Based on the multi-input and multi-output model of a supercharged direct injection engine and vehicle， which was provided by the internal combustion power intelligent control algorithm challenge，a joint decoupling controller was developed by combining back propagation neural network(BPNN)，particle swarm optimization(PSO) and proportional-integral-derivative(PID) control. In order to reduce the coupling error and achieve both target tracking and decoupling control for the exhaust gas recirculation(EGR) valve and the turbine vent valve，the PSO algorithm was employed to optimize the PID control parameters to control the opening of the EGR valve. Meanwhile，the BPNN algorithm was utilized to iterate the PID control parameters of the equivalent flow diameter of the turbine vent valve through the method of error backpropagation. The control effectiveness of the combined decoupling controller with respect to different driving speeds was evaluated on the Simulink platform. The results show that compared with the BPNN-PID algorithm，the mean integrated absolute error of the boost pressure and EGR rate of the combined BPNN-PID and PSO-PID algorithm is reduced by 31.86% and 2.46% respectively，and the controller demonstrates excellent performance under different driving speeds.