A Model-Free Diagnosis Approach for Intake Leakage Detection and Characterization in Diesel Engines

摘要: Feature selection is an essential step for data classification used in fault detection and diagnosis processes. In this work, a new approach proposed, which combines feature algorithm neural network tool leak characterization tasks diesel engine air paths. The Chi square classifier as the based on Levenberg-Marquardt system behavior modeling. obtained characterization. model learned validated using generated by xMOD. This again testing. effectiveness of proposed illustrated simulation when operates low speed/load considered affecting path very small.