Intelligent forecasting of electricity consumption for hospital outpatient buildings by FOA-SVR hybrid strategy

Introduction To address the dilemma that the small sample size of hospital energy consumption data makes it difficult to predict short-term electricity consumption, a combination of the Firefly Optimization Algorithm (FOA) and the Support Vector Regression (SVR) algorithm, i.e., FOA-SVR, was proposed in this work. Methods By combining standardized sample data with the FOA to optimize the hyperparameters of the SVR model, the proposed approach enhances the model’s ability to capture the variation characteristics of hospital electricity consumption. Results The FOA-SVR hybrid strategy achieves an optimal balance between prediction accuracy and computational efficiency when the number of fireflies was 30. The prediction accuracy indicator (Coefficient of Determination, R 2 ) was 0.855, respectively. Under these conditions, the hybrid strategy has the dual advantage of running faster than the existing Sparrow search algorithm, and the traditional seagull optimization algorithm, with run times reduced by 21.192 s and 14.612 s, respectively. When the length of electricity consumption data was greater than or equal to 36, R 2 of the FOA-SVR hybrid strategy was higher than 0.849. Discussion The FOA-SVR hybrid strategy realizes a kind of efficient prediction of power consumption in medical office buildings with a small sample data volume, which provides theoretical and data support for the reasonable optimization of hospital energy use structure and has practical significance for the intelligence of hospital energy management.