Method and device for compressing high-latitude missing time sequence supporting similarity retrieval

Abstract

The invention discloses a method and device for compressing high-latitude missing time sequence supporting similarity retrieval, and the method comprises the following steps: collecting a high-dimensional data sequence array, and carrying out the coding through employing a binary stepped identification; performing compression conversion on a high-dimensional time sequence of the high-dimensional data sequence array, and determining missing data points; and storing the stepped high-compression-ratio data with different accuracies in a clustered storage mode or an unclustered storage mode. According to the method, rapid compression conversion of a high-dimensional time sequence can be completed, recording of missing data points is supported, stepped high-compression-ratio storage with different accuracies is supported, a binary index technology widely used in the field of graphs is supported, and the method is simple and easy to implement.

Description

technical field [0001] The invention relates to the technical field of compression coding, in particular to a compression method and device for high-latitude missing time series supporting similarity retrieval. Background technique [0002] The data collected by the industrial Internet of Things system has time tags, so the collected data belongs to a high-dimensional data sequence array within a certain period of time. The process of processing on the big data processing platform involves data acquisition, storage and retrieval. At present, the time series supported by the big data platform use traditional database technology to support the operation of "points" in the time series, such as finding the maximum value in a time series, taking the mean value of a time period, etc. However, current artificial intelligence algorithms rely on the overall time series similarity comparison that can be achieved in the underlying data system. [0003] Similar to this technique are t...

AI Technical Summary

Problems solved by technology

[0005] Disadvantages of iSAX technology: 1. Does not support missing data processing; 2. Assume that the data points are all normalized; 3. This encoding technology requires a special index structure; 4. Although binary code conversion is used to represent each cardinality space and support ladders Type accuracy, but because the average value is used to represent each piece of data, there is a problem of low accuracy; 5. A special data structure is required to support the index

[0006] Disadvantages of Clipped Code: 1. Does not support stepped precision; 2. Does not support missing data processing; 3. Only supports simple data model and Hamming distance calculation, and cannot identify the subtle differences between two data sequences

Images