附 錄 附錄 A 外文文獻(xiàn) ELECTRONIC STABILITY PROGRAM Feedback control of the vehicle motion is possible by extending the traction control system with four additional sensors: steering wheel angle,brake pressure, yaw rate and lateral acceleration. Since the nominal trajectory desired by the driver is unknown, the drivers inputs are taken to obtain nominal state variables that describe the intended vehicle motion instead. These inputs are the steering wheel angle, the engine drive torque as derived from the accelerator pedal position and the brake pressure. The handling performance of the car can be improved if in dependence of the steering wheel angle the yaw moment on the car can be controlled.The main task of ESP as an active safety system is, however, to limit the slip angle of the vehicle in order to prevent vehicle spin. ESP can control the yaw moment on the car by controlling the value of the slip at each wheel. This can be shown by the influence of some brake slip value 0 at the left front tire of a free rolling car in a right turn (Fig. 1). )0(FR = is the lateral force on the free rolling tire. Because of the brake slip 0 the lateral force will be reduced to )(F0S where it is assumed, that neither the normal force NF nor the tire slip angle 0 are changed. As a result of the brake slip the brake force )(F0B is generated. )(F0R is the resultant force on the tire, which is the vectorial sum of )(F 0S and )(F 0B . If the tire friction limit is reached, the magnitudes of )0(FR = and )(F 0R are approximately equal. Fig. 1 Yaw moment change by slip control The influence of brake slip is now obvious: a change in the brake slip value results in a rotation of the resultant force on the tire. As a result of the rotation the yaw moment on the car is changed.However, simultaneously the lateral force and thelongitudinal force on the car are influenced. The control concept determines by what amount the slip at each tire shall be changed to generate the required change in the yaw moment. Usually it is required that the driver must not have the impression that with ESP the car is slower than without ESP. The vehicle dynamics controller part of ESP (Fig. 2) constitutes the upper part of a hierarchical control. Output are the nominal tire slips Noi In the lower part the slip values of the tires are controlled. The vehicle dynamics controller part consists of several processing blocks. On the top left the motion desired by the driver is derived from his inputs by a linear bicycle model (which uses a linear relationship between the slip angle and the lateral force of the tire). On the top right the motion of the car is measured and missing state variables are estimated. Fig. 2 Simplified block diagram of the ESP control This estimate is valid if the pitch and roll angles of the car are neglected and furthermore, if the car moves on a horizontal plane. In this equation ya is the lateral acceleration of the car and xa is its longitudinal acceleration, V is its velocity and is its yaw velocity. If the car velocity is constant and its slip angle is small then the estimate can be readily obtained by a simple time integration Offset and other errors in the sensor and estimated signals may quickly lead to large errors in the estimate. Furthermore, during full braking the car deceleration can not be neglected. Therefore,during full braking an alternative estimate of the slip angle based on an observer