RESEARCH ON ABNORMAL DETECTION OF POWER METERING DATA FOR ELECTROMECHANICAL INTEGRATED ENERGY CONVERTER BASED ON TRANSFORMER

Read full text Summarise with VIRAT
Cited by: 0
References: 24

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

With the rapid development of smart grid electromechanical integration technology, the energy converter, as the core electromechanical integration equipment, presents highly automated operation characteristics, and its supporting power measurement data has both the characteristics of magnanimity, high dimensionality, nonlinearity and electromechanical coupling.The traditional anomaly detection methods face challenges such as insufficient accuracy, weak generalization ability and inability to adapt to the operating laws of electromechanical drivers.At the same time, the monitoring and anomaly determination of electromechanical energy converters must strictly comply with specific industry standards such as IEC 60034 (basic standard for rotating electrical machines), IEC60870-5-104 (transmission protocol for telecontrol equipment and systems), and the existing methods are not integrated into the standard rules, resulting in limited engineering applicability, which is difficult to be widely accepted by the industry.This paper proposes a Transformer based anomaly detection model for power measurement data (T-AD4Power) to meet the demand of electromechanical integrated energy converter for collaborative monitoring of power measurement and electromechanical operation; The multi head self attention mechanism is used to capture the long sequence dependency, combine the time-series mechanics fusion position coding retention time characteristics and electromechanical operation rules, and introduce multiscale feature fusion module to improve the sensitivity to local anomalies.In order to adapt to specific equipment standards and complex operation scenarios, the model innovatively designs a standard compliance constraint layer and a multi-dimensional complexity adaptation module, embedding the mechanical threshold and electrical parameter determination rules in industry standards into the model training and reasoning process.Experiments on SGSC data sets and the actual data of a provincial grid electromechanical energy converter show that the effect is significant compared with the traditional detection model, which provides technical support for data anomaly detection in the power industry, and also provides a new method for equipment monitoring in the cross field of electromechanical integration and applied mechanics.

Fields of Study & Subjects

Anomaly Detection Techniques and Applications Electricity Theft Detection Techniques Machine Fault Diagnosis Techniques

Publication details

Journal / source
International Journal of Mechatronics and Applied Mechanics
Volume
I
Issue
23
Publication date
19 March 2026
Language
EN