关键词: 人工神经网络；无磁芯电流传感器；相间串扰

Abstract: With the increasing power density of motor controllers, more and more extreme size requirements are imposed on current sensors. Although the emerging coreless current sensor solution can decrease the core size of the current sensor in terms of size, the inter-phase crosstalk is a major problem that cannot be ignored. To solve the inter-phase crosstalk problem of coreless current sensors, this paper proposes a rapid optimization design method for inter-phase crosstalk suppression of coreless current sensors based on a back propagation artificial neural network. By constraining the actual application size range of the sensor, the magnetic field simulation is carried out using three-dimensional finite element, and the simulation results are used for neural network training. In practical applications, this method only requires the inter-phase distance of the coreless sensor to be provided to complete the rapid correction of inter-phase crosstalk of the coreless current sensor. Through the comparison of numerical simulation results of different inter-phase distances with neural network results, the rapidity and effectiveness of the rapid optimization method for inter-phase crosstalk suppression based on artificial neural network are verified.

Key words: artificial neural network; coreless current sensors; crosstalk