基于小波神经网络的船舶缆绳载荷预测方法

Prediction of Mooring Load Based on Wavelet Neutral Network

在船舶缆绳载荷准确预测的研究中,由于船舶缆绳载荷不仅受到风、流、浪等环境因素的影响,还受到船舶的船型、受风面积、吃水大小等因素的共同影响,因此具有较强的随机性和复杂性,是一种非平稳的时间序列,传统的神经网络预测模型在进行负荷预测过程中,无法处理这种非平稳信号导致很难进行准确测量。提出一种基于小波神经网络的船舶缆绳载荷预测方法,算法结合小波分析的时频局部特性与聚焦特性和神经网络的自学习、自适应和推广能力,将小波基函数作为神经网络的隐含层节点的传递函数,建立小波神经网络预测模型,以船舶缆绳的采集数据作为模型的输入与输出,利用小波函数处理非平稳信号的能量,解决缆绳负荷的非线性问题,凭借神经网络小区域计算能力,对预测结果进行进一步优化。仿真结果表明,小波神经网络用于船舶缆绳载荷数据处理,精度满足要求,具有良好的适用性。

Research the accurate prediction of ship mooring load.Ship mooring load has strong randomness and complexity,and is a non-stationary time series.The paper put forward a mooring load prediction method based on wavelet neutral network.The model combines the characteristic of wavelet analysis with the advantage of neutral network.The wavelet basis function has been used as the transfer function of neural network＇ s hidden layer nodes.Then the wavelet neutral network forecasting model was built up to realize the forecasting calculation of mooring load data.The simulation results show that the wavelet neutral network algorithm attains high-precision forecast results.It can meet requirements of short-term early-warning of mooring load.