基于遗传神经网络算法后勤指挥决策能力预测  

A Prediction Algorithm for Command Capability of Logistics Command Organs Based on Neural Networks Optimized by Genetic Algorithm

作　　者：周文明[1,2] 范踊跃 宋吉星 张春荣 

机构地区：[1]后勤学院模拟训练中心 [2]国防大学军事后勤与军事科技装备教研部 [3]解放军96265部队 [4]第二炮兵后勤部物资油料部

出　　处：《火力与指挥控制》2015年第7期51-56,61,共7页Fire Control & Command Control

基　　金：中国博士后科学基金(201150M1560);武器装备预研基金资助项目(9140A27070109JB05)

摘　　要：分析了后勤指挥的主要影响因素,运用信息论中的信息压缩原理,建立和优化了后勤机关指挥决策能力预测指标体系。建立了遗传算法优化的BP神经网络后勤机关指挥能力预测模型,消除了BP神经网络存在的初始权值和阈值难以确定的局限性。给出了预测实例,证明了该算法的合理性和有效性。Key link and main influencing factors of the command capability of logistics command organs are analyzed. Using the theory of information compress,the evaluating index system of logistics command are established and optimized. The neural network is optimized by genetic algorithm,and it is used to predict the command capability of logistics command organs. A prediction example is presented. The characteristics of generalization,practicability and validity of the prediction model is verified,and it strongly support command,decision making and system optimum design of logistics command.

关 键 词：遗传算法 神经网络 后勤指挥 能力评估 

分 类 号：E926.4[军事—军事装备学]