基于神经网络的工程机械远程故障诊断技术研究

发布时间：2018-12-12 18:00

【摘要】：远程故障诊断系统是通过GPRS无线技术将现场车载终端和远程技术诊断中心联系起来,实现即时反应、资源共享、远程监测以及远程诊断的一个系统,它既有传统故障诊断服务方式的优点,又克服了时间、地域的局限。工程机械各部件受所处的环境、温度、水蒸气、粉尘和振动的影响很大,液压系统作为工程机械的核心,结构比较复杂,若出现故障,将会直接影响其工作效率,甚至出现重大的事故。对液压系统进行远程故障实时的检测与诊断,能够缩短工程机械的停机时间,提高经济效益。本文以某重型机械公司HB48混凝土泵车主液压系统为研究对象,采用ATmega16单片机为主控制核心,BenQ M22A GPRS模块为传输单元,设计了一种远程数据采集终端;在分析了液压系统故障常见的故障模式及机理以及神经网络的工作原理的基础上,将BP算法、Hopfield优化的BP算法应用于泵车液压系统故障诊断。通过对基于BP、H-BP和PSO三种神经网络的液压系统故障诊断方法的研究与比较,文章提出了一种先采用粒子群算法优化Hopfield网络权值矩阵后的网络,对原始数据预处理,再进行BP算法诊断结果的故障诊断方法,即PSO-H-BP算法,并将该算法应用于液压系统的故障诊断,验证其有效性和精确性。实验表明:采用ATmega16与BenQ M22A组建的数据采集传输终端能够实现实时采集、快速通讯,具有很好的实用性;PSO-H-BP算法的BP、H-BP相比,具有较高的准确性和可靠性。
[Abstract]:The remote fault diagnosis system is a system that connects the onsite terminal and the remote technology diagnosis center through GPRS wireless technology, and realizes immediate response, resource sharing, remote monitoring and remote diagnosis. It not only has the advantages of traditional fault diagnosis service, but also overcomes the limitation of time and region. The components of construction machinery are greatly affected by the environment, temperature, water vapor, dust and vibration. As the core of construction machinery, hydraulic system has a complex structure, and if failure occurs, it will directly affect its working efficiency. There were even major accidents. Remote fault detection and diagnosis for hydraulic system can shorten the downtime of construction machinery and improve economic efficiency. This paper takes the main hydraulic system of HB48 concrete pump car of a heavy machinery company as the research object, adopts ATmega16 single chip microcomputer as the main control core and BenQ M22A GPRS module as the transmission unit, and designs a remote data acquisition terminal. On the basis of analyzing the common fault mode and mechanism of hydraulic system and the working principle of neural network, the BP algorithm and Hopfield optimized BP algorithm are applied to the fault diagnosis of hydraulic system of pump car. Through the study and comparison of fault diagnosis methods of hydraulic system based on BP,H-BP and PSO neural networks, this paper proposes a network that optimizes the weight matrix of Hopfield network by using particle swarm optimization (PSO), and preprocesses the original data. Then the fault diagnosis method of BP algorithm, that is, PSO-H-BP algorithm, is applied to the fault diagnosis of hydraulic system to verify its validity and accuracy. The experimental results show that the data acquisition and transmission terminal constructed by ATmega16 and BenQ M22A can realize real-time acquisition, fast communication and good practicability, and the BP,H-BP of PSO-H-BP algorithm has higher accuracy and reliability.
【学位授予单位】：太原科技大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：TH165.3;TP183

【参考文献】