改良式蝙蝠演算法應用於三動力複合動力車之最佳能源管理控制器開發

Abstract

本研究旨在開發以改良式蝙蝠演算法（Improve Bat Algorithm, IBA）嵌入數位信號處理器DSP TMS320F28335用於三動力複合動力車的能量管理系統。並且真正應用硬體嵌入式系統(Hardware-In-The-Loop, HIL)進行即時(Real-Time)驗證驗算法之可行性。本研究中，整車系統採用三動力源複合動力車，其中包括內燃機引擎、高功率馬達與一體式啟動發電機。在能量管理系統中以HIL進行評估使用改良式蝙蝠演算法（IBA）之三動力源複合動力車系統能量管理與DSP TMS320F28335控制器整合。改良式蝙蝠演算法能量管理控制中，主要有三個步驟進行優化：（1）頻率、（2）脈衝率、（3）響度。總疊代次數為30次，共有100隻蝙蝠進行最佳能量管理。IBA1改良採用人工蜂群演算法的輪盤法；而IBA2使用細菌覓食演算法中的驅散與複製作為改良。IBA與兩控制策略進行FTP-72行車型態之油耗比較：(1)規則庫管理(Rule base):有五種控制模式並根據工程師經驗設定模式切換時機；(2)最小等效油耗策略:搭配全域搜尋將範圍內所有的可能解進行尋找，找出最小油耗時之動力分配比。最後透過HIL模擬IBA於車輛控制單元(Vehicle Control Unit, VCU)之可行性與油耗效益驗證。基本規則庫、ECMS、IBA、Real-time，這五種情狀況在行車型態FTP-72下的等效油耗：[693.3 g、445.9 g、438.3 g、425.1 g、388.6 g]，WLTC等效油耗：[1120 g、863.7 g、834.4 g、818.7 g、753.6 g]， BA、IBA1、IBA2三種狀況與基本規則庫相比在FTP-72的能耗改善百分比是[35.6 %、36.7 %、38.6 %]，WLTC下運行之能耗改善百分比是[22.8 %、25.5 %、26.9 %]。其中將BA、IBA1、IBA2在離線和Real-time兩者在兩個行車型態中等效油耗改善度有高達92-98%的相似度。未來將會實施於真實之三動力源複合動力車輛。This article mainly developed the Improved Bat Algorithm (IBA) and designed the digital signal processor (DSP): TMS320F28335 for the energy management system of the three-power hybrid vehicle as well as the application of the HIL (Hardware-In-the-Loop) system. The real-time environment was to verify the feasibility of the algorithm. In this study, three power sources were included in the hybrid powertrain where the internal combustion engines, the high-power motor and the integrated starter generator were modeled. The HIL was used to evaluate the IBA for the on-line energy management of three-power hybrid powertrain, and the DSP TMS320F28335 controller was integrated. For energy management system, three steps were conducted for IBA: (1) frequency, (2) pulse rate, and (3)loudness. The total number of iterations was 30, and there are 100 bats for optimization. The proposed IBA1 used the dispersal and replication process from the bacterial foraging algorithm; while the IBA2 uses the roulette method of the artificial bee colony algorithm. The fuel consumption of IBAs were compared in the FTP-72 with two control strategies: (1) rule base (RB) control: five control modes and the timing of mode switching were set according to the engineering experience; (2) ECMS (Equivalent Consumption Minimization Strategy): using Global Search Algorithm (GSA) to search for all possible solutions within the ranges to find the power distribution ratio with minimum equivalent fuel consumption. Finally, the feasibility and fuel consumption efficiency of IBAs in the Vehicle Control Unit (VCU) were verified through HIL simulation.The results of equivalent fuel consumption for case of RB, ECMS, IBAs and real-time under FTP-72 driving cycle were: [693.3 g, 445.9 g, 438.3 g, 425.1 g, 388.6 g], while the results for WLTC were: [ 1120 g, 863.7 g, 834.4 g, 818.7 g, 753.6 g]. Compared with the RB case, the improvement percentages in energy consumption of BA, IBA1, and IBA2 under FTP-72 were [35.6 %, 36.7%, 38.6 %], the improvement percentages under WLTC operation were [22.8%, 25.5%, 26.9%]. Among them, cases of BA、IBA1、IBA2 in offline and real-time simulation have a similarity of 92-98% in the improvement of equivalent fuel consumption. In the future, it will be implemented in real three-power source hybrid vehicles.