The dynamic hydropower troubleshooting information based on EMD multi-scale feature entropy extraction. (2017)

• Record Type: Journal Article
• Authors: Lu, Shibao; Wei, June; Bao, Haijun; Xue, Yangang; Ye, Weiwei
• Is Part Of: International journal of mobile communications. Volume 15:Number 6(2017)
• Page Start: 677
• Page End: 692
• Publication Date: 2017
• Subjects: energy index -- extraction of characteristics -- failure mode identification -- mobile communication technology -- multi-scale characteristic entropy; Mobile communication systems -- Periodicals; Wireless communication systems -- Periodicals; 384.53
• Languages: English
• ISSNs: 1470-949X

Abstract

Hydropower is a kind of clean energy which is renewable and pollution-free with low operating costs. However, the vibration of the hydraulic turbine generator which has not yet been effectively resolved has seriously affected the efficiency of hydroelectricity exploitation. This report includes the multi-scale entropy analysis of the fluctuating signals created by pressure within the hydraulic turbine's draft tube. The analysis is based on the empirical model decomposition method, using the mobile communication technology. The signal was resolved into multiple intrinsic mode functions (IMF) situated on a local characteristic time scale. Energy level indexes were then calculated according to these IMFs. These indexes were then used in order to establish the entropy's multi-scale characteristic value. Next, the entropy's value was used as eigenvector for the identification of different failure modes. Tests were conducted using the fluctuations in the pressure signals created through the mobile communication. The results of these tests show that this method is highly accurate and that it is effective when used to extract eigenvectors in the context of hydraulic turbine generator units. The method was relatively accurate where the extraction of highly complex and specific data relating to the dynamic characteristics of a hydraulic turbine generator was concerned.