基于IPSO-DE-MH算法的突发水污染事件预测模型参数识别

PARAMETER IDENTIFICATION OF SUDDEN WATER POLLUTION ACCIDENTS BASED ON IPSO-DE-MH METHOD

预测模型是有效应对突发水污染事件的前提与基础。为了提高预测模型的准确性,提出了一种新的参数识别方法。首先从反问题与贝叶斯估计的视角构建突发水污染事件预测模型;然后在Metropolis-Hastings抽样方法的基础上,引入混沌理论、粒子群算法、微分进化算法等的思想,设计了一种新的参数识别方法,即IPSO-DE-MH算法;最后通过数值分析验证所设计方法的有效性和准确性。结果表明:新方法能较好地识别模型参数,为突发事件应急预测模型的快速构建提供了新思路。

The prediction model is the premise and foundation of effectively dealing with sudden water pollution accidents. To improve the accuracy of the prediction model, a new parameters identification method was proposed in this paper. This paper first built a prediction model from the perspective of the inverse problem and Bayesian, and then designed a new identification method based on the chaos theory, particle swarm optimization, differential evolution and Metropolis-Hastings sampling method, i.e. IPSO-DE-MH. Finally, the effectiveness and accuracy of the designed method were verified by numerical analysis. The results showed that the new method could better identify the model parameters, and provide a new idea for the construction of an emergency prediction model.