城市污水处理厂出水含氮量在线检测方法研究

Municipal wastewater treatment is an effective way to protect the water environment and realize the circulation of water resources. Nowadays, China is facing the severe environmental issues caused by eutrophication of water body. And a safe and stable operation of denitrification process in the municipal wastewater treatment plant is a prerequisite to solve that problem. Consequently, to realize the safe and stable operation of denitrification process, this project will study on the online detection method of effluent nitrogen content in the municipal wastewater treatment after in-depth analysis of the process characteristics. The major research content includes four parts. Firstly, the mechanism and dynamic characteristics of the denitrification process will be analyzed to determine the online detection variables. Secondly, the data mining technology will be discussed to select the characteristic variables of the online detection variables. Then, with the aim realize the online detection of multiple water quality parameters simultaneously, self-organization mechanisms of the modular neural network will be investigated to design a dynamic soft-sensing model. Finally, the online detection system of effluent nitrogen content in municipal wastewater treatment plant will be designed and developed to break through the technical bottleneck of on-line detection. The proposed method will provide new ideas for the modeling problems under open, disturbed and uncertain dynamic environments, which would promote the development of control science. Furthermore, the online detection technology of effluent nitrogen content will become a strong impulse to the municipal wastewater treatment industry and has broad application prospects.

城市污水处理是保护水环境、实现水资源再利用的有效途径。针对我国水体富营养化严重的环境问题，本项目从实现城市污水处理厂脱氮过程安全稳定运行的实际需求出发，在深入分析工艺流程特征的基础上，研究出水含氮量在线检测方法及关键实现技术。具体内容：研究城市污水处理脱氮过程机理特征和动力学特性，确定在线检测变量；研究数据挖掘技术，获取各水质参数特征变量；研究模块化神经网络自组织机制，设计动态软测量模型，解决多水质参数同步在线检测问题；开发城市污水处理厂出水含氮量在线检测系统，突破出水含氮量难以在线检测的技术瓶颈。研究工作为开放、受扰及不确定动态环境中的建模问题提供了新思路，促进了控制科学的发展。获得的出水含氮量在线检测技术推动了城市污水处理行业的发展，具有广阔的应用前景。

城市污水处理是保护水环境、实现水资源循环利用的重要举措。目前，受困于含氮量检测技术，城市污水处理脱氮过程难以高效调控，出水氨氮和出水总氮排放无法稳定达标。针对以上问题，项目开展了城市污水处理过程含氮量智能检测方法与应用研究。首先，提出了基于过程数据和经验知识的变量相关性分析方法，实现了从生化反应过程数据（进水水质、过程变量、运行环境等）中动态提取出水含氮量特征变量；然后，针对城市污水处理过程复杂的非线性特性，提出了一种基于类脑模块化神经网络的软测量建模方法，一方面引入类脑机制实现建模任务分而治之，另一方面协同结构自组织机制和参数混合学习算法设计软测量模型子模块，实现了出水氨氮和出水总氮的实时检测；此外，针对实际工业过程中的非平稳特性，融合神经网络泛化性能和神经元活跃度，提出了一种软测量模型分级更新策，确保了软测量模型在运行过程中的稳定性；最后，形成了一套出水含氮量智能检测系统，并在实际城市污水处理厂进行了应用验证，实现了出水氨氮和出水总氮排放稳定达标。项目研究成果已在IEEE Transactions on Industrial Informatics、IEEE Transactions on Automation Science and Engineering等刊物和会议上发表论文14篇，其中SCI收录8篇；申请发明专利6项，获得授权软件著作权1项。培养副教授1名，培养研究生10名(其中5名博士生、5名硕士生)。