Abnormal Handling Evaluation Model of Electric Energy Metering Device Based on Multi-Source Information Fusion

With the rapid development of new power systems, abnormal problems in electric energy metering devices are becoming increasingly complex. Especially in the environment of multi-source information interaction, higher accuracy and real-time requirements are put forward for the evaluation of abnormal handling. To solve this problem, this paper proposes an abnormal handling evaluation model for electric energy metering devices based on multi-source information fusion. Firstly, data preprocessing and feature selection techniques are used to ensure the integrity and consistency of multidimensional information data. Secondly, a comprehensive evaluation index system based on Analytic Hierarchy Process (AHP) and dynamic optimization of weights is designed to quantify the treatment effect. Then, based on the Transformer network, the temporal dynamic characteristics of the abnormal treatment process are deeply modeled to capture the complex dependencies and interactive characteristics of multidimensional data. Finally, through multimodal data fusion technology, the equipment operating status, environmental parameters and historical records are integrated to achieve a highly robust and adaptable evaluation model. Experimental results show that the prediction accuracy of the Transformer-based model reaches 94.7%, which is significantly better than random forest (89.3%) and support vector machine (SVM) (85.6%); in terms of time response performance, the processing efficiency is more than 50% higher than that of random forest; in the robustness evaluation under noise interference conditions, the accuracy decrease is the smallest and significantly better than the comparison model. The research results have verified the application value of the model proposed in this paper in complex power grid scenarios, which can provide efficient and scientific support for new power systems.