基于模糊小波聚类混合的多目标覆盖质量评估方法  

Multi-objective evaluation method of coverage quality based on hybrid algorithm of fuzzy wavelet and clustering

作　　者：金杉 金志刚 JIN Shan;JIN Zhigang(Electrical, Automation and Information Engineering College, Tianjin University, Tianjin 300072, China;Fire and Rescue Crops, Bureau of Emergency Management of Tianjin, Tianjin 300090, China)

机构地区：[1]天津大学电气自动化与信息工程学院,天津300072 [2]天津市应急管理局消防救援总队,天津300090

出　　处：《系统工程与电子技术》2019年第6期1316-1323,共8页Systems Engineering and Electronics

基　　金：国家自然科学基金(61571318);海南省重点研发计划项目(ZDYF2018006);博士后面上基金(2017M621076);天津大学自主创新基金资助课题

摘　　要：面向双层无线传感器网络覆盖质量评估,设计出基于模糊小波聚类混合的多目标覆盖质量评估方法。建立网络单元概念和双层网络模型,在各汇聚节点开展各子目标预处理。集中建立二次预警机制:设计基于模糊小波神经网络的分析融合子系统,实现一次预警,选出显著低效覆盖单元;构建决策输出子系统,设计基于k均值聚类算法的多等级网络单元评价体系,实现二次预警,并呈现全部低效覆盖单元。实验表明,该方法从覆盖面积、能耗均衡、传输便利等方面综合评估,能够精确判定低效覆盖单元,有助于及时重部署,维护网络健康运行。For resolving the problem about coverage quality evaluation of wireless sensor networks (WSNs) with 2 layers, a multi-objective evaluation method of coverage quality based on hybrid algorithm of fuzzy wavelet and clustering is proposed. Firstly, the conception of network unit is proposed, and a model of wireless sensor networks with 2 layers is established. Secondly, every sub-objective is pre-processed at the gathing endnodes. Thirdly, twice-alarmed mechanism is introduced, which includes two parts. One part is analysis and fusion sub-system based on fuzzy wavelet neural network. It realizes the first alarm, which checks out the ultra-inefficient coverage units. The other part is decision and output sub-system. In this part, multi-degree evaluation architecture for network units is proposed, which is based on k -means clustering algorithm. Meantime, it achieves the second alarm and shows all of inefficient coverage units. The experimental results validate the proposed method from comprehensive evaluation about coverage, consumption, and transmission. This method can determine inefficient coverage units accurately. Also, it conduces to redeployment timely for healthy running WSNs.

关 键 词：覆盖质量评估 多目标 小波神经网络 模糊推理 K均值聚类 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]