Probability density prediction of peak load based on mixed frequency noise-assisted multivariate empirical mode decomposition.

Accurate peak load forecasting is crucial to ensuring the reliable operation of the power system. However, existing prediction models often neglect full explanations for peak loads by factors that are often complex and fluctuating. Moreover, different factors are usually sampled at diverse frequencies, and existing processing methods have substantial limitations in mining irregular mixed-frequency multi-series information, leading to reduced accuracy. Consequently, this paper proposes an innovative ensemble peak-load prediction model that combines noise-assisted multivariate empirical mode decomposition algorithm for mixed-frequency sampling data with the quantile regression neural network to effectively tackle the aforementioned challenges. Firstly, noise-assisted multivariate empirical mode decomposition algorithm for mixed-frequency sampling data is employed to transform several mixed-frequency fluctuating time series into multiple sets of stationary sub-sequences. Secondly, the quantile prediction of peak load is performed for each group of subsequences using the quantile regression neural network. Finally, conditional quantiles under each quantile are accumulated as the samples for kernel density estimation to complete probability density forecasting. The case study validates the superiority and stability of the proposed model. Extensive experimental results on two real-world datasets from the VT and the Houston in America show that our proposed model is significantly superior to other benchmark methods regarding peak load especially on extremes. [ABSTRACT FROM AUTHOR]