Abstract:Electricity load prediction plays a vital role in energy management. The necessity of combining scattered power generation channels of sustainable energy and other substitute resources into distribution grid networks is increasing because of the 4.1 percent rise in world electricity since 2021. The utilization of probability power generators and energy storage is maximized by short-term load prediction methods, which make it possible to anticipate potential energy usage. To facilitate the growth of energy systems in quite a sustainable way, urban energy planners utilize a range of methods and technological advances, including simulation, probability optimization, comprehensive improvement, and modelling techniques for predicting power requirements. Predicting energy consumption at granular time scales (e.g., hourly or sub-hourly intervals) is common practice for short-term energy load prediction. For similar time-series forecasting challenges, the authors have decided to use transfer learning (TL) with light gradient-boosting machine (LightGBM). 1 This model can benefit from models trained on larger data sets through TL, which might boost prediction accuracy. Energy prediction and effective resource utilization largely support the accurate transition of energy systems. In this paper, the authors propose a hybrid model for short-term energy load prediction. Based on TL to optimize LightGBM (OLGBM) for smart grids, The proposed model first applies data pre-processing using an abnormal supplement strategy with an immediate deviation selection technique. This phase eliminates the missing values and extracts the required features. The second phase uses a TL-OLGBM with hyperparameters. The TL method helps to learn the dynamic time scale and complex data patterns, and the hyperparameters are tuned via the Bayesian optimization technique, which addresses the challenge of short-term load forecasting (STLF) and increases forecasting accuracy. Finally, an optimized LightGBM model is applied, combining the time and energy features to predict effective energy forecasting. The proposed model gained a 2.834589 mean absolute percentage error (MAPE) value and a 0.9706495 GINI index after execution on the prescribed data set. We compare the performance of the proposed model with other comparable models, and the simulation outcomes show that the model can generate the best results for MAPE, accuracy, and root-mean-square deviation (RMSE).

On the practical aspects of machine learning based active power loss forecasting in transmission networks

Robust and Automatic Data Cleansing Method for Short-Term Load Forecasting of Distribution Feeders

Analysis and Prediction of Power Distribution Network Loss Based on Machine Learning

Analysis for Non-Residential Short-Term Load Forecasting Using Machine Learning and Statistical Methods with Financial Impact on the Power Market

Model for 400 kV Transmission Line Power Loss Assessment Using the PMU Measurements

Forecasting Day-Ahead Electricity Metrics with Artificial Neural Networks

Data-Driven Techniques for Short-Term Electricity Price Forecasting through Novel Deep Learning Approaches with Attention Mechanisms

Forecasting of Electrical Energy Consumption in Slovakia

Local Cold Load Pick-up Estimation Using Customer Energy Consumption Measurements

Deep-learning-assisted power loss management for active distribution networks

A Machine Learning Model Ensemble for Mixed Power Load Forecasting across Multiple Time Horizons

A Machine Learning Application for the Energy Flexibility Assessment of a Distribution Network for Consumers

An Asymmetric Loss with Anomaly Detection LSTM Framework for Power Consumption Prediction

Optimized Data-Driven Models for Short-Term Electricity Price Forecasting Based on Signal Decomposition and Clustering Techniques

Smart Grid Forecasting with MIMO Models: A Comparative Study of Machine Learning Techniques for Day-Ahead Residual Load Prediction

A Hybrid Model for Short-Term Energy Load Prediction Based on Transfer Learning with LightGBM for Smart Grids in Smart Energy Systems

A comparative assessment of deep learning models for day-ahead load forecasting: Investigating key accuracy drivers

In Search of Deep Learning Architectures for Load Forecasting: A Comparative Analysis and the Impact of the Covid-19 Pandemic on Model Performance

Machine Learning Model Development to Predict Power Outage Duration (POD): A Case Study for Electric Utilities

A Global Modeling Approach for Load Forecasting in Distribution Networks

An advanced framework for net electricity consumption prediction: Incorporating novel machine learning models and optimization algorithms