Guidance and control of a two-dimensional model of an air defense missile using deep learning and fuzzy adaptive control

Conventional method of designing missile guidance and control, due to not taking into account the interference of the guidance and control subsystems and also due to the delay caused by the difference in working frequency of the subsystems, normally leads to instability or unacceptable miss distance. It is advisable to consider the states of the guidance and control subsystems simultaneously in the design process to increase the accuracy and overall performance of the system in the final phase. This can improve efficiency, and it saves time and effort. In the integrated design method of the guidance and control system, limitations of the subsystems are considered in whole during the design, and as a result, the performance of the system will be improved. This article describes the process of designing and simulating the performance of a deep and fuzzy adaptive controller, proposed to guide the missile in a two-dimensional problem of minimizing the collision time and the distance to the target. In the design of the controller, the deep learning neural network controller is first designed offline and used as a gain table in the adaptive controller. Next, this controller is improved by adding fuzzy control. The performance of both controllers is evaluated in the presence of disturbance. Based on simulation results, it is shown that the use of these proposed controllers and the application of the integrated guidance and control model leads to improved miss distance as well as collision time compared to similar results obtained using a PID controller.