Table 1 Comparative relationship between image recognition and power grid feature recognition.

From: Research on high proportion of clean energy grid-connected oscillation risk prediction technology based on CNN and trend feature analysis

 

Image recognition

Grid measurement data

Compression and segmentation

Compress images at different resolutions and divide them into N * N regions to find key features

There is also a need for compression and segmentation of power grid measurement data on a time scale. Different disturbances have different durations and frequencies in the time dimension, requiring segmentation to explore feature correlation relationships 33

Feature extraction

It is a gradient relationship in the image

In the power grid, it is the time domain characteristic, slope characteristic, and mutation characteristic of the monitoring quantity

Classification

Distinguish between the background and the target

Finding fault data fragments in massive monitoring data

Back to article page