Research on Health State Evaluation Method for IoT-Based Smart Electric Meters Based on LSTM

With the widespread application of IoT technology in power systems, the reliability and health status of smart electric meters directly affect grid security and management efficiency. This study focuses on the health degradation issues of IoT -based smart electric meters during long-term operation. To address the limitations of traditional evaluation methods in capturing the nonlinear and temporal features of multidimensional monitoring data, a health state evaluation method based on LSTM networks is proposed. The goal is to establish a high-precision and interpretable evaluation model to improve the timeliness and accuracy of anomaly warnings, thereby providing data support and decision-making basis for the full lifecycle management of equipment. Through data cleaning and feature engineering, key operational indicators such as voltage, current, active power, temperature, and communication quality are extracted to construct a feature matrix. The LSTM model is then used to model the time series data, and a health index (HI) is incorporated to build an evaluation system that enables quantitative assessment of device status. Experimental results show that the LSTM model outperforms the SVM method in terms of prediction error, fitting degree, and trend capturing, achieving dynamic and accurate evaluation of smart meter health status, particularly demonstrating stronger robustness and sensitivity during the mid-to-late degradation stages.