Self-blast state detection of glass insulators based on stochastic configuration networks and a feedback transfer learning mechanism

•The interlaced group convolution strategy is employed to reconstruct the convolutional layers of the Inception network, which reduces the redundancy of convolutional kernels in the channel extent.•To avoid the redundancy of feature maps, the feature space with discrimination is reconstructed.•SCNs ensures the universal approximation property of the model.•The objective function can evaluate the uncertain results of the self-blast states of glass insulator images in real time.•The dynamic transfer learning mechanism improves the detection accuracy of the model.