Valve inner leakage defect type recognition and inner leakage rate calculation method

Abstract

The invention provides a valve inner leakage defect type recognition and inner leakage rate calculation method. The method comprises the step one of carrying out valve inner leakage detection experiments based on the acoustic emission technology and obtaining experimental data; the step two of extracting the data of valve features, process parameters, acoustic emission signal features and the like and constructing high-dimensional feature space; the step three of carrying out locality preserving projection and dimension reduction on the data of the high-dimensional feature space and extracting the low-dimensional space features; the step four of setting up a valve inner leakage defect type recognition model based on classification of support vectors, selecting the RBF kernel function, determining the optimal parameters of the model according to the particle group algorithm, and inputting the low-dimensional space flag data for carrying out training; the step five of marking no-label data samples influencing the model largely based on the active-learning method, and setting up a valve inner leakage rate calculation model of the regressive support vectors; the step six of utilizing the model to forecast the inner leakage defect types and the inner rates of a valve to be tested. According to the method, dependence on the number of the data samples is lowered, and the problem of difficulty of valve inner leakage quantitative detection can be solved effectively.

Description

technical field [0001] The invention relates to the field of valve internal leakage detection, in particular to identification of valve internal leakage defect types and calculation of internal leakage rate. Background technique [0002] In recent years, valve internal leakage detection based on acoustic emission technology has become a research hotspot. Some researchers, starting from the acoustic emission mechanism, proposed the Kaewwaewnoi model and E.Meland model for valve internal leakage rate estimation by analytical method. However, the premise of applying these models is to determine the shape of the leakage hole, which is difficult to obtain directly in the engineering site. Most researchers focus on the characteristics of the acoustic emission signal of the internal leakage of the valve, and have made good progress in the preprocessing and feature extraction of the acoustic emission signal of the internal leakage of the valve. For the prediction of the internal le...

AI Technical Summary

Problems solved by technology

[0003] Aiming at the problems of few data samples and insufficient utilization of data characteristic parameters when detecting valve internal leakage based on acoustic emission technology, the present invention proposes a valve internal leakage defect from the fields of manifold learning dimensionality reduction, active learning, and support vectors Type Identification and Calculation Method of Endoleak Rate

Images