Preset model of screw pump speed based on improved BP neural network


LIANG Hui, WANG Shi-jie, QIAN Cheng

(School of Mechanical Engineering, Shenyang University of Technology, Shenyang 110870, China)


Abstract:  The rotor speed could not be adjusted accurately in real time, due to the rotor speed was affected by many factors in the process of progressing cavity pump production. In order to solve the problem, a presetting model of screw pump speed based on BP neural network was proposed. Firstly, the BP neural network method based on L-M algorithm was used to predict the optimal speed of the progressing cavity pump under the current conditions. And the predicted optimal speed was transferred to the speed adjustment module. Finally, the motor speed was adjusted by PID control to form the preset model of the speed, which could preset the optimal speed of the screw pump based on the real-time measured crude oil temperature, crude oil viscosity, pump end pressure difference and volumetric efficiency. The simulation experiment results show that the model has an average relative error of 0.96% in predicting the optimal speed of the screw pump under various working conditions. The results show that the model has a good effect on the real-time preset of the speed, which lays a good foundation for the real-time adjustment of the speed in the submersible screw pump oil extraction system. The basis of this is conducive to improving the efficiency and economic life of the screw pump.

Key words:  screw pump; speed preset; L-M algorithm; neural network

LIANG Hui, WANG Shi-jie, QIAN Cheng. Preset model of screw pump speed based on improved BP neural network［J］.Journal of Mechanical & Electrical Engineering, 2021,38(9):1197-1201.