, 1992.

[5] Bupp, R., D. Bernstein, and V. Coppola, “Benchmark problem for nonlinear control design: problem statement, experimental test bed, and passive nonlinear compensation,” Proc. Amer. Control Conf., Seattle, pp. 4363-4367, 1995.

[7] Carpentieri, B., M. J. Weinberger, and G. Seroussi, “Lossless compression of continuous-tone images,” Proceedings of the IEEE, vol. 88, pp. 1797-1809, 2000.