基于差分进化算法的河流突发性污染事故溯源

Source Identification of Sudden Water Pollution Accident in River Based on Differential Evolution Algorithm

针对河流上突发的污染事故溯源问题,在可观测数据有限的情况下有效找出污染源信息对于开展有效的治理措施至关重要。为快速精确地得到事故位置、强度和时间的未知污染源参数,建立了最优化溯源模型,该模型考虑了测量误差,基于贝叶斯推理通过后验概率来度量模型优化目标。为求解优化模型,提出了改进的差分进化算法。首先,利用部分观测数据训练得出流速等水文参数;然后,采用自适应差分进化算法优化污染源参数。实例与仿真结果表明:与传统优化算法相比,改进的差分进化算法可以在更大的源项参数取值范围内有效提高河流突发性污染事故溯源的精度。

It was essential to effectively identify the source information of sudden pollution accidents when the observable data was limited.To obtain the unknown pollution source parameters such as location,intensity and time,an optimization identification model was proposed,which considers the measurement error and measures the model optimization target by posterior probability based on Bayesian inference.To solve the proposed model,an adaptive differential evolution algorithm was developed.Then,part of observation data was used to train the hydrological parameters.After that,adaptive differential evolution algorithm was used to optimize the pollution source parameters.Finally,computational results show that the proposed algorithm can effectively improve the accuracy to find the pollution source parameterswithin larger range of source term.