Fault Diagnosis of NOx Emissions of a China VI Heavy-Duty Diesel Engine Based on Time Window and SVM

NOx emissions fault diagnosis is one of the most important on-board diagnostic(OBD) functional requirements. In this paper，the NOx emissions fault diagnosis was abstracted into classification problem of time series，and an algorithm based on time window and the support vector machine(SVM) was proposed for the fault diagnosis of NOx emissions. Then，one unified fault diagnosis model is established for combining the NOx emission test data of world harmonized transient cycle(WHTC) and vehicle actual road driving measurement with portable emission measurement system(PEMS). Combined with the diagnostic release and delay time conditions，the input feature was dispersed by using time window and the time window was classified by using the SVM algorithm. The model was trained and verified with test data. The results show that high fault diagnosis accuracy is achieved in both WHTC cycle and road spectrum. When the WHTC cycle test data is taken as the training set，and the fault diagnosis accuracy on the road spectrum test data reaches 96.99%. When the WHTC cycle and the actual spectrum test data are mixed as the test set and the training set，the fault diagnosis accuracy can reach more than 99%. The results indicate that the fault diagnosis model has good generalization performance. © 2023 Chinese Society for Internal Combustion Engines. All rights reserved.