护理监测数据的Hadoop集群动态可视化模型仿真  

Dynamic Visualization Model Simulation of Hadoop Cluster for Nursing Monitoring Data

作　　者：常振云[1] 胡碧金 李小红[1] 赵琳[1] CHANG Zhen-yun;HU Bi-jin;LI Xiao-hong;ZHAO-lin(Tianjin Tianshi College,Tianjin 301700,China;Tianjin Polytechnic University,Tianjin 300387,China)

机构地区：[1]天津天狮学院,天津301700 [2]天津工业大学,天津300387

出　　处：《计算机仿真》2020年第8期153-156,251,共5页Computer Simulation

基　　金：天津市大学生创新训练计划项目“基于压力传感矩阵的人体部位识别系统”(项目编号:201910859005)。

摘　　要：实现病人护理监测数据的动态可视化,需要完成大量含噪数据的准确实时分析和处理,为此,提出了基于分布式集群的监测数据可视化模型。在采集融合层,护理监测系统通过传感器采集病人的各项生理数据,为降低噪声影响,同时考虑到目标信号幅值波动和幅值较小情况,设计了与噪声均方差相关的自适应门限滤波。分析层收集采集融合后的数据,针对特征提取的非线性,采用小波分解得到特征参数和分布图谱之间的关联函数,根据时间序列分布对特征提取模型进行重构,并将算法部署到分布式集群上,利用将护理监测数据特征提取拆分成多个子任务,同时将各子任务部署到不同的服务器上执行,形成分布式并行处理,最终在应用层得到病人护理实时准确的监测结果。通过仿真,验证所设计模型能够自适应过滤数据中的噪声信号,快速准确的完成护理监测数据的分析,为动态可视化提供可靠的后端处理支持。To realize the dynamic visualization of patient nursing monitoring data,it is necessary to complete the accurate real-time analysis and processing of a large number of noisy data.Therefore,a visualization model of monitoring data based on Hadoop distributed cluster was proposed.In the collection and fusion layer,the nursing monitoring system collects all physiological data of patients through sensors.In order to reduce the influence of noise,and considering the fluctuation and small amplitude of target signal,an adaptive threshold filter related to noise mean square deviation was designed.The analysis layer collects the data after fusion.In view of the nonlinearity of feature extraction,the correlation function between feature parameters and distribution map was obtained by wavelet decomposition,and the feature extraction model was reconstructed according to the distribution of time series.The algorithm was deployed to the Hadoop distributed cluster,and the feature extraction of nursing monitoring data was split into multiple sub tasks by MapReduce,and each sub task was deployed to different servers for execution,forming a distributed parallel processing.Finally,the real-time and accurate monitoring results of patient care were obtained in the application layer.Through simulation experiments,it was verified that the designed model can filter the noise signal in the data adaptively,complete the analysis of nursing monitoring data quickly and accurately,and provide reliable back-end processing support for dynamic visualization.

关 键 词：护理监测数据 自适应门限 特征提取 分布式集群 动态可视化 

分 类 号：TP392[自动化与计算机技术—计算机应用技术]