Research On Electromagnetic Sensor Excitation Structure Design Method Based On Neural Networks

In the practice of oil flow measurement, high precision electromagnetic flowmeter is an important guarantee for correctly estimating production of oil well, and the major parts of the electromagnetic flowmeter is electromagnetic sensor. As an usual way, structure of the sensor is designed by s simple simulation with software, and it can’t avoid making some errors with the insufficient data. In this paper, in order to develop a more accurate electromagnetic flowmeter, This thesis propose a new design method of electromagnetic sensor, combining non-linear neural network modeling, finite element simulation to electromagnetic field, and advantages of cuckoo search algorithm for searching global optimization solution with high speed.Firstly, introducing the theoretical model of electromagnetic flowmeter and weight function, ensure rational calculation to follow. Then, using ANSYS finite software to model the magnetic field with different sensor structure, and record data for analysis later.Secondly, This thesis introduced the basic theory of Radial Basis Function(RBF) Neural Network and cuckoo search algorithm, how to optimize the Network, and using it to learn the data for building nonlinear system model.Finally, to turn the multi-objective into a single objective optimization problem, This thesis improve the cuckoo search algorithm with two cuckoo to solve the above optimization problem. In the last, we get the optimized structure parameters.