Torque Identification And Prediction Of Crank Shaft Of Pumping Unit

In the late mining period,the working state of many pumping units has changed.According to the traditional torque calculation method,there are many phenomena of over torque in pumping unit.If you are not sure whether the pumping machine really super torque,we cannot adjust the appropriate parameters of the pumping unit,and we can not determine whether the pumping unit can continue to work.Therefore,it is urgent to determine the actual torque of pumping unit.Although the test method is accurate,it can not test every pumping unit under working conditions,so it is necessary to study the prediction model of torque.We have carry out contrast experiment imitated field test work indoor.We analyzed the reliability and feasibility of the test method,and defined test methods and data processing methods.It will provide reliable data for developing torque prediction and detecting prediction results.Test shows,it is feasible to adopt the wireless torque testing method,but there are some test errors.In this paper,torque prediction based on click parameter test is studied.we have researched the proposed model of motor efficiency,and the pumping unit transmission efficiency.According to the torque prediction analysis of pumping units,the result shows that the prediction model has good torque prediction ability.We establish a prediction model for polished rod load according to the beam pumping unit.We analyze and correct the original formula,and we put forward the SPN torque prediction model.Comparing with the experimental data,the calculation error of Rama formula is larger,and the SPN model has better prediction accuracy.In order to improve the prediction accuracy,the ZHF prediction model is proposed based on the comprehensive analysis of the prediction model and SPN prediction model based on the motor parameters.The torque of pumping unit is a nonlinear complex system relating to many influencing factors.In order to further improve the identification ability of torque,we established a neural network prediction model,and completed preliminary training.The model can accomplish torque identification.