基于DWAE和GRUNN方法的船舶轮机传动系统运行状态识别  

Operational Status Recognition of Marine Turbine Transmission System Based on DWAE and GRUNN Methods

作　　者：曾志伟 Zeng Zhiwei(School of Marine Engineering,Guangxi Vocational and Technical College of Communications,Nanning Guangxi 530001,China)

机构地区：[1]广西交通职业技术学院航海工程学院,广西南宁530001

出　　处：《机械管理开发》2025年第3期201-203,共3页Mechanical Management and Development

基　　金：2024年广西向海经济人才培养支持专项项目。

摘　　要：为了更好识别齿轮箱运行状态分析能力,结合门控循环单元神经网络(GRUNN)与深度小波自动编码器(DWAE)的优点,设计了一种基于DWAE与GRUNN方法的齿轮箱故障状态识别方法,并使用dropou处理来提升深度学习模型泛化效果。开展实际齿轮箱运行测试,研究结果表明:随着齿轮故障程度逐步加剧,状态识别准确性呈现出显著提升的趋势,平均准确率提升至99%,表现出来很高的稳定性。与其他方法相比,DWAE-GRUNN在齿轮故障状态识别中的准确率明显提高,实现小样本条件下齿轮故障状态识别。该研究有助于提高齿轮箱的工作稳定性,也可拓宽到其他的机械传动领域。In order to better identify the operating state analysis ability of gearbox,a gearbox fault state recognition method based on DWAE and GRUNN was designed,combining the advantages of gated recurrent unit neural network(GRUNN)and deep wavelet automatic encoder(DWAE),and dropou processing was used to improve the generalization effect of deep learning model.The actual gear box operation test was carried out,and the results showed that:as the gear fault degree gradually intensified,the accuracy of state recognition showed a significant trend of improvement,and the average accuracy rate increased to 99%,showing high stability.Compared with other methods,the accuracy of DWAE-GRUNN in gear fault state identification is significantly improved,and the gear fault state identification can be realized under the condition of small samples.This research is helpful to improve the working stability of the gearbox and can also be extended to other mechanical transmission fields.

关 键 词：齿轮箱运行 故障状态识别 编码器 神经网络 准确率 

分 类 号：TH137[机械工程—机械制造及自动化]