基于决策树算法的多联机气液分离器插反故障诊断

发布时间：2018-04-01 14:27

本文选题：决策树算法　切入点：故障诊断　出处：《制冷学报》2017年05期

【摘要】：本文将决策树算法应用于多联机气分插反故障诊断中,搭建了多联机实验平台采集数据,根据专家知识及数据变化模型验证选取了建模的特征变量,采用决策树C5.0算法构建气分插反故障诊断模型,进一步对由模型分类规则生成的最优变量即过冷器的EEV(电子膨胀阀)进行深入分析和验证。结果表明:将决策树算法应用于多联机气分插反故障诊断的方法,准确率为96%,此诊断方法能满足多联机故障诊断实际运用的需要,并可及时处理多联机发生气分插反故障时,系统过热度降低,从而保证多联机系统的制冷效果和能效比,可通过增大过冷器EEV开度调节。
[Abstract]:In this paper, the decision tree algorithm is applied to the multi-on-line gas division inverse fault diagnosis, a multi-on-line experimental platform is set up to collect data, and the characteristic variables of the model are selected according to the expert knowledge and the data change model. The decision tree C5.0 algorithm is used to construct the gas division interpolation reverse fault diagnosis model. Furthermore, the EEV (electronic expansion valve) of the supercooler, the optimal variable generated by the model classification rule, is analyzed and verified in depth. The results show that the decision tree algorithm is applied to the fault diagnosis method of multi-on-line gas division interpolation. The accuracy is 96. This diagnosis method can meet the needs of the practical application of multi-on-line fault diagnosis, and can deal with the multi-on-line gas division back fault in time, the system superheat is reduced, so as to ensure the refrigeration effect and energy efficiency ratio of the multi-on-line system. It can be adjusted by increasing the EEV opening of the supercooler.
【作者单位】：华中科技大学能源与动力工程学院;珠海格力电器有限公司;北京建筑大学供热供燃气通风及空调工程北京市重点实验室;合肥通用机械研究院压缩机技术国家重点实验室;
【基金】：国家自然科学基金(51576074&51328602) 2013年压缩机技术国家重点实验室开放基金(0214120035)资助项目供热供燃气通风及空调工程北京市重点实验室研究基金课题(NR2013K02)项目资助~~
【分类号】：TB657.2

【相似文献】