GPSとマルチレートカルマンフィルタを用いた電気自動車の安定化制御のための車体横滑り角推定に関する研究

The main contribution of this thesis is the proposal of advanced multi-rate Kalman filter for vehicle sideslip angle estimation using GPS. This method is applied to the stability control system of electric vehicle. The other contribution is the thorough evaluation of lateral tire force sensors for sideslip angle estimation. This thesis is organized in five chapters and a special part as the followings. Chapter 1 presents the background and motivation of the research. Electric vehicles (EVs) will be the means of transportation for human society in near future. Using in-wheel motors, EVs have become the novel motion control system. In order to maintain the stability and safety of EVs, stability control system should be designed. In this system, sideslip angle estimation is the main problem because the sideslip angle sensor is very expensive for practical applications. A number of previous researches on this topic are reviewed in chapter 1. Chapter 2 discusses the lateral dynamics of electric vehicle in which lateral tire force characteristic is the main point. The linear bicycle model is introduced for state estimation and control design in the following chapters. Chapter 3 proposes the application of GPS in sideslip angle estimation. The promising scenario of GPS in vehicle motion control is introduced. Then, the experiment setup of GPS receiver CCA-600 by Japan Radio Company in electric vehicle COMS is presented. The course angle of vehicle measured by GPS receiver is utilized as a measurement input for estimation using multi-rate Kalman filter. However, the conventional multi-rate Kalman filter faces with the problem of low update rate of GPS signal, the time-variant parameters, and the effect of external disturbances to electric vehicle. Therefore, the quality of multi-rate estimation is degraded during inter-samples (between two continuous updates of GPS). In order to enhance the inter-samples performance of multi-rate estimation, this chapter proposes two advanced methods. The first is the prediction of GPS course angle measurement residual for more accurate inter-samples estimation. The second is the general disturbance estimation for compensating the effect of both model error and external disturbance. Both simulations and experiments are conducted in order to evaluate the proposed methods. Chapter 4 proposes the vehicle stability control system based on sideslip angle estimation using GPS and direct yaw moment control. LQR method is utilized for designing the yaw moment control to manage both sideslip angle and yaw rate. Both simulation and experiment tests are conducted with electric vehicle COMS. Special part is a thorough evaluation of estimation based on tire lateral force sensors for vehicle stability control system. A new method of on-line identification of cornering stiffness without sideslip angle is proposed thanks to the lateral tire force sensors produced by NSK Ltd. Based on the difference between the left and right side tire lateral forces, sideslip angle estimation without cornering stiffness is proposed. By combining the influence of load transfer and non-linear tire force as a disturbance term to be estimated, we can achieve an advanced configuration of sideslip angle estimation. Experiments using electric vehicle Kanon are conducted to evaluate the proposed methods. The two-input two-output controller for vehicle stability control system is also proposed by integrating yaw moment control with active steering control. The simulation tests are conducted in order to proof the higher robustness in comparison with conventional decoupling control method. Chapter 6 gives the summary and conclusions on the works done in this thesis. The future research plan on application of GPS in electric vehicle control is also stated