【作者】 张华；

【导师】 刘冲； 徐征；

【作者基本信息】 大连理工大学 ， 机械电子工程， 2012， 硕士

【摘要】 目前,钢轨变形监测研究渐渐受到学术界和工程界的重视。项目组在分析长线式钢轨监测需求及特点的基础上,采用分布式传感器网络实现钢轨应力变化的监测。分布式传感器网络,能够集成大量不同类型智能传感器,采用自组织网络拓扑结构分布,便于协同完成全局任务的监测。本文主要研究传感网系统的远程监控平台设计及相关监控策略,实现对现场系统设备的远程控制与管理、数据监控与存储。为实现钢轨监测系统的分布式可靠监控,本文针对节点差异监控、节点协同监控、信息传输时效控制、分布式数据入库存储和网络架构与链路管理问题开展研究。综合分析分布式钢轨监测需求及问题,本文利用接入服务器管理网关(Sink节点)和监控端的网络连接链路,并实现两者信息透传。依据多节点同步采集及监测需求,远程监控端采取分组策略,依组实现多节点的独立或协同监控；采用超时故障处理和定时查询网络状态等策略引导控制流程的顺利执行,提高信息传输可靠性；建立高效内存管理和动态调配机制,保证分布式数据快速存储入库,并借助逻辑标识使得数据库中数据拥有物理意义且不因设备更换而丧失,提高分布式数据的可靠性及稳定性。基于上述策略开发实现接入服务器和分布式监控软件,并在铁路现场长期测试。监控策略保证现场节点依据不同监控需求实现并发监控,在不排除现场设备不稳定、现场环境干扰和网络异常前提下,基本保障数据完整性达到95%左右,整个监测过程基本实现自动化监控,避免人工修复。更多还原

【Abstract】 At present, the research on rail deformation monitoring gradually has got more and more attention from academics and engineering. Based on analysis of requirements and characters of long term rail monitoring, Distributed Sensor Network (DSN) is chosen to realize global monitoring for variation of rail stress. DSN has the abilities of integrating multiply series intelligent sensors, distribution by self organization network topology structure, monitoring of global task by co-operation.In this paper, the design of remote monitoring platform and relative supervision strategies in sensor network system are introduced in detail, which refer to remote control and management of system devices, data monitoring and storage. In order to distributed reliable supervision of rail monitoring system, the research on several problems is conducted, which include difference supervision for nodes, cooperative supervision for nodes, aging control of information transmission, database storage of distributed data, network architecture and network link management.Synthetically analyzing distributed rail monitoring requirements and problems, access server is adopted to manage the network links from gateway (Sink Node) and supervision terminal and ensure information transmission between them. On the basis of synchronous collection and control requirements, supervision terminal proposes grouping strategy to achieve independent or synchronous supervision of multiply nodes in the form of group; The strategies of timeout failure disposal and querying network state regularly are set up to make the control flow work well, which obviously improves the reliability of information transmission; High-efficiency memory management and dynamic allocation mechanism ensure fast storage for distributed data, in particular, physical significance is given to each data record from database by logical identifier, and remains in place under the situation of device replacement, which remarkably advances reliability and stability of distributed data.Finally access server and DSN supervision software based on above-mentioned strategies are finished and applied to railway field experiment. The strategies ensure concurrent control for nodes by different supervision requirement, and under the foundation of unstable equipment, ambient interference and network anomaly, data integrality reaches approximately95%, the whole process automatically is finished without any manual restoration.更多还原