As a new generation of artificial intelligence technology

DeepSeek-like (DS-like) paradigm has the advantages of open source

strong reasoning ability

low arithmetic demand

and local deployability

which meet the needs of secure

reliable evaluation and edge deployment of digital power equipment fault diagnosis data. We first introduced the development status of DS-like models and the basic architecture of DS-like paradigm

and analyzed its model advantages of high computing efficiency

low cache occupation

superior reasoning capability

open source model sharing

and Chinese optimization. Then

for the basic theory of power equipment fault diagnosis

we studied the mining model of power equipment fault parameter generation mechanism based on DS-like paradigm and the model of power equipment multi-state parameter coupling mechanism understanding; for the application of power equipment fault diagnosis technology

we investigated the multi-modal joint diagnosis technology of power equipment

the lightweight deployment technology of power equipment fault diagnosis model

and the performance self-checking technology of large-scale multi-parameter sensors. Finally

the challenges faced by the DS-like paradigm in the power industry are proposed

and the future development direction of the DS-like paradigm is summarized.