Rod pumped well fault diagnosis method and system based on residual neural network

Abstract

The invention relates to the technical field of oil-gas exploration and development, in particular to a rod-pumped well fault diagnosis method and system based on a residual neural network. The method comprises the following steps: carrying out normalization by using an actual load extreme value of a current indicator diagram, and for a current indicator diagram data set, obtaining k normalization scales by using a clustering algorithm so as to obtain k + 1 normalized indicator diagrams; and secondly, binarizing the indicator diagram, and taking an 18-layer residual convolutional neural network of a k + 1 input channel as an image recognition network model based on pytorch. According to the method, an indicator diagram multi-scale normalization method is combined, an indicator diagram classification model of a multi-channel deep residual convolutional neural network is constructed, deep learning neural network training technologies such as BN and Relu are used, training and testing are carried out under a data set obtained through the multi-scale normalization method (k = 10), the convergence robustness of the model is good, and the testing accuracy reaches 95.6%.

Description

technical field [0001] The invention relates to the technical field of oil and gas exploration and development, in particular to a fault diagnosis method and system for a pumping well based on a residual neural network. Background technique [0002] The fault diagnosis of pumping wells has always been the difficulty and focus of oilfield production. In the past few decades, through the efforts of scientific researchers, the fault diagnosis technology of pumping wells has been greatly improved, and some phased results have been achieved. In recent years, the development of artificial intelligence technology has brought new vitality to the research of fault diagnosis technology. [0003] Among the widely used pumping well fault diagnosis technologies at present, the more commonly used methods are: 1) Fault expert system, a computer system that simulates human experts to analyze the working conditions of pumping units. This method combines computer and artificial intelligence, ...

AI Technical Summary

Problems solved by technology

[0003] Among the widely used pumping well fault diagnosis technologies at present, the more commonly used methods are: 1) Fault expert system, a computer system that simulates human experts to analyze the working conditions of pumping units. This method combines computer and artificial intelligence, according to multiple The working condition analysis experience provided by the experts makes inferences and judgments on the working conditions of the pumping unit, but the expert system is only for the working conditions reflected in the specific pumping unit indicator diagram written into the program, and it needs to be redesigned for different types of pumping units The implementation program cannot be promoted and used; 2) The machine learning method has achieved good accuracy in the classification and recognition of the dynamometer diagram, but it still cannot meet the needs of actual production. It is necessary to expand the data set, improve the classification and recognition accuracy, and shorten the model. Training time method; 3) deep learning method, the continuous update and improvement of deep learning model, the development of computer hardware, especially GPU, has greatly improved the accuracy of image classification and recognition, but there are few applications suitable for pumping unit fault diagnosis, and cannot take full advantage of them

[0004] In oilfield production, there are more than 20 types of dynamometer diagrams measured in pumping wells, and some working conditions are very complicated. The commonly used pumping well working condition diagnosis model based on dynamometer diagram recognition is accurate in the field diagnosis of various working conditions. Both the recall rate and the recall rate under complex working conditions are low, and the uncertainty of diagnostic results is prominent

Images