Research on the intelligent diagnosis method of the server based on thermal image technology
Autor: | Bao Chenchen, Xianlin Song, Hang Liu, Shan Gao, Ting Xie, Weina Wang |
---|---|
Rok vydání: | 2019 |
Předmět: |
Contextual image classification
business.industry Computer science Principle of maximum entropy Feature extraction ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION 02 engineering and technology 021001 nanoscience & nanotechnology Condensed Matter Physics 01 natural sciences Atomic and Molecular Physics and Optics Electronic Optical and Magnetic Materials 010309 optics Support vector machine Homomorphic filtering Server 0103 physical sciences Pattern recognition (psychology) Computer vision Segmentation Artificial intelligence 0210 nano-technology business |
Zdroj: | Infrared Physics & Technology. 96:390-396 |
ISSN: | 1350-4495 |
Popis: | Balancing the uneven temperature distribution of data centers can reduce cooling energy consumption significantly, as the overheated surface of servers is one of the root causes of the uneven distribution. This paper presents a method for intelligently diagnosing server operating status based on the heat distribution of the server surface. Five types of server operation can be diagnosed: normal status, main fan failure, vice-fan failure, air vent blockage and low-load status. The method involves signal processing and pattern recognition techniques such as thermal image enhancement, region segmentation and image classification. First, thermal images of server outlets in running status are captured as data; second, the images are preprocessed for standardization; third, after homomorphic filtering enhancement, the images are subjected to one-dimensional maximum entropy segmentation to obtain server hotspot images; fourth, morphological features, texture features and statistical features are extracted from hotspot images; finally, the server status is diagnosed by a support vector machine. Experiment results show that the method achieves a diagnosis accuracy of 91.111%. This method can solve the problem of uneven heat distribution in data centers effectively. |
Databáze: | OpenAIRE |
Externí odkaz: |