面向自動需求響應的智能家居大數(shù)據(jù)處理技術研究
發(fā)布時間：2019-02-24 11:42
【摘要】：目前電網(wǎng)負荷峰谷差越來越大,如何削峰平谷提高用電效率,對智能電網(wǎng)提出新挑戰(zhàn)。需求響應通過引導用戶合理用電,來緩解高峰時的電網(wǎng)壓力,實現(xiàn)削峰平谷并提高電力資源利用率。在用戶側(cè)響應終端中,智能家居作為大功率負荷,對電呈柔性需求,其用電功率可以動態(tài)調(diào)整,合理調(diào)節(jié)其用電功率可以緩解電網(wǎng)壓力。但智能家居種類多、數(shù)量大、且分散,難以管理,需要構(gòu)建一個平臺將智能家居聚合起來,實現(xiàn)對智能家居的統(tǒng)一控制和管理。將不同廠商的不同設備聚合起來會面臨諸多問題,如通訊協(xié)議的不一致、高并發(fā)系統(tǒng)設計、TB級數(shù)據(jù)存儲和處理等。本論文以智能家居為研究對象,根據(jù)自動需求響應系統(tǒng)的業(yè)務需要,設計了需求響應能力、聚合商認購能力、節(jié)電效果評估和用戶行為分析等模型,以及這些模型的分布式解決方案。這些模型需要分析和挖掘TB級的歷史數(shù)據(jù),采用傳統(tǒng)方式無法滿足業(yè)務對空間和時間上的要求,因此本論文提出并實現(xiàn)智能家居大數(shù)據(jù)平臺。該平臺為自動需求響應系統(tǒng)提供大數(shù)據(jù)技術支持,由需求響應終端系統(tǒng)、業(yè)務數(shù)據(jù)處理系統(tǒng)和可視化系統(tǒng)三個子系統(tǒng)組成。需求響應終端系統(tǒng)采用MINA開發(fā),支持高并發(fā)網(wǎng)絡通訊,并將采集到的數(shù)據(jù)交給Kafka緩存起來,解決數(shù)據(jù)接收和處理速度的不一致。業(yè)務數(shù)據(jù)處理系統(tǒng)基于Lambda架構(gòu)理念實現(xiàn),整合了Hadoop和Spark;利用Spark從Kafka獲取實時數(shù)據(jù)流,做實時預測分析;使用MapReduce實現(xiàn)并行化KNN算法,根據(jù)用戶的用電記錄,對用戶分類;使用MapReduce實現(xiàn)認購能力模型和節(jié)電效果評估模型的分布式算法。可視化系統(tǒng)調(diào)用數(shù)據(jù)處理后的結(jié)果,為數(shù)據(jù)的查詢和展示提供可視化界面。
[Abstract]:At present, the load peak and valley difference of power network is more and more big. How to cut peak and level valley to improve power efficiency is a new challenge to smart grid. The demand response can relieve the pressure of the power grid at the peak by guiding the user to use electricity reasonably, realize the peak cutting and level valley, and improve the utilization ratio of power resources. In the user-side response terminal, smart home is regarded as a high-power load, and it has flexible demand for electricity, its power can be dynamically adjusted, and its power can be adjusted reasonably to relieve the pressure of power grid. However, there are many kinds of smart home, large quantity, scattered and difficult to manage, so it is necessary to build a platform to aggregate smart home to realize the unified control and management of smart home. The aggregation of different devices from different manufacturers will face many problems, such as inconsistent communication protocols, high concurrent system design, TB level data storage and processing, and so on. In this paper, the smart home as the research object, according to the business needs of the automatic demand response system, design the demand response ability, aggregator subscription ability, power saving effect evaluation and user behavior analysis model. And distributed solutions for these models. These models need to analyze and mine the historical data of TB level, and the traditional way can not meet the requirements of space and time. Therefore, this paper proposes and implements big data platform of smart home. The platform provides big data technical support for the automatic demand response system, which consists of three subsystems: the demand response terminal system, the business data processing system and the visual system. The demand response terminal system is developed with MINA to support high concurrency network communication, and the collected data is cached by Kafka to solve the inconsistency between data receiving and processing speed. The business data processing system is realized based on Lambda architecture, which integrates Hadoop and Spark; to obtain real-time data stream from Kafka and use Spark to do real-time prediction and analysis, uses MapReduce to realize parallel KNN algorithm, classifies users according to user's power consumption record. MapReduce is used to realize the distributed algorithm of subscription ability model and power saving effect evaluation model. The visualization system calls the results of data processing and provides a visual interface for the query and display of data.
