The Research Of Aircraft Object Clustering Application Based On Cluster Analysis

For the two sides during the fighting, military decision-making is an important factor in the decision parties of the game results. In the military decision-making process, situation assessment is a very important part, while in the situation assessment, target grouping is the important and difficult part. The importance of the target clustering is that the result of grouping can decide the relationship between the elements of the situation, and it is also a key base for the commanders' strategy-making. Clustering algorithm is a commonly used algorithm to solve the target grouping, there is now quite kinds of clustering algorithms, but each kind has its own restrictions so it can not be applied to various scenarios freely. This thesis gives a detailed introduction about the target clustering as well as the related concepts; and mainly researches the clustering algorithms used to solve the problem of flying fighter target grouping. This is the main contents:(1)This thesis systematically introduces the basic concepts and algorithms of cluster analysis, and mainly focuses on the partition-based clustering algorithm, and uses the classical K-means algorithm and the improved K-means algorithm, and focuses on the implementation of these algorithms on the problem of flying fighter target grouping.(2)This thesis has a detailed description of the basic concepts of Fuzzy C-means algorithm and the related algorithms. On this basis, this thesis introduces an improved algorithm based on fuzzy C-means algorithm and studies the specific implementation processes on the problem of flying fighter target grouping.(3)In order to meet the requirements of real-time performance and stability, this thesis designed a system for flying fighter target grouping. Firstly, this system uses the improved fuzzy C-means clustering algorithm to get the best the number of clusters. Then it uses the improved K-means clustering algorithm to cluster the data sets. This system selects the strategy that the coordinates of three-dimensional space and speed account for a certain weight respectively. And finally, heading dimension is used to correction. The experiment results show that the system solved this problem better.In the first and second parts, this thesis solved the flying fighter target grouping problem by using the optimized the existing clustering algorithms. By simulating the UCI data sets, the better experiment results improved that the optimized algorithm is effective. The third part mainly works on software designing of the system of the flying fighter target grouping, which can be used as a preliminary software of target recognition.