Free Piston Energy Converter Motioncontrol Based On Artificial Neural Network

Free piston energy converter (FPEC) is a highly efficient motor. The general design consists of two opposite combustion cylinders with an integrated linear alternator in between, compare with the conventional ones are simpler. The major advantages of such a design over the conventional ones are the simple engine layout and much improved conversion efficiency. It has a promising future technology for hybrid electric vehicle.We briefly introduced the free piston motors'history. The FPEC is inherently a hybrid system, according to the basic structure and working principle of the FPEC, making a lot of researches on the discrete event systems, the continuous variable dynamic systems and the transition triggers between the discrete event states.This paper makes a new contribution by applying a trajectory sensitivity analysis to piston motion control. Firstly we use feed forward neural network (FFNN) to the FPEC objective function modeling, and to compute the gradients, and then we use conjugate gradient method to the optimal control of FPEC motion control. By the simulation results, it is showed that FPEC motion control is satisfactory in the optimal control of electro-magnetic force and overall heat input.