改进遗传神经网络算法在自动借阅图书系统故障检测中的应用研究  

Application of improved genetic neural network algorithm in automatic book borrowing system fault detection

作　　者：白国政[1] BAI Guozheng(Shaanxi Polytechnic Institute,Xianyiang Shaanxi 712000,China)

机构地区：[1]陕西工业职业技术学院,陕西咸阳712000

出　　处：《自动化与仪器仪表》2023年第5期65-69,共5页Automation & Instrumentation

基　　金：2021年度陕西省高等职业教育教学改革重点项目《中国特色高水平计算机应用技术专业群的研究与实践》(21GZ013)。

摘　　要：为进一步提高图书借阅服务系统的故障检测性能,提出一种改进的神经网络预测模型进行图书自动借阅系统的故障诊断。其中,使用遗传算法对神经网络进行改进,解决神经网络存在的易陷入局部最小值和梯度消失问题,并进行图书自动借阅系统的电路故障预测。实验结果表明,与其他模型相比,本研究提出的遗传神经网络具有更好的性能,AUC值和准确率分别达到了0.71127、85.3%;在进行图书自动借阅系统的电路故障预测时,对于两个电路元件的故障预测误差分别为1.3625E-07和3.2592E-05。以上结果表明,使用本研究的遗传神经网络模型进行图书自动借阅的故障检测时能够取得良好的效果,具有一定的实际应用价值。In order to further improve the fault detection performance of the service robot,an improved neural network prediction model is proposed for fault diagnosis of the automatic book borrowing system.Among them,genetic algorithm is used to improve the neural network to solve the problem that the neural network is easy to fall into the local minimum and the gradient disappears,and to predict the circuit failure of the automatic book borrowing system.The experimental results show that compared with other models,the genetic neural network proposed in this study has better performance,with AUC value and accuracy reaching 0.71127 and 85.3%respectively;During the circuit fault prediction of the automatic book borrowing system,the fault prediction errors of the two circuit components are 1.3625E-07 and 3.2592E-05 respectively.The above results show that using the genetic neural network model in this study to detect the failure of automatic book borrowing can achieve good results,and has certain practical application value.

关 键 词：神经网络 遗传算法 故障检测 

分 类 号：TP392[自动化与计算机技术—计算机应用技术]