Development of cyclic variation prediction model of the gasoline and n-butanol rotary engines with hydrogen enrichment

To investigate the influence of operation parameters on the cyclic variation of the Wankel rotary engine with hydrogen enrichment, an intelligent regression model based on the support vector machine (SVM) was implemented to predict the cyclic variation. For modeling the prediction model, the cyclic variation of speed (CoV_n) and cyclic variation of the main combustion duration (CoV_CA10-90) were used as an evaluator for idle and part load conditions, respectively. The operation conditions including main fuel type (gasoline and n-butanol), hydrogen volume percentage (β_H2), excess air ratio (λ), ignition timing (IT)and speed were used as independent variables. When optimizing the prediction model, the data processing method, kernel function, loss function and optimization method on the prediction performance were discussed in detail. The results indicated that an optimized model can be obtained by using genetic algorithm combined with [0, 1] data processing method, and the coefficient of determination, mean square error and mean absolute percentage error of CoV_n were 0.9904, 0.0783 and 0.3845%, corresponding to CoV_CA10-90 were 0.9972, 0.0197 and 1.1729%, respectively. For the CoV_n, gasoline as the main fuel was lower than the n-butanol at the same operating condition. The CoV_n at high speed was greater than that at low speed. When operating at part load conditions, the CoV_CA10-90 decreased with the increasing β_H2, and first decreased and then increased with advancing IT.