Prototype of Edge Computing IPM with Hardware Artificial Neural Network Soft Sensor and Controller for Parallel Connected IGBT Current Distribution

This work proposes the edge computing intelligence power module (EC-IPM) with hardware artificial neural network (ANN) current soft sensor (SS) and controller for parallel connected IGBT current distribution to solve the problem of imbalance current. The proposed EC-IPM evolves the technologies of semiconductor power device physics, active gate control, machine learning and field bus. The full prototype of EC-IPM and related test power system were built on Matlab Simulink platform with the realization of hardware algorithm of ANN current SS and controller. The results show that the power system with different current rating EC-IPMs can distribute the proper current to each EC-IPM full-autonomously to suppress the IGBT current unbalance in each EC-IPM within 6.5%, compared to 96% that without the proposed technology in the same scenario. Moreover, the EC-IPM can also extend the electric energy internet of things (IoT) to power device level, especially for upcoming popularization of 5G communication environment.