【作者】 彭伟；

【导师】 周孑民； 黄学章；

【作者基本信息】 中南大学 ， 热能工程， 2007， 硕士

【摘要】 喷嘴是构成发动机的一个主要元件,其性能优劣直接影响发动机的运行效率和安全性。在一定压力和温度下喷嘴的喷雾锥角和喷雾分布不均匀度是衡量喷嘴性能的重要指标。长期以来,一直采用人工目测法检测喷嘴的这两项性能,劳动强度大、效率低。因此,开发出具有自动化水平高、检测方便的试验器非常重要。本文以涡扇离心式喷嘴为研究对象,综合比较各种现代检测方法,并选用了基于计算机视觉技术的智能分析方法,成功地检测出喷嘴的喷雾锥角和分布不均匀度。首先,以Fluent 6.2为平台,对喷嘴出口的喷雾场进行了数值仿真,仿真的结果与实际情况较为一致。对如何利用计算机视觉技术自动获取喷雾锥角和分布不均匀度进行了深入的探讨。在经过增强、边缘检测、细化等基础上得到喷雾锥角的二值化图像,利用经典的Hough变换获取喷雾边界线,从而识别出喷雾锥角;运用视频帧差提取出燃油收集器中液位的大概位置后,采用互相关测度法得到更准确的液位信息,通过标定好的数据计算出对应的分布不均匀度参数。实际测试数据表明这种非接触式检测方法有着较高的测量精度。为了保证检测时系统有着稳定的压力,考虑到被控量的特性和现场的环境,采用了基于BP神经网络整定的压力PID控制,并对控制系统进行了MATLAB仿真。实际运行表明,这种控制方案的效果优于常规PID控制。最后,完成了试验器的电控喷油系统、计算机监控系统和图像采集分析系统的设计,并结合虚拟仪器技术开发出相应的监控和图像处理软件。更多还原

【Abstract】 As a key element of engine, the performance of nozzle has directly effects on engine’s operating efficiency and security. The atomization angle and uniformity tested under definite pressure and temperature are two most important indexes of nozzle performance. For a long time, these two indexes are always measured by eyeballing, which result in high labor intensity but low efficiency. Therefore, it’s extremely important to develop new equipment with high level automation and convenience.In the dissertation, fanjet swirl nozzle was taken as research object. Several modern examination methods were compared and finally an intelligent analysis method based on computer vision technology was used successfully to obtain atomization angle and uniformity.Firstly, the spray outlet field distribution condition was simulated by Fluent 6.2, and the result was consistent with actual situation.Then, the dissertation gave a thorough discussion on how to automatically get atomization angle and uniformity using computer vision technology. After a series of dealing such as image enhancement, edge detection and edge thinning, binary image of atomization angle was obtained. Then Hough transform was used to get image borderline, from which atomization angle was identified; after getting the rough position of liquid-level using video frame, more accurate information was obtained by correlation measure method, and then atomization uniformity was calculated based on an emaciated table. The actual test data indicated this non-contact method has excellent examination precision.In order to guarantee stable system working pressure, PID controller based on BP neural network adjusting was adopted considering the property of controlled parameter and working situation. Then a corresponding simulation was carried out on MATLAB. The simulation result and actual test indicated that new control scheme is better than conventional PID controller.Finally, electronic-controlled injection system, computer control system and image processing system of nozzle tester were designed. Meanwhile, the corresponding monitoring and image processing program were developed combined with the concept of virtual instrument.更多还原