基于改进ACO-BP算法的弹药贮存可靠性评估  

Evaluation of Ammunition Storage Reliability Based on Improved ACO-BP Algorithm

作　　者：刘芳[1] 王宏伟 宫华[1] 许可[1] LIU Fang;WANG Hongwei;GONG Hua;XU Ke(College of Science, Shenyang Ligong University, Shenyang 110159, China;Liaoning Huaxing Mechanical and Electrical Co., Ltd., Jinzhou 121017, China)

机构地区：[1]沈阳理工大学理学院,沈阳110159 [2]辽宁华兴机电有限公司,辽宁锦州121017

出　　处：《兵器装备工程学报》2019年第4期177-181,共5页Journal of Ordnance Equipment Engineering

基　　金：辽宁省高等学校基本科研项目(LG201715);辽宁省科学技术计划项目(20170540790);沈阳市中青年科技创新人才支持计划项目(RC170392)

摘　　要：针对BP算法易陷入局部最优和收敛速度慢的问题,提出两种改进蚁群优化BP神经网络的可靠性评估算法(自适应蚁群优化BP神经网络评估算法和精英蚁群优化BP神经网络评估算法优化网络的初始配置;实验结果表明:两种智能模型都显著提高了BP网络的精度和稳定性,减少了网络的迭代次数;前一种算法在评估的精度和迭代次数方面优于后一种算法,而后一种算法比前一种算法更稳定。BP algorithm is easy to fall into the problem of local optimum and slow convergence. Two reliability evaluation algorithms were proposed. They were adaptive ant colony optimization BP neural network evaluation algorithm (AACA-BP) and elite ant colony optimization BP neural network evaluation algorithm (EACO-BP). Two algorithms were used to optimize the initial configuration of the network. The experimental results show that the accuracy and stability of BP network are improved and the number of iterations is reduced with two intelligent algorithms. EACO-BP algorithm is superior to AACA-BP algorithm in evaluation accuracy and iteration times. AACA-BP algorithm is more stable than EACO-BP algorithm.

关 键 词：弹药 贮存可靠性 评估算法 BP神经网络 蚁群 自适应蚁群 精英蚁群 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]