灰狼演算法應用於複合電力電動車輛系統之控制器設計

Control Unit Design Using Gray Wolf Algorithm for a Multiple-Electric-Energy Vehicle

戴仲瑜(Tai Chung-Yu)

指導教授 ： 洪翊軒

國立臺灣師範大學/科技與工程學院/工業教育學系/碩士(2020年)

https://doi.org/10.6345/NTNU202000879

摘要

本研究應用灰狼演算法(Gray Wolf Algorithm, GWA)進行複合電力車之電力系統控制器設計。選定Tesla Model 3作為目標車輛，並透過動力輸出馬達、傳動系統、目標車輛動態參數、駕駛行為參數，將目標車輛數學化。並透過鋰電池模型、燃料電池模型與超級電容模型，建立一含三電力源之複合電力電動車模型。基於目標車輛與標準行車型態，根據不同控制策略：規則庫控制(Rule-based)、最小等效油耗控制策略(Equivalent Consumption Minimization Strategy, ECMS)、人工蜂群演算法(Artificial Bee Colony Algorithm, ABC)、GWA進行電力系統控制器設計。分析控制策略於行車型態下所消耗之電力進行比較，並使用快速雛型控制器(Rapid Prototyping Controller)進行即時(Real time, RT)控制，測試即時控制策略於實際車輛之可行性。ECMS、ABC、GWA於NEDC與規則庫控制之電力消耗比較，電力消耗改善為[33.8%、25.8%、32.5%]，於FTP-72改善為[32.5%、25.1%、30.2%]，灰狼演算法有較佳改善。GWA、GWA(RT)、GWA(HIL)電力消耗累積於NEDC下為[4270仟焦、4430仟焦、4483仟焦]，FTP-72為[5183仟焦、5197仟焦、5251仟焦]，其具有高度相似，未來可應用於實際車輛上。

關鍵字

控制器設計 ； 灰狼演算法 ； 規則庫管理 ； 人工蜂群演算法 ； 複合電力

並列摘要

To design the power system of an electric vehicle control unit, we used the gray wolf algorithm(GWA). According to the target vehicle-Tesla model 3, we built the traction motor, transmission system, target vehicle dynamics, driving behavior parameters to formulate a target vehicle simulation system. Then, we built the electric power source system with the lithium battery model, fuel cell model and supercapacitor model. The target vehicle simulation system and the electric power source system were combined to form a multiple-electric energy vehicle system and the vehicle control unit was employed to calculate the power consumption. Based on the target vehicle and standardized driving cycles, we adopted different control strategies such as rule-based control, ECMS, artificial bee colony algorithm (ABC), gray wolf algorithm (GWA) to design the power system control unit. By analyzing the power consumed of each algorithm under different driving cycles for comparison. We used a rapid prototype controller on the real-time control platform to test the feasibility of the control strategy in an actual vehicle.
Comparing the power consumption of ECMS, ABC, GWA in NEDC driving cycles with the rule base control, the power consumption is improved by [33.8%, 25.8%, 32.5%]. The improvement for the FTP-72 case is [32.5%, 25.1%, 30.2%], which shows respectively that the gray wolf algorithm has better power consumption improvement. Comparing the power consumption of GWA, GWA(RT) and GWA(HIL) in NEDC driving cycle, the power consumption is [4272(kJ), 4430(kJ)], and [5183(kJ), 5197(kJ), 5251(kJ)] in FTP-72 driving cycle respectively. It has a very high similarity compared with real-time control, which can be used on the actual electric vehicle in the future.

並列關鍵字

control unit design ； gray wolf algorithm ； rule-based management ； artificial bee colony algorithm ； multiple electric power

參考文獻

[1] Bp plc, BP Statistical Review of World Energy, UK: British Petroleum, 2017.

[2] 蔡鴻德、楊鍇行、黃星富， “空氣汙染防制新作為”，國土及公共治理季刊，第5卷，第3期，2017。

[3] S. M. Lukic and A. Emadi, “Effects of drivetrain hybridization on fuel economy and dynamic performance of parallel hybrid electric vehicles,” IEEE Trans. Veh. Tech., vol. 53, no. 2, pp. 385-389, 2004.

[4] H. YOO, S. K. SUL, Y. Park, J. Jeong, “System integration and power-flow management for a series hybrid electric vehicle using supercapacitors and batteries,” IEEE Transaction on Industry Applications, vol. 44, pp. 108-114, 2008.

[5] Y. Tang, W. Yuan, Ma. Pan, Z. Wan, “Experimental investigation on the dynamic performance of a hybrid PEM fuel cell/battery system for lightweight electric vehicle application,” Applied Energy, vol. 88, pp. 68-76, 2011.