Imbalanced data based fault diagnosis of the chiller via integrating a new resampling technique with an improved ensemble extreme learning machine. (1st July 2023)

• Record Type: Journal Article
• Authors: Zhang, Hanyuan; Yang, Wenxin; Yi, Weilin; Lim, Jit Bing; An, Zenghui; Li, Chengdong
• Is Part Of: Journal of building engineering. Volume 70(2023)
• Publication Date: 2023-07-01
• Subjects: Chiller -- Fault diagnosis -- Imbalanced data -- Hybrid resampling -- Selective ensemble -- Extreme learning machine; Building -- Periodicals; 690.05
• DOI: 10.1016/j.jobe.2023.106338
• Languages: English
• ISSNs: 2352-7102

Abstract

Abstract: Fault diagnosis of the chiller is essential to guarantee chiller's safe operation and reduce building energy consumption. However, the existing fault diagnosis methods rarely consider the chiller's imbalanced data conditions, which always leads to low diagnosis accuracy of the minority class samples. To figure out the issue of imbalanced fault pattern data during the chiller fault diagnosis, a hybrid resampling based improved extreme learning machine (HRIELM) is developed in our work. A new hybrid resampling technique (HRT) is first proposed to balance the majority and minority classes in the imbalanced fault pattern datasets. The HRT is then carried out repetitively to obtain multiple diverse rebalanced training datasets using different benchmark datasets. Subsequently, various primary extreme learning machine (ELM) models are established utilizing these rebalanced training datasets. Furthermore, an improved ELM model based on a novel selective ensemble learning strategy is presented to enhance the effectiveness of the chiller fault diagnosis. The basic ELM models with superior performance for diagnosing the neighbor subset constructed according to the fault snapshot sample are first picked out, and a weight factor is further defined to build the final selective ensemble ELM model. Finally, the fault pattern of the snapshot sample is identified according to the weighted voting mechanism. Detailed experimental results on the ASHRAE Research Project RP-1043 experimental datasets certify the effectiveness of the presented HRIELM scheme for the chiller fault diagnosis under the imbalanced data environments. Graphical abstract: As illustrated in Fig. 11, our proposed HRTELM based chiller fault diagnosis scheme involves two portions: the HRT based multiple basic ELMs offline modeling stage and the pattern of fault snapshot sample online diagnosis stage. Specifically, the HRT based multiple basic ELMs offline modeling stage consists of the hybrid resampling module and the basic ELM models training module, while the pattern of fault snapshot sample online diagnosis stage contains the selective ensemble ELM model construction module and the fault pattern identification module. The details of the developed HRIELM based fault diagnosis strategy is formulated in Section 6. Image 1 Highlights: The HRT technique is developed to rebalance the fault pattern datasets of the chiller. The standard ELM is employed as the basic diagnosis model to reduce the computational burden. An improved ELM is proposed by choosing and fusing the basic models with superior performance. The weighted voting mechanism is proposed to identify the fault pattern. Detailed experiments are made to verify the superiority of the proposed HRIELM scheme.