Synchronization Control of Nonlinear Chaotic Systems with Deep Reinforcement Learning Algorithm

The control of complex chaotic systems is a challenging problem due to the rapidly changing character of chaotic strange attractors. In this paper, we have presented a controller designed utilizing deep deterministic policy gradient (DDPG) method to deal with the synchronization control problems of chaotic systems. The result shows that under the effect of deep network controller, two chaotic systems with different unknown input signals can achieve synchronization. Moreover, the algorithm is robust with the changing parameters or disturbance in chaotic systems. This algorithm can be utilized in the control of complex nonlinear chaotic systems with unknown characters of disturbances.