Hybrid Fuzzy Logic System and Pattern Search Algorithm (FLS-PSA) for Hospital Electricity Cost Prediction Considering Patient Visit and Ambient Temperature

This study proposes a novel Hybrid Fuzzy Logic System and Pattern Search Algorithm (FLS-PSA) for predicting electricity costs at the Ulin General Hospital, one of the large-scale hospitals in Banjarmasin, South Kalimantan, Indonesia. This study involves patient visit and ambient temperature data, as well as daily electricity consumption data, collected over a six-month period, which were then used for a rigorous five-month rolling-origin evaluation. The proposed method is implemented in the energy management system of Ulin General Hospital with three electricity costs: a peak-load period, an off-peak period, and the associated reactive power demand. Based on validation using the monthly electricity bills issued by the electricity utility company, FLS-PSA produced a better average Mean Absolute Percentage Error (MAPE) of 8.38% than FLS-Particle Swarm Optimization (8.49%), FLS-Genetic Algorithm (8.68%), FLS-Simulated Annealing (12.27%), and FLS Type-1 (10.25%). Moreover, the performance evaluations using R2, Root Mean Squared Error (RMSE), and Mean Absolute Error (MAE) are also presented. These results indicate that the FLS-PSA is more precise in predicting electricity costs at Ulin General Hospital than the other methods.