居住建筑用电数据分析与随机模型构建

Analysis and Stochastic Modelling of Electricity Consumption in Urban Residential Buildings

近年来居住建筑能耗呈快速上升趋势,而居住建筑全年用电曲线体现出明显的季节性差异;同时由于不同住户行为模式的差异,用户的用电模式和用电水平存在一定的随机性。为了指导夏季和冬季用电高峰电力系统的容量设计与调节、应对蓄能系统的全年容量设计与调节,需要提出一个可以反映用户全年用电特征和随机特性的模型。本文以江苏六市的用电数据为例,基于大数据的分析方法提出了一个模拟全年逐月用电数据的随机分布模型,采用K-means聚类算法和Calinski-Harabaz指标将用户按照用电水平和用电特征划分为16个子类,并针对每一个子类将用电特征参数进行分布拟合,得到了全样本的随机分布模型。根据随机分布模型可以随机生成符合样本特征的用户用电曲线,通过参数检验、模拟检验和交叉检验的方法验证的模型的可靠性,并探讨了模型的实践应用。

Energy consumption of the residential building in China has been in a rapidly-growing trend in recent years.And strong seasonal characteristics are shown in the curve of electricity utilization in the all-year round of the residential building.Meanwhile,the variation of household*s energy-related behavior leads to the stochastic distribution in the level and pattern of electricity utilization.In order to cope with the capacity design and wholeyear operation control for the winter and summer electricity peak power supply system and the energy storage system,it is necessary to propose a model that can represent the features and fluctuations of monthly electricity consumption for residential buildings.This paper presents a stochastic distribution model for monthly electricity consumption for residential buildings based on big data analysis with the electricity utilization data of 6 cities in Jiangsu as an example.In this model,K-means clustering method and Calinski-Harabaz indicators are adopted to divide all samples to 16 sub-categories.And then,distribution fitting is conducted on the electricity characteristic parameters of each sub-category to establish the stochastic distribution model.Based on this model,electricity utilization curves conforming to sample characteristics can be randomly generated.The model is validated through parametric validation,simulation validation and cross-validation.At last,the practical application of the model is also discussed.