【作者】 唐鹏； 陈志文； 骆伟超； 曹婷； 梁骁俊； 任浩； 张超波；

【Author】 Tang Peng;Chen Zhiwen;Luo Weichao;Cao Ting;Liang Xiaojun;Ren Hao;Zhang Chaobo;Peng Cheng Laboratory;School of Automation,Central South University;

【机构】 鹏程实验室； 中南大学自动化学院；

【摘要】 空调系统是能源消耗的重要来源,冷水机组作为空调系统核心部件,对其进行故障检测与诊断具有重要意义。由于冷水机组的故障数据较难获取,同时冷水机组运行负载频变,为冷水机组故障检测与诊断带来困难。本文采用了条件变分自编码器实现对变工况运行数据的表征,然后构建检测统计指标进行故障检测,最后通过ASHRAE RP-1043数据集证明了所提算法能够有效检测出不同类型故障。更多还原

【Abstract】 Air conditioning system is the important source of energy consumption. As the core component of air conditioning system, it has significant meaning to achieve fault detection and diagnosis for chillers. The difficulty to capture the fault data of chillers and the frequent fluctuation of operating load bring difficulty for fault detection and diagnosis of chillers. The conditonal variational autoencoder model is used to represent the data features with frequent changes of operating conditions. Then, the detection statistical indexs are built for fault detection. Finally, The ASHRAE RP-1043 dataset is used to prove that the proposed method can effectively detect different kind of fault.更多还原