NON-INVASIVE AI SENSING METHOD FOR HUMAN THERMAL COMFORT

摘要

Disclosed is a non-invasive AI sensing method for human thermal comfort, comprising: acquiring and preprocessing data, defining and extracting a skin sensitivity index, constructing a deep learning network for non-invasive skin-sensitive thermal comfort, and training same to generate a network model. Specifically, said method comprises: acquiring human skin image data by means of computer vision, processing the data, and then constructing an address mapping table between pictures of regions of interest and skin temperatures; introducing a difference in the degree of skin sensitivity of different human bodies to external cold and heat stimuli as a weight coefficient; performing feature extraction on the address mapping table and an SSI, fusing the features and further training same, storing the obtained network models, and selecting an optimal network model for skin temperature prediction. The AI sensing method of the present invention overcomes three challenges, i.e. micro-change of skin change, inter-individual variation, and intra-individual time variation in human thermal comfort measurement, achieves energy optimization, energy saving, and environmental protection, and has high operability.