【作者】 段洁；

【导师】 安志勇；

【作者基本信息】 长春理工大学 ， 仪器科学与技术， 2014， 博士

【摘要】 中高空面阵CCD航空相机是用于军用飞机进行空中侦察的重要设备,其任务就是从中高空实时获取高分辨率可见光目标图像信息,处理后将侦察信息实时传输给地面控制中心,完成地面目标采集和战场目标打击效果的评估。而操纵系统是中高空面阵CCD航空相机重要的控制与检测设备,可实现对航空相机供电电源的实时故障诊断和状态监测与控制,以及完成航拍图像的并行实时无损压缩。论文结合某型中高空面阵CCD航空相机的设计方案,设计了一款满足基本侦察任务要求的航空相机操纵系统,并对所涉及到的关键技术进行了深入研究,为研制功能更强大、更完善的航空相机提供了一定的理论依据和工程实践经验。因此,开展本课题研究对促进我国航空相机事业发展,提高我国航空侦察能力具有非常重要的意义。本文针对现有通用的Linux操作系统功能模块繁多的情况,其不仅占用内存资源,而且影响执行速度和任务响应速度。为此,提出了基于PC104总线的移植与裁剪技术,根据应用需求,对Linux系统进行实用性改造,提高了系统的任务响应速度；针对航拍图像数据量大,存储困难等问题,给出了基于软硬件并行技术的图像压缩子系统设计方案,实现图像的并行、实时和无损压缩,其图像压缩时数据流的处理速率达到60MB/S,图像压缩比达到10：1；针对航空相机供电电源故障率高,传统故障诊断方法不足而使得相机维护难度大的实际问题,进行了基于PSO-RBF神经网络的故障诊技术研究,设计了基于PSO-RBF的航空相机电源的故障诊断系统,实现了对航空相机供电电源的实时、全自动化的故障诊断。其研究的主要内容有：在全面分析某型中高空面阵CCD航空相机工作原理基础上,论述了航空相机的系统组成。依据航空相机操纵系统的性能指标要求,进行了航空相机操纵器系统的总体方案设计,设计了操纵系统的硬件电路模块,并分析了其功能；给出了航空相机操纵系统软件模块的构成,设计了操纵系统总体工作流程；针对系统所涉及到的关键技术,即基于PC104总线的Linux系统的裁剪技术,图像实时无损压缩的软硬件并行技术和基于PSO-RBF神经网络的故障诊断技术,提出了航空相机操纵系统的相关技术指标,并进行了深入研究。(1)研究了基于PC104总线的Linux系统的裁剪技术.研究了Linux裁剪技术与方法,选择了操纵系统的主控计算机主板,分析了PC104主板的硬件资源。针对其硬件资源进行了Boot Loader移植,将移植后操作系统采用粗粒度和细粒度的内核裁剪方法进行了实用性改造,增强了Linux内核的抢占性；通过改善Linux内核实时调度器的调度策略,构建了一个具有实时处理能力的嵌入式系统,以满足系统的实时性需求。实现了任务响应时间不大于20ms,系统启动时间不大于5s。(2)研究了图像实时无损压缩的软、硬件并行技术在分析图像压缩基础知识和小波变换原理的基础上,研究了适用于图像无损压缩9/7-M小波提升方法,给出了航空相机操纵系统图像并行无损压缩子系统的总体设计方案,分别设计了图像并行无损压缩子系统的软硬件。采用DSP+FPGA结构,将一幅4056×5356的大图用两块结构完全相同的图像压缩板并行处理；采用三级整数提升小波变换技术,对小波变换的高频和低频系数分别进行了编码,并对编码数据进行打包处理,减少了对内存的访问时间。实现了航拍图像的并行、实时和无损压缩,其图像压缩时数据流的处理速率达到60MB/S,图像压缩比达到10：1。(3)研究了基于PSO-RBF神经网络的故障诊断技术。在分析粒子群优化算法基本原理基础上,建立了RBF神经网路模型,并提出了PSO-RBF神经网络算法,依据此算法,设计了基于PSO-RBF神经网络的航空相机供电电源的故障诊断子系统方案。搭建了故障诊断子系统的硬件平台,编写了PSO-RBF算法程序。利用己搭建的软硬件平台对航空相机供电电源进行测试,结果表明在不依赖标准设备和附加测试点的条件下,可实现航空相机供电电源的实时、全自动化芯片级的故障诊断,并可自动完成故障定位。故障现象的检测覆盖率达到了100%,准确率达到了95%以上,故障元器件定位率达到98%。更多还原

