混合动力船舶负载功率预测模型研究  

Research on Load Power Prediction Model of Hybrid Power Ship

作　　者：高迪驹[1] 潘康凯 王天真[1] GAO Di-ju;PAN Kang-kai;WANG Tian-zhen(Key Laboratory of Marine Technology and Control Engineering Ministry of Communications, Shanghai Maritime University,Shanghai 201306, China)

机构地区：[1]上海海事大学航运技术与控制工程交通运输行业重点实验室,上海201306

出　　处：《控制工程》2019年第2期362-367,共6页Control Engineering of China

基　　金：国家自然科学基金项目(61304186);上海海事大学研究生创新基金项目(2016ycx032)

摘　　要：为了实现混合动力船舶的各动力源之间的最佳负载功率分配,构建了一种基于多分辨率小波神经网络(MRA-WNN)的混沌时间序列短期预测模型。将小波函数与尺度函数共同应用于网络基函数之中,首先从较大尺度上逼近时间序列的整体趋势,然后根据负载功率波动的大小,在不同尺度上逐层加入细节逼近,提高预测精度。由多分辨率解确定小波基函数的平移和伸缩参数,并结合多分辨率学习算法,能减少训练参数,提高计算速度。实验结果表明,MRA-WNN具有较高的预测精度,是混合动力船舶负载功率预测的一种有效方法。The short term prediction model of chaotic time series based on multi-resolution wavelet neural network(MRA-WNN) is set up to realize the optimal power allocation between the power sources of the hybrid power ship. The wavelet function and the scaling function are used as the network basis function. First,the overall profile of the time series is approximated in a large scale. And then, according to the different degrees of load power fluctuation, the approximation of the layer by layer is added to the details for improving the prediction accuracy. The translation and scaling parameters of wavelet basis functions are determined by the multi-resolution solution, and the number of training parameters can be decreased and the calculation speed can be improved by combining the multi-resolution analysis learning algorithm. The experimental results show that the MRA-WNN has high prediction accuracy, and it is an effective method for the prediction of the load power of hybrid power ships.

关 键 词：混合动力船舶 负载功率 多分辨率小波神经网络 混沌时间序列 短期预测 

分 类 号：TP27[自动化与计算机技术—检测技术与自动化装置]