Application Research Of Intelligent Fault Diagnosis System For Gas Equipment

Gas pressure regulator is a important equipment of gas transmission and distribution system.It is obviously important that the reliable and stable operation state of gas pressure regulator determine the safe and stable operation of the natural gas supply system.At present,the gas pressure regulator of a gas group is faced with the following problems.The first problem is that the fault identification is often completed by technical engineers dependent on previous experiment.This method is subjective,time-consuming and not intelligent.The second problem is that the maintenance of the gas regulator is carried out by the fixed point timing.The regulator will disassemble and repair regardless of whether it is running well,which greatly increases the maintenance tasks and the cost of maintenance.In this way,the worst is that the good gas pressure regulator will be wrecked in sometimes.The excessive maintenance wastes a lot of manpower and money.Therefore,it is necessary to put forward a method of intelligent fault diagnosis technology research based on gas pressure regulator.In this thesis,the intelligent diagnosis technology is introduced to the fault system of gas pressure regulator.It can improve the stability and reliability of gas transmission and distribution system,and reduce the waste of resources such as manpower and material.In this thesis,the fault diagnosis system of gas pressure regulator and the characteristics of fault diagnosis are understood fully.After that,the following work has been carried out:Because the actual fault data collected from the actual operation of Beijing Gas Group is limited,the radial basis function neural network algorithm is used to classify the 4 kinds of faults in the operation of gas pressure regulator.Its diagnostic recognition rate is 62.5%.In order to improve the diagnostic effect,the method of combining principal component analysis with RBF(PCA-RBF)was used.The principal component analysis is used to reduce the dimension of the sample data of gas pressure regulator.The fault diagnosis recognition rate increased to 87.5%,and the diagnostic effect was improved greatly.In addition,in order to achieve a better effect of fault diagnosis in the case of small sample data,the method of gas pressure regulator fault diagnosis based on compressive sensing theory is proposed.The contrastive study between PCA-RBF and the sparse representation of compressed sensing theory is made in this thesis.The results show that the latter is better.Its classification accuracy rate is 92.8%.On the basis of the comparison and analysis of the fault algorithm,the fault diagnosis software system of gas pressure regulator is designed in this thesis by using GUI of MATLAB.The development process of each function module is introduced,and the algorithms are implemented on the software.It can be more intuitive and convenient to show the diagnostic effect of a variety of algorithms.Users can choose the appropriate diagnostic methods to achieve a better diagnostic function according to the actual situation.It can be applied to the intelligent fault diagnosis of gas pressure regulator better.