Machine Allocation via Pattern Recognition in Harmonic Waves of Manufacturing Plants

Non-intrusive load monitoring is currently used to analyze changes in the energy consumption of households. Due to the number of electrical consumers, the associated superpositions and the variety of harmonic waves on the shop floor, current proceedings are not applicable in industrial environment. In this paper, patterns in harmonic waves of four manufacturing plants are analyzed in the time and frequency domain. For machine allocation, features were extracted and classified by k-means and support vector machines with an accuracy of 97.3 and 97.9%. For comparison, convolutional neural networks were trained with the harmonic profiles in the time domain with an accuracy of 98.7%.