is used. The observer is based on a full four wheel model of the car and uses two dynamic equations, one for the yaw velocity and the other for the lateral velocity of the car. These equations are rearranged and discretized to be used as the model for a Kalman filter. Since the yaw velocity is measured, the solution of the differential equation of the yaw velocity is used to derive the measurement equation. Here pc denotes a known brake constant, whlp denotes the brake fluid pressure in the brake wheel cylinder, R denotes the known tire radius, CaHalfM denotes half of the engine torque at the axle, whlJ denotes the known moment of inertia of the wheel about its axis of rotation and whlV denotes the wheel speed which is the product of the wheel angular velocity and the tire radius.The engine torque value can be obtained from the engine management system, while the rotational wheel velocity is measured by the wheel speed sensor. Finally by modeling the hydraulic unit the wheel brake pressure is estimated at each wheel. The side forces are not readily available.Therefore a tire model is used. Specifically, the HSRI tire model is used which allows for a simple relation between the lateral and the longitudinal force. The estimate of the lateral velocity by the Kalman filter is robust to tire changes as only the ratio of the lateral and longitudinal tire stiffness is used.For winter tires the ratio is nearly the same as for summer tires. The same is true for new and worn tires, conventional and wide tires etc. Thus both evaluations of the slip angle are more or less insensitive to changes in the tire properties. Unfortunately the vehicle slip angle estimation is not always sufficiently accurate and the confidence level of its value is sometimes low. Therefore, the vehicle dynamics controller uses additionally a model following control for the yaw velocity of the car, for which the already mentioned linear bicycle model is taken. Output of the linear bicycle model is the nominal value of the yaw rate No. Thus a first value for the nominal yaw velocity No is obtained (Fig. 3). The wheel base l is a simple geometric paameter while the vehicle forward velocity xV is estimated by the brake slip controller. Fig. 3 Nominal yaw velocity from the linear bicycle model The characteristic speed chV depends mainly on the lateral tire stiffness C of the tires. Therefore, the nominal yaw velocity changes with the tire type, make and state (new or worn). This change may occur suddenly if new tires are mounted. The model following control is thus sensitive to changes in the tire stiffness and ESP may suddenly change its behavior. This will be shown below. ESP must therefore be checked to correctly perform with all released tires. Since the lateral acceleration of the car can not exceed the maximum coefficient of friction between the tire and the road , the nominal yaw velocity must be limited to a second value by the following relation (see the hyperbola in Fig. 3). VNo g/V For summer tires the nominal yaw velocity is different from that of winter tires (Fig. 4). Similarly,for worn tires the yaw velocity is different from that of new tires. The vehicle becomes oversteer if on the front axle worn and on the rear axle new tires are mounted (Fig. 5). In such cases the vehicle behavior deviates significantly from the behavior of the linear bicycle model (Fig. 3) and ESP interventions can be expected