Research on Early Warning of Small Leakage Faults in Primary Circuit Based on Data Mining

Abstract

In order to prevent the emergence of serious accidents caused by the excessive evolution of small leakage in the primary loop, an improved Gaussian mixture model- Grey relational analysis-Entropy weight method (GMM-GRA-EWM) based on multi-characteristic parameter synthesis is proposed. First, analyze the mechanism of the dynamic operating characteristics of the small leakage of the primary circuit, and determine the early warning characteristic parameters. Secondly, based on the determined early warning characteristic parameters, a multi-parameter comprehensive early warning model is established by using the method of combining entropy weight method and gray correlation degree. Finally, correlation analysis and improved Gaussian mixture model algorithm are used to effectively learn the statistical characteristics of a large amount of data, so that the early warning threshold has an adaptive ability under different working conditions. The verification and comparison results show that the method can still achieve an effective, stable and accurate early warning effect under variable operating conditions, which can provide a reference for realizing the status monitoring of the primary loop system.