Research On Multi-faults Intelligent Diagnosis System For Pumping Well

Grasping the pumping system’s operational state in time is crucial to the petroleum extraction sector.Rapid and efficient identification of indicator diagrams is a direct method of determining the operating status of pumping wells.The different shapes of the indicator diagram indicate the different working conditions of the pumping system.However,in traditional industrial production processes,the characteristics of the indicator diagram are extracted from the personal experience of experts.As an alternative,the classification of pumping system operating conditions is achieved by using images containing indicator diagram curves to the model.These problems increase the operation and maintenance costs of pumping wells while reducing the system fault classification efficiency and classification accuracy.In this work,a new method of multimodal attention learning using superimposed images of the indicator diagram images and its frequency domain maps combined with numerical data of the pumping system is proposed.By processing the specified type of data through each branch network of the model in this method,the model could acquire the characteristics of various modal data,and the attention mechanism introduced can efficiently capture the detailed information of the data and increase the accuracy of fault diagnosis.The proposed method is validated by applying it to an actual data set from an oil field in northwest China.The experimental results demonstrate that the average classification accuracy of the model with attention mechanism added to the convolutional neural network branch and BP neural network branch is increased by 1.17 percent compared to the simple multimodal learning model.The average accuracy of identifying the faults of the indicator diagram using the CNN-BP multimodal attention learning model reaches 93.05 percent,which is 10.47 percent higher than other models.In addition,the average fault classification accuracy of the model is improved by 2.46 percent after superimposing the frequency domain maps on the indicator diagrams.What’s more,the current operating status of oil wells is transmitted to the cloud via a 4G module,besides,the information is revealed by a visualization terminal built by a third-party open-source library Kivy.The information is also saved to a My SQL database to provide data support for later management and maintenance of the pumping system.The system design scheme of the multi-faults diagnosis model of the pumping well and the visualization terminal proposed in this paper can effectively realize real-time monitoring,remote management and efficient maintenance of oil wells,and this intelligent diagnosis system has certain references and guidance for the subsequent construction of intelligent oil fields.