基于RBF神经网络优化的装备保障系统效能评估  

Effectiveness Evaluation of Equipment Support System Based on Optimized RBF Neural Network

作　　者：杜秀丽[1] 周敏[1] 吕亚娜 邱少明 DU Xiuli;ZHOU Min;LüYana;QIU Shaoming(Key Laboratory of Communication and Network,Dalian University,Dalian,Liaoning 116000,China)

机构地区：[1]大连大学通信与网络重点实验室,辽宁大连116000

出　　处：《计算机工程》2021年第9期282-287,296,共7页Computer Engineering

基　　金：装备发展部装备预研领域基金(61400010301)。

摘　　要：针对现有装备保障系统效能评估方法存在依赖专家经验、主观性强等问题,提出一种基于径向基函数(RBF)神经网络优化的装备保障系统效能评估方法。通过完备交叉与预变异策略对遗传算法(GA)进行改进,利用改进的GA对RBF神经网络中心、宽度及输出权值进行全局寻优,并采用优化的RBF神经网络实现装备保障系统效能评估。仿真结果表明,该方法相比AGA-RBF和GA-RBF评估方法能更准确地评估装备保障系统效能值,并且可在一定程度上降低评估过程中主观因素的影响,确保评估结果更客观真实。The existing methods for evaluating the effectiveness of equipment support systems rely on expert experience,and are thus influenced by subjective factors.To address the problem,a new evaluation method is proposed based on optimized Radial Basis Function(RBF)neural network.The method employs a complete crossover and pre-mutation strategy to improve the Genetic Algorithm(GA),which is then used to optimize globally the center,width and output weights of RBF neural network.Finally,the optimized RBF neural network is used to evaluate the effectiveness of equipment support systems.Simulation results show that the proposed method improves the accuracy of effectiveness evaluation for equipment support systems than AGA-RBF and GA-RBF methods.Also,it reduces the influence of subjective factors on evaluation to a certain extent,and ensures the objectivity of evaluation results.

关 键 词：装备保障系统 效能评估 径向基函数神经网络 遗传算法 适应度函数 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]