Structured Thermal Representation of a Brushless DC Motor Based on Infrared Image Analysis

Background. The article proposes a method for forming a structured thermal representation of a brushless DC (BLDC) motor based on analysis of infrared thermograms in three‑dimensional red-green-blue (RGB) space. The relevance of the study is due to the need to increase the reliability of monitoring the technical condition of BLDC engines, which are widely used in the designs of unmanned aerial vehicles. Materials and methods. The paper implements a synthesis of computer vision and statistical analysis methods: clustering of color patterns of thermograms is performed using the KDE and Mean Shift algorithms. Based on the selected clusters, the key color moments are calculated, which make it possible to quantify the thermal state of the critical zones of the engine. Results and conclusions. A methodological scheme for the analysis of thermograms has been developed, including the decomposition of the engine into thermal zones according to the spatial and color criterion and the calculation of color moments as compact numerical features.