基于嵌入式的柴油机在线故障诊断系统设计

发布时间：2018-08-28 10:40

【摘要】：柴油机作为工业制造、车辆、船舶、采矿的主要动力源,其故障诊断的技术成为近几年研究的热点之一。本文设计了一种下位机采集与上位机测试一体化的在线诊断的方法,并通过LMD-SVM算法实现柴油机的故障诊断。本诊断系统的下位机设计,主要应用了信息技术,设计了一种以ARM11内核的S3C6410为处理器、以ADXL345为振动传感器、以Linux为操作系统的数据可存储的采集装置,且可以通过串口实现与上位机的数据通信。本诊断系统的上位机设计,主要应用了MATLAB工具,利用GUI功能设计了一个用于人机交互的用户界面,用于振动数据的动态显示和逻辑控制,并嵌入了LMD-SVM算法,用于模式分类和故障诊断。LMD-SVM算法是将原始信号进行局域均值分解成多个PF函数分量,求解每个PF函数的近似熵,并将近似熵组合作为特征向量,先将训练样本输入支持向量机进行训练,建立SVM数学模型,再将测试样本输入SVM进行模式分类,最终实现故障诊断。最后,以R6105AZLD型号的柴油机为研究对象,选取柴油机的6种工况,搭建柴油机试验平台,调试下位机和上位机,实现信号的在线采集,分别用180组测试样本进行试验,测试通过率高达91.67%,试验表明,此方法能基本满足工程应用。
[Abstract]:As the main power source of industrial manufacture, vehicle, ship and mining, the fault diagnosis technology of diesel engine has become one of the hotspots in recent years. In this paper, an on-line diagnosis method of the integration of lower computer acquisition and upper computer testing is designed, and the fault diagnosis of diesel engine is realized by LMD-SVM algorithm. In the design of the lower computer of the diagnosis system, the information technology is mainly used, and a kind of data acquisition device, which uses S3C6410 of ARM11 kernel as processor, ADXL345 as vibration sensor and Linux as operating system, is designed. And can realize the data communication with the host computer through the serial port. In the design of the upper computer of the diagnosis system, the MATLAB tool is mainly used, and a user interface for human-computer interaction is designed by using the GUI function, which is used for dynamic display and logic control of vibration data, and the LMD-SVM algorithm is embedded. LMD-SVM algorithm is used for pattern classification and fault diagnosis. LMD-SVM algorithm decomposes the local mean of the original signal into several PF function components, solves the approximate entropy of each PF function, and combines the approximate entropy as the eigenvector. The training sample is input into support vector machine to train, the SVM mathematical model is established, then the test sample is input into SVM to classify the pattern, and finally the fault diagnosis is realized. Finally, taking the diesel engine of R6105AZLD model as the research object, selecting six working conditions of the diesel engine, setting up the diesel engine test platform, debugging the lower computer and the upper computer, realizing the on-line acquisition of the signal, the test is carried out with 180 groups of test samples, respectively. The passing rate of the test is up to 91.67. The experiment shows that this method can basically meet the engineering application.
【学位授予单位】：中北大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TK428

【参考文献】