Neural-network-based demand forecasting in a deregulatedenvironment

The traditional approach to load forecasting is based on processing time series of load and weather factors recorded in the past. In the dynamic environment of the deregulated power industry, historical load data may not always be available. This paper explores the possibility of an alternative approach toward load forecasting based on indirect demand estimation from available customer data. This approach requires utilization of demand models for different customer categories. This paper presents a neural network-based method of demand modeling. Neural networks are designed and trained based on the aggregate demands of the groups of surveyed customers of different categories. The performance of such models depends on the neural network design and representativeness of the training data. The forecast accuracy is also affected by the forecasted group size, customer characteristics, customer classification system, and the extent of demand survey. This paper discusses the issues of neural network design and illustrates the proposed method by its application to forecasting demand of residential customers     

The application of artificial neural networks to substation load forecasting

The substation loading is highly correlated with the customers served. The substations in a distribution system can be categorized as residential, commercial and industrial. Each type has a different power consumption pattern. The substation loading will be varied according to the combination of the above three types of customers. In this paper, a supervisory functional artificial neural network (ANN) technique is applied to solve the load forecasting of three Taipower substations which serve the different customer types. The load forecasting accuracy is enhanced by considering the temperature effect on the substation load demand. With the converged ANN models derived by a training procedure, the temperature sensitivity of the substation load demand is easily obtained by the recall process. It is suggested that the substation load forecasting can be performed efficiently by the proposed method to support distribution operation effectively.     

考虑新用户负荷发展的中短期负荷预测方法

区分新接电用户与老用户负荷增长的不同规律,使中短期负荷预测更符合实际.介绍了考虑新接电用户负荷发展的中短期负荷预测方法的基本思路,数学模型及负荷预测表达式,新接电用户需用系数计算方法.通过地区中短期负荷预测实例的结果分析,建议了准确度最高的负荷预测公式,并证明该预测方法较原有的方法明显提高了负荷预测的精确程度.     

Derivation of class load pattern by field test for temperature sensitivity analysis

This paper proposes systematic procedures to derive the load pattern of various customer classes in a utility company. The questionnaires are adopted to find the power consumption of key electrical appliances. The customer load information is obtained through intelligent equipment which records customers' electricity demand on a 15-minute interval basis throughout the year. Five hundred meters are installed on statistically selected samples from the various customer classes. By the proposed sampling theory, the customer load characteristics will be derived with a sufficient confidence level. Statistical analysis is then performed to find the typical load pattern of each customer class based on the power measurements of field tests. The temperature effect on the power consumption of each customer class is then solved by investigating the relationship between customer power consumption and the ambient temperature. The proposed procedure has been adopted by Taipower Company to determine the customer load pattern to provide valuable information for better distribution planning and to design better load management programs to enhance system operating efficiency.     

配网故障快速复电系统建设及应用

文章提出了一种实现配网故障快速复电的信息系统模型。该信息系统模型以营配一体化平台为基础,集成了计量自动化、配网自动化、调度自动化、客服等系统的信息,实现了故障监控、报障过滤、故障预判、指挥调度、智能分析、移动作业等功能,为故障快速响应、沟通和复电提供信息支持。经过在深圳供电局有限公司1年多的应用证明,该系统在提升故障管理水平、减少用户故障停电时间和提高客户服务水平方面发挥了重要作用。     

基于最大信息熵原理的短期负荷预测综合模型

引入了信息理论来研究和处理负荷变化的不确定性，提出了基于最大信息熵原理的短期负荷预测综合模型，该模型将各种单一预测模型的预测结果以及历史预测误差分布作为约束信息，利用最大熵原理得到预测结果的分布。文中阐述了新模型的应用背景、思路和理论，给出了具体的实现方案和算法，并在实际电网中得到了应用。针对实际电网的算例研究表明，对于随机性较大的电网负荷，传统综合预测模型存在明显的过拟合现象，而新模型则有效地提高了预测精度。     

Using AMR data for load estimation for distribution system analysis

With the development of distribution automation (DA) and other advanced applications in distribution systems, the real-time monitoring and control of distribution systems becomes possible. Now there are only a limited number of real-time measurements on the distribution systems. The load monitoring and estimation of customers can be an important source of information used by the distribution analysis applications. In recent years, an increasing number of automated meter reading (AMR) systems have been installed. AMR can provide customer consumption information and other data such as confirmations for outages and restoration. In this paper, a load estimation algorithm is discussed. The proposed algorithm makes use of the above information that AMR provides as its input. It also incorporates time series forecasting method and the use of the customer load curves to improve the accuracy of individual customer real-time load estimates. This method with the use of AMR data has excellent load estimation results. This method demonstrates how AMR data can be used for other functions besides billing.     

