Adaptive PID controller using deep deterministic policy gradient for a 6D hyperchaotic system

Abstract

This article introduces a method for the adaptive control of a six-dimensional (6D) hyperchaotic system using a multi-input multi-output (MIMO) approach, leveraging the deep deterministic policy gradient (DDPG) algorithm. The states and tracking errors of the hyperchaotic system are amalgamated to form an input image signal. This signal is then processed by a deep convolutional neural network (CNN) to extract profound features. Subsequently, the DDPG determines the coefficients of the proportional–integral–derivative (PID) controller based on the features discerned from the CNN. The proposed approach exhibits robustness to uncertainties and varying initial conditions, attributed to the DDPG’s ability to learn from the input image signal and adaptively adjust control policies and PID coefficients. The results demonstrate that the proposed adaptive PID controller, integrated with DDPG and CNN, surpasses conventional controllers in terms of synchronization accuracy and response speed. The paper presents the following: a 6D hyperchaotic system’s dynamic model, a CNN-based DDPG’s structure, and how it performs and compares to traditional methods. Then, it summarizes the main findings.