基于VMD和贝叶斯优化LSTM的母线负荷预测方法

A Bus Load Forecasting Method Based on VMD and Bayesian Optimization LSTM

新型电力系统背景下,为提升母线负荷预测的精确性与稳定性,针对母线负荷噪声,提出一种考虑变分模态分解(variational mode decomposition,VMD)降噪优化和长短期记忆网络(long short-term memory,LSTM)的母线负荷预测方法。通过VMD将母线负荷分解为多个平稳的固有模态函数余项,将其分解项去除噪声后进行重组,达到降噪优化效果;对降噪后的母线负荷序列构建基于LSTM的时序预测模型,利用贝叶斯优化方法对网络初始超参数进行优化,以提高时序预测模型的精度。算例研究结果表明:利用VMD对母线负荷进行降噪优化后再进行预测,有利于预测结果更加稳定,且贝叶斯优化寻参解决了因初始参数设置不当而使预测结果精度不高的问题。该文方法可运用于母线短期负荷预测,并为电网调度运行提供了决策依据。

In the context of the new power system,a bus load forecasting method Is proposed by combining variational mode decomposition(VMD) and long short-term memory(LSTM)network to improve the accuracy and stability of bus load forecasting. Firstly,the bus load sequence Is decomposed into several intrinsic mode functions and remainder by VMD,and the bus load Is reconstructed by deleting remainder to reduce the noise of bus load. Secondly,the bus load is predicted by LSTM and the initial parameters of LSTM are improved by Bayesian optimization theory to improve the accuracy of prediction. The results of the case analysis shows that the prediction of the bus load that using VMD to reduce the noise can get more stable prediction results and the Bayesian optimization thesis has solved the problem of low accuracy caused by random initial parameter. This forecasting method has accurate and stable prediction results that can be applied to short-term bus load forecasting and provide effective decisionmaking basis for the power grid operator.