基于改进PSO_BP神经网络的干燥机可靠性预测  

RELIABILITY PREDICTION OF DRYER BASED ON IMPROVED PSO_BP NEURAL NETWORK

作　　者：文昌俊[1,2] 陈哲 邵明颖 陈立 徐云飞 WEN ChangJun;CHEN Zhe;SHAO MingYing;CHEN Li;XU YunFei(School of Mechanical Engineering,Hubei University of Technology,Wuhan 430068,China;Hubei Province Key Laboratory of Modern Manufacturing Quality Engineering,Wuhan 430068,China;Hubei Sanjiang Aerospace Xianfeng Electronic Information Co.,Ltd.,Xiaogan 432000,China)

机构地区：[1]湖北工业大学机械工程学院,武汉430068 [2]湖北省现代制造质量工程重点实验室,武汉430068 [3]湖北三江航天险峰电子信息有限公司,孝感432000

出　　处：《机械强度》2023年第2期504-508,共5页Journal of Mechanical Strength

基　　金：国家自然科学基金项目(51875180)资助。

摘　　要：针对BP(Back Propagation)神经网络模型对谷物干燥机进行可靠性预测时,模型存在收敛速度慢和易陷入局部最优等问题,采用改进的粒子群算法对BP神经网络模型进行优化,建立PSO_BP神经网络的谷物干燥机可靠性预测模型,并与BP网络模型和GA_BP网络模型获得的MAERMSEMAPE指标进行对比。研究结果表明,采用改进的PSO_BP网络模型预测时,与BP网络模型相比三项指标分别降低了0.0518、0.0479和28.04%;与GA_BP网络模型相比,三项指标分别降低了0.0004、0.0002和0.61%,说明其具有更小的误差和较好的预测能力。为实现谷物干燥机可靠性精准预测提供方法和思路。When the BP neural network model is used to predict the reliability of the grain dryer,the model has problems such as slow convergence speed and easy to fall into local optimum.An improved particle swarm algorithm is used to optimize the BP neural network model and establish the PSO_BP neural network The reliability prediction model of grain dryer is compared with MAERMSEMAPE index obtained by BP network model and GA_BP network model.The research results show that when the improved PSO_BP network model is used for forecasting,the three indicators are reduced by 0.051 8,0.047 9 and 28.04% respectively compared with the BP network model;the three indicators are reduced by 0.000 4,0.000 2 and 0.61% respectively compared with the GA_BP network model,Which shows that it has smaller errors and better predictive ability.The methods and ideas for realizing accurate prediction of the reliability of grain dryers are provided.

关 键 词：谷物干燥机 粒子群算法 BP神经网络 可靠性预测 

分 类 号：TB114.35[理学—概率论与数理统计]