基于投影寻踪法的某火炮选型研究

Study on the Selection of Guns Based on Projection Pursuit Classification

火炮选型是某装备论证的重点工作之一,论证时采取了选型评价指标定量对比、指标间综合分析的方法。但该方法缺乏对备选火炮定量评价数据的充分挖掘,不能反应评价数据特征。为提高火炮选型的准确性和客观性,采用投影寻踪法进行定量分类评价。通过分析火炮选型原则及装机适应性和火力性能要求,建立了由选型原则、选型评价指标、指标特征值分层的火炮选型评价指标体系,形成了由备选火炮选型指标特征值为样本集的评价模型;采用制约函数法构建投影指标优化模型,为避免初值和初始搜索方向引起的局部收敛问题,用多种群遗传算法求解;模型求解获得了最佳投影方向及投影值,得到了各备选火炮的选型定量评价排序以及评价指标权重。排序结果与论证时定性分析对比结果较为相符,且权重较高的评价指标与论证时确定的重点指标一致,验证了投影寻踪法可用于类似武器选型问题。

The selection of a gun is one of the key tasks in the demonstration of a certain type of equipment.Quantitative comparison of selection evaluation indexes and comprehensive analysis of the indexes were adopted in the demonstration.However,this method did not make full use of the quantitative evaluation data of the alternative gun and could not reflect the characteristics of the evaluation data.In order to improve the accuracy and objectivity of the gun selection,projection pursuit classification was used for quantitative classification evaluation.Principles of gun selection,adaptability to installation and firepower requirements were analyzed,and the evaluation index system of gun selection was established based on the selection principles,the evaluation indexes and the characteristic values of the index.An evaluation model was set up with the characteristic values of the selection indexes of candidate guns as the sample set.In order to avoid the local convergence problem caused by initial va-lues and initial search directions,a projection index optimization model was constructed by using the constraint function method and it was solved by using the multi-population genetic algorithm.The optimal projection direction and projection value were obtained by solving the model,and the selection and quantitative evaluation ranking and evaluation index weight of each candidate gun were obtained.The ranking results are in good agreement with the comparison results of the qualitative analysis during the demonstration,and the evaluation indexes with higher weight are consistent with the key indexes determined during the demonstration,which verifies that this method can be applied to solve similar weapon selection problems.