Electricity demand load forecasting of the Hellenic power system using an ARMA model

Effective modeling and forecasting requires the efficient use of the information contained in the available data so that essential data properties can be extracted and projected into the future. As far as electricity demand load forecasting is concerned time series analysis has the advantage of being statistically adaptive to data characteristics compared to econometric methods which quite often are subject to errors and uncertainties in model specification and knowledge of causal variables. This paper presents a new method for electricity demand load forecasting using the multi-model partitioning theory and compares its performance with three other well established time series analysis techniques namely Corrected Akaike Information Criterion (AICC), Akaike's Information Criterion (AIC) and Schwarz's Bayesian Information Criterion (BIC). The suitability of the proposed method is illustrated through an application to actual electricity demand load of the Hellenic power system, proving the reliability and the effectiveness of the method and making clear its usefulness in the studies that concern electricity consumption and electricity prices forecasts.     

基于负荷细分的差异化用户基线负荷预测

电力用户基线负荷(CBL)预测精度会极大影响需求响应效果的评估。本文基于负荷细分,考虑多维用电行为及其影响因素,通过精细化用户用电行为特征,提出一种考虑用户用电模式差异化的基线负荷预测方法。首先采用Ward-模糊C均值(FCM)聚类法,并结合负荷特性指标,改善用户负荷曲线聚类分析的效果;然后,分析气象、时间等多维影响因素,建立考虑温湿度和气温累积效应等城市微气象因素及节假日社会行为因素的差异化用电行为分析模型,提出温度敏感型、节假日敏感型以及两者均不敏感的精细化用电模式;最后,提出不同用电模式的CBL预测方法,建立综合评估方法分析其预测准确度。算例结果表明,所提算法能进一步提高CBL预测精度。     

计及需求响应的Elman-NN短期负荷预测模型研究

通过频谱分析研究了需求响应负荷的基本特性,并以此为依据建立了计及需求响应的Elman神经网络(Elman-NN)预测模型。Elman-NN具有处理动态信息能力强、训练时间短、全局寻优性强的优点。通过实际算例,对比在Elman-NN模型中计及需求响应因素前后的预测性能,结果显示计及需求响应因素可显著提高Elman-NN模型预测精度。本文证实了在模型中计及需求响应因素的重要作用,为需求响应负荷的预测研究奠定了必要的理论基础。     

基于模糊信息粒化与多策略灵敏度的短期日负荷曲线预测

针对气象变化时负荷曲线预测精度低、预测模型不能完全适应气象变化的情况,提出了一种基于模糊信息粒化与多策略灵敏度的短期日负荷曲线预测方法。提出了完全气象因子序列的概念,建立气象粒化集;采用空间多元回归及滞后模型结合多策略灵敏度分析法,建立了针对复杂气象条件下的极值预测模型;基于改进的K-means聚类分析法查找并获取气象特征日,计算初步预测曲线,主动判断预测曲线畸变概率并进行优化修正,得到最佳预测日负荷曲线;利用动态数据流对模型参数进行更新,实现精细化预测。最后采用该方法对我国南方某地区全年负荷曲线进行预测,验证了模型在多种气象条件下的预测准确性,尤其适用于短期内气象存在复杂变化的情形。     

Long-Term Load Forecasting Using Grey System Theory

Supply and demand in power system planning and operation is required to be balanced. An operational reserve for protection against faults or accidental demands also is required. Therefore load forecasting is one of the most important fields and various load forecasting methods have been applied. In this paper the grey system theory which mats uncertain information is applied to the long-term load forecasting from three aspects: the point prediction; the interval prediction; and the topological forecasting. In the point prediction, the annual total demand is predicted, in the interval prediction, the annual peak demand is predicted, and in the topological forecasting, the date where a yearly maximum peak demand would occur is predicted. The grey dynamic model (abbreviated as GM model) is adopted as the predicted model. The GM model is a differential equation model which is different from most forecasting models. The GM model is quite powerful when combined with the preliminary transformation called the accumulated generating operation (AGO). This paper proposes a new method for the long-term load-forecasting problems involving uncertainty. The predicted results have been found to be very satisfactory. The grey system theory is a new tool which is very efficient for load forecasting.     