【Abstract】 As an important equipment of aerial recommaissance for military aircraft, Medium-high altitude array CCD aerial camera is used to capture the high resolution visible light image information,after images processing,the reconnaissance imformation can be transmissted to the groun control center, such the evaluations about the targets acquisition and the effect of battlefield targets strike are accoomplished. The operating system is the best equipment of above camera for control and detection, which can diagnose faults, monitor condition and contrl real-time the power supply of aerial camera, and realizes aerial images lossless compression real-time by parallel trammission.Combined with the design of a certain type of medium-high altitude array CCD aerial camera,the aerial camera operating system met requirements the basic reconnamissance mission,which of the technologies researched in deeply further.that provides the theory basis and the practical experiences for aerial camera with the more powerful and perfect performance.Therfore,It is the very important significance to promote the aviation camera business and improve the aciation reconnaissance capability by this paper research.In view of problem of function module variety in the existing general Linux operating system, which not only memory resources oppupied, but also the spped of execution and response speed influenced.Therefore, the transplantation and cutting technology is presented based on PC104bus, and according to the application demand, the Linux system is carried on a practical transformation for improving the response speed to the system task. In view of memory problem because of large image date,the design of image compression subsystem based on hardware and software prarllel technology is given that can achieve image parallel,real-time and lossless compression.The processing rate of data stream and the image compression ratio can reach60MB/S and10:1during the image compression processing. In order to solve the problems of the high failure rate on the aerial camera power supply resulting from the camera maintenance difficult applying of the traditional fault diagnosis methods, the research of fault diagnosis technology is put forward based on PSO-RBF neural network.The fault diagnosis system of the aerial camera power supply based on PSO-RBF is designed to realize the aerial camera fault diagnosis in real-time, full automation.The main research contents include:In a comprehensive working principle analysis of a certain type of medium-high altitude area array CCD aerial camera, the system of the aerial camera is discussed. According to the performance requirements of aerial camera control system, the overall design of aerial camera manipulator system is given.The hardware circuit module of operation system is designed and its functions are analyzed. The software module design of aerial camera control system is given too.The system overall workflow is presented.In view of the key technologies, namely the Linux system cutting technology based on PC104bus, the software and hardware real-time parallel imagelossless compression technology and the fault diagnosis technology based on PSO-RBF neural network, the relevant technical indexes of aerial camera operating system are put forward, and nd conducted in-depth study.(1) The cutting technology on the Linux system based on PC104bus is researched.The cutting technology and method on the linux system is researched.The master control computer board of operating is selected. The hadware resources of PC104motherboard is analyzed. After Carrying on the Boot Loader transplantation for the hardware resource of system, this system is practical transformated through the method using coarse-grained and fine-grained to enhance the preemption of the Linux kernel.By improving the scheduling strategies of kernel real-time scheduler, the embedded system with the capabilities of real-time processing is constructed to meet the needs of system real time. Through above job,the system can reach the task response time<20ms, startup time≤5S.(2) The real-time imge lossless compression and soft-hardware parallel technology are researched.Based on the analysis of based knowledge about image compression and the theory of wavelet transform,the overall design scheme of the parallel image conmpressin subsystem of aerial camera operation system is given after the studing9/7-M wavelet lifting method applied to image lossless compression.The software and hardware is designed of this system.Using DSP+FPGA structure,size of4056×5356large image using two pieces of the same structure compression plate is parallel compressed.The high and low frequency coefficients of wavelet transform are coded respectively by the technology based on the three stage integer lifting.The coding date is packed to reduce the memory access time.The aerial image parallel, real-time and lossless compression can be realized finally,.The compression data stream reaches60MB/S,and the image compression ratio can reach10:1.(3) The fault diagnosis technology is researched based on PSO-RBF neural network.In the analysis of the basic principle of the particle swarm optimization algorithm, the RBF neural network model is established, and the PSO-RBF neural network algorithm is put forward. A fault diagnosis system for aerial camera power supply based on PSO-RBF neural network is designed by above algorithm. To build the hardware platform of fault diagnosis subsystem, the program of the PSO-RBF algorithm is written.The power supply aerial camera is tested by above platform.The results show that this system can realized real-time,automatic chip level fault diagnosis without relying on standard equipment and additional test conditions and fault location automatically.The fault coverage rate of detection reaches100%, the accuracy rate reaches more than95%, the fault components localization rate reaches98%.更多还原