气流环境下无线传感器网络分布式气体释放源定位方法研究

For the requirements of high precision and high reliability and low energy-consumption, heterogeneous wireless sensor network coverage control, distributed robust odor source parameter estimation, and distributed odor substance probability distribution numerical model decoding, have become the main bottlenecks for the development of environment monitoring missions relating to locating the source of anomalous odor substance diffusion, which are important to the national economy and the people's livelihood..To this end, this project systematically carry out critical infrastructure studies in airflow environments as follows. Firstly, the odor source sensing model relating to different airflow environments will be investigated, and a novel heterogeneous wireless sensor network multiple coverage control method will be proposed utilizing the multi-objective intelligent optimization algorithms. Secondly, the odor source parameter estimation problem with non-Guassian distribution measurement noise in homogenous steady airflow environments will be investigated, and a novel distributed robust odor source parameter estimation method with high parameter estimation capability will be proposed. Finally, the problem of odor source localization in dynamic airflow environments, where the odor concentration distribution cannot be formulated as analytic models, will be investigated, and a novel distributed odor substance probability distribution hidden Markovian numerical model decoding method will be proposed..This study can provide scientific support to establish the early-warning and emergency response mechanism for the toxic/harmful-gas-leak accidents and forest-fire accidents, and it has great practical significance for the protection of people’s lives and property and the promotion of sustainable national economy development.

为了保障气味物质异常扩散的源头定位任务中无线传感器网络面临的高精度、高可靠性、低能耗等需求，异构传感器网络覆盖控制、分布式稳健气味源参数估计、气味物质概率分布数值模型分布式解码等关键问题，成为制约此类关系国计民生的环境监测任务发展的主要瓶颈。.为此，本项目针对气流环境系统地开展以下关键基础问题研究：研究与气流环境相关的气味源感知模型，并利用多目标智能优化算法提出异构传感器网络多重覆盖控制方法；研究均匀分布稳态气流环境下存在非高斯分布浓度测量噪声的气味源参数估计问题，提出具有较强参数估计能力的分布式稳健气味源参数估计新方法；研究动态气流环境下气体浓度分布无法通过解析模型描述的气味源定位问题，提出气味物质概率分布隐马尔可夫数值模型分布式解码方法。.本研究可为毒害气体泄漏、森林火灾等事故预警和应急处理机制的建立提供科学支撑，对于保障人民生命及财产安全，促进国民经济的可持续发展具有重大现实意义。

针对无线传感器网络(WSN)分布式气体释放源定位难题，本课题组系统地开展了三个方面的主题研究：“气流场模拟及基于模拟气流场的WSN气体源感知覆盖”、“气体扩散模型多未知参数分布式稳健估计研究”、“气流环境下气体来源方向辨识及WSN分布式气味源定位”。本课题取得的代表性研究成果包括：自主研发了具有高频气体浓度采集和数据处理能力的无线传感节点、小型风场生成装置、封闭腔式气体浓度采集及辨识系统等实验装置；建立了基于气流速度概率分布的传感节点气味源覆盖模型，基于多目标智能优化算法实现了传感器网络的确定性k重覆盖；基于分形理论分析了轴流风机生成气流传播过程中的空间相关性；提出了基于梯度提升树算法的异构气体传感系统；基于人工蜂群算法和最小二乘估计方法，实现了气体扩散模型多未知参数稳健估计；提出了基于神经网络的气体来源方向辨识及WSN分布式气味源定位方法。以上所提方法的有效性和可行性均通过大量仿真或真实WSN实验进行了验证。本研究的成果预期可为有毒有害气体泄漏源定位、毒品等带有挥发性气体的违禁物品搜查，地震等灾害废墟下幸存者救援等提供理论和技术支撑。