Multivariable time sequence classification method and system based on time-frequency mixed contrast learning

• Main Authors: LIU MIN, MORIE, QIN LE, RAO XINPING, LUO YONG, WU FUYING, YI YUGEN, HUANG XIAOMEI
• Format: Patent
• Language: Chinese; English
• Published: 28.05.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; PHYSICS

The invention provides a multivariable time sequence classification method and system based on time-frequency mixed contrast learning, and the method comprises the steps: inputting a multivariable time sequence into a time domain encoder and a frequency domain encoder, and obtaining a time domain first feature and a frequency domain first feature, performing nonlinear transformation to obtain corresponding second time domain features and second frequency domain features, performing time-frequency mixed contrast learning to obtain a self-supervised contrast loss function and a supervised contrast loss function, and performing a label propagation algorithm on the second time domain features and the second frequency domain features to obtain a time sequence sample with a pseudo label and a classification loss function; and finally, constructing an overall loss function according to the self-supervised comparison loss function, the supervised comparison loss function and the classification loss function, and clas