Search Parameter Optimization of Phased Array Radar with Multi-antenna Array Based on Proximal Policy Optimization

An optimization algorithm of multi-antenna array radar search parameters based on proximal policy optimization is proposed to solve the problem that the parameters of traditional single-antenna array radar optimization model is hard to solve in complex multi-antenna array scenario. Firstly, considering the constraints of multi-antenna array scenario, combined with the actual search task requirement of airborne radar, an optimization model of multi-antenna array radar search parameters based on the maximum expected discovery distance of the target is established. Secondly, the dual observation space and composite reward function with discount factor are introduced to transform the above optimization problem into a dynamic decision problem of reinforcement learning, in which the solution of model parameters is realized by executable action output by the policy network. Using the Actor-Critic algorithm structure, combined with methods for solving extreme value of function, the weight of each antenna array and the corresponding sub-airspace search resource allocation coefficient are updated online. Finally, the simulation results show that the proposed algorithm can quickly make accurate autonomous decision based on the airspace-target set covering model and the threat of each target, and the performance of proposed algorithm is significantly better than the traditional algorithm in the multi-antenna array scenario.