Performance Prediction of Hydrogen-enriched Compressed Natural Gas Engine Based on Support Vector Machine

A hydrogen-enriched compressed natural gas (HCNG) fuel engine with hydrogen volume fraction of 20% is experimentally studied and its performance is analyzed to improve the calibration efficiency of HCNG fuel engine and accurately predict the engine parameters. An engine parameter association model is established by using support vector machine (SVM) based on the steady-state experimental calibration data under high speed and low load conditions, and the different optimization algorithms are used to improve the prediction accuracy of engine parameters. Results show that the engine has the minimum brake specific fuel consumption (BSFC) and ideal brake specific NOx consumption level (BSNOx) at the minimum advance for best torque. The external characteristics is obviously improved when increasing the hydrogen ratio especially. The SVM model can describe the nonlinear relationship between the input and output parameters of the engine, and has high prediction accuracy and strong correlation between independent variables and dependent variables (all coefficients of determination R2 are all greater than 0.97). The optimal prediction model derived using the genetic algorithm has a high generalization capability, with mean absolute percentage errors of only 1.23%, 1.98%, and 5.43% for torque, BSFC, and BSNOx, respectively. © 2022, Editorial Board of Acta Armamentarii. All right reserved.