【學位授予單位】：華北電力大學(北京)
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：TP311.13;TU855
[Abstract]:At present, the load peak and valley difference of power network is more and more big. How to cut peak and level valley to improve power efficiency is a new challenge to smart grid. The demand response can relieve the pressure of the power grid at the peak by guiding the user to use electricity reasonably, realize the peak cutting and level valley, and improve the utilization ratio of power resources. In the user-side response terminal, smart home is regarded as a high-power load, and it has flexible demand for electricity, its power can be dynamically adjusted, and its power can be adjusted reasonably to relieve the pressure of power grid. However, there are many kinds of smart home, large quantity, scattered and difficult to manage, so it is necessary to build a platform to aggregate smart home to realize the unified control and management of smart home. The aggregation of different devices from different manufacturers will face many problems, such as inconsistent communication protocols, high concurrent system design, TB level data storage and processing, and so on. In this paper, the smart home as the research object, according to the business needs of the automatic demand response system, design the demand response ability, aggregator subscription ability, power saving effect evaluation and user behavior analysis model. And distributed solutions for these models. These models need to analyze and mine the historical data of TB level, and the traditional way can not meet the requirements of space and time. Therefore, this paper proposes and implements big data platform of smart home. The platform provides big data technical support for the automatic demand response system, which consists of three subsystems: the demand response terminal system, the business data processing system and the visual system. The demand response terminal system is developed with MINA to support high concurrency network communication, and the collected data is cached by Kafka to solve the inconsistency between data receiving and processing speed. The business data processing system is realized based on Lambda architecture, which integrates Hadoop and Spark; to obtain real-time data stream from Kafka and use Spark to do real-time prediction and analysis, uses MapReduce to realize parallel KNN algorithm, classifies users according to user's power consumption record. MapReduce is used to realize the distributed algorithm of subscription ability model and power saving effect evaluation model. The visualization system calls the results of data processing and provides a visual interface for the query and display of data.
【學位授予單位】：華北電力大學(北京)
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：TP311.13;TU855
【參考文獻】
相關期刊論文 前10條
1 王德文;楊力平;;智能電網(wǎng)大數(shù)據(jù)流式處理方法與狀態(tài)監(jiān)測異常檢測[J];電力系統(tǒng)自動化;2016年14期
2 宋夢;高賜威;蘇衛(wèi)華;;面向需求響應應用的空調(diào)負荷建模及控制[J];電力系統(tǒng)自動化;2016年14期
3 楊濱;田娜;劉建宇;劉金華;曹筱歐;宋杰;;基于能源管理的電力需求響應支持系統(tǒng)架構(gòu)設計與應用[J];電子技術與軟件工程;2016年11期
4 Jin WANG;Xinrong MA;;Some Notices on Mina Matrix and Allied Determinant Identities[J];Journal of Mathematical Research with Applications;2016年03期
5 孫建偉;唐升衛(wèi);劉菲;王丹;戚野白;王冉;;面向需求響應控制的家用電熱水器建模和控制策略評估[J];電力系統(tǒng)及其自動化學報;2016年04期
6 張晶;孫萬s，
本文編號：2429518
本文鏈接：http://sikaile.net/jianzhugongchenglunwen/2429518.html
最近更新
教材專著
