基于Boosting的多联机制冷剂充注量故障诊断集成模型

多联机空调系统被广泛用于各种公共建筑物,一旦发生故障会导致舒适性降低,能耗增加。制冷剂充注水平是影响空调系统高效运行的重要参数。本文提出一种基于Boosting集成算法的故障诊断模型,以制冷剂充注量故障为研究对象,将逻辑回归、决策树、随机森林、支持向量机和BP神经网络等5个基分类器集成,使用卡方检验进行特征选择,并使用制冷、制热模式的实验数据建立诊断模型。结果表明:基于Boosting的集成模型能高效检测多联机制冷剂充注量的故障,准确率高达96. 8%,相比于传统故障检测方法,大幅提高了诊断模型的响应速度、准确度和实用性。

Boosting-based Refrigerant Charge Fault Diagnosis Integration Model of Variable Refrigerant Flow System

Variable refrigerant flow (VRF) air-conditioning systems are widely used in various public buildings. If a fault occurs, it will result in reducing comfort and increasing energy consumption. The refrigerant charging level is an important parameter affecting the efficient operation of the air-conditioning system. In this paper, a fault diagnosis model based on Boosting integrated algorithm is proposed by taking refrigerant charge fault as the research object. Five basic classifiers, such as logistic regression, decision tree, random forest, support vector machine and BP neural network, are integrated. Chi-square test was used for feature selection, and the diagnostic model was established with experimental data for cooling and heating modes. The results show that the Boosting-based integrated model can efficiently detect the fault of VRF refrigerant charge, and the accuracy rate of the model is up to 96.8%. Compared with the traditional fault detection method, the proposed model greatly improves the response speed, accuracy and practicability of the diagnostic model.