for vehicle maneuvers which are well within the physical limit. Fig. 4 Nominal yaw velocity from the full four wheelmodel with nonlinear new and worn summer and wintertires (steering wheel angle 60) Fig. 5 Nominal yaw velocity from the full four wheel model with nonlinear summer and winter tires, with worn tires at the front axle and new tires at the rear axle(steering wheel angle 60) A first nominal limit value for the slip angle of the car (Fig. 5) is chosen as discussed using the Beta method in dependence of the coefficient of friction between the tires and the road. This value is reduced in dependence of the velocity of the car to a second value No, in order to improve the support for the driver at higher speeds. If the state of the car as described by its yaw velocity and its slip angle differs from its nominal state, then the vehicle dynamics controller checks if this difference is within some tolerable dead zone. If not, a yaw moment is generated to reduce this difference to within this tolerable dead zone. 附錄 B 外文文獻(xiàn)中文翻譯 通過(guò)在牽引力控制系統(tǒng)上擴(kuò)展方向盤(pán)轉(zhuǎn)角、制動(dòng)壓力、橫擺角速度和側(cè)向加速度四個(gè)傳感器，就可以實(shí)現(xiàn)對(duì)車輛運(yùn)動(dòng)的反饋控制。 由于駕駛員所希望的名義軌跡是未知的，需要采集駕駛員的輸入變量來(lái)獲得能描述期望車輛運(yùn)動(dòng)的名義狀態(tài)變量。這些輸入變量包括方向盤(pán)轉(zhuǎn)角、通過(guò)加速踏板獲得的發(fā)動(dòng)機(jī)驅(qū)動(dòng)轉(zhuǎn)矩和制動(dòng)壓力。 如果汽車獨(dú)立于方向盤(pán)轉(zhuǎn)角的橫擺運(yùn)動(dòng)得到控制，汽車的操縱 性能就會(huì)得到提升。然而， ESP 作為主動(dòng)安全系統(tǒng)，其主要任務(wù)是限制車輛的知心側(cè)偏角 來(lái)防止車輛側(cè)翻。 圖 1 由側(cè)偏角控制引起的橫擺運(yùn)動(dòng) ESP 能通過(guò)控制每個(gè)車輪上的側(cè)偏角的值來(lái)控制汽車的橫擺運(yùn)動(dòng)。在向右轉(zhuǎn)向的自由滾動(dòng)的汽車上，左前輪的制動(dòng)側(cè)偏角0的作用可以說(shuō)明這一點(diǎn) ，如圖 1所示 。 )0(FR = 為作用在自由滾動(dòng)輪胎上的側(cè)向力。由于制動(dòng)側(cè)偏角0， 側(cè)向力會(huì)減小到假定值 )(F0S，法向力NF和輪胎側(cè)偏角0都不變。由于制動(dòng)側(cè)偏，車輛產(chǎn)生了制動(dòng)力 )(F0B。 )(F0R是輪胎上的縱向力，是 )(F0S和 )(F0B的矢量和。如果到達(dá)輪胎的摩擦極限， )0(FR = 和 )(F0R的值近似相等。 現(xiàn)在，制動(dòng)側(cè)偏角 的作用很明顯：制動(dòng)側(cè)偏角的變化會(huì)造成輪胎上合力的旋轉(zhuǎn)。由于該旋轉(zhuǎn)，汽車的橫擺運(yùn)動(dòng)發(fā)生變化。但與此同時(shí)，汽車上的側(cè)向力和縱向里也會(huì)受到影響?？刂圃砣Q于每個(gè)輪胎上的側(cè)偏角需要變化多大才能產(chǎn)生期望的橫擺運(yùn)動(dòng)的變化。通常駕駛員不能有這樣的想法：裝配有 ESP 的汽車比沒(méi)裝配的要慢。 圖 2 ESP 控制的簡(jiǎn)單框圖 ESP 系統(tǒng)的車輛動(dòng)力學(xué)控制器部分組成分層控制的上層部分 ，如圖 9 所示 。輸出是輪胎側(cè)偏角Noi。在下層部分控制輪胎側(cè)偏角。車輛動(dòng)力學(xué)控制器部分包括一些過(guò)程模塊。在左上方，駕駛員期望的運(yùn)動(dòng)通過(guò)現(xiàn)行車輛模型由他的輸入得到 (該模型使用車輪側(cè)偏角和側(cè)向力的線性關(guān)系 )。在右上方，測(cè)量車輛的運(yùn)動(dòng)并估計(jì)實(shí)際的狀態(tài)變量。 第一種估計(jì)車輛側(cè)偏角方法用到側(cè)偏角的導(dǎo)出公式： ）（ s ina-co saV1= xyV+ 如果忽略汽車的前傾角和搖擺角，并且如果汽車在水平面上行駛，這個(gè)估計(jì)就是合理的。在這個(gè)等式中，ya是汽車的側(cè)向加速度 ， xa 是縱向加速度，V是車速， 是橫擺角速度。如果車速是常量并且側(cè)偏角較小，可以通過(guò)對(duì)時(shí)間積分很容易地得到估計(jì)值。 傳感器的補(bǔ)償和其他誤差以及估算信號(hào)可能會(huì)很快的導(dǎo)致估計(jì)中很大的偏差。另外，在全速制動(dòng)過(guò)程中，汽車的減速度不能被忽略。因此，在全速制動(dòng)過(guò)程中，需要用到另外一個(gè)基于監(jiān)測(cè)器的側(cè)偏角估計(jì)量。 該監(jiān)測(cè)器建立在汽車四輪模型基礎(chǔ)上，使用兩個(gè)等式：一個(gè)是橫擺角速度，另一個(gè)是汽車的側(cè) 向速度。這些等式被重新整理和離散化用作卡爾曼濾波器模型。由于橫擺角速度是測(cè)量的，該微分方程的解用來(lái)推導(dǎo)估計(jì)等式。 這些等式中都需要每個(gè)輪胎上的縱向加速度 BF ， BF 可以通過(guò)下面的等式估算出來(lái)。 w h l2w h lC a H a l fw h lpB Vdtd*RJRM-Rp*c=F + 其中，pc指一個(gè)已知的制動(dòng)常數(shù)，whlp指的是制動(dòng)缸內(nèi)的制動(dòng)液壓力， R 代表已知的輪胎半徑，CaHalfM指發(fā)動(dòng)機(jī)在車軸上轉(zhuǎn)矩的一半，whlJ指車輪相對(duì)于其轉(zhuǎn)動(dòng)軸線的轉(zhuǎn)動(dòng)慣量，whlV車輪速度即車輪角速度和輪胎半徑的乘積。發(fā)動(dòng)機(jī)轉(zhuǎn)矩可以通過(guò)發(fā)動(dòng)機(jī)管理系統(tǒng)得到，而車輪轉(zhuǎn)速是通過(guò)輪速傳感器獲得的。最后通過(guò)液壓?jiǎn)卧哪Ｐ凸浪忝總€(gè)車輪上的制動(dòng)壓力。 側(cè)向力不是直接就能用的，需要一個(gè)輪胎模型。特別地，要用到文獻(xiàn) 15中描述的 HSRI