Adaptive integration method of transformer state big data based on deep hash algorithm

To solve the problem of data omission and low storage efficiency of big data, an adaptive integration method of transformer state big data based on deep hash algorithm is designed. Quantitatively evaluate the fluctuation degree of big data, judge the fluctuation degree of big data of transformer status, and formulate a stress transformer status big data acquisition scheme. Design a semantic retrieval algorithm based on the deep hash algorithm, configure the CNN network, design the loss function, train the CNN network through the loss function and the input semantic data, and realize the semantic retrieval of transformer status big data. Transform the retrieved isomerization semantics into metadata to solve the problem of data isomerization in big data integration. The adaptive integration model of transformer state big data is designed to realize multi-source adaptive integration of big data. The test results show that when the heterogeneous semantic data accounts for 30% and 60%, respectively, the data operation and processing time under the proposed method is 3589 ms and 5068 ms. The adaptive integration error of this method is low, fluctuating around 0.10%. The big data storage efficiency is higher than 95%, which has certain significance for transformer maintenance.