Abstract:

Aiming at the problem of lacking of knowledge about air combat decision-making behind the abundant combat training data, a method for constructing air combat decision-making knowledge is proposed. Firstly, this method analyzes the process of decision-making knowledge based on the influencing factors of air combat decision-making, and chooses to use production rules to represent air combat decision-making knowledge. Secondly, because it is difficult to meet the discrete measurement requirements of data mining algorithms due to the presence of noisy data and attribute data in combat training. The k-means clustering algorithm is applied to detect outliers and the minimum description length principle algorithm discretizes the continuous attributes. Finally, based on the preprocessed data, rough set is used to extract the minimum knowledge of air combat decision rules, and the reasoning and application of air combat decision-making knowledge are realized by building an fuzzy logical inference system. Simulation shows that this method can effectively extract the knowledge of air combat decision rules and the reasoning and application of knowledge.

Key words: decision-making knowledge, production rules, k-means clustering algorithm, minimum description length principle, rough set fuzzy logical