Missing monitoring data filling method based on multi-scale space-time memory sharing network

Abstract

The invention provides a missing monitoring data filling method based on a multi-scale space-time memory sharing network, and the method specifically employs the multi-scale space-time memory sharing network to achieve the prediction and filling of monitoring data. According to the multi-scale space-time memory sharing network, data characteristics corresponding to monitoring data of different time scales in historical monitoring data of a large-scale rotating unit under different working condition states are recorded through pre-training; firstly, time scale information and corresponding data features of test monitoring data are extracted, then a monitoring data prototype corresponding to a time scale with matched data features is inquired from historical monitoring data, and a monitoring data prediction value after the test monitoring data is predicted according to the monitoring data prototype. Because the authenticity and correctness of the historical monitoring data are easy to control and guarantee, generation of abnormal prediction data is avoided, and the accuracy of data filling is better guaranteed.

Description

technical field [0001] The invention relates to the technical field of engineering application and big data acquisition and detection, in particular to a method for filling missing monitoring data based on a multi-scale space-time memory sharing network. Background technique [0002] The data acquisition and monitoring system can realize remote real-time monitoring, control and diagnosis of large rotating units, ensuring the healthy operation of large rotating units. The data acquisition and monitoring system collects and stores data through sensors deployed on various components, but it is often affected by uncontrollable factors such as sensor failure and network congestion, resulting in data loss. The lack of data has a great impact on some fan condition monitoring and fault diagnosis methods that require time series data. Therefore, it is necessary to design a solution to deal with the missing monitoring data of large rotating units. [0003] Monitoring data is multi-d...

AI Technical Summary

Problems solved by technology

[0006] In view of the deficiencies of the above-mentioned prior art, the purpose of the present invention is to provide a method for filling missing monitoring data based on a multi-scale spatio-temporal memory sharing network, so as to more accurately repair and fill missing monitoring data of large rotating units to solve the problem of current problems. There are technologies to fill in the missing monitoring data of large rotating units and repair the problem of insufficient accuracy

Images