林巍聳臺灣大學:電機工程學研究所曲先平Chu, Hsien-PingHsien-PingChu2010-07-012018-07-062010-07-012018-07-062009U0001-1606200905223200http://ntur.lib.ntu.edu.tw//handle/246246/188084本研究針對線傳控制汽車發展一套自評煞車控制系統，目標是優化車輛的牽引力和循跡控制性能。該控制系統以雙啟發規劃法(dual heuristic dynamic programming)為基礎，搭配自組型徑向基底函數網路(radial basis function network)和梅基克式(Magic Formula)輪胎模型，達成用單一控制器整合防鎖控制、牽引力控制、煞車力分配和循跡控制的功能，使車輛能夠在自評過程中自動優化煞車控制器的性能。自評煞車控制系統具有自動追求優化的能力，能夠自動調整非線性控制器的參數，自組型徑向基底函數網路設有自動增減類神經元的訓練法則，可以自動組合最適用的網路結構。本研究完成自評煞車控制系統的各項設計，並且建立該控制器和汽車的整合模擬系統。藉由模擬系統檢驗自評煞車控制系統的性能表現，結果顯示在各種測試條件下，自評煞車控制系統都可以有效的達成優化車輛性能的目標。This research develops an adaptive critic braking control system to optimize the traction and yaw control performance of by-wire braking cars. Underlying techniques are mainly the dual heuristic programming adaptive critic design, the self-organizing radial basis function network, and the Magic-Formula tire model. The adaptive critic braking control system is characterized by integrating the functions of antilock brake control, traction control, electronic brake-force distribution, and electronic stability program. The generalized growing-pruning mechanism of radial basis function networks and the adaptive critic design enable the system pursuing optimal performance. This thesis shows the detailed design of the adaptive critic braking control system. A simulation system involving an adaptive critic braking controller and a prototype car is developed. Simulation results show that, under variable test conditions, adaptive critic braking control system is significantly effective in optimizing traction and yaw performance.摘要......................................................iBSTRACT................................................iiihapter 1 .................................................1ntroduction..............................................1.1 Brake-by-wire technology..............................1.2 Adaptive Critic Design (ACD)..........................7.3 Motivation and Contribution...........................9.4 Organization of this thesis..........................10hapter 2 ................................................11ehicle Active Safety Systems............................11.1 Antilock Brake System (ABS)..........................11.2 Brake Assistant System (BAS).........................15.3 Electronic Brake-force Distribution (EBD)............18.4 Traction Control System (TCS)........................20.5 Electronic Stability Program (ESP)...................22.6 Summary..............................................23hapter 3 ................................................25daptive Critic Design and Dual Heuristic Programming....25.1 Adaptive Critic Design (ACD).........................25.1.1 Preliminary of Mathematical Representation.........26.1.2 Overview of Adaptive Critic Design.................27.1.3 Mathematical Formulation of ACD....................31.1.4 Categories of ACD..................................33.2 Dual Heuristic Programming...........................37.2.1 Basic Concept of DHP...............................37.2.2 DHP Update Process.................................38.2.3 Summary of DHP Procedure...........................41.3 Radial Basis Function Network........................42.3.1 MLP and Batch Training.............................42.3.2 Online Incremental Learning RBFN...................44.3.3 GGP-RBFN...........................................46.4 Simulation Results of MLP and GGP-RBFN...............50.4.1 Simulation Results without Training................56.4.2 Simulation Results with Complete Training..........67.4.3 Simulation Results with Partial Training: Case I...77.4.4 Simulation Results with Partial Training: Case II..87.4.5 Simulation Results with Different Neuron Width and Discount Factor..........................................97.5 Summary.............................................104hapter 4 ...............................................107ehicle Dynamic Model and Tire Model....................107.1 Vehicle Dynamic Model...............................107.1.1 Vehicle Axis System...............................107.1.2 Load Transfer.....................................109.1.3 Equations of Motion...............................111.1.3 Ackermann Principle...............................113.2 Tire model..........................................115.2.1 Categories of Tire Models.........................116.2.2 The Magic Formula Tire Model......................118.3 Reference States and Control Inputs.................130hapter 5 ...............................................135daptive Critic By-wire Braking Control System Design...135.1 Architecture and Updating Process of Complete System..................................................135.1.1 Architecture......................................135.1.2 Updating Process..................................137.2 Simulation Studies..................................137.2.1 Parameters........................................138.2.3 Training Process..................................141.2.4 Performance of Acceleration.......................149.2.5 Performance of Brake..............................152.2.6 Brake on Split Friction Coefficient Road Surface..155.2.7 Performance of Double Lane Change.................160hapter 6 ...............................................165onclusion..............................................165eferences..............................................1673031157 bytesapplication/pdfen-US雙啟發規劃法車輛控制類神經網路自評控制輪胎模型dual heuristic programmingvehicle controltire modelneural networkthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/188084/1/ntu-98-R95921001-1.pdf