基于BP神经网络模型的电弧织物防护性能预测  

Protection Performance Prediction of Arc Fabric Based on BP Neural Network Model

作　　者：蒋伟 徐韵扬 顾虎 秦鑫 JIANG Wei;XU Yunyang;GU Hu;QIN Xin(Hangzhou Institute of Quality&Metrology,Hangzhou 310019,China)

机构地区：[1]杭州市质量计量科学研究院,杭州310019

出　　处：《纺织科技进展》2024年第8期17-20,共4页Progress in Textile Science & Technology

基　　金：国家市场监管总局科技项目(2021MK139)。

摘　　要：电弧防护性能的预测关系着相关作业人员的人身安全问题。试验建立了ATPV值和EBT值的BP神经网络预测模型,探究不同的输入层节点个数、隐含层节点个数和训练次数对模型预测精确度的影响。通过BP网络模型试验验证,实测值和预测值的对比分析,BP模型预测的ATPV值与实际测试值的结果误差率小于5%,EBT值的BP模型预测值与实际测试值的结果变化趋势吻合度达到95%以上。试验结果显示,BP神经网络模型对于ATPV值和EBT值的预测准确度满足防电弧织物性能评价。The prediction of arc protection performance is related to the personal safety of relevant operators.A BP neural network prediction model with ATPV value and Er value was established,while the influence of different input layer node number,hidden layer node number and training times on the model prediction accuracy was explored.Through the BP network model test and the comparative analysis of the measured value and the predicted value,the error rate of the ATPV value predicted by the BP model and the actual test value is less than 5%,and the agreement of the predicted value of the BP model and the actual test value is more than 95%.The experimental results show that the prediction accuracy of BP neural network model for ATPV value and Eπvalue can satisfy the performance evaluation of anti arc fabric.

关 键 词：BP神经网络 ATPV值 EBT值 模型预测 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TS106[自动化与计算机技术—控制科学与工程]