Neural Network Identification of Operating Modes of a Robotic Platform Electric Drive

Abstract

This paper considers an approach to identifying electric motor operation modes by means of neural networks. It is shown that the amplitude–frequency response (AFR) of current in the motor windings can serve to achieve this goal. The two neural network models that are proposed cover AFR approximation and mode identification, respectively. Potential emergency situations are considered that can be identified using the suggested approach. The proposed neural network is taught to recognize these situations by the current consumption characteristics of the platform’s motors. It is shown that a motor’s running mode affects the wave characteristics of its windings, which allows creating a model to identify the motor’s running mode by measuring the values associated with the processes in the windings.