Research On State Estimation And Control Of Multi‐axle Steering Vehicle

The multi‐axle vehicle needs multi‐axle steering, and steering control at high speed is one of the key technologies in the field of active safety control of multi‐axle steering vehicle. Due to steering of many axles, wheels of the axles are easy to produce movement interference each other and movement resistance, and tires are easy to wear out. When relations of steering angles satisfy kinematic principles, wheels of all axles make rolling motion purely. Pure Rolling of all wheels when steering can reduce the movement interference and resistance and tire wear, which can decrease application cost and fuel consumption, and improve economy, to multi‐axle steering vehicle. Steering control when driving at high speed can obtain execllent response characteristics of yaw rate and side slip angle of mass center and improve steering performance, which is beneficial to driving assist and alleviation of driver nervousness and improvement of handling stability and driving safety.When studying the control with state feedback of multi‐axle steering vehicle, because of existence of system and measurement disturbance and some states that can not be measured, states of vehicle need to be estimated. The axle loads of multi‐axle steering vehicle are large, and dynamic behavior of tires is easy to enter the nonlinear region, and there exists various random disturbance, so research on Bayesian statistical filtering based on nonlinearity and random disturbance to estimate states plays important significance to guarantee control implementation smoothly and control performance.Because loading capacity of multi‐axle steering vehicle is large, loading of different goods leads to large variation of vehicle parameters easily and causes large uncertainty of vehicle model. Based on comprehensive consideration of the influencing factors of tire nonlinearity, model uncertainty and external disturbance, it is necessary to study the related control to improve robustness of control system to multi‐axle steering vehicle.Control of tracking yaw rate can make multi‐axle steering vehicle obtain the referenced steering performance. Tracking control without static error based on internal model principle can make tracking static errors become zeros and make tracking robustness be improved. Tracking control without static error plays important roles to improve tracking characteristics and steering performance to multi‐axle steering vehicle.Fuzzy T‐S model is more precise than linear model. Robust control algorithms are nearly based on linear models. Robust control algorithms can be applied to local linear models of fuzzy T‐S model and local robust controllers can be obtain, then total steering robust controller can be derived using local robust controllers and fuzzy T‐S inference. It plays an important role to combine Fuzzy T‐S model with robust control in improving robustness of steering control system and steering performance to multi‐axle steering vehicle.In short, research on state estimation based on Bayesian statistical filtering and control of tracking yaw rate without static error and robust control based on T‐S fuzzy model of multi‐axle steering vehicle can increase control system robustness and improve steering performance when steering at high speed, which has important theoretical and practical value.Main research work of seven aspects that were finished in the dissertation are summarized as follows:(1) The method that multi‐axle steering input angles of vehicle model were simplified as input angles of the front‐axle steering angle and the second‐axle steering angle was proposed. Tires of vehicle are easy to worn out at high speed, so all wheels ought to keep pure rolling when steering. Based on the theory of rigid body kinematics, wheels of all axles of multi‐axle steering vehicle ought to steer round the same steering center. According to Ackerman principle, there exists constraint relationship of all wheel angles of all axles. When front‐axle steering angle is determined by the manipulation of driver and active steering control, the steering center can be determined by any rear axle steering angle, and the other rear axle steering angles can be determined at the same time.(2) Based on the theory of tire nonlinear dynamics of Pacejka "magic formula", the approximate nonlinear models of two degrees and three degrees of freedom were derived through separating states and inputs of multi‐axle steering vehicle. Based on the fuzzy T‐S inference theory, selecting the states of vehicle as inputs of the fuzzy inference system, selecting points of the state vector in normal range when steering at high speed as the linguistic terms of input variables and points of linearization, the local linear models and corresponding membership functions of the fuzzy sets of the linguistic terms were determined and fuzzy T‐S model of multi‐axle steering vehicle when steering was derived.(3) Based on linear model of multi‐axle steering vehicle, state feedback controller of pole assignment was designed. Considering the factor that side slip angle of vehicle can not be measured, full‐dimension state observer was also desiged in the closed‐loop control system. Design of Controller was to achieve the performances of desired closed‐loop poles of multi‐axle steering vehicle in tire working area of linearity and weak nonlinearity.(4) Based on internal model principle, the internal model controller of tracking yaw rate without static error was designed, which achieved robust tracking of reference yaw rate without static error against internal and external interference of multi‐axle steering vehicle.(5) Statistical filtering algorithms based on Bayesian theory were applied to estimate the states of nonlinear stochastic system of multi‐axle steering vehicle,and the method applying particle filtering to estimate the states of nonlinear stochastic system of multi‐axle steering vehicle was presented.Due to nonlinearity of the system, state observer based on linear model results in the degradation of estimating performance of states. Unscented Kalman filtering based on Bayesian theory can estimate the states of nonlinear stochastic system of multi‐axle steering vehicle effectively,but it is only applicable to Gaussian system. Particle filtering is suitable for nonlinear and non‐Gaussian stochastic system, which can improve precision of the state estimation and closed‐loop control performance of of multi‐axle steering vehicle.(6)In view of the influencing factors of model parameter uncertainty, external interference and tire nonlinearity of multi‐axle steering vehicle, in order to improve control precision and control system robustness, based on fuzzy T‐S inference theory and fuzzy T‐S model of the vehicle, H∞ robust controllers and guaranteed cost robust controllers of region pole assignment were designed for the local linear models of each state point. Then inference antecedents of fuzzy T‐S model of the vehicle were applied repeatedly to infer and derive the total controllers of steering control.(7) Semi‐physical simulation experiment system of four‐axle steering model vehicle drived by remote control was developed,and the simulation experiment platform to verify the vehicle models and control algorithms was established. Based on the simulation experiment system, experiment methods were designed, and vehicle model parameters were measured, and vehicle models and control algorithms were verified through the semi‐physical simulation experiment. Verification results indicate the effectiveness of fuzzy T‐S model of vehicle, state estimation method based on particle filtering algorithm, tracking control method without static error based on internal model principle and H∞ robust control method based on fuzzy T‐S model of the vehicle.