论文部分内容阅读
为提高大坝安全监测数据预测精度,将差分自回归移动平均模型(ARIMA)与广义回归神经网络(GRNN)相结合,建立了ARIMA-GRNN预测模型。以前期实测值和ARIMA拟合值作为GRNN网络的输入,后期实测值作为网络输出,以平均平方误差最小为原则寻找光滑因子,建立最佳的预测模型,并运用熵权法和标准离差法对各模型进行多指标综合评价。结果表明,ARIMA-GRNN模型预测精度较ARIMA模型明显提高,可应用于大坝安全监测。
In order to improve the prediction accuracy of dam safety monitoring data, the ARIMA-GRNN prediction model is established by combining the ARIMA with GRNN (Generalized Regression Neural Network). The pre-measured value and ARIMA fitting value are used as the input of GRNN network, and the post-measured value is output as network. The smoothing factor is found based on the principle of average square error minimum, and the best prediction model is established. The entropy method and standard deviation method The multi-index comprehensive evaluation of each model. The results show that the prediction accuracy of ARIMA-GRNN model is significantly higher than that of ARIMA model and can be applied to dam safety monitoring.