Fault Monitoring and Diagnosis of Actuators in Electromagnetic Valve-Train Based on Neural Networks Optimization Algorithm

Abstract

In order to realize fully flexible variable valve lift and improve the intake efficiency of the engine, a new electromagnetic valve-train (EMVT) was designed to replace the traditional CAM in the engine valve-train. However, as the durability of electromagnetic linear actuator (ELA) in the EMVT was lower than that of CAM, and there might be some problems during a long running time. This paper presented an improved online fault diagnosis method combining BP Neural Networks with Grey Relation Analysis (GRA) to analyze current signals, which could realize the monitoring, diagnosis of electromagnetic linear actuator faults and early warning so as to prevent unnecessary situations such as accidental cylinder stop of the engine. The results showed that this method had a high fault diagnosis rate, high speed, reliability and practicability.