优化短期余水位组合预测模型

Combination prediction model of optimized short-term residual water level

针对现有非稳定非线性余水位预测模型较少和精度较低的问题,本文研究基于MEEMD算法与遗传优化BP神经网络的余水位组合预测模型。利用夏威夷岛4个长期验潮站获取的余水位时序数据,首先采用遗传算法MEEMD对余水位时序数据进行处理分析,得到较为稳定的余水位IMF分量;然后将经过遗传算法优化后分解的较为稳定的各个IMF分量作为BP神经网络预测模型的输入变量,分别建立12、24、48 h短期余水位的MEEMD遗传算法优化BP神经网络预测模型。通过与非优化BP神经网络预测模型结果进行对比分析,结果表明,优化前后均方根误差的偏差最高达2.03 cm,验证了预测24 h内的短期余水位仍保持其相关特性。该组合预测模型对于分析余水位变化规律和潮汐预报的精度、水位改正等均有重要意义。

In order to solve the problem that the existing non-stable and non-linear residual water level prediction models are less and low accuracy,a combined residual water level prediction model based on MEEMD algorithm and genetic optimization BP neural network is studied.Based on the time series data of residual water level obtained from four long-term tidal stations in Hawaii island,the genetic algorithm MEEMD is firstly used to process and analyze the time series data of residual water level,and a relatively stable IMF component of residual water level is obtained.Then,the stable IMF components decomposed by genetic algorithm optimization are taken as the input variables of BP neural network prediction model,and the prediction models of BP neural network optimized by MEEMD genetic algorithm for 12,24 and 48 h short-term residual water levels are established respectively.By comparing with the results of the non-optimal BP neural network prediction model,the results show that the deviation of the root mean square error before and after optimization is up to 2.03 cm,which verifies that the short-term residual water level within 24 h is still maintained its relevant characteristics.The combined prediction model is of great significance to the analysis of the variation law of residual water level,the accuracy of tide prediction and the correction of water level.