Research On Lithology Change Prediction Method Based On Neural Network

The prediction of lithology change is one of the basic works of logging interpretation,and it is the core link to identify the geological structure and oil-bearing condition of oil and gas layers.In particular,the prediction of key volatile attributes such as porosity and permeability will directly affect the subsequent development and production of oil and gas fields,so it is highly valued by logging interpretation,geological analysis,and other departments.At present,the method of lithology change prediction has been gradually transferred from the conventional artificial mathematical calculation to the stage of partial intelligent technology.However,there are still the following problems: First,the current lithology change prediction method can only macroscopically calculate the porosity,permeability,and other parameters of the whole well,and there is a lack of stratified prediction method;Second,lithology prediction based on single good data is not accurate enough,which requires adjustment and revision of relevant Wells in the region.However,the process of regional good selection has always relied on expert experience for manual screening,which is not only a huge amount of calculation,time-consuming and laborious,but also has been questioned for its accuracy.Aiming at the above problems,this paper proposes a prediction method of lithology change based on the neural network based on an in-depth investigation of the actual working process of front-line production units.Firstly,the real workflow of lithology prediction is studied,the key problems are analyzed,the intelligent prediction flow of lithology change is designed,and the prediction model of lithology change based on neural network is established,and its composition,working mechanism,key technologies,and feasibility are described.Secondly,the physical property parameters of lithology change are selected to clarify the rationality of lithology physical property parameters reflected by logging curves.The logging curves are selected and the characteristics of logging curves are extracted as the input parameters of the neural network.The calculation method of physical property parameters of lithology change based on neural network is designed to realize the calculation of physical property parameters of selected lithology.Thirdly,after the establishment of the BP neural network model,the experience of expert well selection is summarized and the expert system of regional good selection is constructed to realize regional intelligent well selection.Meanwhile,the fitting method of the regional stratified lithology change curve is studied to realize the change trend prediction of key lithology attributes of a single well and region.Finally,based on the above theoretical basis,the lithology change prediction platform is constructed to verify the reliability of the proposed method in the way of the real application system.Experimental tests and field application results show that the prediction method of lithology change based on the BP neural network can accurately calculate the stratified porosity and permeability of a single well.On this basis,the regional well selection expert system can automatically select effective Wells and can fit the single well stratification time-series change curve into the regional stratification time-series change curve,and the prediction results meet the field expectations,which has a certain theoretical research significance and practical value.