Conventional chaos control methods need model information of the chaos system, but model parameter is always unknown in engineering practice and the dynamic uncertainty of model-building often appears inevitably. In this case it is impossible to acquire best control performance index of optimization. To resolve this problem an adaptive control algorithm of chaotic systems based on model free method is proposed. It is a low cost controller. With this method, we can easily know a priori knowledge of chaos system only be adjusting a few parameters online without training before hand. We prove that this algorithm has a good stability mathematically. Simulation results show the effectiveness of this theory.