採用輪轂直流無刷馬達直接驅動車輪的電動車可以減少傳動鏈的功率消耗,使動力系統發揮最高效率,但是電動車的牽引控制系統必須抵抗滾動阻力、空氣阻力和爬坡阻力等非線性變動性負載的干擾,以至於固定參數的線性控制器無法讓牽引控制系統在不同車速之下都保持高牽引效率。本論文以車輪轉速和負載轉矩做為模糊輸入,透過平行分配補償法設計牽引控制系統的模糊PID控制器,接著用適應最佳控制法結合模糊PID控制器構成一個電動車牽引控制系統的自優化模糊PID控制器,以最大牽引效率和最小功耗為目標,在各種試驗行駛之中自動優化各模糊規則之下的PID控制器參數。適應最佳控制法把最佳控制的極小值原理解法轉換為順向的循序演算機制,可以用來優化各類型的非線性控制器。本論文用電腦模擬驗證自優化模糊PID控制器可以在UDDS、HWFET與US06模式的試驗行駛之中自動優化各模糊規則的PID參數,對照傳統PID參數設定法的模糊PID控制器,在多種情境之下的模擬結果均顯示自優化模糊PID控制器能以較少功耗確保電動車在各種車速之下的牽引效率。
Electric vehicles propelled by in-wheel brushless DC motors have minimum power consumption in the power chain. However, traction control of the vehicle is subject to variable perturbations resulting from rolling resistance, aerodynamic drag and climbing resistance. The time-varying, nonlinear properties of perturbations frequently fail a fixed parameter linear controller to efficiently control traction forces under variable speed drive. Taking wheel speed and load torque as universes of discourse, this thesis proposes a self-optimizing fuzzy PID controller for the traction control system to optimize traction efficiency while minimizing power consumption. The fuzzy PID controller is a practice of the parallel distributed compensation method of the TSk fuzzy system. The self-optimization ability is achieved by including an adaptive optimal control algorithm into the fuzzy PID controller to draw out good rule PID parameters through reinforcement learning and sequential optimization. The objective of optimization is to minimize a cost function defined to maximize traction efficiency while minimizing power consumption. The proposed design is verified by testing in UDDS, HWFET and US06 test cycles. The results show that the self-optimizing fuzzy PID controller converges on good rule PID parameters rapidly. In contrast to tuning the PID parameters with conventional methods, a traction control system equipped with the self-optimizing fuzzy PID controller attains higher traction efficiency while consuming less electric power.