遗传算法优化神经网络的重金属污染源预测

BP Neural Network Based on Genetic Algorithm for Prediction of Soil Heavy Metal Pollution Source Location

根据土壤重金属的分布特征来确定重金属污染源的位置一直是一个突出的难题,通过将城市划分为不同功能区和测量表层土壤中重金属元素的背景值和分布情况,建立了遗传算法优化BP神经网络的模型,可以求取重金属浓度的最大值点对应的坐标值,实现对可能存在的污染源位置的定位预测。实例计算表明:该模型对于确定污染源位置具有很好的适应性,具有较好的容错容差能力和推广能力,研究结果可为有关环境部门进行环境整治提供一定的理论依据。

How to locate the heavy metal pollution source according to the distribution of heavy metal in soil has been a prom- inent problem. A BP neural network model based on genetic algorithm is established in the light of concentration values of heavy metal in surface soil of different functional areas in the city. The coordinate point compatible with of the maximum value of heavy metal concentration can be calculated to achieve the location prediction of the possible pollution source. The example calculations show that the model has good adaptability for determining the loeation of pollution source with good fault-tolerant capacity and gen- eralization ability. The results can serve as certain theoretical basis for environmental remediation of environmental departments.