论文部分内容阅读
针对钱塘江潮位呈现出的周期性、随机性和波动性,为实现对钱塘江潮位的有效预测,提出一种基于离散小波变换和时间序列的预测方法,即先利用离散小波变换将实测的钱塘江潮位序列进行分解与重构,将非平稳的序列转化为多层较平稳的序列;然后利用时间序列建模方法对分解后的各个序列分别建立时间序列模型,对各层进行动态预测;最后将各层预测值求和作为最终的预测结果。试验表明,所提方法预测的效果明显优于其他混合模型及单一模型,能够提供更加准确的潮位预测。
In view of the periodicity, randomness and volatility of the Qiantang River tidal level, in order to realize the effective prediction of the Qiantang River tide level, a prediction method based on discrete wavelet transform and time series is proposed, that is, firstly, the measured Qiantang Then the non-stationary sequence is transformed into a smoother multi-level sequence by using the time series modeling method. Then the time series model is established for each sequence after the decomposition, and the dynamic prediction is made for each layer. Finally, Summing the forecast values of each layer as the final forecast result. Experiments show that the proposed method is obviously superior to other mixed models and single models in predicting the tide level.