自我建構第二類型模糊類神經網路於電動汽車速度控制之設計

風阻、輪胎與路面之間的摩擦阻力、驅動馬達特性、路面的坡度以及其他非線性動態變化之因素，皆會嚴重影響電動汽車之性能。本文採用自我建構第二類型模糊神經網絡控制器，可即時地追蹤EV之速度並計算控制直流馬達的轉矩。第二類型模糊神經網絡控制器係以強健的第二類典型模糊神經網路控制器為主軸，加上以自我建構參數學習之演算法與在線學習演算法，計算馬達運轉之角速度之責務進而控制EV。其結構和參數學習皆能於線上自動完成。自我建構學習演算法，以馬氏距離方法，以確定規則增生/消減與否。線上學習演算法係基於倒傳遞方式來更新參數（平均值，標準偏差和權重）的使用誤差法則。最後，速度控制系統考慮爬坡坡度的不同，並與比例積分微分控制器比較，說明第二類型模糊神經網絡控制器控制器可更有效地控制電動汽車速度。

關鍵字

權重；電動汽車；模糊系統；類神經網路；第二類型；自我建構；倒傳遞

並列摘要

The forces of drag, tire and road surface friction resistance, the drive motor characteristics, the hill climbing angle and other non-linear dynamic factors tremendously effect the performance of electric vehicles (Electric Vehicle, EV). The proposed design, self-construction of the type-2 fuzzy neural network (SCT2FNN) controller, based on robust typical type-2 fuzzy neural network (T2FNN) controller, with the help of self-construct parameter learning algorithms, and online learning algorithm to estimate the angular velocity of the motor operation to control the EV, can promptly track the speed of the EV and estimate the torque control of a DC motor. Applying the Mahalanobis distance (M-distance) method in the self-constructing learning algorithm to determine whether the T2FNN rules are generated or not, and the online learning algorithm, basing on back propagation method to adjust the parameter (mean, standard, deviation and weight) errors generated from T2FNN. The proposed SCT2FNN controller is identified more efficient while controlling the speed of EV, by comparing the PID controller , considering the difference of the climbing slope.

並列關鍵字

Electric Vehicle ； Fuzzy system ； Neural network ； Type-2 membership function ； Self-constructing ； Back-propagation

參考文獻

[2] D. E. Nye, Electrifying America: Social Meanings of a New Technology. MA: MIT Press, 1992.

[3] J. J. Flink, The Automobile Age. MA: MIT Press, 1990.

[5] F. A. Wyczalek, “Hybrid electric vehicles: year 2000 status,” IEEE AES Systenrv Magazine, vol. 16, pp. 15-25, Mar. 2001

[6] V. D. Colli, G. Tomassi, and M. Scarano, “Single wheel longitudinal traction control for electric vehicles,” IEEE Transactions on Power Electronics, vol. 21, no.3, pp. 799-808, May 2006.

[8] L. R. Ray, “Real-time determination of road coefficient of friction for IVHS and advanced electric vehicle control,” in Proc. 1995 American Control Conference, Seattle, WA, USA, June 21-23, 1995, vol. 3, pp. 2133-2137.

國際替代計量

自我建構第二類型模糊類神經網路於電動汽車速度控制之設計

未授權

主題瀏覽