Assessment of Static Voltage Stability Margin for Power Systems with Causal Decomposition and Reverse Feature-Enhanced Temporal Convolutional Networks

Rapid and accurate static voltage stability assessment is crucial for the safe operation of large-scale interconnected power grids. Existing methods have failed to effectively filter key input variables in massive measurement data, limiting assessment accuracy. This study proposes an assessment method based on causal decomposition and reverse feature-enhanced temporal convolution networks. Initially, using bottom-layer measurement data as input, key variables are selected through causal decomposition, reducing the original feature set and enhancing computational efficiency. Subsequently, a reverse feature-enhanced temporal convolution model is constructed to fully mine data characteristics and achieve high precision assessment. Validation in an improved four-machine two-area simulation system demonstrates that this method significantly enhances estimation accuracy and algorithm performance while reducing the dimensionality of original variables.