【作者】 杨波；

【导师】 王志文；

【作者基本信息】 大连海事大学 ， 工程硕士（专业学位）， 2023， 硕士

【摘要】 气压传动系统广泛应用于工业生产部门。随着智能制造和绿色制造的深入发展,气动技术在能效低、故障诊断智能化水平低等方面的缺点日益凸显。由于气压传动系统的强非线性和强耦合性,其故障诊断较液压传动、电气传动和机械传动系统更为困难。本研究中,提出一种基于能量?和机器学习的气动系统故障诊断方法,以单气缸内外泄漏故障诊断为例,验证了方法的可行性,希望为后续开发基于?的气压传动系统能量-健康-质量融合管理范式奠定基础。本文的主要研究工作包括以下几个方面:(1)对气压传动系统故障诊断研究进行了总结分析,基于第一性原理分析提出了传动系统故障-能量-质量关联框架,通过对能量?数据进行分析,就可以实现对气压传动系统的能量、健康和质量的融合管理。针对气压传动系统的能量与健康融合管理,提出了气动元件/系统的能耗模式与故障模式之间存在映射关系的基本假设,以单气缸泄漏故障为例进行实验设计与假设验证。(2)搭建了单气缸泄漏故障诊断气动回路实验系统,开发了基于Lab VIEW的数据采集与控制系统,实现了对监测点压力和流量信号实时采集处理以及元件控制;通过处理缺失值、重复值、分割循环等步骤对原始数据进行了数据预处理;对比了栈式自编码器、自编码器与核主成分分析的特征提取性能,证明了栈式自编码器特征提取的优势;通过单气缸内外泄漏实验,对比了以压力、流量与?为指标时,单气缸故障的分类准确率,证明了机器学习方法对气压传动系统故障诊断的适用性以及利用?数据实现气缸状态监测的可行性。(3)运用了不同的机器学习方法,通过对压力、流量与?数据分析实现对气缸内外泄漏故障的诊断,结果表明使用压力和流量数据时的诊断准确率高度依赖于系统运行工况,而使用?数据时的诊断准确率则对系统运行工况不敏感,基于?数据的诊断准确率相比于压力和流量数据更加稳定。此外,通过对原始数据在时域和时频域的不同处理分析,完成了对单气缸未知外泄漏流量范围的判定。更多还原

【Abstract】 Pneumatic transmission systems are widely used in industrial production departments.With the deepening development of intelligent manufacturing and green manufacturing,the shortcomings of pneumatic technology in low energy efficiency and low level of intelligent fault diagnosis are becoming increasingly prominent.Due to the strong nonlinearity and strong coupling of pneumatic transmission system,its fault diagnosis is more difficult than that of hydraulic transmission,electrical transmission and mechanical transmission systems.In this study,a method combining energy based pneumatic system maintenance and fault diagnosis is proposed.Taking the diagnosis of internal and external leakage faults in a single cylinder as an example,the feasibility of the method is verified.It is hoped to lay the foundation for the subsequent development of an energy based pneumatic transmission system management paradigm.The research work of this paper mainly includes the following aspects:(1)The research on fault diagnosis of pneumatic transmission system was summarized and analyzed.Based on first principles analysis,the correlation framework of transmission system fault-energy-quality was put forward.Exergy data analysis can realize the integrated management of energy,health and quality of pneumatic transmission system.Aiming at the integrated management of energy and health of pneumatic transmission system,the basic assumption that there is a mapping relationship between energy consumption mode and failure mode of pneumatic components/system is proposed.Taking single cylinder leakage fault as an example,the experimental design and hypothesis verification are carried out.(2)The pneumatic circuit test system for single cylinder leakage fault diagnosis was built,and the Lab VIEW-based data acquisition and control system program was developed to realize real-time acquisition and processing of pressure and flow signals of a single upstream monitoring point of the cylinder.The original data is preprocessed by processing missing value,duplicate value and segmentation cycle.According to internal and external indexes,the feature extraction performance of stack autoencoder,autoencoder and kernel principal component analysis is compared,and the advantages of stack autoencoder feature extraction are proved.By conducting internal and external leakage tests on a single cylinder,the classification accuracy of single cylinder faults was compared using pressure,flow rate,and energy as indicators,proving that traditional machine learning methods are suitable for the pneumatic field and that using energy to monitor cylinder status is feasible.(3)The cylinders with internal and external leakage were tested and different machine learning methods were used to monitor the cylinders by pressure,flow rate and exergy indicators.It is found that the diagnostic accuracies when using pressure and flowrate data are highly dependent on operating conditions.While the diagnostic accuracy when using exergy data is always high regardless of operating conditions.While Exergy was more stable than other cylinders.By processing the original data in time domain and time frequency domain,the unknown external leakage range of single cylinder is determined.更多还原