Hierarchical Prediction of Infrared interference Environment based on Mutual Information and Random Forest

In order to realize the classification prediction of environment in the process of combat, we set up the classification prediction model. First, we divided data set into blocks because of the big scale, wide range of influencing factors and strong nonlinear relationship among them. The missile hit rate is reflected in the miss distance, clustering each block data to analysis the correlation factors that affect the miss distance. Second, according to the mutual information theory, the strong correlation factors affecting the miss distance are identified from three aspects which are throwing strategies, maneuver mode and the relative situation by calculating the mutual information matrix. Third, the strong correlation factors as the sample attributes of node splitting in the decision tree, a hierarchical prediction model of infrared interference environment is constructed. This method can better identify the strong correlation factors that affect the miss distance, and avoid the adverse impact on the classification prediction due to the difference of correlation factors. The results show that the method is suitable for the processing of miss distance data. The prediction accuracy of infrared interference environment classification is about 98%, and the generalization ability of the model is strong.