Fault Diagnosis Of ESP System Based On Unknown Input Observer

ESP is also known as electronic stability program or Dynamic Stability Control system.It can improve the stability of the vehicle through detecting vehicle running state andreducing the loss of driving force. It is developed on the basis of ABS and ASR. In a movingvehicle, ESP get actual operation data of the vehicle through steering wheel angle sensor, wheel speed sensor, yaw angular velocity sensor and roll angle sensor and transfer theinformation to ECU. Using the information, ECU can determine the driver’s intention. Aftercomparing the intention to the actual operating state, through certain control logic ECU willcalculate how much yaw moment is needed to keep the car’s stability. ESP will adjust thewheel cylinder pressure through hydraulic modulator and get the required yaw moment. When necessary, it may even change the driving force of the engine by the enginemanagement system to stabilize the vehicle.This article describes the methods of fault diagnosis, including hardware redundancymethod and analytical redundancy method. Taking the system structure, the weight of thesystem and the space required by hardware redundancy method into account, we have chosenthe method based on unknown input observer in analytical redundancy method.The most important thing of method based on unknown input observer is to build anaccurate vehicle model. This article introduces the eight-degree-of-freedom nonlinear carmodel first. Eight-degree-of-freedom model takes the non-linearity of the tire into account. Itis close to the actual state of operation of the vehicle, but its complexity is not suitable toestablish an unknown input observe. We simplify the eight-degree-of-freedom model tofour-degree-of-freedom model with lateral movement, longitudinal movement, yawmovement and cornering movement. After establishing the four-degree-of-freedom vehiclemodel in MATLAB-Simulink, we estimated the yaw moment, the front and rear wheelsteering stiffness in it and verified its validity.Unknown input observer is established based on the four-degree-of-freedom vehiclemodel. After that we do modeling and fault diagnosis to stuck, drift increase, outputsaturation and lower output accuracy of the yaw angular velocity sensor. It Confirmed the validity and accuracy of the observer we have design. At the same time, take the yaw angularvelocity sensor fault as a target fault and the roll angle sensor fault as a non-target failure,after analysis of the simulation results generated due to yaw rate sensor failure and roll anglesensor failure, it proved that our approach can achieve a decoupling of these two fault.