Abstract:Electricity load forecasting plays an important role in the energy planning such as generation and distribution. However, the nonlinearity and dynamic uncertainties in the smart grid environment are the main obstacles in forecasting accuracy. Deep Neural Network (DNN) is a set of intelligent computational algorithms that provide a comprehensive solution for modelling a complicated nonlinear relationship between the input and output through multiple hidden layers. In this paper, we propose DNN based electricity load forecasting system to manage the energy consumption in an efficient manner. We investigate the applicability of two deep neural network architectures Feed-forward Deep Neural Network (Deep-FNN) and Recurrent Deep Neural Network (Deep-RNN) to the New York Independent System Operator (NYISO) electricity load forecasting task. We test our algorithm with various activation functions such as Sigmoid, Hyperbolic Tangent (tanh) and Rectifier Linear Unit (ReLU). The performance measurement of two network architectures is compared in terms of Mean Absolute Percentage Error (MAPE) metric.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the non - linear and dynamic uncertainty issues in short - term power load forecasting (STLF). Specifically, the main challenges faced by power load forecasting in the smart grid environment include: 1. **Non - linearity**: Power load data has complex non - linear characteristics, which makes it difficult for traditional linear models to accurately capture its variation patterns. 2. **Dynamic uncertainty**: Power loads in the smart grid environment are affected by multiple factors, such as weather, time, holidays, etc., which increase the difficulty of prediction. To solve these problems, the author proposes to use deep neural networks (DNN) to model the complex non - linear relationships between power loads and various influencing factors. By introducing two deep neural network architectures - deep feed - forward neural network (Deep - FNN) and deep recurrent neural network (Deep - RNN), and combining different activation functions (Sigmoid, Tanh and ReLU), the aim is to improve the accuracy of short - term power load forecasting. ### Specific objectives: - **Improve prediction accuracy**: By using DNN, especially Deep - FNN and Deep - RNN, to better capture the non - linear relationships in power load data, thereby improving the accuracy of prediction. - **Evaluate the effects of different network architectures and activation functions**: Compare the performance of different network architectures and activation functions in short - term power load forecasting tasks and select the optimal model configuration. - **Verify the performance of the model in different seasons**: By testing data from different seasons, evaluate the robustness and adaptability of the model under various conditions. ### Key points of the solution: - **Data pre - processing**: Pre - process the power load data and meteorological data obtained from the New York Independent System Operator (NYISO), and extract features that have a significant impact on power consumption, such as temperature, working days/non - working days, time effects, lagged load effects, etc. - **Model training**: Use the Levenberg - Marquardt back - propagation algorithm to train the Deep - FNN and Deep - RNN models, and minimize the loss function by adjusting weights and biases. - **Performance evaluation**: Use the mean absolute percentage error (MAPE) as a performance evaluation metric, and select the model with the lowest MAPE value as the final prediction model. Through the above methods, the author hopes to achieve higher accuracy and better robustness in short - term power load forecasting, thereby providing strong support for demand - side management in smart grids, energy storage operations, peak load reduction, etc.

Short Term Load Forecasting Using Deep Neural Networks

Short term power load forecasting using Deep Neural Networks

A Hybrid Short-Term Load Forecasting Method Based on Improved Ensemble Empirical Mode Decomposition and Back Propagation Neural Network

Short-Term Electrical Load Demand Forecasting Based on LSTM and RNN Deep Neural Networks

Deep Neural Network and Long Short-Term Memory for Electric Power Load Forecasting

Data-driven short term load forecasting with deep neural networks: Unlocking insights for sustainable energy management

An Ensemble Broad Learning Scheme for Short-Term Load Forecasting

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

Short-Term Load Forecasting of Integrated Energy Systems Based on Deep Learning

Short-term Load Forecasting Using a Long Short-Term Memory Network.

Electricity load forecasting and feature extraction in smart grid using neural networks

Building Energy Load Forecasting using Deep Neural Networks

A Deep Learning Framework for Short-term Power Load Forecasting

Machine and deep learning approaches for forecasting electricity price and energy load assessment on real datasets

Application of Fuzzy-RBF-CNN Ensemble Model for Short-Term Load Forecasting

A Novel Two-Stage Framework for Mid-Term Electric Load Forecasting

Deep Learning Algorithms and Parallel Distributed Computing Techniques for High-Resolution Load Forecasting Applying Hyperparameter Optimization

Short-term Load Forecasting with Deep Residual Networks

Short term residential load forecasting using long short-term memory recurrent neural network

Short-Term Load Forecasting in Smart Grids Using Hybrid Deep Learning

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