Font Size: a A A

Quantitative Study Of Valve Internal Leakage Rate Based On Factor Analysis And Wavelet-BP Neural Network

Posted on:2020-11-02Degree:MasterType:Thesis
Country:ChinaCandidate:Y YuFull Text:PDF
GTID:2381330614965318Subject:Mechanical engineering
Abstract/Summary:PDF Full Text Request
Valve as an important accessory for control pipeline transportation,its operation involves a wide range.It may have a certain impact on pipeline transportation,once the valve leaks.Meanwhile,acoustic emission technology,as an emerging non-destructive testing technology,can be used in the performance testing of valves in operation.The main work of this paper is as follows:(1)Fully investigate the relevant technical progress of valve internal leak detection,and optimize the design of a better performance internal valve leakage detection instrument based on the original research.(2)Based on SPSS 22.0(Statistical Product and Service Solutions),this paper aims to reduce the dimensionality of these variables by factor analysis to obtain lower-dimensional sample feature sets for the high-dimensional characteristic variables of the leaking acoustic emission signals collected by the laboratory-developed testing instruments.(3)Based on the low-dimensional sample feature set obtained by factor analysis,wavelet three-layer decomposition is performed to increase the number of samples and improve the stability of the valve internal leakage quantification model.This paper uses the trial and error method to determine the structure of the neural network model.By comparing the predictive performance of the three models(the wavelet-BP neural network model based on the sample feature set obtained by dimensionality reduction,the BP neural network model(error Back Propagation Neural Network)based on the reduced-dimensional sample feature set,and the wavelet-BP neural network model based on the feature set of the unmet-dimensioned sample are built into the valve.Leak rate quantification model).The wavelet-BP neural network valve leakage rate quantification model constructed by the optimal feature-based sample feature set based on dimensionality reduction is selected as the valve internal leakage rate quantitative prediction model.(4)Finally,laboratory simulation experiments and field tests were carried out on the valve internal leak detection instrument to verify the performance of the instrument,to guarantee the safety for the safe operation of the gas pipeline.
Keywords/Search Tags:Valve Internal Leak Detection, Factor Analysis Dimensionality Reduction, Wavelet-BP Neural Network, Quantification of Valve Internal Leakage Rate
PDF Full Text Request
Related items