摘要
地下水位波动预测对水资源管理具有重要意义。利用人工神经网络(ANNs)、自适应神经模糊推理系统(ANFIS)和支持向量机(SVM)模型对黄河三角洲地下水位波动进行了预测。选取历史地下水位作为影响地下水位波动的唯一因素,输入向量由地下水位时间序列的自相关函数(ACF)和偏自相关函数(PACF)确定。结果表明,ANNs,ANFIS和SVM模型能够利用历史地下水位数据成功预测地下水位。ANNs模型和SVM模型具有较好的稳定性,预测精度较高,而ANFIS对于地下水位突变信息的预测效果最好。
Predicting groundwater level fluctuation is very important for water resources management.Groundwater level fluctuation in the Yellow River Delta was predicted using artificial neural networks(ANNs),adaptive neurofuzzy inference system(ANFIS),and support vector machine(SVM)models.The historic groundwater level was selected as the only factor affecting groundwater level fluctuation.The input vectors were consisted of the past groundwater level and determined by the autocorrelation function(ACF)and partial autocorrelation function(PACF)of the groundwater level time series.Based on the comparisons,it concluded that ANNS,ANFIS and SVM models can successfully predict groundwater levels using historical groundwater level data.The ANNs and SVM models performed more stable,and obtained desirable predictive accuracy;ANFIS has the best prediction effect on the information of groundwater level mutation.
出处
《环境保护与循环经济》
2020年第12期41-45,共5页
environmental protection and circular economy
