Research Of Situation Data Management And Target Clustering Technology Based On Big Data

With the rapid development of electronic information technology,more and more communication technologies have been put into the battlefield,which has led to the increasingly complex and huge information acquired by the battlefield monitoring system.The processing system based on the first-level data fusion mainly provides the operational information of the target of each entity in the battlefield,and the decision-makers need to understand the group combat information at all levels.In order to help decision-makers quickly grasp the current battlefield situation,efficient and timely information on the massive The analysis and finishing has become the research focus,and it has become the research difficulty of the situation assessment technology.In this thesis,the target intent recognition technology and the target grouping technology in the situation assessment system are the main research objects,focusing on optimizing the problem that the existing target grouping technology has,including unknown cluster number,high initial cluster center sensitivity and single distance metric.At the same time,deep learning is introduced into the target intent recognition algorithm to improve the accuracy and real-time of the evaluation.The research content of the thesis is mainly divided into five parts.1.The three functional modules of situational awareness,situational understanding and situational prediction in the situation assessment system are analyzed.The framework of the situation assessment system is given according to the battlefield requirements,and the two main technical modules in the framework are explained.The module's classical algorithm was analyzed and compared.2.The iterative self-organizing data analysis algorithm is analyzed.It is found that the algorithm can dynamically adjust the number of clusters,but it is sensitive to the initial value and the distance metric is single.Therefore,the algorithm is optimized,and the iterative self-organizeing data analysis algorithm based on the maximum and minimum manifold distance is proposed.The manifold distance is used instead of the commonly used Euclidean distance,and the initial clustering center is selected by pre-clustering.Finally,the improved algorithm is compared with the iterative self-organizing data analysis algorithm through two sets of data sets.It is proved by experiments that the improved algorithm has better performance when clustering small formation targets.3.The fuzzy C-means algorithm is analyzed.The problem that the number of cluster centers cannot be dynamically adjusted and the problem is easy to fall into local optimum is improved.Based on the algorithm,the particle swarm optimization fast search fuzzy C-means algorithm based on gap statistic is proposed.Finally,it is proved by two sets of experiments that after introducing the particle swarm algorithm and Gap Statistic algorithm,the improved algorithm can better identify the formation,which is superior to the fuzzy C-means algorithm in performance.4.The traditional target intent recognition method is no longer used,and the deep learning algorithm is applied to the battlefield to identify the target intent.The Bidirectional long-term and short-term memory network algorithm based on attention mechanism is adopted to alleviate the problem that the hidden layer of the long-and shortterm memory network algorithm cannot be too deep.The Bi-directional long-term and short-term memory network algorithm is mainly used to solve the regression problem,and the Softmax module is added to solve the classification problem.Finally,the algorithm is compared with the Bi-directional long-term and short-term memory network algorithm.It is proved by experiments that the Bi-directional long-term and short-term memory network algorithm based on attention mechanism is better in performance..5.According to the specific battlefield scenario,the situation assessment system was designed and completed on the development platform.The target clustering module adopts the particle swarm optimization algorithm based on interval statistics.The target intent prediction module adopts the attention mechanism-based Bi-directional long-term and short-term memory network algorithm.The situation display module uses the humancomputer interaction interface based on the front-end development.Finally,the simulation data set test confirms the system implements the basic functions that the situation assessment system needs to complete to meet the battlefield needs.