多联式空调制冷系统堵塞故障位置辨识仿真

Simulation of Blocking Fault Location Identification in Multi Connected Air Conditioning Refrigeration System

采用当前方法辨识制冷系统堵塞故障位置时,由于没有对相关数据进行预处理,导致辨识结果召回率低、误报率高、漏检率高。于是提出多联式空调制冷系统堵塞故障位置辨识方法。将待辨识的多联式空调制冷系统堵塞故障位置设为观测点,利用迭代检验方法对观测点时间序列进行清洗,消除对故障位置辨识贡献较低的数据,提高召回率。利用数据的能量熵、近似熵特征构建堵塞故障数据特征集合,完成堵塞故障特征融合提取,以降低误报率,采用基于交叉验证的SVM故障分类模型完成堵塞故障位置的辨识。实验结果表明,所提方法的召回率高、误报率和漏检率较低,实际应用效果更好。

When the current method is used to identify the blocking fault location of refrigeration system, due to the lack of preprocessing of relevant data, the identification results have low recall rate, high false alarm rate and high missed detection rate. Therefore, we proposed a method for identifying the location of blockage fault in multi connected air conditioning refrigeration system. The location of the blockage fault of the multi-connected air conditioning refrigeration system to be identified was set as the observation point. According to the iterative test method, the time series of observation points were cleaned to eliminate the data with low contribution to fault location identification, improving the recall rate. In order to reduce the false alarm rate, the energy entropy and approximate entropy features of data were applied to establish the feature set of blocking fault data, thus achieving the fusion extraction of blocking fault features. SVM fault classification model based on cross validation was introduced to identify the location of blocking fault. The experimental results show that the method has high recall rate, low false alarm rate and missing detection rate, and excellent practical application effect.