全自动血型分析系统关键技术的研究

发布时间：2018-09-12 19:35

【摘要】：全自动血型分析系统作为高端血液分析设备的代表，其研究涉及到许多重要的关键技术，包括运动控制技术、微量移液技术、故障诊断技术和血型图像识别技术等。因此，，解决好各个关键技术的研究工作，是实现全自动血液分析系统的前提。本文针对目前国内对全自动血型分析系统的需求，在分析目前血型分析技术和国外血型分析系统研究现状的基础上，以解决好血型分析系统各个关键技术为基础，研制了国内首台全自动血型分析系统。全自动血型分析系统的微量移液机构是一个多轴协调运动的复杂机械结构。本文研究了微量移液机构的机械结构、控制策略、系统模型、单轴运动控制器、多轴运动控制器和基于DSP芯片、FGPA芯片的运动控制主板。由于采用了改进的单神经元PID速度控制器和改进的模糊PID位置控制器，弥补了传统PID算法的不足，单轴的定位精度可达0.05mm，最大运动速度可达1.6m/s，最大加速度可达6m/s2，极大的提高了微量移液机构的运行速度和稳定性，最终实现了每小时80个样品的仪器检测速度。全自动血型分析系统的微量移液技术是微量移液机构实现精确加样的前提。本文研究了一种精密微量泵的机械模型、加样精度检测、取样针堵塞识别和液面分层探测等问题。通过应力分析和模态分析，证明了柱塞选择ZrO2材料的有效性，其一阶振型最大位移小于1um。通过对微量泵加样精度检测机构的光学改进，把荧光采集效率提高了48.19%。通过分析取样针导管压力数据的幅值分布和频谱特性，有效的识别了血块堵塞和遇到泡沫两种情况。最后，采用CD4046芯片设计的液面传感器，研究了红细胞和血清分层探测时的不同导电特性。全自动血型分析系统自动化要求的实现需要分析系统具有较低的故障率。本文研究了离心机的机械结构、故障检测平台、振动信号软件滤波、特征提取和小波BP神经网络故障诊断器。通过自适应率和遗传算法对小波BP神经网络的改进，使小波BP神经网络训练的迭代次数由463代降低到了112代，故障诊断正确率由82.5%提高到了95%，有效的提高了小波BP神经网络的收敛速度和故障诊断正确率。全自动血型分析系统运行的最终目的是为了正确地识别血型结果。本文研究了血型试剂卡的倾斜校正、图像分割、种类识别和结果识别四个问题。通过采用两次匹配算法，使血型试剂卡的模板匹配时间由9.342s降低到了0.946s，有效地减少了图像处理所需的时间。课题的研究工作得到了国家高技术研究发展计划（863计划）（2011AA02A104）、苏州市科技计划项目（ZXG201135）等基金的支持。
[Abstract]:As the representative of high-end blood analysis equipment, automatic blood type analysis system involves many important key technologies, including motion control technology, micro-fluid transfer technology, fault diagnosis technology and blood group image recognition technology. Therefore, it is the premise of automatic blood analysis system to solve the research work of every key technology. This paper aims at the demand of automatic blood group analysis system in China, based on the analysis of the current blood group analysis technology and the current research status of blood group analysis system abroad, based on solving the key technologies of blood group analysis system. The first automatic blood type analysis system in China has been developed. The micro fluid transfer mechanism of automatic blood type analysis system is a complex mechanical structure with multi-axis coordinated motion. In this paper, the mechanical structure, control strategy, system model, uniaxial motion controller, multi-axis motion controller and motion control board based on DSP chip are studied. The improved single neuron PID speed controller and the improved fuzzy PID position controller are adopted, which make up for the deficiency of the traditional PID algorithm. The positioning accuracy of single axis can reach 0.05 mm, the maximum velocity of motion can reach 1.6 m / s, and the maximum acceleration can reach 6 m / s 2. The running speed and stability of the micro fluid transfer mechanism are greatly improved. Finally, the instrument detection speed of 80 samples per hour is realized. Automatic blood type analysis system of micro-liquid technology is the premise of micro-fluid transfer mechanism to accurately add samples. In this paper, the mechanical model of a kind of precision micro pump, the precision detection of adding sample, the identification of sampling needle clogging and the detection of liquid level stratification are studied. By means of stress analysis and modal analysis, it is proved that the plunger is effective in selecting ZrO2 material, and the maximum displacement of the first order mode is less than 1 um. The efficiency of fluorescence collection is improved 48.19 by improving the precision detection mechanism of micropump. By analyzing the amplitude distribution and spectrum characteristics of the pressure data of the sampling needle catheter, two kinds of blood clot blockage and foam are effectively identified. Finally, the liquid level sensor designed by CD4046 chip was used to study the different conductivity characteristics of erythrocyte and serum in delamination detection. The realization of automatic blood type analysis system requires a low failure rate. In this paper, the mechanical structure of centrifuge, fault detection platform, vibration signal software filtering, feature extraction and wavelet BP neural network fault diagnosis are studied. Through the improvement of wavelet BP neural network by adaptive rate and genetic algorithm, the iterative times of wavelet BP neural network training are reduced from 463 generations to 112th generation. The correct rate of fault diagnosis is improved from 82.5% to 95%, which effectively improves the convergence speed of wavelet BP neural network and the correct rate of fault diagnosis. The final purpose of the automatic blood type analysis system is to correctly identify blood type results. In this paper, four problems, such as skew correction, image segmentation, category recognition and result recognition, are studied. By using the twice matching algorithm, the template matching time of the blood group reagent card is reduced from 9.342s to 0.946s, which effectively reduces the time needed for image processing. The research work is supported by the National High Technology Research and Development Program (2011AA02A104) and Suzhou Science and Technology Project (ZXG201135).
【学位授予单位】：中国科学院研究生院（长春光学精密机械与物理研究所）
【学位级别】：博士
【学位授予年份】：2012
【分类号】：R197.39

【相似文献】