Research on anti-interference transmission algorithm for sensitive images in smart grids based on the combination of tent chaotic mapping and deep learning

Abstract Image monitoring and transmission systems widely used in smart grids face problems such as complex electromagnetic interference, privacy leakage and inefficient retrieval. To address these challenges, this paper proposes an anti-interference transmission algorithm for sensitive images of smart grid based on the combination of Tent chaotic mapping and deep learning. The method improves NSGA-II optimization algorithm through Tent chaotic mapping to achieve global convergence enhancement and local optimal jumping out ability; meanwhile, it integrates DenseNet deep features with traditional HOG, BOW, ColorSpace features, and adopts PCA dimensionality reduction to improve the anti-interference expression ability of the features. In order to safeguard privacy security and retrieval efficiency, this paper designs a fine-grained access control based on CP-ABE and ρ -stable LSH secure indexing mechanism to support efficient similarity retrieval in ciphertext state. The experimental results show that the method can significantly reduce the BER, stabilize the DC bus voltage, maintain the fast recovery of three-phase current and voltage waveforms, and outperform the existing schemes in terms of image retrieval accuracy and computational efficiency under the extreme interference conditions, such as high-voltage overlay discharge and power switching. This study provides a new theoretical support and engineering realization path for the safe and efficient transmission of sensitive images in smart grids.