Intelligent Terminal State Recognition in the Distribution Network

This paper proposes a state recognition algorithm for the intelligent terminal in the distribution network to combine the two functions of state assessment and fault recognition in condition-based maintenance. Secondly, this paper utilizes the probability distribution to reflect the likelihood of the terminal being in the normal state or a specific failure, which indicates the probability of the terminal transitioning to other states and enhances failure warning. Moreover, since there exists the performance limitation caused by the difficulty of collecting fault data in practical scenarios, this paper utilizes supervised comparative learning to address the few-shot problem and learns a representation space characterizing the state probability distribution. Finally, the effectiveness of the proposed algorithm in state recognition and fault warning is validated through simulation experiments.