Abstract
In view of the lack of the existing insulator self-blast state detection method and the scale defects of deep neural network structure, imitated the human cognitive model, and learn from the closed-loop control theory, this paper explores a feedback recognition method of insulator self-blast state with multi-scale convolutional neural network. Firstly, for the pre-processed insulator images, based on ResNet-18, branches with different network structure are added to improve the network ability to adapt to different resolutions. At the same time, the multi-scale information fusion module is added at the end of the network. Secondly, the multi-scale feature vector is sent to stochastic configuration networks (SCN) with universal approximation ability to establish the classification criterion of the self-blast state of insulator images with strong generalization ability. Finally, an imitation of human thinking patterns is employed that exhibits repeated deliberation and comparison. Consequently, based on generalized error and entropy theories, the performance index is defined to evaluate the uncertain results of the insulator self-blast states in real time. Then, the regulation mechanism is given to realize the self-optimizing of fixed feature space and the reconstruction of classification criteria, which renders the insulator self-blast states is re-recognized with feedback mechanism. Experimental results show that, compared with other open-loop and closed-loop algorithms, the proposed method enhances the generalization ability and improves the recognition accuracy of the model.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have