Autonomous Role Design of Game Based on Neural Network Algorithm and Artificial Intelligence

At present, the interaction mode of role behavior in most games is deterministic interaction behavior. Such deterministic behavior often fails to reflect the autonomy of the role, and it also leads to the monotonous action of the role, which is easily predicted by the player and reduces the playability of the game. In this paper, the design of game autonomous role based on neural network algorithm and AI (artificial intelligence) is developed. Based on cognitive role modeling, this paper puts forward the design idea of game autonomous role by combining BPNN (BP neural network) and GA (genetic algorithm). Using GA to optimize BPNN, applying GA in BPNN can realize automatic optimization of learning rules, and improve the learning efficiency and the optimization speed of weight coefficient of BPNN. The research results show that using the optimized neural network algorithm to evolve the autonomous role of the game can reduce the computational complexity by about 25% compared with using the traditional evolutionary neural network algorithm. The running results of virtual characters in the system show the effectiveness of the behavior model; The effectiveness of the optimized neural network algorithm is verified by experimental analysis.