LAUSR

The reliable operating state of smart electricity meters is significant in industrial applications. Faulty meters or meters in a poor measurement state will seriously impact both customers and stakeholders. However, the maintenance personnel now still use the manually periodic sampling inspection from the batch of smart electricity meters to evaluate the state of the entire batch, which has limitations of blindness, poor real time, inadequate, and insufficient examination. With the population of smart meters, it is feasible to evaluate the health condition of these devices with big data and artificial intelligence technology. One significant contribution of this article is first proposing an operating state evaluation method of smart electricity meters based on the transformer–encoder and bidirectional long-term and short-term memory. The evaluation indicators and preprocess of meters’ data are carefully selected. A deep neural network is constructed and trained, the experimental verification is carried out, and the performance of the proposed method is compared with that of other traditional methods. The results show that the average classification accuracy of the proposed neural network model is 99.5%. Besides, compared with conventional machine learning and deep learning models, the proposed model is suitable for the operation state evaluation of smart electricity meters. From the experimental result, the potential benefit of the proposed method is that it could improve the accuracy and robustness of state evaluation.1