双碳背景下结合Informer和IPOA算法的电力负荷预测

Combining Informer and IPOA algorithms for power load forecasting under the dual carbon background

摘要

在双碳目标背景下,准确预测电力负荷对于实现能源优化和电网稳定运行至关重要,提出了一种结合I时间序列预测模型和改进鹈鹕优化算法的电力负荷预测模型。首先对原始数据进行清洗和标准化处理,通过改进鹈鹕优化算法优化时间序列预测模型的超参数。引入的时间失真指数提高对时间序列形状和时间扭曲的捕捉能力。为验证所提模型的有效性,将该模型与标准鹈鹕优化算法等开展了对比测试,并在标准测试函数和实际电力负荷数据集上对误差和拟合优度进行了全面评估。实验结果显示,提出的改进模型预测误差范围从标准鹈鹕优化算法模型的-7.5 ~5.8 kW·h显著降低至-4.3~4.3 kW·h,决定系数值提高了2.53%~6.6%。该结果验证了模型在提高预测精度和稳定性方面的显著效果,对双碳目标下优化电力资源配置具有重要的理论和实践意义。

Abstract

Accurately predicting power load is crucial for achieving energy optimization and stable operation of the power grid in the context of the dual carbon target. This article proposes a power load forecasting model that combines an I time series forecasting model with an improved Pelican optimization algorithm. This article first cleans and standardizes the raw data, and optimizes the hyperparameters of the time series prediction model by improving the Pelican optimization algorithm. The introduced time distortion index improves the ability to capture the shape and time distortion of time series. To verify the effectiveness of the proposed model, this paper conducted comparative tests with standard Pelican optimization algorithms, and comprehensively evaluated the error and goodness of fit on standard test functions and actual power load datasets. The experimental results show that the improved model proposed in this paper significantly reduces the prediction error range from -7.5 kW·h to 5.8 kW·h of the standard Pelican optimization algorithm model to -4.3 kW·h to 4.3 kW·h, and increases the determination coefficient value by 2.53% to 6.6%. This result validates the significant effect of the model in improving prediction accuracy and stability, and has important theoretical and practical significance for optimizing power resource allocation under the dual carbon target.

关键词

双碳目标 / 电力负荷 / 预测 / Informer / IPOA算法 / 机器学习

Key words

dual carbon targets / power load / forecasting / Informer / IPOA algorithm / machine learning