The Further Study Of Fault Diagnosis For Gas Regulator

Along with the rapid development of gas industry, gas transmission and distributionsystem has been expanded quickly all over the city. Gas regulator is a key component of thegas transmission and distribution system. Its operation directly affects the safety andreliability of gas transmission and distribution systems. Gas regulator fault diagnosistechnology research has very important social and economic significance. Wavelet neuralnetwork fault identification technology is an important research direction of mechanicalequipment fault diagnosis field in recent years.Accordingly, this paper studied fault diagnosis technology research status at home andabroad, highlighting the wavelet analysis and neural network integration approach.Meanwhile, it is studied that the basic theory of the wavelet and wavelet packet analysis, andsystematic analysis of the gas regulator failure mechanism and failure characteristics based onthe working principle of the gas regulator. Wavelet transform with time-frequency localizationproperties can focus the object to the details. The wavelet packet transform based on thewavelet transform is better for the local time-frequency analysis of the high-frequency signal.In this paper, the regulator outlet pressure signal was transferred to believe that the signalfrequency band by wavelet packet decomposition technology. By the corresponding change inthe proportion of bands in the energy regulator extract the fault feature.Neural network is a powerful self-organizing, self-learning and self-memory patternrecognition technology, and is more widely in the machinery equipment fault diagnosis field.BP network is a multi-layer feed-forward neural network universal approximation and can beany nonlinear mapping form input to output. Because it uses error back-propagation to adjustweight. The fault characteristics extracted by wavelet packet energy detection technique isfeed in the neural network. BP neural network is established. Then, the neural network is usedto identify the three states of the regulator. It is test that the wavelet packet analysis withneural network for gas regulator line of fault diagnosis method is effective.