Research On Prediction Of Downhole Pipe Sticking Based On Neural Network

Nowadays,petroleum is the most important non-renewable energy sources in the world.The main method of petroleum exploration and development is drilling operations.Downhole stuck drill accidents caused by subjective and objective reasons have become one of the main factors affecting drilling economic benefits and on-site operation safety.However,restricted by the geological environment,the technical level of drilling,and the drilling equipment,stuck drill accidents are still difficult to eliminate fundamentally.Therefore,how to predict the accidents in advance to improve the production efficiency and intelligence level of drilling has become the current research focus.Based on the analysis of a large amount of drilling site historical records,the principle,classification,characteristics and influencing factors of stuck drill accidents are summarized.According to the parameter change law in the development process of stuck drill accidents,a BP neural network stuck drill accidents prediction model is established.Based on the analysis of the historical stuck accidents data in drilling,the structure and related parameters of the prediction model are designed.The model is trained and tested by using sample data in the same area and the same stratum.According to the analysis of the test results,the model is in line with the expectations.Experiments show that the model can detect abnormal parameter changes and realize the prediction of stuck drill accidents.In order to improve the accuracy of the stuck drill accidents prediction model,the optimization algorithm is combined with BP neural network to optimize the initial weight and threshold of BP neural network.Two neural network stuck drill accident prediction models based on GA-BP and CPSO-BP are established.By comparing the test results of these models,the feasibility of GA and PSO to improve the accuracy of BP neural network model is verified.It is concluded that the CPSO-BP model has higher accuracy and generalization ability,and can realize accurate prediction of the stuck drill accidents.Based on the above theoretical analysis and experiments,in order to complete the stuck drill accidents prediction system,the overall process structure of the prediction system is designed,the discrimination methods of the stuck drill accidents types are described,and the prevention and resolution suggestions of stuck drill accidents are given.