Key Laboratory of Integrated Regulation and Resource Development on Shallow Lake of Ministry of Education, College of Environment, Hohai University, Nanjing 210098, China E-mail:
Water Sci Technol. 2014;69(4):687-93. doi: 10.2166/wst.2013.740.
For solving the multi-parameter identification problem of a river water quality model, analytical methods for solving a river water quality model and traditional optimization algorithms are very difficult to implement. A new parameter identification model based on a genetic algorithm (GA) coupled with finite difference method (FDM) was constructed for the determination of hydraulic and water quality parameters such as the longitudinal dispersion coefficient, the pollutant degradation coefficient, velocity, etc. In this model, GA is improved to promote convergence speed by adding the elite replacement operator after the mutation operator, and FDM is applied for unsteady flows. Moreover the influence of observation noise on identified parameters was discussed for the given model. The method was validated by two numerical cases (in steady and unsteady flows respectively) and one practical application. The computational results indicated that the model could give good identification precision results and showed good anti-noise abilities for water quality models when the noise level ≤10%.
为了解决河流水质模型的多参数辨识问题，解析法求解河流水质模型和传统的优化算法都很难实现。为了确定水力和水质参数，如纵向离散系数、污染物降解系数、速度等，构建了一种基于遗传算法（GA）和有限差分法（FDM）的新的参数识别模型。在该模型中，通过在变异算子后添加精英替换算子，对 GA 进行了改进以提高收敛速度，并将 FDM 应用于非稳态流。此外，还讨论了观测噪声对给定模型中识别参数的影响。该方法通过两个数值算例（分别在稳态和非稳态流中）和一个实际应用进行了验证。计算结果表明，当噪声水平≤10%时，该模型能够给出良好的辨识精度结果，并且对水质模型具有良好的抗噪声能力。
