The model of a real FCE(fuel cell engine) is set up by using neural network identification.The validity and accuracy of the model are proved by the simulation results.Then based on the model.In order to shorten the time of warming machine and improve the poweroutput of FCE after cold start as well as to raise temperature smoothly from idle to accelerating state and reach the rating poweroutput as soon as possible,a new method for solving the optimized control of FCE problem is proposed by using the idea of ant colony.The model test results show that it basically achieves the goal and show the advantage through comparing with traditional PID method.