基于ARM的航空结构冲击载荷监测系统研究

发布时间：2019-03-20 09:16

【摘要】：随着航空飞行器性能的提高,其结构日趋复杂,在服役过程中这些结构将不可避免受到各种形式的冲击且极易造成不同程度的损伤,而这些损伤会成为结构安全与使用寿命潜在的威胁,因此,结构健康监测技术中的冲击载荷在线实时监测研究十分有现实意义。本文基于这一迫切工程需求,开发出一套基于ARM的在线实时监测系统,主要研究工作如下:首先简要介绍了声发射技术的概念,压电传感器的采集原理以及声发射信号的传播特性,后根据上述介绍选择小波变换作为声发射信号处理技术,并分析利用此技术获取波达时间的方式,同时提出了非正交小波变换的快速算法;其次,介绍了互相关函数以及神经网络两种定位算法的基本原理,并根据实际系统监测需求对两种算法进行改进;随后,重点介绍了冲击载荷在线实时监测系统的软硬件设计,该系统选择STM32F407为主控设备,拥有小型化便携式(179.5mm×140mm)、低功耗、低成本特性。最后,通过对实际航空结构某飞行器机翼后缘模型进行冲击实验,分别运行两种算法,以此验证监测系统的功能可行性以及比较两种算法的优劣性。实验结果表明,在600×600 mm的监测区域内,本系统能有效识别冲击载荷(互相关算法定位误差小于20mm;BP神经网络算法非边界处定位误差小于10mm),运行时间短(不超过5s),定位界面图像显示清晰。
[Abstract]:With the improvement of the performance of aeronautics and vehicles, their structures are becoming more and more complex. In the course of service, these structures will inevitably be impacted by various forms and will easily result in different degrees of damage. These damages will become a potential threat to structural safety and service life. Therefore, the research on-line real-time monitoring of impact loads in structural health monitoring technology is of great significance. Based on this urgent engineering requirement, a set of on-line real-time monitoring system based on ARM is developed in this paper. The main research work is as follows: firstly, the concept of acoustic emission technology is introduced briefly. The acquisition principle of piezoelectric sensor and the propagation characteristics of acoustic emission signal are analyzed. According to the above introduction, wavelet transform is selected as the processing technology of acoustic emission signal, and the way to obtain the arrival time by using this technique is analyzed. At the same time, a fast algorithm of non-orthogonal wavelet transform is proposed. Secondly, the basic principles of cross-correlation function and neural network are introduced, and the two algorithms are improved according to the requirement of real system monitoring. Then, the hardware and software design of the on-line real-time monitoring system for impact load is introduced. The system chooses STM32F407 as the main control device and has the characteristics of miniaturization and portable (179.5mm 脳 140mm), low power consumption and low cost. Finally, two algorithms are run to verify the functional feasibility of the monitoring system and to compare the advantages and disadvantages of the two algorithms through impact experiments on the wing trailing edge model of an aircraft with actual aeronautical structure. The experimental results show that the system can effectively identify the impact load in the monitoring area of 600 脳 600 mm (the location error of the cross-correlation algorithm is less than 20 mm). The non-boundary location error of BP neural network algorithm is less than that of 10mm, the running time is short (no more than 5 s), and the image of location interface is clearly displayed.
【学位授予单位】：南京航空航天大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：V215;TP274

【参考文献】