A granularity data method for power frequency electric and electromagnetic fields forecasting based on T–S fuzzy model

Abstract

The impact of electromagnetic radiation generated by signal transmission base stations and power stations to meet the needs of communication equipment and energy consumption on the environment has caused people concerns. Monitoring and prediction of electric and magnetic fields have become critical tasks for researchers. In this paper, we propose a granularity data method based on T–S (Takagi–Sugeno) fuzzy model, named fuzzy rule-based model, which utilizing finite rules that are determined by the deviations between the predicted values and the true values after the data goes through a granulation-degranulation mechanism, to predict the intensity of power frequency electric field and electromagnetic field. A series of experiments show that fuzzy rule-based models have better robustness and higher prediction accuracy in comparison with several existing prediction models. The improvement of the performance of the fuzzy rule-based model quantified in terms of Root Mean Squared Error is 20.86%, 51.91%, 62.28%, 65.10%, and 71.92% in comparison with that of the Ridge model, Lasso model, and a family of support vector machine model with different kernel functions, including linear kernel (SVM-linear), radial basis function (SVM-BRF), polynomial kernel (SVM-polynomial) respectively, on the electromagnetic field testing data, and 37.42%, 55.16%, 58.79%, 59.28%, 64.27% lower than that of the Ridge model, Lasso model, SVM-linear model, SVM-BRF model and SVM-polynomial model on the power frequency electric field testing data.