Evaluation Of Weapon Operational Effectiveness And Its System Realization Under Marine Environment

The evaluation of weapon operational effectiveness is an important part of the auxiliary military decision system. Under marine environment, the meteorological and hydrological elements that influence the weapon operational effectiveness are numerous and complex, making evaluation more difficult. At present, the theoretical framework of weapon operational effectiveness evaluation is not complete, the analytical assessment method, analytic hierarchy process, expert-based evaluation method and battle simulation assessment method are commonly used to solve the problem of weapon operational effectiveness evaluation, but these evaluation methods are mostly linear methods and affected by subjective factors, which make it unable to achieve satisfactory results for this study. So, it has significant research value to establish objective and efficient weapon operational effectiveness evaluation model. In this thesis, according to the problem of weapon operational effectiveness evaluation under marine environment, based on the analysis of marine environmental factors that affect the weapon operational effectiveness, two operational effectiveness evaluation models are proposed for the evaluation of weapon operational effectiveness under certain meteorological and hydrological conditions.Firstly, the evaluation model of weapon operational effectiveness based improved K-nearest neighbor is established. As a popular machine learning regression method, K-nearest neighbor regression can be easily applied to solve the research problem. Traditional K-nearest neighbor regression has low efficiency, and feature weights are ignored when calculating the similarity. After determining the evaluation index, similarity based on attribute weighted is computed and the k nearest neighbors are searched, then operational effectiveness value under specific conditions is evaluated for the given instance through the distance weighted method. Because the K-nearest neighbor regression model has a negative learning characteristic that postpones the evaluation process, k-d tree structure is utilized to ensure the evaluation result obtained in a shorter time.Then, the evaluation model based on improved ant colony clustering algorithm and RBF neural network is established. RBF neural network has remarkable advantages in function approximation, which has excellent learning capability and strong generalization ability. For the basic RBF neural network model, k-means clustering algorithm is used to determine data centers, which has high efficiency, but the clustering result is greatly influenced by selection of initial cluster centers. Ant colony clustering algorithm has high accuracy, but it takes long time to search for an optimal solution, so k-means clustering algorithm and ant colony clustering algorithm are combined, and local search is added to the ant colony clustering. Based on the improved clustering algorithm, a new RBF neural network evaluation model is established. Firstly, initial cluster centers of sample data for weapon operational efficiency are obtained through k-means clustering, and then ant colony clustering is used for further clustering, final clustering result is applied to determine data centers of RBF neural network and the corresponding extension constants, then RBF neural network connection weights are obtained through training, finally the evaluation of weapon operational effectiveness under specific meteorological and hydrological conditions is completed. Through certain weapon operational examples, the feasibility and validity of the evaluation model proposed are validated.Meanwhile, the overall design framework and functional requirements analysis of the weapon operational effectiveness evaluation system are introduced in detail, and then the system model is established through UML. Subsequently, class libraries are designed and developed. Finally, the development of weapon operational effectiveness evaluation system is completed based on the.NET platform, Oracle databases, ADO.NET database access technology and three-layer component model structure. In the conclusion part, the main work of this thesis is summarized and some problems for further study are considered.