基于特征数据聚类的紫外线消毒系统运行状态实时监测方法

Real-Time Monitoring Method of Ultraviolet Disinfection System Operating State Based on Feature Data Clustering

目的为提高对紫外线消毒系统运行效果的监测质量,提出基于特征数据聚类的紫外线消毒系统运行状态实时监测方法。方法首先对紫外线消毒系统框架进行分析,完成喷射器的消毒过程进行建模,然后基于卷积神经网络提取紫外线消毒系统历史、技术监测信息等属性的运行状态特征,最后利用优化的K均值聚类算法,对紫外线消毒系统运行状态进行实时监测。结果实验结果证明,该方法的聚类性能和异常检测性能均较好,聚类准确率在96%以上,异常检测的平均检测率为97.49%,误检率小于1.89%。将其应用于医院进行实证,全天可对监测对象监测3000余次。结论基于特征数据聚类的紫外线消毒系统可有效监测喷射器出口温度变化情况、耐高温菌群消杀效果,保证医院各个科室消杀运行稳定情况,具有较好的应用价值。

Objective In order to improve the monitoring quality of ultraviolet disinfection system operation efect,to propose a realtime monitoring method of ultraviolet disinfection system operating state based on feature data clustering.Methods First,the framework of the ultraviolet disinfection system was analyzed,and the disinfection process of the ejector was modeled,and then the operating state characteristics of the history and technical monitoring information of the ultraviolet disinfection system were extracted based on the convolutional neural network.Finally,the optimized K-means clustering algorithm was used to conduct realtime monitoring of the operating state of the ultraviolet disinfection system.Results The test results showed that the clustering performance and anomaly detection performance of this method were both good.The clustering accuracy rate was more than 96%,the average detection rate of anomaly detection was 97.49%,and the false detection rate was less than 1.89%.It was applied to hospitals for demonstration,and the monitored objects could be monitored more than 3000 times throughout the day.Conclusion The ultraviolet disinfection system based on feature data clustering can efectively monitor the temperature change at the outlet of the injector and the sterilizing efect of high-temperature resistant bacteria,and ensure the stable operation of disinfection and sterilization in each department of the hospital,which has great application value.