矿用对等无线监测数据集成系统运行时自验证机理

With mining progress, Data integration system of mine wireless monitoring network often changes in structure, amount of data, etc., which maybe impair the system. It is difficult to maintain the system off-line because monitoring data of mine environment must be collected at any one time. In order to solve this problem, an effective online approach to check the system automatically after it changes is proposed in this project. Its main contents include rules to discover and abstract security parameters combining wireless monitoring data integration system with mining engineering, security properties to be verified and its formal description, method of building P2P trust clustering model of wireless monitoring data integration system for collecting detection data, self-building method of probabilistic temporal automata to describe real-time states transition, runtime self-verification method of security property checking on probabilistic temporal automata. According to our prior researches, to identify security and dynamic change of wireless monitoring data integration system, we would redefine security properties by combining with mine safety regulation in the system, and collect detection data based on P2P trust clustering model to update probability values of state transition. In order to check correctness of real-time change in the system at running time, we propose a novel runtime approach of self-verification method to check security property on probabilistic temporal automata. The project would make a progress on the security theory of wireless sensors network in mine and enrich runtime verification theory. It also improves reliability of wireless data integration system in mine and reduces maintenance costs. It is a novel cross research among mine safety, runtime verification and wireless sensor monitoring network, which has a wide application prospect in mine.

随着采矿工程进展，矿用无线监测数据集成系统会在结构、数据量等方面发生变化，由此可能导致系统原有功能受损。针对该问题，本项目研究一种能够根据系统变化且不间断进行自动在线检验的方法。内容包括：矿山环境下，融合矿山安全的系统安全性表征参数的发现与抽取规则，待验证安全性质及其形式化描述，收集检测数据的对等信任群落模型建立方法，描述系统实时状态迁移的概率时间模型自构建方法，概率时间模型上的系统运行时自验证方法。为描绘矿用无线监测数据集成系统安全，拟融合矿山安全规范对安全性质重定义，使用对等信任群落模型实时更新系统状态迁移概率，表征系统动态安全状态；为验证系统实时变化的正确性，拟提出在概率时间模型上检验系统安全性质的运行时自验证方法。本项目是矿山安全、运行时验证与无线传感器网络有机结合的新颖交叉研究，研究成果将深化和丰富现有矿用传感器网络安全和运行时验证的理论，提升矿用无线网的可靠性，降低维护成本。

本项目首先在前期研究的基础上，针对矿用传感器数据与web GIS的结合问题，提出一组对矿山物联数据访问控制的规则和一种分级管理的安全控制模型，该模型对远程访问矿山监测数据进行安全保障。其次，通过总结无线网络中的数据融合方法，并针对无线传感器网络的网络特性，提出群落主干线性网络模型。依据模型，结合现有数据融合算法，通过群体合作，对感知模糊的数据利用soft函数进行分级，合作评估分级辅助数据和衡量权重来决策数据融合结果，扩大了数据收集能力，提升了数据融合决策的性能。随着应用的开展和深入，考虑融合计算的需要，探索了基于矿山机器视觉数据对人体轮廓检测方法，尝试了机器视觉数据融入矿山无线网络数据决策的基本方法。针对群落主干线性网络模型和矿用无线网络的特点，利用群智的基本理论，探索和改进了PSO优化网络能耗的计算方法，协调和平衡了数据融合的负载性能。在这些工作的基础上，结合提出的窄长空间下无线网络延拓算法，利用运行时检验结果，探讨了利用群体协商引导移动载体自主和自适应的网络重塑方法的基本机制，获得了一种运行时无线传感器网络可信重塑方法。