Research On Military Situation Analysis Based On Clustering Analysis And Implementation

Information technology has become a key point in modern wars, in the changing and complicated situations, it plays an important role to help the commanders to quickly seize the information. Computer science especially AI technology will bring great advantage to conquer this task. Due to the complication of battlefield situation, there isn't a single method can handle this problem, systems that have been made are always based on multiple AI methods. This paper focuses on the function and implementation of clustering analysis in situation analysis.Chapter 1 introduces the research background and present status of this field. Chap.2 introduces the JDL model of situation analysis, briefly reviews the main methods of this field.Chap.3 first introduces the concept of clustering analysis, after explaining some major methods, like k-means, fuzzy c-means, argues that these methods work well on sphere-like data set, but not well on irregular data set. This paper proposes a method in which the similarity between points is measured using a graph-based method, and use the center of the graph to replace the centroid in clustering methods. Experiments show that this method can improve the result of clustering.Chap.4 focuses on DBScan algorithm (Density-Based Spatial Clustering of Applications with Noise), this is an effective clustering method can be used on irregular data set. This paper shows a method to eliminate the parameter choice step in this algorithm and finds a criterion to evaluate the result. Finally, a dynamic version of DBScan is explained.Chap.5 shows implementation of a situation system including the architecture, interface and other implementing details.