Research On Performance Test And Fault Diagnosis Of Gas Pressure Regulator

In recent years,Chinese natural gas industry has developed rapidly,and the total natural gas consumption has been increasing year by year.How to effectively realize the safe transmission of natural gas has become a major issue that needs to be overcome in the further development of the natural gas industry.As an important part of today’s urban gas delivery,gas pressure regulator achieves a stable output of downstream pressure through the balance between the pressure inside the valve diaphragm and the force of the spring.Any failure during its operation may lead to the leakage of flammable gas.Which in turn caused immeasurable economic losses and irreparable casualties.Therefore,performance testing before leaving the factory and fault diagnosis after leaving the factory have become the current focus of research in the field of voltage regulators.Based on this,the thesis designed a gas pressure regulator performance detection system,and based on the system to carry out data measurements,carried out the gas pressure regulator fault diagnosis research.(1)The working principle of the gas pressure regulator is briefly described,and the technical parameters of the tested pressure regulator are introduced.Combined with the current national standard "Urban Gas Pressure Regulator"(GB 27790-2011),the industry standard "Urban Gas Shut-off Valve and Relief Valve"(CJ/T 335-2010)and enterprise requirements,the functional requirements of the detection system have been specifically formulated,Analyzed the detection principle of each detection link,and designed the overall performance detection program based on this.(2)A complete gas pressure regulator performance testing system was designed and built to realize the performance testing of the regulator before leaving the factory.The detection system uses PLC as the main control unit,combined with an industrial touch screen to achieve good human-computer interaction,which can realize automatic detection of the static characteristics,closing pressure,release pressure,and air tightness of the gas pressure regulator.(3)An "end-to-end" gas pressure regulator fault diagnosis algorithm based on the improved convolutional neural network was designed to realize the fault diagnosis of the operating state of the pressure regulator.First,use a one-dimensional convolutional neural network as a feature extraction network,and use the original one-dimensional regulator outlet pressure as an input signal.secondly,use a multi-scale convolution kernel in the first convolution layer to extract more adequate faults in parallel Feature.Then,the SVM classifier is introduced to replace the Softmax classifier in the traditional convolutional neural network,which not only solves the difficult and tedious problem of traditional SVM feature extraction,but also strengthens the classification ability of the model under a small number of samples.finally,based on the pressure adjustment A data acquisition system was built on the performance testing platform of the regulator,and the effectiveness of the algorithm was verified under the obtained voltage regulator fault data set.(4)A fault diagnosis algorithm based on W(Weighted)-Dense Net is designed to realize fault diagnosis under unbalanced samples of the voltage regulator.First,the dense convolutional neural network is used as the basic feature extraction network.secondly,the weighted cross-entropy loss function is combined with the corresponding penalty coefficients for different types of samples to achieve a weighted average of the unbalanced sample errors.then,through the dimensional comparison experiment,it is determined Using the two-dimensional grayscale image reconstructed from the original one-dimensional outlet pressure signal data of the voltage regulator as the algorithm input can enable the algorithm to obtain a higher recognition accuracy.finally,the voltage regulator failure data set under different balances The classification performance experiment was carried out on the above,and the results show that compared with other mainstream algorithms,W-Dense Net can effectively reduce the difficulty of classification of fault categories caused by unbalanced samples.(5)In order to verify the superiority of the performance of the detection system,a batch of gas pressure regulators were tested based on the detection system.The test results show that the detection system can accurately judge whether the performance of the tested voltage regulator is qualified or not,and the machine capability index calculated by repeated tests meets the requirements.Aiming at the fault diagnosis research of the pressure regulator after delivery,a set of gas pressure regulator fault diagnosis software system with good interaction and simple operation has been developed.The software system incorporates the proposed "end-to-end" fault diagnosis algorithm and unbalanced sample fault diagnosis algorithm.The effectiveness of the proposed algorithm is verified by software system classification experiments.