人工蜂群演算法應用於三電力電動車系統之最佳能量管理
No Thumbnail Available
Date
2018
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
本研究旨在開發人工蜂群演算法（Artificial Bee Colony Algorithm, ABC）用於三電力電動車輛的能量管理系統，並且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。使用HIL進行評估並以人工蜂群演算法(ABC)之三電力電動車輛的能量管理系統策略控制。車輛子系統包括110 kW燃料電池、192 kW馬達、32 kW超級電容和50 kW-h鋰電池，車重1,370 kg。在能量管理系統ABC控制中，主要有三個步驟（1）工蜂階段、（2）觀察蜂階段、（3）偵察蜂階段。總疊代次數為30次，共有50隻工蜂進行最佳能量管理。
ABC與兩種控制策略進行NEDC與FTP-72行車型態之油耗比較：(1)規則庫管理(Rule base):有四種控制模式(純鋰電模式、混合電力模式、延距模式及超級電容輔助模式)，根據經驗設定模式切換時機；(2)最小等效能耗策略（Equivalent Consumption Minimization Strategy, ECMS）:搭配全域搜尋(Global Search Algorithm, GSA)將範圍內所有的可能解進行尋找，找出最小耗氫之電力分配比。最後透過HIL模擬ABC於車輛控制單元(Vehicle Control Unit, VCU) 即時模擬之可行性與油耗效益驗證。基本規則庫、ECMS、ABC及Real-time，這四種情狀況在NEDC的等效耗氫分別為：[1177 g、667g、665.7 g、375.3 g]，FTP-72等效耗氫分別為：[1402 g、808.7 g、806.6 g、429.2 g]。其中，ECMS、ABC、Real-time三種狀況與基本規則庫相較下NEDC的能耗改善分別為[43.3 %、43.4 %、68.1 %]，FTP-72下運行之能耗改善分別為[42.3 %、42.5 %、69.4 %]。；ABC與Real-time兩者在兩個行車型態中總耗氫量改善度有高達99.7%的相似度，皆僅次於ECMS最佳解。未來將可實施於真實之三電力源複合電能車輛。
The purpose of this study is to develop the artificial bee colony algorithm (ABC) by applying it to the energy management strategy system of a three-energy-source hybrid powertrain. Furthermore, this study was practical in nature, as it used the real-time simulation Hardware-in-the-Loop (HIL) to verify the algorithm’s feasibility. This study employs HIL to assess the influence that using ABC will have on the energy management strategy control of a three-energy-source hybrid powertrain. The vehicle weighs 1,370 kilograms and its subsystems include a 110kW fuel cell, 192kW motor, 32kW supercapacitor, and a 50kW-h lithium battery. There are three primary steps for the energy management system and ABC energy management control: 1) employee bee phase, 2) onlooker bee phase, and 3) scout bee phase. The overall number of iterations was 30, and 50 bees were used carry out optimal energy management. ABC and two control strategies were used to carry out a comparison of fuel consumption with the NEDC (New European Driving Cycle) and FTP-72 (Federal Testing Procedure) driving pattern. 1) Rule-based management: There are five control modes, which are system preparation, battery charging mode, electric mode, hybrid power mode, and extended range mode; the engineer used his experience to determine when to set and change modes. 2) Equivalent consumption minimization strategy (ECMS): By incorporating the global search algorithm (GSA), we searched for all the scope’s possibilities in order to find the most minimal fuel consumption for power distribution ratio strategy. At the end of the study, we used HIL to simulate the feasibility and verify fuel consumption benefits of ABC on vehicle control units (VCU) in real time. A basic rule base, ECMS, ABC, and real-time were the four conditions for the equivalent consumption with the NEDC driving pattern: 1177g, 667g, 665.7g, and 375.3g were their respective values. The equivalent consumption values with a FTP-72 driving cycle were 1402g, 808.7g, 806.6g, and 429.2g. ECMS, ABC, and real-time were compared with a basic rule base when using a NEDC driving pattern to determine percentage values for improvement in energy consumption: 43.3%, 43.4%, and 68.1%. Percentage values for improvement in energy consumption for a FTP-72 driving cycle were 42.3%, 42.5%, and 69.4%. The improvement in equivalent hydrogen consumption values for ABC and real-time for the NEDC driving pattern and FTP-72 driving cycle were 99.7% similar, and they were only outperformed by ECMS, which was the optimal solution. In the future, this experiment will be used to test a three-energy-source hybrid-powered vehicle.
The purpose of this study is to develop the artificial bee colony algorithm (ABC) by applying it to the energy management strategy system of a three-energy-source hybrid powertrain. Furthermore, this study was practical in nature, as it used the real-time simulation Hardware-in-the-Loop (HIL) to verify the algorithm’s feasibility. This study employs HIL to assess the influence that using ABC will have on the energy management strategy control of a three-energy-source hybrid powertrain. The vehicle weighs 1,370 kilograms and its subsystems include a 110kW fuel cell, 192kW motor, 32kW supercapacitor, and a 50kW-h lithium battery. There are three primary steps for the energy management system and ABC energy management control: 1) employee bee phase, 2) onlooker bee phase, and 3) scout bee phase. The overall number of iterations was 30, and 50 bees were used carry out optimal energy management. ABC and two control strategies were used to carry out a comparison of fuel consumption with the NEDC (New European Driving Cycle) and FTP-72 (Federal Testing Procedure) driving pattern. 1) Rule-based management: There are five control modes, which are system preparation, battery charging mode, electric mode, hybrid power mode, and extended range mode; the engineer used his experience to determine when to set and change modes. 2) Equivalent consumption minimization strategy (ECMS): By incorporating the global search algorithm (GSA), we searched for all the scope’s possibilities in order to find the most minimal fuel consumption for power distribution ratio strategy. At the end of the study, we used HIL to simulate the feasibility and verify fuel consumption benefits of ABC on vehicle control units (VCU) in real time. A basic rule base, ECMS, ABC, and real-time were the four conditions for the equivalent consumption with the NEDC driving pattern: 1177g, 667g, 665.7g, and 375.3g were their respective values. The equivalent consumption values with a FTP-72 driving cycle were 1402g, 808.7g, 806.6g, and 429.2g. ECMS, ABC, and real-time were compared with a basic rule base when using a NEDC driving pattern to determine percentage values for improvement in energy consumption: 43.3%, 43.4%, and 68.1%. Percentage values for improvement in energy consumption for a FTP-72 driving cycle were 42.3%, 42.5%, and 69.4%. The improvement in equivalent hydrogen consumption values for ABC and real-time for the NEDC driving pattern and FTP-72 driving cycle were 99.7% similar, and they were only outperformed by ECMS, which was the optimal solution. In the future, this experiment will be used to test a three-energy-source hybrid-powered vehicle.
Description
Keywords
人工蜂群演算法, 規則庫管理, 最小等效能耗策略, 混合電力, artificial bee colony algorithm, rule-based management, equivalent consumption minimization strategy, hybrid power