Similarity Query Method for Time Series Subsequences Based on Pathologic Matching of DTW

Time series of fuzzy concept with random noise and complex deformation cause the difficulty of subsequence similarity query. To solve the problems of subsequence query, a similarity query method for time series subsequences based on pathologic matching of DTW is proposed. Firstly, boundary curves are built up and down by using query sequence sample sets. Lower bounding distance based on the sequence of samples is proposed with a proof correctness to build lower bounding algorithm for filtrating subsequences to increase efficiency. Then, the invalid subsequence field is recognized and deleted for determining the valid subsequence field of matching by pathologic matching of DTW. The results of experiments prove that the proposed method can quickly and accurately find similarity subsequences.