Machine Learning-Based Load and Price Forcasting for Energy Management Systems

Abstract

An intelligent load forecasting (LF) model is proposed for residential loads using a novel Machine Learning (ML)-based approach, achieved by assembling an integration strategy model with the Mean Absolute Percentage Error (MAPE) optimizer. In this proposed method, the time-series-based auto-regression schemes were carried out to collect historical data and set the objective functions of the proposed model. An algorithm with seven different auto-regression models was also developed and validated through a feed forward adaptive-network-based fuzzy inference system (ANFIS) model based on the ML approach. Moreover, a binary genetic algorithm (BGA) was deployed for the best feature selection, and the best fitness score obtained with Principal Component Analysis (PCA). A unique decision integration strategy is presented that led to a remarkably improved transformation in reducing MAPE. The model then tested by using a one-year Pakistan Residential Electricity Consumption (PRECON) dataset, and the attained results verify the validity of the proposed model with promising values in MAPE of 1.70%, 1.77%, 1.80%, and 1.67% for summer, fall, winter, and spring seasons, respectively. The overall improvement percentage is 17%, which represents a substantial increase for small-scale decentralized generation units. In the second part, a novel and improved technique is introduced to forecast electricity prices. The data of various power producers, Capacity Purchase Price (CPP), Power Purchase Price (PPP), Tariff rates, and load demand from National Electric Power Regulatory Authority (NEPRA) is considered for MAPE reduction in PF. Eight-time series and auto-regression algorithms are developed for data fetching and setting the objective function. The feed-forward ANFIS based on the ML approach and space vector regression (SVR) is introduced to forecast price forecasting (PF) by taking input from time series and auto-regression algorithms. Best feature selection is made by adopting the BGA-PCA approach, which reduces the repeated, irrelevant, and unnecessary data that ultimately minimizes the complexity and computational time of the model. The proposed integration strategy computes the MAPE according to the above-mentioned steps, which exhibited significant improvement of the system. Finally, the third part presents demand-side management (DSM) based on the Firefly algorithm (FA) has been presented. The peak load exerts extra stress on the grid; therefore, Abstract ix the FA has been established to forecast load to minimize peak load demand. By implementing FA, the cost of electricity has been reduced 21%, 19%, and 20% for building 1, 2, and 3, respectively. It utilizes the different types of the forecasted load for three residential buildings to verify the system's capability under different load scenarios. Similarly, it has also reduced the cost of electricity by implementing the data set of PF very efficiently.