Year: 2022
Type: conference-paper
Abstract: This paper presents the results of research on a neural classifier system for the recognition of electrical drive disturbances and its speed. The classification task is performed by a convolutional neural network implemented on an artificial intelligence accelerator. The operation of the drive is monitored using intelligent accelerometers. Based on the sensor data, it is possible to create spectrograms of the signal using Short Time Fourier Transformation. Since convolutional neural networks are excellent at recognizing images and detecting objects, the signal spectrograms were used as input data for learning the network. Transfer learning method has been used, to create a model of neural network.
Institution: AGH University of Krakow
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4281748199