Research And Design Fault Monitoring System Of Pumping Unit And Driving Motor Based On Fuzzy Neural Network

Pumping unit is one of important equipment in oil filed system.However,the driving motor is the key part of the whole pumping unit.Once the failure of pumping unit and the driving motor happened,no timely diagnosis,which not only will result in the waste of energy,and even lead to serious accidents.Therefore,it is very important to diagnose the fault of the pumping units and the driving motor timely and accurately,which means lot to the production efficiency and the energy saving of the oil field.According to the rod pumping unit as the research object,this paper puts forward variable learning step of golden section method fuzzy neural network diagnosis method to judge the typical fault diagnosis and designs the fault monitoring system of pumping unit based on fuzzy neural network.First,this paper summarizes the research status of pumping unit fault monitoring system and its drive motor fault diagnosis as well as the problems in the current monitoring system of pumping unit,on the basis of explaining the operation principle of the rod pumping unit,the generation of the indicator card,the sort of the fault indicator card,the characteristic of the fault indicator card and the corresponding solution measures of each fault are discussed particularly,and this paper extracts the corresponding feature parameters of each fault indicator card.And then,the fault mechanism of the driving motor is studied deeply,and the feature parameters of its fault are extracted.Second,this paper use the traditional BP neural to diagnose the fault of the pumping unit and its driving motor.Based on the data of the experiment,the results of training and simulation reflect the BP neural network in fault diagnosis is slow convergence speed,limitation of knowledge interpretation,error precision and local minima.Third,aiming at the disadvantages of the training and simulation of BP neural network,this paper puts forward the fuzzy neural network structure which by neural network mainly and fuzzy theory as a useful complement based on fuzzy theory and neural network which are more mature in the application.According to the error tendency dynamically,this paper puts forward BP algorithm based on variable learning step of golden section method to train the neural network,it achieves the learning step of adaptive adjustment.The results of algorithm compares with the traditional BP training results,obtain more ideal diagnostic results.It can effectively reduce the training time,reduce the error and avoid the local minimum value.Finally,the fault diagnosis of the driving motor is verified,and the superiority of the improved algorithm is proved.Last,this paper designs the overall of monitoring system of pumping unit.The system realizes the function of data collection,data transmission and data management.System applies the mixed programming of Lab VIEW and MATLAB based on the specific implementation process of fault diagnosis to develop upper computer of the pumping unit monitoring system.At last,the whole system is tested.It is proved that the fault monitoring system of pumping unit can diagnose the fault type of pumping unit and the driving motor accurately and efficiently,besides it can judge the severity level of fault easily,and this has a certain reference value for the system upgrade later.