A condition assessment method of key equipment in substation based on DBN-DFF

Given the issues of scattered maintenance data, differing knowledge structures, and the difficulty of accurately representing the overall state of equipment using single-source information in substations, it is necessary to conduct state assessments of critical substation equipment. Accurate evaluation of equipment status can reduce downtime during substation operation and maintenance. This paper proposes a substation equipment state assessment method based on DBN-DFF. First, multi-source heterogeneous data is processed, and a Deep Belief Network (DBN) is used to perform dimensionality reduction and fusion of high-dimensional multi-source heterogeneous data to extract feature vectors. Then, feed the treated feature information into a deep feedforward network (DFF). Utilizing the powerful processing capabilities of DFF for nonlinear data, this method addresses the weak correlation issues in historical data and improves the accuracy of fault diagnosis. The obtained feature vectors are concatenated and fused to achieve equipment state perception. Finally, the accuracy of the proposed method is verified using historical data from a specific substation as an example. The results indicate that the substation equipment state evaluation method based on DBN-DFF achieves high diagnostic accuracy, enabling more accurate assessment of critical substation equipment states.