Basic Error Prediction for Smart Meters Based on ISSA-LSSVM With Combination Weighting

Aiming to the practical need for achieving the online monitoring and operation warning of on-grid smart meters in extreme environment stresses, an analysis method is proposed to predict the basic error of smart meters based on improved sparrow search and optimized least-squares support vector machine with combination weighting. Firstly, redundant quantities in the environmental stresses are eliminated through Pearson correlation analysis, and the environmental stresses highly correlated with the basic error of smart meters are extracted as input features of the least squares support vector machine (LSSVM). Subsequently, an improved sparrow search algorithm (ISSA) is established by combining chaotic mapping and Gaussian variation methods to achieve the adaptive optimization of the parameters of LSSVM prediction model. Additionally, in order to further enhance the prediction performance of LSSVM, the input samples are subjected to kernel fuzzy C-means clustering (KFCM) analysis, and the ISSA-LSSVM combined weighting model is constructed based on the membership degree between the test set and the KFCM subset. Example analysis of real-time basic error data from smart meters in the high dry heat environment test base in Xinjiang demonstrates that the average root mean square error of the proposed prediction model is 0.0157, with an average goodness of fit of 98.755%, indicating higher prediction accuracy compared to prevalent prediction methods.