1、配电网大数据技术分析与典型应用案例（英文） 王璟 杨德昌 李锰 范征 Mark Chew 国网河南省电力公司经济技术研究院 中国农业大学信息与电气工程学院 北京艾能万德智能技术有限公司 美国 AutoGrid 公司 摘 要： 以美国 Auto Grid 公司及其 2 个产品(能量数据平台 EDP 和需求侧优化管理系统DROMS)为例,分析了大数据技术在配电网数据分析中的应用。首先,简要介绍了主动配电网中大数据的典型特点;然后列举了 EDP 和 DROMS 的关键技术和主要功能,并介绍了 Auto Grid 公司的 4 个典型应用案例;最后,结合配电网的设备水平和运行水平,总结归纳了配电网大数据
2、应用的关键技术,提出配电网大数据应用的实施路线图。关键词： 大数据; 主动配电网; 能量数据平台; 需求侧优化管理系统; 作者简介：王璟(1973),女,高级工程师,主要研究方向为智能电网、电力系统规划、分析与控制;作者简介：杨德昌(1983),男,博士,主要研究方向为主动配电网的状态估计和微型能源系统的容量配置;E-mail:;作者简介：李锰(1984),男,硕士,工程师,主要研究方向为主动配电网的规划运行与控制;作者简介：范征(1984),男,工程师,主要研究方向为新能源接入对配电网的影响分析;作者简介：Mark Chew(1978),男,工程师,主要研究方向为大数据技术在配电网中的应用。
3、收稿日期：2015-07-15基金：Project Supported by National Natural Science Foundation of China(51407186)Analysis of Big Data Technology in Power Distribution System and Typical ApplicationsWANG Jing YANG Dechang LI Meng FAN Zheng Mark Chew State Grid Henan Economic Research Institute; College of Information an
4、d Electrical Engineering, China Agricultural University; Beijing Energywende Intelligent Technologies Co., Ltd.; Auto Grid Co., Ltd.; Abstract： Auto Grid Company and its two products, Energy Data Platform(EDP) and Demand Response Optimization and Management System(DROMS), are taken as examples. Firs
5、tly, characteristics and features of active distribution system data are summarized. Then, key technologies and main functions of EDP and DROMS are described in detail. Four study cases are employed to illustrate their applications. Finally, prospective application of big data technology is analyzed
6、 based on development of active distribution system. Some useful advices and suggestions are also proposed to ensure safe, reliable and economic operation of distribution system.Keyword： big data; active distribution system; energy data platform; demand response optimization and management system; R
7、eceived： 2015-07-150 ForewordWith development of smart meters, utility companies are dealing with more data coming from more connected nodes than any industry, especially in the distribution system. Recently, many published works have explored the big data application in distribution systems. The ma
8、in achievements can be classed into two aspects: 1)Summaries of big data theory. Paper 1 analyzed the basic features of the big data in distribution system. The typical applications in load forecasting, operating condition evaluation and power quality monitoring Velocity contains two meanings: one i
9、s the disposing speed, most of the measured data in ADS is near real time; the other is the different time scales and sampling intervals. Value stands for that most of the data contains low and repeating information.From the aspect of variety, the big data includes the device data, operating data an
10、d management data. Different data should be analyzed by different algorithm14-15. The illustration is shown in Fig.3.图 3 主动配电网中大数据的多样性 Fig.3Variety of the big data in ADS 下载原图1.3 Analysis of Data MiningThe core of the application of big data is to extract the patterns from existing disparate sets to
11、:1) predict outcomes and optimize solution; 2)execute in real-time to improve operational decision-making.A variety of algorithms is commonly used to mine the data, including the clustering, K-mean, neural network, correlation analysis16-18.Comparedwith the traditional data systems, the data in ADS
12、contains new characteristics. The explanations are summarized as follows: 1)Wide sources, complex relation, fine-grained, variety of crystal structure and fast production can be seen as the first feature of big data in ADS1-2,19. 2)Massive size, rich information, hard processing and so on, are the s
13、econd main feature of big data in ADS. From above analysis, it is necessary to seek for applied technology of big data which is compatible with the characters of distributionsystem 20-21.Although many theoretical analyses of big data have been done, the real application of big data to serve the util
14、ity and customer are just at the beginning. There is a lack of information on how to implement the real application in the society. In this paper, one typical company named Auto Grid and their products are proposed in the next chapter22.2 Introduction of Products of Auto GridIn the last decade, part
15、icipants in the electricity supply chain have been processing thousands of times more data than before, mostly from smart meters,distributed generation and other building and grid sensing technologies. In this context, Auto Grid was founded in 2011, with the vision of turning the big data into knowl
16、edge and effective action.Auto Grids productsare to organize the exponentially increasingamount of energy data produced from a networked andautomated grid and make it actionable for electricitygenerators and providers, grid operators, and electricity users.It has built a highly scalable software sys
17、tem that is capable of processing the petabytes of data being generated,making real-time predictions and running complex optimization algorithms across millions of variables.The customers of Auto Grid can be grouped into four parts22:1)Large Utilities. By mining operational trends and consumption pa
18、tterns, Auto Grids Energy Data Platform can forecast demand days, minutes or even seconds in advance. With our tools, the emergency capacity can be reduced by the big data analysis.2)MUNIS, CO-OPS.Auto Grid delivers services over the cloud, reducing the cost of demand management systems by 90 percen
19、t. Auto Grid is also100-percent hardware neutral: high capital and operational costs are no longer a barrier.3)Service Providers. With our software, ESCOs,demand response providers, electricity retailers and others can employ demand management tools.Auto Grid increases power “yields” by 30 percent o
20、r more.4)Facilities.High levels of security, rock-solid stability, compatibility with standards and a rapid ROI mean that manufacturers and real estate managers can follow through with plans to adopt energy efficiency strategies.2.1 Energy Data Platform (EDP)Energy Data Platform (EDP) can analyze pe
