面向物联网应用的环境变化事件感知方法研究

发布时间：2018-08-26 15:33

【摘要】：物联网作为继计算机互联网之后第三次世界信息产业浪潮,通过各种信息传感设备与互联网联系在一起,按约定的协议,让所有物品都能够被感知和控制,实现智能化的识别、定位、跟踪、监控和管理。无线传感器网络是物联网的核心之一,主要负责信息的采集与获取,帮助物联网系统提高对周边环境的感知能力。目前很多研究从感知当前环境状态的角度把传感器的监测数据转换成环境上下文作为提供给物联网应用的环境信息。但是,物联网应用系统的运行不单取决于环境状态,还取决于环境状态的变化即环境变化事件。因此,如何对传感器的原始监测数据处理以形成有效可靠的环境变化事件,实现高效的信息感知,是物联网要解决的重要问题。本文针对上述问题,提出一种面向物联网应用的环境变化事件感知方法。该方法主要是基于信息融合的思想,根据无线传感器网络的监测数据对预定义的环境变化事件进行感知。有些环境变化事件仅涉及到某些特定的环境指标,而实际上很多环境变化事件更加复杂,它们的发生还与其他事件有关,即事件之间可能具有时序关系或者因果关系,分析环境变化事件之间关联性有助于提高感知效率。因此,本文根据环境变化事件的特点将它们分为三类,并按照环境变化事件的类别有针对性地采用不同的感知方法。首先,导致单环境指标变化的环境变化事件,这类事件采用基于规则判别的感知方法,根据实时的环境状态数据序列与规则曲线拟合程度判别环境变化事件。其次,导致多环境指标变化的环境变化事件,这类事件采用BP神经网络的感知方法,利用历史环境变化事件的数据集训练BP神经网络,获得相应的基于BP网络环境变化事件感知模型,再根据实时环境状态数据序列感知环境变化事件。再次,导致多环境指标变化的复合环境变化事件,这类事件采用动态贝叶斯网络的感知方法,根据专家领域经验构建动态贝叶斯网络结构,再利用历史环境变化事件的数据集学习动态贝叶斯网络参数,进而用于实时环境变化事件感知。最后,分别设计并实现了这三种环境变化事件感知方法,从感知结果的准确性及数据量两个角度对方法进行分析和评价,验证本文提出方法的有效性。
[Abstract]:The Internet of things, as the third wave of the world information industry after the computer Internet, connects with the Internet through a variety of information sensing devices. According to the agreed protocol, all items can be sensed and controlled to realize intelligent identification. Position, track, monitor and manage. Wireless sensor network is one of the core of the Internet of things, mainly responsible for information collection and acquisition, to help the Internet of things system to improve the perception of the surrounding environment. At present, many researches have transformed sensor monitoring data into environmental context from the perspective of sensing the current state of the environment as environmental information for Internet of things applications. However, the operation of the Internet of things application system depends not only on the state of the environment, but also on the change of the state of the environment, that is, the event of environmental change. Therefore, how to process the original monitoring data of the sensor to form an effective and reliable environment change event and to realize efficient information perception is an important problem to be solved in the Internet of things. In order to solve the above problems, this paper presents an event sensing method for the application of the Internet of things (IOT). This method is mainly based on the idea of information fusion, according to the monitoring data of wireless sensor networks to perceive the predefined environmental change events. Some environmental change events involve only certain environmental indicators, but in fact many environmental change events are more complex, and their occurrence is also related to other events, that is, there may be a temporal or causal relationship between events. Analyzing the correlation between environmental change events is helpful to improve perception efficiency. Therefore, according to the characteristics of environmental change events, this paper divides them into three categories, and adopts different perceptual methods according to the types of environmental change events. First of all, the event of environmental change which leads to the change of single environmental index, which adopts the perceptual method based on rule discriminant, discriminates the event of environmental change according to the fitting degree of the real-time environmental state data sequence and the regular curve. Secondly, the event of environmental change, which leads to the change of multiple environmental indicators, uses the BP neural network to perceive the event, and trains the BP neural network by using the data set of the historical environment change event. The corresponding event-aware model based on BP network environment change is obtained, and then the environment change event is sensed according to the real-time environment state data sequence. Thirdly, the complex environmental change events, which lead to the change of multiple environmental indicators, adopt the perceptual method of dynamic Bayesian network and construct the dynamic Bayesian network structure according to the experience of expert field. Then the dynamic Bayesian network parameters are learned by using the data set of historical environment change events, and then used for real-time environment change event awareness. Finally, the three environmental change event sensing methods are designed and implemented, which are analyzed and evaluated from the perspective of the accuracy of perception results and the amount of data to verify the effectiveness of the proposed method.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP391.44;TN929.5

【相似文献】