基于TRIZ提高小样本下轴承齿轮故障诊断模型的泛化能力  

Generalization Ability Improvement of Small Sample Bearing Gear Fault Diagnosis Model Based on TRIZ

作　　者：黄文静 李志农[1] HUANG Wenjing;LI Zhinong(Key Laboratory of Nondestructive Testing of Ministry of Education,Nanchang Hangkong University,Nanchang 330063,China)

机构地区：[1]南昌航空大学无损检测技术教育部重点实验室,南昌330063

出　　处：《机械工程师》2024年第12期61-65,共5页Mechanical Engineer

基　　金：科技部创新方法工作专项(2019IM010100);江西省高等学校教学改革研究课题(JXJX-21-8-31)。

摘　　要：轴承和齿轮的故障诊断对旋转机械的正常运行影响很大。由于当前的深度学习模型更多只是针对单一机械故障进行识别,同时需要大量的数据来支持模型的训练。为了解决这些问题,应用TRIZ创新方法工具,提出了基于TRIZ提高小样本下轴承齿轮故障诊断模型泛化能力的方法。运用TRIZ工具生成了15个方案,经过综合分析和评估,最终得到了一种组合解,不仅可以识别多种故障类型,还可以提高在小样本条件下轴承齿轮故障诊断模型的泛化能力。Troubleshooting of bearings and gears has a significant impact on the proper operation of rotating machinery.Because the current deep learning model is only used to identify a single mechanical fault,a large amount of data is required to support the training of the model.In order to solve these problems,this paper proposes a method based on TRIZ to improve the generalization ability of small sample bearing gear fault diagnosis model by using TRIZ innovative method tool.Fifteen conceptual schemes are generated by using TRIZ tool.After various methods of analysis and evaluation,a combined solution is finally determined,which can not only identify multiple fault types,but also improve the generalization ability of bearing gear fault diagnosis model under the condition of small samples.

关 键 词：轴承齿轮 小样本 故障诊断 泛化能力 TRIZ 

分 类 号：TH133.3[机械工程—机械制造及自动化]