Abstract: A double-Wiebe function calibration method based on immune particle swarm optimization（IMPSO）algorithm was proposed to predict combustion parameters for a turbocharged diesel engine. The heat release at 25% and 100% loads was calibrated respectively,and an improved algorithm and immune algorithm were introduced to solve the multiple problems during fitting. The optimization results show that R~2 and the stability are improved to be more than 0.998 and 0.700,respectively. The heat release was also calibrated for other operating conditions of the diesel engine. The results show that fitting data are consistent to the experiments with a good generalization. In addition,a double-Wiebe function prediction model using back propagation（BP）neural network was established,prediction was conducted by selecting the engine speed,fuel mass per cycle,inlet pressure and temperature as input. Results show that R~2 is greater than 0.950 and the predicted value is consistent with the calibrated value,indicating that the prediction model can be used to predict the heat release law of diesel engines under multiple operating conditions.