CIESC Journal ›› 2011, Vol. 62 ›› Issue (S2): 112-119.DOI: 10.3969/j.issn.0438-1157.2011.z2.019

Previous Articles     Next Articles

SVM-based FDD of multiple-simultaneous faults for chillers

GU Bo,HAN Hua,HONG Yingchun,KANG Jia   

  • Received:2011-03-17 Online:2011-12-20 Published:2011-12-20

基于SVM的制冷系统多故障并发检测与诊断

谷波,韩华,洪迎春,康嘉   

  1. 上海交通大学制冷与低温工程研究所
  • 通讯作者: 谷波

Abstract: Heating, ventilation, air-conditioning, and refrigeration (HVAC&R) systems operating under faulty condition often result in extra energy consumption (up to 30% for commercial buildings) and cost, less comfort control and bad indoor/outdoor air quality, especially when multiple faults happened simultaneously. Because of the diversity of individual faults in reality and the possible synergistic effect, detection and diagnosis of multiple-simultaneous faults (MSF) is one of the puzzles encountered by fault detection and diagnosis (FDD) experts. This study presents a novel hybrid strategy that combines support vector machine (SVM) and multi-label (ML) technique for the automated detection and diagnosis of multiple-simultaneous faults (MSF), and elaborates its application to a building chiller. One of the great advantages ML has against the mono-label (mL) technique is that no MSF data are needed for model training while good FDD performance for MSF could be obtained. Two individual chiller faults and one of their combinations (an MSF) were investigated. Detailed studies on the use of 8 fault indicative features from the previous study and the training of the model with/without normal or/and MSF data were conducted and compared with the mL-SVM model. The results showed that the ML-SVM model performed well, even without MSF samples for model training, and it could still detect and diagnose MSF effectively with a correct rate (CR) as high as 99. 902%. The performance of the detection and identification of MSF was even better than that trained with MSF data. That demonstrated a promising effort in searching for the solution to the MSF puzzle.

Key words: multiple-simultaneous faults, fault detection;fault diagnosis, support vector machine, refrigeration

摘要: 因现实中单发故障的多样性,以及各故障并发时可能存在的协同作用,使并发故障成为故障诊断界难点之一。利用支持向量机(SVM)优良的模式识别能力,分别与单标识(mL)及多标识(ML)技术结合,构建可用于并发故障检测与诊断的模型,应用于制冷机组双故障并发时的检测与诊断。结果表明,ML-SVM模型表现突出,训练时无需并发故障数据,却可用于并发故障的检测与诊断,且性能优良,总体诊断准确率(CR)达99.902%,故障检测及对并发故障的识别率甚至高于采用并发故障训练时的模型,具有良好应用前景。

关键词: 多故障并发, 故障检测, 故障诊断, 支持向量机, 制冷

CLC Number: