DTW-based multivariate time sequence characteristic prediction method and system

• Main Authors: LI TONG, LIU LIANGFENG, WANG SHAOLIN, DAI HAISHAN, FAN JUNJIE, LOU MINGJING
• Format: Patent
• Language: Chinese; English
• Published: 22.03.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention provides a DTW-based multivariate time sequence feature prediction method and system, and the method comprises the steps: carrying out the segmentation processing of satellite telemetry data, and obtaining an original time sequence; extracting features respectively, and combining to generate a multivariate feature vector as an initial sample data set for preprocessing; performing time mark alignment equal-interval interpolation processing on the processed feature sequence to generate a sample time sequence feature set; taking the initial sample data set and the corresponding classification label set as the input of an LSTM network, and carrying out model training; generating a feature template of a corresponding type according to the feature sequence of the sample data and the corresponding classification label; inputting the trained LSTM model, and obtaining a classification label; calculating the similarity between each section of feature template and the moving feature sequence in sequence; a