Electricity Consumption Prediction Model for Southern Luzon State University-Main Campus Using Artificial Neural Network

Neural Neural networks have been used in many applications over the years, from image classification, text classification, object detection, weather forecasting, parameter prediction, and load forecasting. This study presents two load forecasting models of SLSU based on long short term memory (LSTM) neural network architecture using a single variable as input (univariate) and using multiple variables as input (multivariate). Factors that significantly affect the electricity consumption of SLSU were determined using correlation analysis. Based on the result of correlation analysis, wind speed, number of students and employees, temperature, and number of days with classes significantly affects the electricity consumption of SLSU. These parameters were used to develop the multivariate LSTM model, while the univariate LSTM model used the historical electricity consumption of SLSU alone. Upon testing and validation, the multivariate LSTM model proves to be more accurate in electricity forecasting of SLSU by having a MAPE of 7.97% and RMSE of 3809.75 compared to the MAPE and RMSE 10.45 and 3992.98 respectively for the univariate model.