基于PSO优化BP神经网络的GNSS PWV和气象数据的尾矿库降雨预测研究

Study on Rainfall Prediction of Tailings Pond Based on GNSS-PWV and Meteorological Data of PSO Optimized BP Neural Network

随着GNSS技术在气象领域的不断发展与进步,如今许多尾矿库和矿山都配备了GNSS设备,通过监测大气中的水汽含量、大气延迟量,可以更准确地预测降雨天气的发生。但当多种气象因素综合作用时,该技术的预测仍然存在一定的错报率。BP神经网络能够处理复杂的非线性问题,能够根据多种自然因素进行自适应的学习,提高模型的泛化能力,本文对BP神经网络进行了PSO优化,引入多种优化参数,使用NCEI数据平台对福建省区域的气象数据进行降雨预测模型的构建,为提高预测模型在尾矿库环境下的预测效果,本文结合尾矿库现场GNSS设备测量的PWV大气可降水量值,分析其变化量、变化率与降雨的关系,将两者引入降雨预测模型,构建更适应尾矿库环境的降雨预测模型。实验结果表明,针对尾矿库的特有情况,其正确率较国际方法能够提高7%~8%的效果。

With the continuous development and progress of GNSS technology in the meteorological field,many tailings ponds and mines are now equipped with GNSS equipment,which can more accurately predict the occurrence of rainfall weather by monitoring the water vapor content and atmospheric delay in the atmosphere.However,when a variety of meteorological factors are combined,the prediction of this technology still has a certain error rate.BP neural network can deal with complex nonlinear problems,can conduct adaptive learning according to various natural factors,and improve the generalization ability of the model.In this paper,the BP neural network has been PSO optimized,and a variety of optimization parameters have been introduced.The NCEI data platform is used to construct the rainfall prediction model based on the regional meteorological data of Fujian Province.In order to improve the prediction effect of the prediction model in the tailings pond environment,Based on the PWV atmospheric precipitable water measured by GNSS equipment at the tailings pond site,this paper analyzes the relationship between its change amount,change rate and rainfall,introduces the two into the rainfall prediction model,and constructs a rainfall prediction model more suitable for the tailings pond environment.The experimental results show that the accuracy of this method can be improved by 7%~8%compared with the international method according to the specific situation of tailings pond.