The day-ahead load forecast is essential for the efficient planning and operation of electric power systems, especially in the context of smart grids. This task is becoming increasingly important with the growing integration of variable renewable energy sources. Among the various machine learning-based load forecasting methods, Long Short-Term Memory (LSTM) networks have shown to be particularly effective. This paper analyses the impact of reactive load as an exogenous variable on active load forecasting and vice versa, employing LSTM networks with hyperparameters optimized through Gaussian Process Regression (GPR). The results, validated using dataset from Bangalore, India, demonstrate that including exogenous variables enhances forecasting accuracy. Additionally, the effect of different training/(validation+test) percentage ratios on prediction performance is evaluated finding that a 70%:30% ratio yields a satisfactory balance of accuracy and training efficiency. Finally, a combined forecasting model is used to analyse the forecasting accuracy of a model that is trained using data from one location (Bangalore) and tested using data from another location (Itanagar), proving there is no overfitting in the forecasting model.

The composite load model is one of the most comprehensive and widely used load models, as it includes and differentiates between static and dynamic load components. The simulation results, in which various load models were used, showed that the use of this model provides a good agreement between the simulated and measured responses. In order to obtain information about the composition of the load for the day ahead, a simple but improved artificial neural network (ANN) was used. It requires forecast active and reactive load data and gives as output the participation of each component of the composite load model. Forecast values of total active and reactive demand were obtained using another ANN which has the same settings as the one for load decomposition, but with different input and target. To show how much the forecast values of active and reactive demand affect the accuracy of the forecasted components of the composite load model, a load decomposition forecast was made for 7 days. The results showed that the forecast values of the total active and reactive demand do not proportionally affect the load decomposition error and depend on the variability of daily consumption and the use of the most recent historical data.

Efficient work of grids with the maximum potential utilization, together with supply and modern consumer satisfaction, as well as unpredictability of distributed energy sources, represents the challenge of successful load management. This paper proposes the load management framework and analyses state-of-the-art research activities from the area of load management in smart grids. It addresses three steps in load management framework: modelling and prediction, measurement and monitoring, and optimization and control. The original contribution of this meta-analysis and state-of-the-art review is a multi-factor approach to the process of load management with the identification of key influence factors from groups: system, context and user.