A Method of Parametric Abnormal Pattern Detection for Missing Satellite Data

Abstract

A method for detecting abnormal parameter patterns of missing satellite data, including: data preprocessing, completing normalization and standardization of satellite time series data; abnormal detection, based on artificial neural network models, loading satellite time series data containing missing data, setting Batch processing parameters, group satellite time series data according to the parameter value, and then perform anomaly detection on each group of satellite time series data that contains missing data; visualize the results, save the detection results, and graphically display the abnormal detection results. This method uses a new neural network model, combined with the inherent time-dependent characteristics of satellite time-series data, to realize the process of accurately detecting satellite abnormal patterns in the case of a large number of missing data, and to ensure the stability of satellite fault diagnosis and health management systems sex.

Description

technical field [0001] The invention belongs to the field of intelligent information processing and computer technology, and in particular relates to a method for detecting abnormal parameters of missing satellite data. Background technique [0002] Due to reasons such as receiving blind spots or interference, the satellite timing data will cause the lack of original timing data, that is to say, there are no observation data at some time points or time periods. Missing time series data will cause great difficulties for applications such as pattern recognition, anomaly detection, and fault diagnosis. [0003] Over the past few decades, researchers have developed various methods to address the case of missing values ​​in time series data. A simple solution is to simply ignore the missing data and only analyze the observed data, but this method does not provide good performance when the missing rate is high and insufficient samples are retained in the data. Another approach i...

AI Technical Summary

Problems solved by technology

[0004] In short, the research on time series data with missing values ​​has gradually begun to be valued by the public. So far, the processing of time series data with missing values ​​is to use evaluation methods to guess missing values, which makes the data have a certain degree of deviation , when the deviation is particularly large, it will even affect the judgment of time series data

Images