樊树强 万俊杰（安科瑞电气股份有限公司，上海市 201801）

Research on the Prediction of Enterprise Energy Consumption

Based on VMD - GWO - HKELM

FAN Shuqiang WAN Junjie(Acrel Electric Co.,Ltd.,Shanghai 201801，China)

Abstract：Energy consumption prediction is the major concern of an enterprise and is one of the important tasks for an enterprise to establish an energy consumption management and control platform,which plays a significant role in ensuring safe and stable energy consumption. An energy consumption prediction model of optimized hybrid kernel extreme learning machine(HKELM)is proposed based on variational mode decomposition(VMD)and grey wolf optimizer (GWO). First,VMD is employed to decompose energy consumption data sequence and obtain sub‑sequences with different characteristic rules,so as to reduce the randomness of raw data of energy consumption. Then,Gaussian kernel function and polynomial kernel function are adopted to establish HKELM model with a larger generalization ability. Meanwhile,considering the difficulty in the selection of parameters of HKELM model,GWO is employed to optimize the selection of parameters and establish GWO - HKELM model. Finally,the data after decomposition are input into GWO - HKELM model, and the prediction results of all sub‑sequences are added together to obtain the final prediction result. The data of actual electric energy consumption of a plant in Zhejiang Province is used as an example to verify the effectiveness and feasibility of such model.

Key words：energy consumption prediction;energy consumption management and control;VMD;GWO;HKELM;data decomposition;Gaussian kernel function;polynomial kernel function

摘　要：能耗预测是企业关注的重点，对于企业保障用能安全稳定有重大作用，是企业建立能耗管控平台的重要任务之一。为完善企业能耗管控平台并且高效准确地进行企业能耗预测，提出一种基于变分模态分解(VMD)与灰狼算法(GWO)优化混合核极限学习机(HKELM)的能耗预测模型。首先利用VMD对能耗数据序列进行分解，获得具有不同特征规律的子序列，以降低能耗原始数据的随机性；然后，结合高斯核函数和多项式核函数构建更具有泛化能力的HKELM模型，同时针对HKELM模型中参数难以抉择的问题，采用GWO对其参数进行优化选择，并构建GWO - HKELM模型；最后，将分解后的数据输入到GWO - HKELM模型中，并将每个子序列的预测结果相加得到最终的预测结果。以浙江某厂房的实际电力能耗数据为例，验证该模型的有效性和可行性。

关键词：能耗预测；能耗管控；VMD;GWO;HKELM；数据分解；高斯核函数；多项式核函数

中图分类号：TU111.19 + 5     　  文献标识码：A

doi：10.3969 / j. issn.1003 － 8493.2024.04.014