|
| 基于现场传感器参数及SVM的冷水机组故障诊断 |
| Research on chiller fault diagnosis based on field sensor parameters and support vector machine |
| 投稿时间：2018-11-16 |
| DOI：10.13259/j.cnki.eri.2021.03.004 |
| 中文关键词: 冷水机组 支持向量机 故障检测与诊断 现场传感器 模型 |
| 英文关键词:chillers support vector machine fault detection and diagnosis field sensors model |
| 基金项目:国家自然科学基金资助项目（51506125） |
|
| 摘要点击次数: 1977 |
| 全文下载次数: 2311 |
| 中文摘要: |
| 为了使冷水机组故障诊断模型在现场运行中得以推广，采用冷水机组现场运行的传感器参数，基于支持向量机（SVM），对离心式冷水机组的7类故障建立诊断模型SVM−site，并与采用原始64个参数的SVM64模型进行对比分析。结果表明，SVM−site模型耗时减少，且具有较佳的诊断性能，基本满足现场诊断需求。若继续增加润滑油供油压力和供油温度两个传感器，可有效降低虚警率，显著提升润滑油过量和冷凝器结垢故障的诊断性能。可见，现场传感器参数基本可以满足故障在线诊断的需求，所建立的模型具有较高诊断性能。适当增加传感器参数，可使诊断模型的表现更加出色，从而具有更加良好的应用前景。 |
| 英文摘要: |
| In order to promote the application of chiller fault diagnosis model to field operation, a diagnostic model named “SVM-site” for seven typical faults of centrifugal chillers was established using the sensor parameters of the on-site operation chiller and support vector machine (SVM) in this study, which was compared with the original SVM64 model with 64 parameters. It showed that the model SVM-site was time-saving and had better diagnostic performance, which basically met the requirements of on-site diagnosis. Two parameters of PO_feed and TO_feed from the field sensors could effectively reduce the normal false alarm rate and significantly improved diagnostic performance of ExcsOil and ConFoul faults. Therefore, the parameters from the sensors installed in the existing chillers could basically meet the requirements of fault diagnosis online. The proposed model had high diagnostic performance. The increased sensor parameters could make the performance of the diagnostic model even better and had a good application prospect. |
| HTML 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
