The Design and Realization of Dynamic Evaluation Strategy of Pieces in Military Chess Game System

• Published in: 2019 Chinese Control And Decision Conference (CCDC) pp. 6287 - 6292
• Main Authors: Xiaochuan, Zhang, Wanwan, Wang, Qin, Li, Tianao, Wang, Hao, Song
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-06-2019
• Subjects: Artificial intelligence; Cost accounting; Dynamic valuation function; Exploration and income; Games; Heuristic algorithms; History; history heuristics; Incomplete information game; Military computing; Monte Carlo methods; Q-learning
• ISSN: 1948-9447
• DOI: 10.1109/CCDC.2019.8832609

In the system of military chess, The accuracy of the evaluation directly affects the function of the system. Because of the military chess game is incomplete information game with high complexity, the accuracy of guessing the enemy chess piece is not high in a limited time. Therefore, this paper uses history heuristic tables to guide the Monte Carlo algorithm to simulate enemy chess pieces, which greatly saves simulation time. And on the evaluation of our chess pieces, Q-learning algorithm based on the idea of exploration and income equalization was proposed to solve the problem of the evaluation based on the multiple kinds of military chess pieces and lack of timely evaluation, Improve the effectiveness of the evaluation. and constructs a complete game system based on dynamic estimation function with four factors by competing with static evaluation system of military chess game the winning rate of our approach reached 60%, And won the first prize in the 2018 National University Computer Game Competition. which verified the effectiveness of dynamic evaluation strategy.