A Dynamic Harmonic Source Identification Method Integrating Power Grid Physical Topology Mechanism and Spatiotemporal Diagram Residual Analysis

Abstract

This invention discloses a dynamic harmonic source identification method that integrates power grid physical topology mechanisms with spatiotemporal graph residual analysis, relating to the field of power quality monitoring and analysis technology. By synchronously collecting voltage and current waveform data from multiple nodes in the power grid, a high-dimensional spatiotemporal feature tensor containing multiple harmonic components is constructed. The method deeply integrates power grid physical mechanisms, using an exponential decay function to map line length or impedance as edge weights, constructing a physically weighted adjacency matrix. A spatiotemporal prediction model is built to capture the spatiotemporal evolution of harmonics in the network and generate predicted values ​​for the entire network state. A multi-dimensional positioning decision function is used to weighted score and rank nodes. This method effectively overcomes the limitations of single indicators, achieving high-precision dynamic harmonic source identification even under strong background noise and multi-source interference, thereby further improving the spatiotemporal positioning accuracy of multi-source harmonic source identification.

Images