Application Research Of Intelligent Fault Diagnosis System For Gas Equipment

In the gas pipeline network system,gas regulator plays a relatively important role,but the gas group in the operation and maintenance of gas regulator,common problems are: 1)the gas regulator fault diagnosis technology is relatively backward,intellectualized degree is not high;2)gas group take regular maintenance strategy,lead to greatly increase the cost of operation and maintenance.Based on this,according to gas pressure regulator,fault intelligent diagnosis research,it is necessary.In this dissertation,the traditional grey relation entropy theory and the advanced algorithm,the depth belief network has been applied to the gas regulator in the study of fault diagnosis,and the simulation test.The main research content is as follows:First of all,using the EMD data analysis,energy characteristics of the original data are extracted,and then combined with the grey relation entropy theory,put forward the gas regulato fault diagnosis model based on the EMD-the grey relation entropy theoryr,and to build the model according to the diagnostic process.First of all,through the analysis of experts to classify the fault type data,by the EMD method,extract the energy torque characteristics of the data.Based on this,the establishment of all kinds of fault reference sequence and compared sequence,through calculating the correlation method,realizes the fault classification,and the instance simulation test.At the same time based on PCA principal component analysis method,is put forward the gas regulator fault diagnosis model based on PCA-grey relation entropy theory,the diagnosis effect comparison of the two models.According to the existing inside the team results and conclusions,comprehensive comparison of diagnosis algorithm.Grey relation theory in the selection of reference sequence,is the emphasis and difficulty in the entire model,and affected by human factors.Therefore,put forward the gas regulator fault diagnosis model based on the theory of the deep belief network,based on deep belief networks(DBN)fault diagnosis method.Compared with the traditional diagnosis methods.Compared with the traditional diagnosis methods,the advantage is that the characteristics of the data extract has significant functions,effectively reduced the dependence for experience,reduce artificial participation leads to the emergence of uncertainty;Network model can effectively reflect the depth,the data characteristics and the complex corresponding relationship between operation condition.Using DBN greedy unsupervised training,the network architecture of the underlying data distribution,reassembled into a higher level of distribution characteristics,implement the data feature extraction,using supervised learning,finished the fault training.Finally,using the data of test group,in the simulation platform,for instance simulation testing,and the feasibility of this method is verified by simulation results.Finally,this chapter studies the network depth and the relationship between the diagnostic accuracy,and the related conclusion.For further study of factors affecting effect voltage regulator,in view of the FL series gas regulator mathematical model is established.First to introduce FL series of gas pressure regulator,including internal structure and regulating principle,List the motion equation and pressure control equation,under the condition of assumption,For gas regulator from the view of gas dynamics,analyzes the running state,And the simulation model is set up on the Simulink as well and try to take advantage of the simulation model for failure data,and all kinds of factors on the pressure regulating effect.