Study On Observer-based Fault Diagnosis For Vehicular ECAS Sensors

Electronically controlled air suspension(ECAS) system can control suspension states to meet vehicle’s driving condition requirements based on electronic control unit(ECU), suspension’s states are measured by sensors installed around ECAS system,measurement information is processed by ECU and control signal is transferred from ECU to actuator. However, vehicle’s driving conditions are complex and changing,pavement and vibration caused by engine have great influence on reliability of ECU, in which sensors are the most vulnerable, sensor faults are liable to occurred. Sensor faults can lead ECU malfunction, which has adverse effect on air suspension’s performance. Thus, in order to achieve functional safety and reliable performance of ECAS system, sensor fault diagnosis is studied this paper.Combined with vehicle dynamics and aerothermodynamics, ECAS system model with sensor fault is established. Sprung mass’ s motion and unsprung mass’ s motion are analyzed, road input and vibration of unsprung mass are viewed as random disturbance applying on suspension dynamic displacement, air suspension reduced order model is established by using vehicle system dynamics. Meanwhile, charging and discharging of air spring is deemed as changeable mass thermodynamics process, its opening changeable mass model is obtained by using first law of thermodynamics. ECAS sensors includes height sensor, acceleration sensor and air pressure sensor, sensor fault features are analyzed, faults are divided into three types: constant gain fault, constant bias fault and fix value fault, ECAS system models under different fault conditions are established.Fault detection and isolation(FDI) method for 1/4 ECAS sensor fault is studied.Observers are established to estimate state information of ECAS system, ECAS’s output residual is obtained combined with sensor measurement output and estimated information. Further, fault detection index, fault isolation index and optimal threshold are calculated to realize the FDI of ECAS sensor.Due to strong non-linearity of ECAS system, its dynamic adjustment is disturbedby road input, based on Kalman filter theory, cubature Kalman filter(CKF) is used to design the observer, four types of sensor fault conditions are chosen, simulation and experiment about sensor FDI are carried out. Meanwhile, three methods: extended Kalman filter(EKF), strong tracking filter(STF) and CKF are used to design the observers, contrast simulation and experiment about sensor FDI are carried out.Results show the three methods can realize FDI of ECAS sensors respectively and the CKF method has the best performance according to global response indexes.Linearization error and parameter uncertainty error in whole vehicle ECAS system model are analysed, ECAS sensor fault diagnosis method based on self-adaptive threshold is proposed. Combined with CKF method, observers for whole vehicle ECAS system are designed by building relationship between threshold, control input and parameter variation range, sensor FDI simulation is finished. Finally, whole vehicle test bench is built, road simulator is used to simulate real pavement,experiment is carried out to verify the proposed FDI method on real vehicle, reliability and safety of ECAS system are improved.