基于动态数据驱动的突发水污染事故仿真方法

Sudden water pollution simulation method based on dynamic data-driven techniques

提出基于动态数据驱动的误差修正方法.结合城市供水水质安全预警系统仿真分析服务功能的建立,阐述动态数据驱动的突发水污染事故预测误差修正基本原理,以正向建模、模型封装与调用、初始仿真和模型修正等为主线研究技术实现的方法.研究模型边界更新法、模型参数更新法、模型结果校正法3种实现模型校正的技术.采用2个试验例子进行有效性验证.试验结果表明,由于引入了反馈机制,水质污染演化模拟仿真结果得到了实时修正,减少了不确定因素对仿真输出的影响,结果的准确性和可靠性得到了提高.

A simulation errors correction method was proposed based on dynamic data-driven techniques.Basic principle of the method was analyzed combined with the development of the simulation analysis service for a urban water quality early warning system.The implementation of the key steps,including forward modeling,model encapsulation and calling,the initial simulation and model correction,were investigated.Three model correction methods were focused on,including the model boundary updating method,model parameters updating method and model results correction method.The application of the model correction techniques was demonstrated using two test examples.Experimental results showed that the simulation results were optimized owning to the feedback mechanism.The impact of uncertainty was reduced,and the accuracy and reliability of simulation results were improved.