Research On Soft-Measurement Of Vehicle Status And Integrated Navigation System Fault Detection Methoc

With the development of integrated navigation system on vechicle and different kinds of filter mehods, integrated navigation system becomes more and more complicated. On the one hand that how to use the other methods to effectively overcome the shortage of Inertial Navigation System (INS) itself and enhance its performance need to be considered; on the other hand, this complex system brings the problems that how to make sure its dependability. Besides, safety of vehicle running and precision of location have tight relation with integration navigation system. Lastly, status estimation of running vehicle by the outside sensors is guarantee to the active safety of vehicle and the passanger.On the basis of constructing the vehicle reference frame and global frame, this thesis are mainly about the fault detection of integration navigation system, stable control, and status measurement technology of autonomous vehicle. Main research work and innovative achievements in this thiesis are as follows:Error model of INS under the reference frame in the paper is set up, and difference of GPS coordinate transformation is discussed. Kinematics characteristics of wheeled mobile vehicle are analyzed, and kinematics model is set up without considering slide and brake of vehicle. Ackerman principle is then added to the basic kinematics model, and the math formulas about turning angle, steering angle and turning radius are given to declare the movement status of vehicle.For the reasons that the vehicle is influenced by influences such as:side slip, brake or slider, and the nonholonomic restriction of vehicle is always destroyed.In the paper, this nonholonomic restriction condition is brought to the ideal kinematics model, and mapping relations of error vector between local coordinate frame to global coordinate frame are analyzed.Use this kinematics model to analyse lateral operation stability of vehicle.Then design a method to verify the ideal kinematics model and dissucuss how to get the parameters in this model. By the way, a method used to mearsure the front turning angle on line is proposed.To make sure the control target of Vehicle Stability Control (VSC) system, it must get the running status with high reliability.However, due to the fault of vehicle or complexity of circumstances (vibration and side slip), vehicles are always happen to poor control stability, which can result in accidents in a short time intervals if the situation are not estimated in time.Aiming to the active safety problems of vehicle, a soft measurement method based on the above kinematics model is presented, and this method is testified through the successful estimation of longitudinal velocity, lateral velocity, side slip angle of vehicle and wheel. Estimation equations of velocity errors of the vehicle are given out to estimate velocity errors of side and forward according to the gesture information and velocity information. So the stability of the whole vehicle could be judged by velocity errors of the vehicle, and the precision is given out by the bais between velocity along the forward direction and estimated velocity. Conclusion was validated through the vehicle experiment. This method is based on GPS/INS integrated navigation system, and can provide the foundation for fault detections in unmanned autonomous vehicles.When designing the integrated navigation system, a corresponding simulation system is necessary, however, now the simulation system is designed aiming to aerial vehicle. The characters of wheeled vehicle are more influenced by the ground and terrain. To solve this problem, the principle of common track generators is analyzed, and combined the vehicle kinematics model with the principle of dead reckoning, a new vehicle track generator is proposed, which can provide process parameters of carrier vehicle in all times and achieve simulation of typical actions of moving vehicle. Then Inertial Measurement Unit (IMU) elements are simulated, including gyroscope and accelerometer. Track generator and IMU compose the integrated navigation simulation system. Different kinds driving actions are used to test this simulator. Two kinds Karlman filtering algorithms are simulated in this system, both errors with inhibition and without inhibition experiments show that it is of validity.Precision of Integrated navigation system is not only decided by filter algorithm, but also by the precison of obversation information. The fault data in GPS obversation information will bring deep contamination to navigation system.To solve this problem, a fault detection method based on varying-length scanning model is proposed. This method does minus operation on the time series data which are waiting to be detected and then get the corresponding difference minus sequence. The statistic characters of two neighbor difference minus sequences are affected due to the change point. So it estimates the bound of the next time's value through current difference minus sequence, and then realizes fault detection according to estimate whether the next time's value is in this interval. Considering the complexity of algorithm, a time window size mechanism is designed, which aims to bring the balance between the detection speed and precision. The questions that how to select the confidence interval and time window size are discussed by a simulation experiment. This fault detection method is used in GPS/INS integrated system at last. The experiment compares with traditional detection methods, χ2-proof-test, and shows that it can detect the change point in the data effectively and enhance the precision of filter.