医用高压氧舱氧气浓度控制装置的设计
发布时间：2019-05-09 01:04
【摘要】：高压氧医学已经是一门比较成熟的学科,并且已经被广泛应用在临床医学上。目前新建的大型氧舱中有很多已经采用了计算机控制技术,实现控制过程中的自动控制。但是此项控制技术还只是停留在比较浅显的PID控制的层面,并没有将计算机的潜在的巨大的能量发挥出来。随着神经元网络技术及模糊控制的发展,我们可以尝试着研发出新型的控制技术,以便使高压氧舱的工作效率达到最佳值。本文对高压氧舱中氧气的浓度控制做了深入的研究,完成了几项主要的工作：(1)对高压氧舱中的氧气系统建立了动态数学模型,根据数学模型可以总结出氧气的放散与制氧机参数之间的关系。(2)对氧气的短期负荷进行了研究,确定了人工神经网络的结构及算法,选择了历史数据对网络进行了训练,得到了令人满意的训练结果；(3)对高压氧舱的高压环境内氧气的浓度进行控制,最终确定了了模糊控制算法可以使输出结果达到较好的控制精度；(4)对高压氧舱所用的氧气传感器进行了故障检测,具体的实现是采用了人工神经网络的算法,经过反复多次训练,取适当的学习系数,可使均方误差达到最小。试验与实践证明,该网络具有良好的收敛性和稳定性。
[Abstract]:Hyperbaric oxygen medicine is a mature subject and has been widely used in clinical medicine. At present, many of the new large oxygen tanks have adopted computer control technology to realize automatic control in the process of control. However, the control technology is still in the superficial level of PID control, and does not bring the potential of the computer into full play. With the development of neural network technology and fuzzy control, we can try to develop a new control technology in order to optimize the working efficiency of hyperbaric oxygen chamber. In this paper, the control of oxygen concentration in hyperbaric oxygen chamber has been deeply studied, and several main tasks have been completed: (1) the dynamic mathematical model of oxygen system in hyperbaric oxygen chamber has been established. According to the mathematical model, the relationship between oxygen release and oxygen generator parameters can be summarized. (2) the short-term load of oxygen is studied, the structure and algorithm of artificial neural network are determined, and the historical data are selected to train the network. Satisfactory training results have been obtained; (3) to control the oxygen concentration in the high pressure environment of the hyperbaric oxygen chamber, and finally determine the fuzzy control algorithm which can make the output result reach a better control precision; (4) the fault detection of oxygen sensor used in hyperbaric oxygen chamber is carried out. The algorithm of artificial neural network is adopted. After repeated training and appropriate learning coefficient, the mean square error can be minimized. The experiment and practice show that the network has good convergence and stability.
【学位授予单位】：长春工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：R459.6;TP273
本文编号：2472359
[Abstract]:Hyperbaric oxygen medicine is a mature subject and has been widely used in clinical medicine. At present, many of the new large oxygen tanks have adopted computer control technology to realize automatic control in the process of control. However, the control technology is still in the superficial level of PID control, and does not bring the potential of the computer into full play. With the development of neural network technology and fuzzy control, we can try to develop a new control technology in order to optimize the working efficiency of hyperbaric oxygen chamber. In this paper, the control of oxygen concentration in hyperbaric oxygen chamber has been deeply studied, and several main tasks have been completed: (1) the dynamic mathematical model of oxygen system in hyperbaric oxygen chamber has been established. According to the mathematical model, the relationship between oxygen release and oxygen generator parameters can be summarized. (2) the short-term load of oxygen is studied, the structure and algorithm of artificial neural network are determined, and the historical data are selected to train the network. Satisfactory training results have been obtained; (3) to control the oxygen concentration in the high pressure environment of the hyperbaric oxygen chamber, and finally determine the fuzzy control algorithm which can make the output result reach a better control precision; (4) the fault detection of oxygen sensor used in hyperbaric oxygen chamber is carried out. The algorithm of artificial neural network is adopted. After repeated training and appropriate learning coefficient, the mean square error can be minimized. The experiment and practice show that the network has good convergence and stability.
【学位授予单位】：长春工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：R459.6;TP273
【参考文献】
相关硕士学位论文 前2条
1 陈传虎;基于自适应神经模糊推理的传感器在线故障检测与预测[D];苏州大学;2004年
2 苏小红;基于人工神经网络的燃气短期负荷预测研究[D];重庆大学;2005年
，本文编号：2472359
本文链接：https://www.wllwen.com/huliyixuelunwen/2472359.html
最近更新
教材专著
