Research And Application Of Oil Well Dynamometer Acquisition And Working Condition Diagnosis Metho

Petroleum plays an important role in modern economies and is critical for the interests of nations and their people.There are two main methods for oil extraction:spontaneous oil production and mechanical oil production.Mechanical oil production equipment can be divided into two types: rod and rodless.The beam pumping unit is the most commonly used rod-type mechanical oil production equipment.In the process of oil well exploitation,diagnosing the well conditions is a key step in ensuring safe and efficient oil extraction.Currently,the primary method for diagnosing well conditions and detecting faults is through the use of dynamometer cards.Therefore,the acquisition and identification of dynamometer cards have become the key technologies for diagnosing well conditions.A dynamometer card is a closed graphic formed by the change in load on the rod with displacement,which reflects the working condition of the pump and the downhole liquid supply.In order to achieve automatic acquisition of dynamometer cards,a displacement-based indirect measurement method is proposed.Based on the analysis of the mechanical structure of the beam pumping unit,the method indirectly measures the displacement of the rod using mechanical relationships and kinematic formulas.An accelerometer sensor is used to sense the motion state of the rod and calculate the period,and the dynamometer card is formed by combining the data collected by the load sensor.Based on this method,an automatic dynamometer card acquisition device for oil wells is designed.The main control chip uses an ARM microprocessor,and data transmission uses NB-IoT communication.The power management adopts low-power circuits.In software design,methods such as mean filtering and extremum point discrimination are used to calculate the motion period and dead point position of the rod.Experimental results show that the method has high acquisition accuracy and is easy to implement.Given the complex working conditions and harsh environments of oil wells,the well conditions are often not a single type but a mixture of multiple types.Previous dynamometer card recognition methods have excellent recognition ability for single-type dynamometer cards but often have difficulty recognizing dynamometer cards with multiple types of well conditions.A dynamometer card recognition method based on Res Net is proposed,and a deep learning model is trained.The model treats different well conditions in oil wells as different labels of dynamometer cards and performs binary judgments,outputting the probability information for each well condition label.In data processing,a seed filling algorithm is used to enhance the contour of the dynamometer card,and the dynamometer card dataset is replicated and expanded.During model training,transfer learning is used to quickly train a Res Net binary classification group to recognize different well condition labels separately.Experimental results show that the model can accurately recognize dynamometer cards with multiple types of well conditions and output the probability of different well condition labels.Based on the above hardware and software,a dynamometer card acquisition and well condition diagnosis system is developed and communicated using the NB-IoT network and MQTT protocol.The system can real-time collect and upload dynamometer card data,diagnose well conditions using a deep learning model,and display the results.The system can achieve real-time diagnosis of well conditions,improving the efficiency and accuracy of well condition diagnosis.The research and application of dynamometer card acquisition and well condition diagnosis methods provide new ideas and tools for well condition diagnosis,which is of great significance for ensuring safe and efficient oil extraction in oil wells.