| 船舶轴系滑油中磨损铁屑含量预测方法 |
| |
| 作者姓名： | 徐昌昌 张丹瑞 |
| |
| 作者单位： | 上海船舶运输科学研究所航运技术与安全国家重点实验室,上海200135;上海船舶运输科学研究所航运技术与安全国家重点实验室,上海200135 |
| |
| 摘 要： | 为有效解决船舶轴系滑油中的磨损铁屑含量预测与评价方面的问题,提出一种组合预测方法。为降低测量噪声对预测的影响,利用小波变换对测量序列进行降噪,选择Daubechies 4(Db4)作为小波基,结合软阈值函数对时间序列进行分解和重构,同时利用基于平滑度和均方根误差的复合指标确定最优的分解层数。采用非线性自回归神经网络(Nonlinear Auto-Regressive Neural Network,NARNN)进行预测分析,实现对变化趋势的预测和对保养时间的评估。以某船轴系滑油中的磨损铁屑含量历史数据为样本进行试验,结果表明该方法是有效的。
|
| 关 键 词： | 船舶轴系 小波变换 软阈值 非线性自回归神经网络 |
Method for Predicting the Content of Iron Filings in Lubricating Oil of Ship Shafting |
| |
| Authors: | XU Changchang ZHANG Danrui |
| |
| Institution: | (State Key Laboratory of Navigation and Safety Technology, Shanghai Ship and Shipping Research Institute, Shanghai 200135, China) |
| |
| Abstract: | The measurement data are preprocessed with a wavelet transform denoising method to reduce the adverse impact of measurement noise on prediction.Db4 is selected as the wavelet basis and the soft threshold function is applied to complete the decomposition and reconstruction of time series.The optimal decomposition layer number is determined by using the composite index based on the smoothness and root-mean-square error.NARNN(Nonlinear Auto-Regressive Neural Network)is used to process the de-noised time series,implementing the trend forecast and maintenance time evaluation.The method is verified with the historical data from a ship’s shafting monitoring. |
| |
| Keywords: | ship shafting wavelet transform soft threshold NARNN |
| 本文献已被 维普 万方数据 等数据库收录！ |
|
