EEMD-based time series data abnormal value detection and correction method

Abstract

The invention discloses an EEMD-based time series data abnormal value detection and correction method, and mainly aims at solving the problem that the existing method cannot well detect local abnormal values according to time series change characteristics. The method comprises the following steps of: obtaining original time series data, sorting the original time series data according to a time sequence, and filling the missing data by using the value 0; carrying out preliminary abnormal value detection on the original data by using a median method, and replacing the detected abnormal values by using the value 0; carrying out refined abnormal value detection on the preliminarily detected data by using an EEMD method, and replacing the detected abnormal values by using a value 0; and after carrying out abnormal value detection for twice, all the abnormal values are replaced by the value 0, filling the value 0 by using a local curve fitting method, namely, correcting the abnormal values. Through abnormal value detection and correction, data closer to objective reality is obtained. The method can be used for the abnormal value detection and correction of one-dimensional time series data, and is wide in application fields such as water resource monitoring data, traffic flow data, meteorological monitoring data and financial data.

Description

technical field [0001] The invention relates to a method for detecting and correcting abnormal values ​​of time series data based on EEMD and belongs to the field of data processing. Background technique [0002] For the research on time series outlier detection methods, predecessors have proposed many algorithms, such as outlier detection methods based on statistics, clustering, distance, and density. However, these methods do not consider the time series variation characteristics of time series data, but consider the complete set of data, and it is difficult to detect local outliers. [0003] Empirical Mode Decomposition (EMD) method was proposed by Norden E.Huang et al. in 1998. EMD has been widely used in mechanical fault diagnosis, geophysical detection, biomedical analysis and so on. There is no literature for time series outlier detection. EMD can separate fluctuations or trends of different scales from the original signal step by step. EMD is suitable for analyzi...

AI Technical Summary

Problems solved by technology

[0007] The technical solution of the present invention is to overcome the deficiencies of the prior art and provide a method for detecting and correcting abnormal values ​​of time series data based on EEMD, which is used to solve the problem of unreliable data in monitoring data

Images