Research On The Identification Method Of Production Working Condition To Swimming Beam Type Pumping Machine Based On DBN

Among the commonly used oil extraction methods,the swimming beam pumping machine is one of the most widely used types of pumping machines in oil fields due to its simple structure,flexible operation,strong applicability,good durability,safety and reliability.Oil wells generall Among the commonly used oil extraction methods,the swimming beam pumping machine is one of the most widely used types of pumping machines in the oilfield because of its simple structure,flexible operation,strong applicability,good durability,safety and reliability.Oil wells generally work hundreds or even thousands of meters deep underground,and the working environment is complex.When a well fails,if the type of failure cannot be accurately determined in a timely manner,it may lead to downtime or safety accidents,affecting the normal,safe and efficient production of the oilfield.The well workover diagram is intuitive and contains a lot of information about well production,so this thesis uses the workover diagram as the basis for well condition identification.To address the problems of unbalanced number of samples,inaccurate extraction of feature values,and poor classification effect of the classification method in the present stage of work condition identification method based on the schematic diagram,this paper firstly adopts ADASYN oversampling method to enhance the schematic diagram data and balance the number of schematic diagrams between different categories,and secondly,selects the depth belief network algorithm to establish the depth belief network Secondly,the deep belief network algorithm is selected to establish the deep belief network recognition model,and the deep belief network structure is used to automatically extract the feature values of the power diagram and classify them to achieve the purpose of work condition recognition.Finally,the structural parameters of the deep belief network are optimized by the sparrow search algorithm,so that the structure of the deep belief network working condition recognition model can be optimized.Simulation experiments were carried out to diagnose oil well condition in Liaohe oilfield,and the results showed that the deep belief network model based on sparrow search algorithm improved the accuracy of oil well condition recognition.