21、tabytes of structured and unstructured data,streaming from smart meters, Supervisory Control and Data Acquisition (SCADA) systems, and home,building or industrial energy management systems,along with electrical models of grid assets to create a comprehensive, multidimensional, and dynamic portrait o
22、f individual customers, groups of customers,or the entire grid. The EDP supports a multitude of applications, as shown in Fig.422.图 4EDP 的主要功能模块 Fig.4Main functions of EDP 下载原图By examining the relationship between tens of t h o u s a n d s o f v a r i a b l e s , E D P c a n discoverconsumptionpatte
23、rns, understand customer b e h a v i o r , e s t a b l i s h correlations between pricing and consumption, enable new and personalized services and pricing plans,monitor the health of the overall system in near real-time, and control grid assets to improve reliability and minimize losses across the
24、entire electricity supply chain. Additionally, Auto Grids system architecture performs all of these activities at scale-even in situations with millions of meters. By providing timely, actionable intelligence from the exponentially expanding wave of data, EDPallows creation of new applications that
25、help with short-term tactical operations, as well as long term strategic planning.The diagram of EDP is shown in Fig.5.图 5EDP 系统架构 Fig.5Basic diagram of EDP 下载原图2.2 Demand Response Optimization and Management System (DROMS)Auto Grids DROMS is the worlds most advanced demand management system. Harnes
26、sing big data analytics and open standards, DROMS is a cost-effective, cloud-based service for implementing and managing a wide range of power management programs such as direct load control, critical peak pricing, peak-time rebates, grid balancing, spinning reserve and demand bidding22.DROMS can be
27、 accessed through a highly secure public or private cloud, depending on the needs of the customer. Because it is based on open standards, such as Open ADR and Zig Bee, DROMS is compatible with the broad ecosystem of software, networks and hardware already deployed. By using open standards,Auto Grid
28、also mean that we can leverage the advances in security and reliability being achieved through industry-wide initiatives.The basic functions of DROMS are shown inFig.6.图 6DROMS 基本功能 Fig.6 Basic functions of DROMS 下载原图The key features of basic functions are explained as follows:1 ) Enrollment and pro
29、gram management. A complete system of record for all demand management activities.1Easy to use interface;2Compatible with all known demand response programs;3Supports millions of customer participants;4 Manage multiple programs and dual participation;5True operational visibility before, during and a
30、fter events.2 ) Real-time resource forecasting. DROMS provides bottoms-up forecasting allowing aggregation and deeper insight into customer groups.1Better, more accurate representation of opt- outs; Load shed forecast is dropped by the forecasted load shed of the particular customer, rather than an
31、average;2Allows forecasting of load shed under different price signals enabling optimization of a portfolio for hedging, revenue optimization, or reliability;3Best utilizes localized weather data leading to better regional forecasts;4Provide insights based on correlation of data points.3)Real-time R
32、esource Optimization. Optimize demand resources across programs, resources and geographiesthroughcontrolandsituational awareness.1Unified views of all of your programs;2Visualize program constraints and how they interact to avoid double counting resources;3Transition demand response from finite disc
33、rete events/year to a real-time resource;4Optimizes monetization of demand flexibility across programs capacity, energy, reserves,regulation.4)Customer Notification Engine. Fully-featured customer communications platform for voice calls,Emails or Short Messages (SMS)text.1Communicate via voice calls
34、, emails or SMS text;2Highly scalable- support millions of participants;3Multi-language support-communicate in any language;4Allow customers to put-out;5Advanced analytics and reporting to track customer receipts.5)Automated Demand Response. The largest ecosystem of direct, two-way, machine to machi
35、ne communications tothe control and instrumentation(C2Compliant with Open ADR 1.0 and 2.0;3Compliant with Smart Energy Profile 1.x and2.0;4Seamless integration with AMI networks from Itron and Silver Spring Networks;5 Interoperable with dozens of hardware platforms across C6Can integrate with any pr
36、oprietary metering platform.6 ) Post event analytics. Robust real-time analytics to track events, provide feedback to customers, create settlement data and fulfill regulatory Measurement and Verification (M2Create billing determinants for settlement;3Run M4Determine load shed potentials by participa
37、nt class;5 Auditlegacyprogramstoidentify non-performing devices.2.3 Study Cases2.3.1Florida Power:Predictive Asset Maintenance.In this project, the big data is utilized to improve the reliability of Florida Power 2)Substation data,Fault-current indicators, Capacitor bank reports; 3)10other distribut
38、ion automation devices on the feeder.Auto Grids software examines each of data streams as a single time series, allowing FPL to take advantage of patterns that span multiple data sources.2.3.2 Oklahoma Gas 3)One customer can save 300$ in every year.2.3.3 E.ON: Revenue Assurance.E.ON Romania has high
39、 theft rates, every month of undetected theft costs 15% of revenue. Moreover,the big data are collected from many different data sources and systems. The goal of the revenue assurance product is to accurately detect cases of electricity theft activity, allowing E.ON to reduce grid losses. In this situation, Auto Grids system employs machine learning to improve detection accuracy,resulting in a 300% improvement.2.3.4Eneco:Flexibile Asset Management.
