Prediction of Electric Consumption Based On Economic Coefficients Using Walrus Optimization Techniques

Artificial intelligence-driven models demonstrate robustness and comprehensiveness across various application domains and prediction timeframes. Introducing a novel algorithm, Walrus Optimization Techniques (Walrus_OT), inspired by the natural behavior of walruses. This research aims to utilize both linear and non-linear models to predict electricity consumption by leveraging economic coefficients. Focusing on Tamil Nadu as a case study, factors such as population, GDP, and individual income are incorporated as independent variables. Forecasts are extended up to 2031, employing average scenarios to predict future electricity consumption. To assess model performance, It is suggested to use a complete reference range for Mean Absolute Percentage Error (MAPE). The analysis shows that nonlinear models is preferred for yearly power consumption predictions. These models provide the least MAPE for future projections in addition to clarifying the connection between consumption data and influencing variables. This study creates a basis for comparing different electric consumption forecasting techniques and provides insightful information for researchers choosing models.