Threefold Optimized Forecasting of Electricity Consumption in Higher Education Institutions

Energy management benefits both consumers and utility companiesalike. Utility companies remain interested in identifying and reducing energywaste and theft, whereas consumers’ interest remain in lowering their energyexpenses. A large supply-demand gap of over 6 GW exists in Pakistan asreported in 2018. Reducing this gap from the supply side is an expensiveand complex task. However, efficient energy management and distributionon demand side has potential to reduce this gap economically. Electricityload forecasting models are increasingly used by energy managers in takingreal-time tactical decisions to ensure efficient use of resources. Advancementin Machine-learning (ML) technology has enabled accurate forecasting ofelectricity consumption. However, the impact of computation cost affordedby these ML models is often ignored in favour of accuracy. This studyconsiders both accuracy and computation cost as concurrently significantfactors because together they shape the technology environment as well ascreate economic impact. Thus, a three-fold optimized load forecasting modelis proposed which includes (1) application specific parameters selection, (2)impact of different dataset granularities and (3) implementation of specificdata preparation. It deploys and compares the widely used back-propagationArtificial Neural Network (ANN) and Random Forest (RF) models for theprediction of electricity consumption of buildings within a university. In addition to the temporal and historical power consumption date as input parameters, the study also embeds weather data as well as university operationalcalendars resulting in improved performance. The outcomes are indicativethat the granularity i.e. the scale of details in data, and set of reduced and fullinput parameters impact performance accuracies differently for ANN and RFmodels. Experimental results show that overall RF model performed betterboth in terms of accuracy as well as computational time for a 1-min, 15-minand 1-h dataset granularities with the mean absolute percentage error (MAPE)of 2.42, 3.70 and 4.62 in 11.1 s, 1.14 s and 0.3 s respectively, thus well suitedfor a real-time energy monitoring application.