Error prediction method for mutual inductor based on dynamic feature map and space-time graph convolution network

By combining dynamic feature maps with spatiotemporal graph convolutional networks, the problems of insufficient mining of multi-source data topology relationships, weak spatiotemporal coupling feature extraction capability, and poor noise resistance in the error prediction of mutual inductors are solved, achieving high-precision and robust error prediction that can adapt to non-stationary operating conditions.

CN122241021APending Publication Date: 2026-06-19CHINA THREE GORGES UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
CHINA THREE GORGES UNIV
Filing Date
2026-03-10
Publication Date
2026-06-19

AI Technical Summary

Technical Problem

Existing methods for predicting instrument transformer errors are insufficient in handling the fusion of multi-source heterogeneous data, in-depth mining of spatiotemporal features, and noise resistance, making it difficult to meet the needs of smart grids for accurate evaluation of instrument transformer metering performance.

Method used

A method for predicting the error of mutual inductors based on dynamic feature maps and spatiotemporal graph convolutional networks is adopted. Through multi-source time series data acquisition, data cleaning and preprocessing, dynamic adjacency matrix construction and ST-GCN prediction model construction, deep decoupling and fusion of spatiotemporal features are achieved. The model is trained by combining Huber loss function and Adam optimizer.

Benefits of technology

It significantly improves the accuracy and robustness of transformer error prediction, can adapt to different operating conditions, provides high-precision error prediction results, and enhances the interpretability and noise resistance of the model.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122241021A_ABST
    Figure CN122241021A_ABST
Patent Text Reader

Abstract

The current transformer error prediction method based on dynamic feature graphs and spatiotemporal graph convolutional networks belongs to the interdisciplinary field of power big data and artificial intelligence. It addresses the problems of existing methods neglecting multi-source data topological relationships, insufficient spatiotemporal feature mining, and weak noise resistance. This method first collects historical error data, load current data, and ambient temperature and humidity data of the current transformer. After preprocessing such as denoising with a Savitzky-Golay filter, each physical quantity is defined as a graph node. A dynamic adjacency matrix is ​​constructed by combining Pearson correlation coefficients and adaptive node embedding. Then, an ST-GCN model with alternating stacked "time-graph-time" convolutions is built to extract spatiotemporal coupling features. Finally, the model is trained using the Huber loss function and the Adam optimizer to output the error prediction value. This invention achieves explicit modeling of the dynamic coupling relationship of multi-source variables, improving the prediction accuracy and robustness under non-stationary operating conditions, and is suitable for current transformer condition monitoring and metering performance evaluation in smart grids.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Voltage transformer metering error prediction method, medium and terminal

    CN117932236A

  • Voltage transformer error prediction method and system based on spatio-temporal feature fusion

    CN119537861A