A Time Series Missing Value Filling Method Based on Bidirectional Recurrent Codec Neural Network

A time series and neural network technology, applied in the field of artificial intelligence, can solve problems such as the impact of changes in the filling effect of time-space relations, achieve the effect of weakening gradient explosion and dispersion phenomena, and improving interpretability

Active Publication Date: 2022-01-07
HANGZHOU DIANZI UNIV
View PDF7 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

It can overcome the shortcomings of existing time series filling methods that it is difficult to correctly model the spatio-temporal relationship in time series with missing values, and the filling effect is greatly affected by the change of missing rate

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A Time Series Missing Value Filling Method Based on Bidirectional Recurrent Codec Neural Network
  • A Time Series Missing Value Filling Method Based on Bidirectional Recurrent Codec Neural Network
  • A Time Series Missing Value Filling Method Based on Bidirectional Recurrent Codec Neural Network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0054] All artificial intelligence-related terms used herein have the same meanings as commonly understood by those of ordinary skill in the technical field to which this application belongs.

[0055] The invention provides a time series missing value filling method based on a two-way loop codec neural network to fill the sensor time series. Examples provided by this application, such as figure 1 Shown is a schematic diagram of the scenario for missing value filling of sensor multidimensional time series. The acquisition device is connected with several monitoring devices, and the data is collected at shorter intervals and uploaded to the server at longer intervals. The data received by the server can be regarded as a time series of equal length. If there is no missing time series, the server will directly store the series into the historical database; if there are missing data, the server will input the series into the filling module, and the filling module will store the fi...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention provides a time series missing value filling method based on a bidirectional cyclic codec neural network. This method combines autoencoders and recurrent neural networks to model time series with missing values; this method uses two training losses to measure the difference between the filling sequence and the label sequence, and reversely updates the encoding in an asynchronous manner decoder and decoder; the method amplifies the network's response to missing data by coordinating gating units. The method of the invention overcomes the problems that the general method cannot correctly model the time-spatial relationship of the time series containing missing values, and the filling effect is sensitive to the change of the missing rate.

Description

technical field [0001] The invention relates to the field of artificial intelligence, in particular to a method for filling missing values ​​of time series based on a bidirectional cyclic codec neural network. Background technique [0002] In the application tasks of multi-dimensional time series in the industrial Internet of Things, such as context recognition, predictive maintenance, anomaly detection, etc., a complete time series is the prerequisite for the smooth progress of the task. However, a large number of equipment accesses and unstable environments lead to the prevalence of missing values ​​in the multidimensional time series of the Industrial Internet of Things. The existing multidimensional time series filling methods include mean filling, clustering filling, regression filling and so on. The mean filling effect depends on the difference between data points, and the filling accuracy is not high, especially when continuous missing occurs, it is easy to cause lar...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Patents(China)
IPC IPC(8): G06F16/2458G06F16/9537G06K9/62G06N3/08
CPCG06F16/2474G06N3/08G06F16/9537G06F18/214
Inventor 邬惠峰丘嘉晨孙丹枫
Owner HANGZHOU DIANZI UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap