基于卡尔曼滤波-反问题模型的入河排口监管方法

A Monitoring Method for Outfalls Along River Based on Kalman Filter-inverse Problem Model

摘要:

水下入河排口的监管是当前城市污水治理提质增效的难点。针对人工排查及物探成像等技术在入河排口的监管中成本高而且实时性差的现状，通过耦合水动力模型和马尔科夫链-蒙特卡洛算法，同时引入卡尔曼滤波策略，建立反问题数学模型，根据河道在线水位和流量监测数据，反演入河排口日排放水量和一天内不同时间段排水的动态变化，并根据入河排口实际调查案例进行验证。结果表明，该模型可有效反演入河排口动态水量排放，反演误差在10%以内，而且相较于未引入卡尔曼滤波的马尔科夫链-蒙特卡洛算法节省了近50%的计算时间。基于该数学模型，可集成一组在线监测设备，实现入河排口的动态监管，为构建基于在线监测技术的入河排口智能监管体系提供技术支撑。

Abstract:

The supervision of underwater outfalls along the river is challenging in improving the quality and efficiency of urban wastewater treatment. In response to the high cost and poor real-time performance of current techniques such as manual inspections and geophysical imaging techniques in the monitoring of outfalls， the mathematical model of the inverse problem was established by coupling the hydrodynamic model and the Markov Chain-Monte Carlo algorithm， and the Kalman filter strategy was introduced to inversely estimate the dynamic discharge rates of the outfalls based on online water level and flow monitoring information. The results indicated that this inverse problem model can effectively invert the dynamic variation of outfall drainage， with an inversion error below 10% and save nearly half of the calculation time compared with the Markov Chain-Monte Carlo algorithm without introducing Kalman filter. Based on this mathematical model， an online monitoring system could be integrated to achieve effective dynamic supervision of the outfalls， thereby providing technical foundation for developing an intelligent outfall monitoring system with the support of the online monitoring technology.