Theoretical model construction and structure optimization of electromagnetic flow transducer based on neural network

Abstract

 The development of Daqing Oilfield in China has entered the middle and late stages of high water cut. At this time, oil-water two-phase flow is ubiquitous, and its flow rate is very difficult to measure accurately. Addressing this issue, the measurement model and simulation model of electromagnetic flow transducer (EFT) with saddle excitation structure is designed in this paper. Then the distribution characteristics of magnetic flux density of different excitation structures are analyzed by finite element simulation. Furthermore, the prediction model between the parameters of different excitation structures and the performance evaluation indexes is established based on RBF neural network. Through normalization and weight assignment on the output of neural network model, the structure optimization factor is constructed. Then the optimum solution of this factor is gotten, and the optimum parameters of EFT’s excitation structure are obtained. In addition, an EFT with the optimum structure is developed and tested in Daqing oilfield, and the experiment results show that the EFT has high precision, especially in the high viscosity wells.