基于贝叶斯网络的电力客户停电敏感度预测

精准预测停电敏感的电力客户群体,能够有效感知客户用电需求,提升客户用电满意度,助力提高电力服务水平。文中提出基于贝叶斯网络构建电力客户停电敏感度预测模型,从95598客服平台、营销业务系统、用电信息采集系统获取分析数据,结合客户基本信息、用电信息、智能电能表计量信息以及用户用电交互行为,定义客户停电敏感度数据标签,对用户的停电投诉进行分析与预测。采用K折交叉验证法对停电敏感度预测模型进行实验验证。实验表明,基于贝叶斯网络构建的电力客户停电敏感度预测模型,在停电投诉分析应用中具备较高的精准度,验证了模型的有效性。     

计及需求响应的主动配电网短期负荷预测

随着分布式电源、电动汽车及储能等广义需求响应资源的接入,用户在电力市场各种激励影响下进行需求响应,将改变负荷特性并影响负荷预测。根据需求响应计划信号的可预知性及季节性基础负荷的独立性,利用小波分解等方法对主动配电网负荷在不同层面上进行了分解,形成季节性基础负荷和需求响应信号及各种气象因素作用的负荷部分,利用时间序列模型对季节性基础负荷进行预测,利用支持向量回归模型对需求响应信号及气象因素影响的负荷部分进行预测,形成组合预测模型,两部分预测负荷叠加得到总负荷。利用线性时变模型仿真的主动配电网负荷数据算例,进行了预测测试与分析,通过与其他方法相比较,证明了所提方法预测计及需求响应的主动配电网负荷的有效性及精确度。     

考虑电力市场机制下的客户侧储能参与需求响应控制策略

当下电力市场机制的改革为客户侧储能电站的建设提供了市场环境和政策支持。在此环境下研究了客户侧储能参与需求响应控制策略,确保实现电网与用户之间的友好交互。首先,阐述客户侧储能系统架构;然后,建立客户侧储能参与需求响应控制策略模型,包括基于信息系统需求响应的控制策略、基于标准模块化的需求响应预案的控制策略和计及不同场景下的储能调控的控制策略;最后,进行案例分析,进而进行成效分析。可以发现利用先进的通信控制技术对多个离散的、规模较小的客户侧储能系统进行集群控制,调控其参与电力需求响应,扩展了客户侧储能电站的应用场景,可显著提高客户侧储能系统的收益,降低企业用能成本。     

电力需求的不确定性分析

电力需求预测与分析是电力系统规划与运行的基础工作。常规的电力需求预测结果一般是确定性的,忽视了预测结果本身的概率特性。引入不确定性的分析思想,实现概率化的预测和分析,对于实际工作具有重要意义。在剖析传统的电力需求分析方法的基础上,引入序列运算理论,形成一个全新的不确定性分析思路,并以电量类指标的预测为突破口,提出了序列化分析方法,可以充分考虑分析结果的概率特性,为电力系统的规划与运行提供直接的决策依据。文中以实例验证了所提出的分析方法的有效性。     

Load forecasting using support vector Machines: a study on EUNITE competition 2001

Load forecasting is usually made by constructing models on relative information, such as climate and previous load demand data. In 2001, EUNITE network organized a competition aiming at mid-term load forecasting (predicting daily maximum load of the next 31 days). During the competition we proposed a support vector machine (SVM) model, which was the winning entry, to solve the problem. In this paper, we discuss in detail how SVM, a new learning technique, is successfully applied to load forecasting. In addition, motivated by the competition results and the approaches by other participants, more experiments and deeper analyses are conducted and presented here. Some important conclusions from the results are that temperature (or other types of climate information) might not be useful in such a mid-term load forecasting problem and that the introduction of time-series concept may improve the forecasting.     

面向有源配电网的公共信息模型扩展

IEC61970和IEC61968标准的公共信息模型(CIM)给出了电力系统设备、用户模型及其相互关系,有效促进了管理系统间数据交互,提高了调度自动化水平。但随着有源配电网中新设备的出现,现有CIM已不能对其进行完整描述。为此本文提出了一种CIM扩展方法,在EnterPrise Architect软件环境下,根据CIM扩展的导则,对有源配电网中的分布式发电、储能设备和电动汽车充电站等新兴设备的模型进行了完整的描述。此外,与交流微网相比,直流微网具有显著的优势而受到越来越多的关注,因此本文建立了直流微网CIM模型,以满足有源配电网管理系统的信息需求。     

基于电压聚类和关联卷积的配电网户变关系识别方法

准确的低压配电网户变关系是电力营销管理和台区线损治理的重要基础,传统的户变关系识别方法排查成本高、识别效果欠佳,无法适用于规模日趋庞大的低压配电网.在此背景下,提出了一种基于智能电表量测数据和用户档案信息的低压配电网户变关系识别方法.首先利用用户地理位置信息实现邻近用户的初步合并,再基于GMM聚类算法对电压时序数据进行聚类划分,用户划分结果作为下一步的迭代初值.然后基于能量供需平衡建立配变与用户的关联卷积识别模型实现低压配电台区户变关系的辨识.最后,在实际的低压配电系统中验证了该方法在提升户变关系识别效率和准确率等方面具有显著优势,具备一定的实践应用价值和工程指导作用.