Abstract: The equipotential switching algorithm of the traditional mechanical static compensator（MSC）

device had many limitations in the operation of the single machine，such as the long operation time of the

microcontroller control chip，the poor convergence of the calculation results had been affected by the hardware，and so on. In recent years，the research on devices that can communicate with the back end through the message

queuing telemetry transpor（t MQTT）protocol through the internet of thing（s IoT）platform has becoming more and more mature，and the development of IoT technology has also made it possible to apply machine learning

algorithms such as neural networks to stand-alone electrical equipment. In the experiment，through a large number

of data sets training，the switching time in different electrical environments could be predicted to achieve the

adaptive equipotential switching effect，the adaptive switching without spark，no inrush，no overvoltage，etc，was achieved in the developed device. The research and development focus includes the equipotential switching

principle and simulation of the device，the IoT communication method between the device and the back-end

software，the implementation process and training results of the BP neural network algorithm，and the experimental results and experimental analysis of the switching experiment.

      Key words: internet of things（IoT）devices；mechanical static compensator（MSC）device；Matlab simulation；BP neural networks；equipotential switching；adaptive switching