基于改进K-SVM的防疫物资生产设备健康度识别方法研究

K-SVM-based health identification algorithm for manufacturing equipment of epidemic prevention materials

目的:为有效识别防疫物资生产设备的工况,提出一种基于改进K-SVM的防疫物资生产设备健康度识别方法。方法:首先,通过K-means算法过滤掉部分不利于支持向量机(support vector machine,SVM)分类器样本训练的数据;其次,利用K-means算法及特征提取算法获取新的SVM分类器训练样本,并依据此样本训练SVM分类器;最后,利用训练好的SVM分类器对初始样本进行分类从而得到最终的预测结果。为验证改进K-SVM算法对设备健康度的识别性能,将该算法与SVM算法、未改进K-SVM算法、XGBoost算法进行对比实验。结果:改进K-SVM算法能较为准确地识别设备故障点,识别准确率为89.79%,优于SVM算法、未改进K-SVM算法和XGBoost算法。结论:提出的改进K-SVM算法能够较好地识别防疫物资生产设备健康度,对保证防疫物资生产设备的工作效率具有重要意义。

Objective To propose a K-SVM-based manufacturing equipment health identification algorithm for epidemic prevention materials to understand the working conditions of the equipment effectively.Methods Firstly,K-means algorithm was used to filter out some data which was not conducive to SVM classifier training.Secondly,new SVM training samples were obtained with K-means and feature extraction algorithms,and SVM classifier was trained accordingly.Finally,the initial samples were classified with the trained SVM classifier to get the final prediction result.The improved K-SVM algorithm proposed was compared with K-SVM algorithm,unimproved K-SVM algorithm and XGBoost algorithm to verify its performance for identifying equipment health.Results The improved K-SVM algorithm detected the fault points of the equipment effectively with an accuracy of 89.79%,which gained advantages over K-SVM algorithm,unimproved K-SVM algorithm and XGBoost algorithm.Conclusion The improved K-SVM algorithm proposed behaves well in identifying the manufacturing equipment health identification,and is of great significance for ensuring the working efficiency of the manufacturing equipment of epidemic prevention materials.