Abstract:Accurate electrical load forecasting is of great importance for the efficient operation and control of modern power systems. In this work, a hybrid long short-term memory (LSTM)-based model with online correction is developed for day-ahead electrical load forecasting. Firstly, four types of features are extracted from the original electrical load dataset, including the historical time series, time index features, historical statistical features, and similarity features. Then, a hybrid LSTM-based electrical load forecasting model is designed, where an LSTM neural network block and a fully-connected neural network block are integrated that can model both temporal features (historical time series) and non-temporal features (the rest features). A gradient regularization-based offline training algorithm and an output layer parameter fine-tuning-based online model correction method are developed to enhance the model's capabilities to defend against disturbance and adapt to the latest load data distribution, thus improving the forecasting accuracy. At last, extensive experiments are carried out to validate the effectiveness of the proposed electrical load forecasting strategy with superior accuracy compared with commonly used forecasting models.

What problem does this paper attempt to address?

The paper aims to address the accuracy issue in power load forecasting, particularly focusing on the increased volatility and complexity brought about by the high penetration of renewable energy and electric vehicles in modern smart grids. The main contributions of the paper include: 1. **Comprehensive Feature Extraction**: The authors comprehensively consider four types of features, including historical time series, time index features, historical statistical features, and similarity features. By utilizing this multi-dimensional information, the accuracy of the forecast can be improved. 2. **Hybrid LSTM Model Design**: A hybrid Long Short-Term Memory (LSTM) neural network model is proposed, which consists of LSTM neural network blocks and fully connected neural network (FCNN) blocks. The LSTM blocks are used for modeling time series features, while the FCNN blocks are used for handling non-time series features. This structure can better capture the complex relationships in the data. 3. **Online Correction Strategy**: An offline training strategy based on gradient regularization and an online model correction method based on output layer parameter fine-tuning are adopted. These strategies enhance the model's resistance to interference and allow it to adapt to the latest load data distribution, thereby further improving forecasting performance. The research background points out that accurate power load forecasting is crucial for the optimal planning and efficient operation of smart grids. With the development of technologies such as renewable energy and electric vehicles, the power system faces greater volatility and complexity, which imposes higher requirements for precise power load forecasting. Although existing forecasting models have made some progress, they still have deficiencies in the comprehensiveness and adaptability of feature utilization. Therefore, this paper proposes a new hybrid LSTM model to address these challenges.

Electrical Load Forecasting Model Using Hybrid LSTM Neural Networks with Online Correction

Long-Short-Term Memory Network Based Hybrid Model for Short-Term Electrical Load Forecasting

Power Load Forecasting Based on the Combined Model of LSTM and XGBoost

Short-Term Load Forecasting Based on Deep Learning Bidirectional LSTM Neural Network

Load Forecasting Method for Building Energy Systems Based on Modified Two-Layer LSTM

A Method of Load Forecasting Based on Temporal Convolutional Network

Correction To: A Decomposition-Based Improved Broad Learning System Model for Short-Term Load Forecasting

Forecasting of Electric Load Using a Hybrid LSTM-Neural Prophet Model

Research and application of a hybrid model based on multi-objective optimization for electrical load forecasting

Load Forecasting Based on LSTM Neural Network and Applicable to Loads of “replacement of Coal with Electricity”

Research and Application of a Hybrid Forecasting Model Based on Data Decomposition for Electrical Load Forecasting

An electricity load forecasting model based on multilayer dilated LSTM network and attention mechanism

Load Forecasting Based on Improved Long Short-Term Memory Artificial Neural Network

Research and Application of Hybrid Forecasting Model Based on an Optimal Feature Selection System-A Case Study on Electrical Load Forecasting

Enhanced Short-Term Load Forecasting: Error-Weighted and Hybrid Model Approach

Electricity Load Forecasting for Distribution Network Based on Long Short-term Memory Recurrent Neural Network

A Novel Hybrid Short-Term Load Forecasting Method of Smart Grid Using MLR and LSTM Neural Network.

Multi-Convolution Feature Extraction and Recurrent Neural Network Dependent Model for Short-Term Load Forecasting

Application of Hybrid-Model Neural Networks to Short-Term Electric Load Forecasting

A CNN-LSTM Hybrid Model Based Short-term Power Load Forecasting

Electric Vehicle Charging Load Prediction Model Based on Improved LSTM Neural Network