At the present work for improving the engine performance and decrease of emissions, a port injection gasoline engine is converted into direct injection. Engine performance behavior was investigated by AVL Fire software with adding hydrogen to natural gas from 0% up to 30%. Validation of the simulated model and experimental results show good confirmation. To determine the relationship between independent variables engine speed, ignition timing, injection timing and H2% versus the dependent variables including engine performance parameters, specific fuel consumption, CO and statistical analysis models were used. Comparison between different errors models shows that Radial basis function model with training algorithm Bayesian regularization back propagation can estimate better engine performance variables. The results showed that adding hydrogen to natural gas cause the output power, torque, fuel consumption efficiency increase and specific fuel consumption drop. Also, CO decreases when ignition and injection timing be advanced and engine speed reaches to its largest.

Keywords