載重車輛的即時狀態估測與參數識別

Abstract

本研究旨在探討狀態估測以及參數識別對於車輛模型的重要性，強調其在加強先進駕駛輔助系統 (ADAS) 性能方面的作用。透過回顧相關文獻中即時估測的方法，包括卡爾曼濾波及其最新進展，本研究評估了這些方法的適用性、優點和局限性。研究中所提出演算法利用一個四自由度的車輛模型，結合Magic Formula輪胎模型，並採用了離散化空間狀態模型來進行狀態與參數估測。由於擴展卡爾曼濾波(EKF)具計算效率佳和適用於複雜系統等優點，因此以雙擴展卡爾曼濾波(DEKF)為框架設計估測演算法。此基於雙擴展卡爾曼濾波的估測方法於MATLAB/ Simulink 環境中進行模擬。該研究模擬使用 dSPACE ASM 模擬軟體，以多體虛擬車輛替代實車測試。本篇研究中詳盡說明測試案例、感測器規格和過程中使用的模擬工況，以及它們與研究目標的相關性。模擬結果顯示，所提出的狀態與參數識別方法能夠改善車輛在不同情況下的狀態估測精度。

This research investigates the significance of state estimation, with a specific focus on parameter identification techniques, emphasizing its use in Advanced Driver Assistance Systems (ADAS). Through a comprehensive review of real-time estimation methods, including Kalman filtering and recent advancements, assessing their applicability, advantages, and limitations in relevant literature. The proposed approach utilizes a 4-degree-of-freedom vehicle model, incorporating the tire magic formula model and employing discretization techniques. The Dual Extended Kalman Filter (DEKF) framework is proposed as the primary estimation method, and is implemented using MATLAB/Simulink environment. The research justifies the use of dSPACE ASM as a surrogate for real-world testing, detailing test cases, sensor specifications, maneuvers, and their relevance to research objectives. Results demonstrate efficacy of the proposed methodology in parameter identification and its enhancing effect on vehicle state estimation accuracy across various maneuvers.