Radar interference game strategy design method based on neural network virtual self-game

Abstract

The invention discloses a radar interference game strategy design method based on neural network virtual self-game. The method comprises the following steps: constructing an extended multi-round game interaction model corresponding to a radar and a jammer; wherein the radar and the jammer are respectively set as subjects in neural network virtual self-match: based on the extended multi-round game interaction model, interaction information between the radar and the jammer is acquired, the interaction information is encoded, and the interaction information comprises game state transition information and optimal reaction behavior information; solving the encoded interaction information based on a neural network virtual self-match to obtain a plurality of game strategy combinations between the radar and the jammer; and obtaining the average utilization degree of each game strategy combination, and determining the game strategy combination corresponding to the minimum value of the average utilization degree as a target game strategy combination. According to the method, the game condition between the radar and the jammer can be efficiently and accurately obtained.

Description

technical field [0001] The invention belongs to the technical field of radar, and in particular relates to a radar jamming game strategy design method based on neural network virtual self-play. Background technique [0002] In electronic warfare, there is a natural confrontation relationship between radar and jamming. How to make decisions to maximize one's own performance is the top priority in the field of radar technology. Game theory is a mathematical tool to study the conflict and cooperation relationship between decision makers. Therefore, it is feasible to analyze the confrontation relationship between radar and jamming and design anti-jamming and jamming strategies based on game theory. [0003] In the prior art, radar jamming strategy design based on game theory is mainly divided into two categories: [0004] The first is to model the radar and jamming based on the static game, and then solve the Nash equilibrium strategy. However, although the existing research ba...

AI Technical Summary

Problems solved by technology

However, in the existing research based on dynamic games, the extended formal game modeling in dynamic games is not used, the modeling method is not universal, the application range is small, and the characteristics of radar and jamming sequence decision-making cannot be described.

[0006] In addition, in the prior art, radar jamming strategy design is usually based on reinforcement learning. This method usually assumes that one of the radar and jamming has the ability to learn, and the other adopts one or several fixed strategies. Based on such asymmetric assumptions, radar and jamming Low gaming ability for interference

Images