摘要
本研究针对水源热泵机组常见的6种热力故障,尝试使用SOFM神经网络进行故障诊断。利用水源热泵机组试验台人为制造制冷剂充注量过多、制冷剂泄漏、膨胀阀开度过大与过小、冷却水管路阻塞、系统含不凝性气体共计6种热力故障,记录机组带故障运行时的运行参数,将收集到的参数进行归纳整理,提取出部分特征值制作成数据集。将数据集划分为训练集合与测试集合,前者用于神经网络的训练,后者用于验证神经网络故障的诊断效果。结果表明,SOFM神经网络对于本次实验人为制造出的6种水源热泵热力故障具有较高的诊断正确率,网络迭代500次,用时2.7 s,在有效诊断的同时具有较快的响应速度。
SOFM neural network is used to diagnose six common thermal faults of water source heat pump units. Using water source heat pump test-bed man-made refrigerants filling quantity is too much, refrigerant leakage, expansion valve opening through big and small, cooling water pipe jam, system contains no non-condensable gas for a total of six kinds of thermal failure, recording unit operation parameters of fault runtime, organizing the collected parameters, and to extract the feature of values into a data set. The data set is divided into training set and test set. The former is used to train the neural network, and the latter is used to verify the diagnosis effect of neural network faults. The results show that SOFM neural network has a high diagnostic accuracy for six kinds of thermal faults of water source heat pump manufactured artificially in this experiment. The network iteration is 500 times and the time is 2.7 s. It has a fast response speed and effective diagnosis.
作者
赵玉清
邵剑峰
ZHAO Yuqing;SHAO Jianfeng(North China University of Technology,Beijing 100144,China)
出处
《建筑节能（中英文）》
CAS
2023年第1期115-118,144,共5页
Building Energy Efficiency
