摘要
对基于神经网络方法的冷水机组故障监测效率取决于训练数据和被测数据的质量问题进行了研究。采用小波变换的方法剔除测量数据中的噪声,提高数据质量,从而提高冷水机组故障诊断效率。结果表明:采用小波变换使得各个水平故障的检测效率均得到提高,尤其水平一的故障检测效率提高明显。故障水平一检测率的提高能够及时的辨别冷水机组的故障,从而采用措施防止故障进一步恶化,对降低能源消耗、提高系统的可靠性以及保证室内舒适性具有重要的意义。通过利用ASHRAE Project提供的数据对故障诊断与检测(fault detection and diagnosis)策略进行验证,检测率明显提高。
Chiller fault detection based on neural network is a data - based analysis method. The fault detection efficiency relies on the quality of the training data and the mesasured data. The wavelet transfer method which can remove the measurement nosise is used to im- prove the detection efficiencies of chiller. The results show that wavelet transfer make the detection efficiencies of fault level improved, es- pecially the first level. The increase of the first level detection rate will be able to timely identify the chiller fault, and take the measures to prevent further deterioration of chiller fault, which is of important significance to reduce energy consumption and improve the reliability of the air-conditioning system and ensure the indoor thermal comfort. The FDD （ fault detection and diagnosis） strategy is validated through using ASHRAE Project data, which shows that the detection rate is improved obviously.
出处
《制冷学报》
CAS
CSCD
北大核心
2016年第1期12-17,共6页
Journal of Refrigeration
基金
国家自然科学基金(51328602)资助项目
2013年压缩机技术国家重点实验室开放基金项目(230031)
供热供燃气通风及空调工程北京市重点实验室研究基金资助课题(NR2016K02)项目资助~~
关键词
冷水机组
故障检测与诊断
神经网络
小波分析
贝叶斯正则化
chiller
fault detection and diagnosis
BP neural network
wavelet denoising
bayesian regularization
