The free piston energy converter (FPEC) is a combination of a combustion engine and a linear electrical machine which develops a new type of power unit for hybrid electric vehicle. The movement component of the FPEC is a multi-input multi-output coupling nonlinear system. As a consequence, an adaptive decoupling control strategy is presented in this paper for controlling the piston motion. Two neural PID controllers in parallel are employed as the decoupling controller, in which PID gains are auto adjusted by supervised Hebbs learning rule. Simulation results show that the utilization of the neural adaptive PID decoupling controller gives rise to stabilize the piston motion.