【摘要】：隨著物聯(lián)網(wǎng)的發(fā)展,接入的感知設(shè)備無論在種類還是數(shù)量上,都在日益增長,導(dǎo)致物聯(lián)網(wǎng)環(huán)境中的數(shù)據(jù)量日益激增。如何從海量的數(shù)據(jù)中高效、智能地發(fā)現(xiàn)我們感興趣的數(shù)據(jù)成為了我們的關(guān)注點。規(guī)則引擎是由基于規(guī)則的專家系統(tǒng)逐步演化而來的,可以通過源源不斷的接收數(shù)據(jù),將其作為事實與事先設(shè)定好的規(guī)則做匹配,從而可以在簡單的數(shù)據(jù)中發(fā)現(xiàn)復(fù)雜事件。然而目前的主流規(guī)則引擎都只能在單機的環(huán)境下運行,缺少能夠分布、并行的實現(xiàn)機制,這樣使得當(dāng)系統(tǒng)的數(shù)據(jù)量較大時,單個計算機的處理能力會成為系統(tǒng)的瓶頸。本文通過對國內(nèi)外工業(yè)界和學(xué)術(shù)界的優(yōu)秀規(guī)則引擎的研究,提出了一種適用于物聯(lián)網(wǎng)EDSOA架構(gòu)環(huán)境的分布式規(guī)則引擎的實現(xiàn)方法。本文首先從系統(tǒng)的總體需求開始分析,對傳統(tǒng)的分布式框架做出改進,并在其基礎(chǔ)上提出了分布式規(guī)則引擎的總體架構(gòu),將通過分解規(guī)則的方式,拆分規(guī)則集合,并將子規(guī)則集部署于不同的工作節(jié)點上,每個工作節(jié)點都將作為獨立的規(guī)則引擎,進行規(guī)則匹配,再由主節(jié)點歸并各個工作節(jié)點產(chǎn)生的中間結(jié)果,以達到并行的匹配規(guī)則的目的。由于需要對單個規(guī)則進行分解操作,本文對規(guī)引擎中的規(guī)則進行了研究。通過對知識的研究,對規(guī)則做出了定義并根據(jù)特性對規(guī)則進行了分類,不同的規(guī)則將會使用不同的方法進行分解。同時發(fā)現(xiàn)不同的規(guī)則集合分解策略會影響到系統(tǒng)的匹配效率,本文又對如何“合理”地分解規(guī)則來提高系統(tǒng)的效率做出了研究。本文使用了 Apriori算法對規(guī)則之間的關(guān)聯(lián)做了分析。最后本文給出了原型系統(tǒng)的實現(xiàn)類圖,以及對系統(tǒng)進行了性能測試來驗證系統(tǒng),測試結(jié)果基本達到了預(yù)期效果
[Abstract]:With the development of the Internet of things (IoT), the number of sensor devices is increasing day by day, which leads to the increasing amount of data in the Internet of things (IoT) environment. How to find the data we are interested in efficiently and intelligently from the massive data has become our focus. The rule engine is evolved from the rule-based expert system. It can be used to match the facts with the rules set in advance by receiving the data continuously, so that complex events can be found in the simple data. However, the current mainstream rule engines can only run in a single machine environment, and lack of distributed and parallel implementation mechanism, which makes the processing ability of a single computer become the bottleneck of the system when the data volume of the system is large. Based on the research of the excellent rule engine in industry and academia at home and abroad, this paper presents a method of implementing the distributed rule engine suitable for the EDSOA architecture of the Internet of things. Based on the analysis of the general requirements of the system, this paper improves the traditional distributed framework, and proposes the general architecture of the distributed rule engine, which will split the rule set by decomposing the rules. The subrule set is deployed to different working nodes, each working node will act as an independent rule engine to match the rules, and then merge the intermediate results generated by each working node by the primary node. In order to achieve the purpose of parallel matching rules. Because of the need to decompose a single rule, the rules in the gage engine are studied in this paper. Through the study of knowledge, the rules are defined and classified according to their characteristics. Different rules will be decomposed in different ways. At the same time, it is found that different decomposition strategies of rule set will affect the matching efficiency of the system. This paper also studies how to decompose the rules reasonably to improve the efficiency of the system. In this paper, Apriori algorithm is used to analyze the association between rules. Finally, the implementation class diagram of the prototype system is given, and the system performance is tested to verify the system. The test results basically reach the expected results.
【學(xué)位授予單位】：北京郵電大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2016
【分類號】：TP391.44;TN929.5

【參考文獻】

相關(guān)期刊論文 前4條

1 孫其博;劉杰;黎，

本文編號：2396252