基于EMD-PSO-BP模型的短期潮流流速预测

Short-Term Tidal Current Speed Prediction Based on EMD-PSO-BP Model

针对潮流流速的随机性和波动性,本研究基于经验模态分解(Empirical mode decomposition,EMD)和粒子群优化(Particle swarm optimization,PSO)算法,改进了反向传播(Back propagation,BP)神经网络的短期潮流流速预测模型。该模型首先对原始流速序列进行EMD分解,得到多个本征模函数(Intrinsic mode function,IMF)和残差。然后,利用PSO改进BP神经网络,对分解所得的IMF和残差分别进行预测。最后,将各个预测结果相结合,得出流速的最终预测结果,从而提高潮流流速的预测精度。本文以江苏省潮流流速为例,分别建立BP、PSO-BP、EMD-BP以及EMD-PSO-BP四类预测模型,以对潮流流速进行预测和对比分析。结果表明,相较于其他模型,EMD-PSO-BP预测模型在潮流流速的预测方面具有更高的精度,为潮流能开发提供重要的数据支撑。

Tidal current speed prediction is of great significance to the development of tidal energy,aiming at the randomness and volatility of tidal current speed,this paper proposes a short-term tidal current prediction model based on empirical mode decomposition(EMD)and particle swarm optimization(PSO)to optimize back propagation(BP)neural network.The original tidal current speed sequence is decomposed by EMD to obtain multiple intrinsic mode function(IMF)and residuals,and then the BP neural network is optimized by PSO to predicts the IMF and residuals obtained by decomposition,and the paper combines all the prediction results to get the prediction result of the final tidal current speed,so as to improve the prediction accuracy of tidal current speed.In this paper,taking the tidal current speed in Jiangsu Province as an example,BP,PSO-BP,EMD-BP and EMD-PSO-BP prediction models are established to predict the current speed.The prediction results show that the prediction based on EMD-PSO-BP model further improves the accuracy of tidal current speed prediction,which can provide important data support for the development of tidal energy.