基于多特征融合的故障电弧诊断方法OA

Aimed at the problem of higher false positive rate and false negative rate in fault detection caused by weak se-ries arc fault signals in industrial load circuit,a fault arc detection method based on multi-feature fusion is proposed in this paper.The short-time Fourier transform and residual neural network are used to extract the deep time-frequency characteristics of the current signal,the deep modal characteristics are obtained by synchronous variational mode de-composition and convolution combined with the long and short-term memory network,and the gated recurrent unit is used to capture the timing characteristics.The improved IVY algorithm is innovatively introduced to optimize the syn-chronous variational mode decomposition parameters,and the multi-head self-attention mechanism is integrated to strengthen the feature association and improve the recognition accuracy.Comparative experiments were carried out on a self-built series arc fault simulation platform,and results show that the proposed method not only shows high accuracy,but also effectively overcomes the limitation of single feature.Compared with the traditional fault diagnosis model,it has stronger robustness,significant advantage and engineering application potential,providing solutions and theoreti-cal support for series arc fault detection under complex industrial scenarios.