智能电网下需求响应机理及其短期负荷预测模型研究

发布时间：2018-01-17 20:36

本文关键词：智能电网下需求响应机理及其短期负荷预测模型研究　出处：《青岛大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着电力市场的不断发展和日益完善,其利益主体逐渐呈现出多元化,电价机制也得到调整和更新。为综合优化整个电力系统的资源配置,缓解电网短期负荷容量不足,需求响应成为电力领域研究热点。在智能电网的技术支撑下,需求响应可通过制定峰谷电价促进电网与用户之间的实时性互动,有利于电力资源合理分配,配合短期负荷预测技术,最终实现双方获益。因此,量化分析峰谷电价下用户需求响应机理和研究计及需求响应的短期负荷预测新方法意义重大。根据消费者心理学原理,建立基于分段函数的需求响应机理模型,对峰谷电价进行量化分析,由于模型未充分考虑非电价因素的影响,因此赋予用户响应模糊属性。引入基于数据挖掘的负荷时间序列聚类方法,以负荷序列间欧式距离和差分序列标准差构成的交集约束作为聚类过程的综合判据,对目标电网全年的历史负荷数据做聚类处理。根据聚类结果,同时从负荷数据大小和曲线形态变化两个方面挖掘识别负荷特性,制定次日动态峰谷电价。通过实际算例,拟合预测日需求响应负荷曲线,结果表明实施动态峰谷电价削峰填谷效果显著,且参与需求响应项目的用户可根据参与度从中获取相应收益。另外,本文以曲线的形式协助描述了需求响应机理,基于时变函数建立了用户实际需求响应模型。通过频谱分析研究了需求响应负荷的基本特性,并以此为依据确定预测模型输入量组成,引入需求响应量化结果,分别建立了计及需求响应的RNN、Elman-NN和RBF-NN预测模型。通过实际算例,对比在三种预测模型中计及需求响应因素前后的预测性能,结果表明RBF-NN模型预测性能最佳,且将需求响应量化结果引入预测模型可显著提高其预测精度。分析上述实际算例结果,结合动态峰谷电价机制,根据消费者心理学原理描述了基于Logistic函数的用户需求响应机理。通过量化电价、用户响应程度以及温度等外界因素,构建了考虑需求响应综合影响因素的RBF-NN短期负荷预测模型。基于Logistic函数的需求响应机理充分考虑了电力用户应对不同电价的心理响应状态,且其响应度曲线在不同电价差分段点处连续可导,与描述需求响应机理的分段函数相比,更符合客观事实。通过实际算例,分析了本文构建模型在不同电价机制下的预测性能,证明了在RBF-NN模型中综合考虑电价、用户响应度等因素的重要性,为计及需求响应的短期负荷预测研究提供了一定的理论依据。
[Abstract]:With the development of power market and the increasingly perfect, the stakeholders gradually diversified, the price mechanism can be adjusted and updated. For the comprehensive optimization of the whole power system of the allocation of resources, alleviate the power short-term load capacity, demand response has become the research focus in the field of electric power. The technical support of the smart grid, demand response by make tou promote real-time interaction between grid and users, is conducive to the rational allocation of power resources, short term load forecasting technology, realize the benefit of both parties. Therefore, the quantitative analysis and Research on the response mechanism and demand response to short term load forecasting method is significant user demand under peak valley electricity price. According to the principle of consumer psychology. To establish a mechanism model of piecewise function response based on the needs of quantitative analysis of the peak valley price, because the model does not take into account the non price for The influence, therefore gives the user response fuzzy attributes. The introduction of load time series clustering method based on data mining, the Euclidean distance between the load sequence and the difference series of standard deviation constitute the intersection of constraints as the comprehensive criteria of the clustering process, the historical load data of the annual target grid clustering processing. According to the clustering results, at the same time from two a load curve data size and morphological changes of mining to identify load characteristics, formulate the dynamic tou. Through practical examples, the fitting and prediction of demand response on load curve, the results show that the dynamic peak electricity peak effect, and participate in the project according to customer demand response participation from which to obtain the corresponding revenue. In addition, this paper describes the demand curve in the form of Assistance Response Mechanism Based on time-varying function to establish the actual needs of users. The frequency spectrum response model Analysis of the basic characteristics of load demand response, and on this basis to determine the model input. The introduction of demand response quantitative results, considering demand response RNN are established respectively, Elman-NN and RBF-NN prediction model. Through practical examples, contrast response prediction performance factors before and after three kinds of prediction model considering demand the results show that the RBF-NN model is the best prediction performance, and the quantitative results into the prediction model can significantly improve the prediction accuracy. In response to the needs of the actual analysis results, combined with the dynamic peak valley electricity price mechanism, according to the principle of consumer psychology describes the Logistic function of the user demand response based on the mechanism. By quantifying the price, external factors and temperature etc. in response to the user, based on the consideration of comprehensive effect of short term load forecasting model RBF-NN factors in response to demand. The Logistic function of the demand response based on machine And take full account of the power users with different price psychological response state, and the response curve of price difference in different sub points continuously differentiable, and describe the demand response mechanism of piecewise function compared to more accord with the objective facts. Through practical examples, this paper constructs the performance prediction model in different price mechanism is analyzed. It is proved that RBF-NN model considering the price, the user response degree of importance and other factors, provide a theoretical basis for the research of demand response to short-term load forecasting.

【学位授予单位】：青岛大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM715

【参考文献】