運用H∞ 強健演算法設計前輪操控車輛之控制器

Abstract

本論文主要是針對前輪操控之車輛 (front-wheeled steering vehicles) 再配合H∞強健控制理論的研究與探討。在前輪操控之二階動態方程式中，可以發現到一路面角度(lane angle)，此角度包含了方向盤角度 (wheel-handling angle) 與輪胎側滑角 (sideslip angle)。如此，車身響應，如在前輪(front wheels)，中心 (center of gravity, CG)與後輪 (rear wheels) 的實際側向位置 (lateral position)、速度(lateral velocity)與加速度(lateral acceleration)亦可個別地得到。 其次，車輛系統的輔助前輪轉向角與輔助駕駛者控制器兩種控制器亦被發現。前者是一合成轉向角控制器，當車輛行經過彎路段時，它能有效地抑制轉向角而完成路徑規劃。因此，就不會有轉向不足 (under-steer) 或過渡轉向 (over-steer)的情形發生。然而，此一合成控制器是無法從二階車輛模型直接求得的，必須藉由輪胎之側滑角和前軸角度概念而得到。為了消除後輪成份之角度，必須再提出另一輔助前輪角度控制器，此一控制器運用車身旋轉(yaw rate)的觀念可以得到。驗證之後確實地可以消除後輪成份。同時亦可將設計的角度控制器放入標準的車輛模擬軟體CarSim中分析並擷取其車輛行駛中的動畫來驗證。再者，車身之估測側滑角可從車身幾何而得到，亦可視為一角度模型。其模型是利用穩態的角度偏轉概念而算得的，此模型亦可運用到第三種車輛操控系統中進行分析。 輔助駕駛者行為之控制器是一前饋之操控控制器 (forward-looking steering controller)，的主要功用是替代駕駛者的行為，它亦可經由理論推導而被提出來。其所使用的方法有三種:1) 使用餘弦定律法可求得其控制器，但是用此方法，不幸地，所求得的控制器是一固定式的增益值而不是一可變式的值。需藉由特殊的前置濾波器(pre-filter)，它可輔助此控制器修正其輸出響應; 2)利用車身在前輪位置的側向加速度概念與配合一比例與積分 (proportional-integral) 補償器進行推導。同樣地，藉由此項方法推衍 3)第三種合成的控制器，主要成份是由比例與積分和比例與微分 (proportional-derivative) 補償器所組成的。 將車輛之側向位移訊號容入標準單一迴授訊號操控系統(unity feedback control steering system)，因此，此架構控制器可運用H∞ 演算法求的。接著，第一類型是將前面所述的前饋操控控制器和方向盤轉角或者是車身旋轉等訊號，其迴授訊號除了側向位移訊號外，同時應用在被提出的進階迴授操控系統上，再運用H∞ 強健演算法以便進行設計控制器。在第二類型的系統架構中，可以清楚地發現到方向盤命令由前輪滑差角(front-wheeled slip angle)與前輪側滑角(front-wheeled sideslip angle)所組成。第三種類型的控制架構涵蓋方向盤與車身中心側滑角，又增加一個方向盤操作增益(handling gain)。此架構亦可使用強健演算法進行分析。第四種類型提出的控制架構由一修正後的駕駛者訊號與前饋操控控制器組成之應用，此應用之演算法推導是基於前輪側向加速度的概念。最後，此四種控制系統都是考慮以路徑為輸入訊號。其路徑資訊取得是藉由CarSim來完成，此套裝軟體亦可以與MATLAB數學模擬軟體相互運作。合理的響應圖證明車輛沿著一需求的路徑能夠準確且平穩地完成規劃。

The dissertation mainly addresses the research of front-wheeled steering (FWS) vehicles and discussion of H∞ robust control algorithm. The lane angle involving wheel-handling and sideslip angles is found in 2-degree-of-freedom (2DOF) dynamics steering vehicle model. Furthermore, the reality results of lateral position, velocity and acceleration are computed respectively in the lateral direction of body at the front wheels, the center of gravity (CG) and the rear wheels. Then, two-type controllers of the auxiliary front-wheel steering angle and chauffeur are developed in the vehicle system. The former is a synthesized controller and it can restrain the angle component of front wheels when the vehicle travels along the curved and straight lane. Therefore, the turning situation in under-steer and over-steer area cannot be appearance. The synthesized controller can be obtained from the sideslip angle that it has the rear-wheel component and the front-axle concepts; however, it is unable to calculate immediately from 2DOF vehicle model. In order to eliminate the rear-wheel component, the front-axle angle controller considering the yaw rate concept is proposed. Indeed, the presented front-axle angle controller can remove the rear-wheel component via the theoretical analysis. Then, the created angle controller is set into a standard simulation tool, CarSim, and validated by the animation. Next, an estimated sideslip angle can be obtained from the geometry of body which can be regarded as a sub-model for which it is calculated by using the steady-state deflection angle concept. As well, this model can be analyzed in the third-type steering control structure. The auxiliary driver behavior controller is a forward-looking steering controller (FLSC) whose objective can substitute for a driver behavior and FLSC can be derived from the theoretic inference. Moreover, the FLSC with three methodologies: 1) uses the cosine law to achieve the FLSC design, unfortunately, the inferred consequence is a regular value but it is not an altered value. Moreover, a unusual pre-filter is needed to form the system performance; 2) utilizes the concept of the lateral acceleration at the front-wheel position and then adds a proportional-integral compensator to perform the FLSC. Similarly, 3) the third compound FLSC, which contains of a proportional-integral and a proportional-derivative compensators, can be derived. The signal of lateral displacement can be exerted in the unity feedback control steering system and consequently the designed controller can be obtained by using H∞ algorithm. Furthermore, the first-type is that the foregoing FLSC and, besides the lateral displacement signal, the signal of the wheel-handling or the yaw rate is involved in the advanced feedback steering system and then the offered controller is manipulated by H∞ algorithm. Obviously, the second-type is that the command of the wheel-handling consists of the front-wheeled slip and sideslip angles in the advanced feedback steering system. The third-type contains of the commands of the wheel-handling, the handling gain and the sideslip angle located at the CG. This system structure can be analyzed by using the H∞ robust algorithm. The forth-type presents an integral control structure is comprised of a modified driver command and the FLSC which the application is described by the formulated of the front-wheel lateral acceleration. Ultimately, all four-type control structures have a common input trajectory: Target path. The input track information can be represented by CarSim for which it can work together with MATLAB. The reasonable performance demonstrates follow a desired target path both accurately and stably under the variations of mass, forward velocity and road-tire contact.