| As the oil field enters the middle and late stage of oil production,the factors restricting the efficient operation and control of the mechanical production system in the old oil field are becoming more and more complex.At present,the system efficiency of pumping wells in the middle area of Daqing oil field is distributed between 18%-30%,and the system efficiency distribution is quite different,so there is a large potential for tapping.However,the traditional pumping well system efficiency mostly takes a single well as the research object,and it is unable to consider the impact of oil-water wells,block development system,production measures and other related factors on the pumping well system efficiency from the whole block,which makes it difficult to fundamentally solve the problem of inefficient operation of pumping wells in the middle and late stage of oil production.With the development of digital oil field,a large number of engineering data and geological data are produced every day,which provides conditions for the analysis of pumping well system efficiency from the perspective of geological engineering integration.From the perspective of data mining,this paper combines the surface data and downhole data to carry out the analysis of influencing factors and prediction of core control factors of pumping well system efficiency,providing a new method for the analysis means and management measures of mechanical production system efficiency.The main research contents are as follows:The water injection volume of the old oil field of water injection development determines the oil well production.In this paper,the grey correlation method is used to determine the best connected injection production well group in the block.The sparrow search algorithm based on sine mapping is improved to BP neural network,and the water injection prediction model is established.The average relative error of prediction is 4.5%.It solves the problem that the traditional methods in water injection and oil production technology can not distinguish the connection between water injection wells and oil production wells,so as to slow down the rise of water content and maintain stable production Provide basis for improving the efficiency of mechanical mining system.The influencing factors of pumping well system efficiency are complex and interrelated.Based on this,firstly,this paper obtains the characteristics of low efficiency wells of pumping units in the block through cluster analysis method,obtains the main influencing factors of pumping well system efficiency in the block by lasso Lars algorithm,and uses DBSCAN algorithm and moisture content method combined with improved K-means algorithm to cluster the characteristic parameters of pumping wells in the block,Experiments show that the parameter characteristics of low efficiency wells in the block obtained by the improved algorithm are more prominent.The research shows that twelve factors such as daily oil production,water cut and oil pressure of an oil well in the middle area of Daqing Oilfield are the characteristic parameters of low efficiency wells in the block.The short-term and long-term memory network(LSTM)is used to predict the time series of the characteristic parameters of the low efficiency wells in the twelve blocks such as daily oil production and water cut obtained from the above cluster analysis.The parameters of the low efficiency wells are adjusted according to the range of reasonable values.Among them,the parameters such as daily liquid production and water cut related to the water injection are used to determine the best connected water injection well through the grey correlation analysis,and the water injection is adjusted through the improved sine SSA BP water injection prediction model,And predict the efficiency of the adjusted mechanical production system,so as to improve the efficiency of the block pumping well system.Based on the above content,the efficiency evaluation and prediction system of oilfield mechanical production system is built.According to the requirements of the system function,it is divided into five modules:user management module,block oilfield data management module,block system efficiency evaluation module,block oil well parameter prediction module,and block oil well parameter adjustment module.According to the basic historical parameters of the block,the system can accurately excavate the hidden links between injection and production wells and the system efficiency characteristics of pumping wells in the block,predict and analyze the characteristic parameters of oil wells in the block,and put forward an effective scheme to improve the efficiency of the block mechanical production system.Based on the data analysis technology and taking the block as the analysis object,this paper establishes the efficiency analysis technology of the oilfield mechanical production system,It provides a new method to improve the efficient operation of mechanical production system in the middle and late stage of oilfield development. |
