US 12,423,326 B2
Method of association inference for time series data based on local variation features
Guohui Zhao, Lanzhou (CN)
Assigned to Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences, Lanzhou (CN)
Filed by Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences, Lanzhou (CN)
Filed on Dec. 11, 2023, as Appl. No. 18/535,945.
Claims priority of application No. 202310108058.7 (CN), filed on Feb. 13, 2023.
Prior Publication US 2024/0273117 A1, Aug. 15, 2024
Int. Cl. G06F 16/28 (2019.01)
CPC G06F 16/284 (2019.01) 9 Claims
OG exemplary drawing
 
1. A method for analyzing a relationship between data sequences, comprising steps of:
acquiring a first data sequence and a second data sequence;
binarizing the first data sequence into a first equivalent sequence according to a first sequence threshold of the first data sequence;
binarizing the second data sequence into a second equivalent sequence according to a second sequence threshold of the second data sequence;
incrementing a current lag duration to obtain a target lag duration;
intercepting a first equivalent segment from the first equivalent sequence according to the target lag duration;
intercepting a second equivalent segment from the second equivalent sequence according to the target lag duration, wherein a time period corresponding to the first equivalent segment is staggered from a time period corresponding to the second equivalent segment by the target lag duration;
obtaining a correlation index of a current segment pair according to the first equivalent segment and the second equivalent segment, wherein the current segment pair comprises the first sequence segment in the first data sequence corresponding to the first equivalent segment and the second sequence segment in the second data sequence corresponding to the second equivalent segment,
wherein if the target lag duration does not meet a termination condition, it is returned to perform the step of incrementing a current lag duration to obtain a target lag duration, until the target lag duration meets the termination condition, so as to obtain the correlation indexes of the plurality of segment pairs;
selecting, according to the correlation indexes of the plurality of segment pairs, a target segment pair with the largest correlation index therefrom; and
obtaining correlation information between the first data sequence and the second data sequence according to the target segment pair.