Cavitation Noise Signal Classification of Hydroturbine Based on Improved Multi-Scale Symbol Dynamic Entropy

Abstract

Cavitation is a phenomenon in the operation of hydroturbine, which is related to the operation efficiency and service life of the turbine. To identify both the cavitation noise signal and the non-cavitation noise signal, prevent damage as soon as possible, and avoid irreversible damage to the hydroturbine, a new paradigm based on multi-scale information entropy is proposed in this paper. The new proposed classification model combines improved multi-scale symbol dynamic entropy (IMSDE) and least square support vector machine (LSSVM). Improved multi-scale symbol dynamic entropy is utilized to learn features from the cavitation noise signal, and then the classifier of the least square support vector machine is used to classification. Multi-scale sample entropy (MSE), multi-scale permutation entropy (MPE) and multi-scale symbol dynamic entropy (MSDE) are selected as the contrast algorithms. According to the experimental results of four different operating conditions, IMSDE has the highest recognition rate. The average recognition rate of IMSDE is higher than that of MSDE, MSE and MPE. There is no significant difference in computational efficiency of IMSDE, MSDE and MPE. In conclusion, the IMSDE method proposed in this paper is superior to MSDE, MSE and MPE, for meeting the requirements of cavitation noise signal classification.