Research on transformer fault diagnosis models with feature extraction

To address the challenge of low accuracy in traditional transformer fault diagnosis algorithms, this paper introduces a novel approach that utilizes the Artificial Hummingbird Algorithm (AHA) to optimize both Kernel Principal Component Analysis (KPCA) and Extreme Learning Machine (ELM). We propose the use of various gas concentration ratio features and apply the AHA algorithm to fine-tune the kernel function parameters of KPCA, thus establishing an AHA-KPCA feature extraction model. This model takes the expanded gas concentration ratio features as input and selects the top N principal components with a cumulative contribution rate above 95% to form the feature vectors for fault classification. Following this, the AHA algorithm is employed to optimize the input weights and hidden layer biases of the ELM, leading to the development of the AHA-ELM fault classification model. Ultimately, the principal components identified by AHA-KPCA serve as inputs for the simulation verification of the AHA-ELM model. Experimental results indicate that the proposed AHA-KPCA-ELM method attains an accuracy rate of 95.73%, surpassing traditional intelligent diagnostic methods and existing advanced algorithms, thereby confirming the effectiveness of our proposed method.