【作者】 徐鹏；

【导师】 刘春宝；

【作者基本信息】 吉林大学 ， 机械设计及理论， 2021， 硕士

【摘要】 随着我国人民对优质医疗服务的需求日益增加,众多医院超声检查科室长期处于高负荷工作状态。体位变换机器人能够缩短每例患者,尤其是失能人群的超声检查体位变换时间,减轻医师的工作负担,优化超声室的就医环境。为满足上述需求,在吉林省重点科技研发项目“多功能仿生智能超声检查体位变换装置研发”(20180201043YY)的资助下,本文旨在开发一种具有智能体位角度及动作参数预测功能的智能超声检查体位变换机器人,并对其机械系统、控制系统及软件系统进行研究,具体的研究内容及工作如下:(1)智能超声检查体位变换机器人的结构设计与分析同吉大一院超声科合作,进行人体姿态分析,获取超声检查体位变换的临床数据,并基于数据对体位变换模式及传动、执行机构展开研究。对超声检查体位需求进行分析,设计具有辅助患者起背、仰卧、曲腿、侧翻、侧倾、升降、移位、转体功能的机器人整体结构。(2)体位变换机构的运动学、动力学分析及参数优化运用ADAMS与MATLAB联合仿真方法对起背、曲腿机构进行了运动学及动力学分析。以机构原动件受力最小与执行机构加速度最小为优化目标、杆长为因素,机构的关键铰点的空间坐标为约束条件,基于Insight(ADAMS)模块对上述连杆机构进行试验设计,优化连杆参数。随后基于MATLAB运用遗传算法再次对上述连杆机构进行优化求解,通过交叉比对验证了试验设计及优化结果的正确性。在满足空间要求的前提下,上述工作将实现执行机构的平稳运动及原动件减载。(3)智能超声检查体位变换机器人硬件开发与集成阐述硬件系统的整体架构及模块选型,确定以树莓派为上位机,Arduino为控制器,基于蓝牙通讯的技术方案。从需求出发,基于定时计数器的原理阐述其在生成PWM信号、变频信号以及采集Hall脉冲的扩展应用。最后对硬件系统的驱动器及传感器模块进行原理解析以及接口分析,为后续上位机软件及控制器程序编写提供研究支撑。(4)智能体位变换机器人控制及软件系统开发进行上位机软件及控制器程序研究。从功能需求出发,分别对上位机软件及控制器程序进行总体设计,着重阐述上位机GUI程序、体位变换逻辑程序及基于BP神经网络的超声检查最优位姿角度预测程序设计。针对神经网络的数据训练需求,提出离线3D扫描体位数据采集方法。首先基于MATLAB通过对体位数据进行分析、处理、人工神经网络拟合,揭示患者性别、身高、体重等基本数据与进行超声检查体位变换过程中的侧翻、屈髋、曲腿角度之间存在映射关系。经测试人工神经网络模型拟合效果良好,测试样本的最小均方差为157.6014,R值为0.922。随后对BP神经网络位姿角度预测程序的C++实现展开研究,并以样机的形式对上述研究进行功能层面的验证。超声体位变换机器人通过合理设计机构及床面,实现辅助人体进行多达7个体位的相互切换。创新设计一种非强迫挤压式侧翻方案,实现了无人辅助下的失能患者的被动侧翻、侧倾,相较于其他具有侧翻身功能的医疗辅助装置,本研究所设计侧翻过程舒适性及安全性均有提升。首次提出基于神经网络的智能体位参数预测,借助神经网络的非线性拟合特性,实现了超声检查过程的体位变换参数精准预测。通过临床的长期应用,机器人的神经网络控制程序能够不断学习,进而提升体位变换的准确性。超声检查体位变换机器人其机械系统、硬件系统及软件系统经第三方检测,能够协调运行,满足安全性、快速响应及精准体位变换的技术要求。更多还原

【Abstract】 With the increasing demand for high-quality medical services in our country,many hospitals’ ultrasound inspection departments have been under heavy workload for a long time.The posture transformation robot can shorten the time of ultrasound inspection posture transformation for each patient,especially for disabled people,reduce the workload of physicians,and optimize the medical environment in the ultrasound room.In order to meet the above needs,under the funding of Jilin Province’s key scientific and technological research and development project "Multifunctional Bionic Intelligent Ultrasonic Inspection Body Transformation Device"(20180201043YY),this article aims to develop an intelligent ultrasonic inspection with intelligent body position angle and motion parameter prediction functions.Posture transformation robot and research its mechanical system,control system and software system.The specific research content and work are as follows:(1)Structural design and analysis of intelligent ultrasonic inspection robot for body position transformationCooperate with the Ultrasound Department of the First Hospital of Jilin University to analyze the human body posture,obtain the clinical data of the body position change of the ultrasound inspection,and conduct research on the body position change mode,transmission and actuator based on the data.Analyze the position requirements of the ultrasound inspection,and design the overall structure of the robot with the functions of assisting the patient to get up,lie on the back,bend legs,roll over,tilt,lift,shift,and rotate.(2)Kinematics,dynamics analysis and parameter optimization of body position change mechanismUsing ADAMS and MATLAB co-simulation method,the kinematics and dynamics analysis of the kick and bent leg mechanism are carried out.Taking the minimum force of the original moving parts of the mechanism and the minimum acceleration of the actuator as the optimization objective,the rod length as the factor,and the spatial coordinates of the key hinge points of the mechanism as the constraint conditions,the above-mentioned linkage mechanism is designed and optimized based on the Insight(ADAMS)module.Connecting rod parameters.Then,based on MATLAB,GA algorithm was used to optimize the abovementioned linkage mechanism again,and the correctness of the experimental design and optimization results was verified by cross-comparison.On the premise of meeting the space requirements,the above work will realize the smooth movement of the actuator and the load reduction of the original moving parts.(3)Hardware development and integration of intelligent ultrasonic inspection robot for body position transformationExplain the overall architecture and module selection of the hardware system,and determine the technical solution based on Bluetooth communication with Raspberry Pi as the host computer,Arduino as the controller.Starting from the demand,based on the principle of the timing counter,it explains its extended application in generating PWM signals,variable frequency signals and collecting Hall pulses.Finally,the principle analysis and interface analysis of the driver and sensor modules of the hardware system are carried out to provide research support for the subsequent programming of the software and controller of the upper computer.(4)Intelligent body position transformation robot control and software system developmentResearch on the host computer software and controller program.Starting from the functional requirements,the overall design of the upper computer software and the controller program is carried out,and the upper computer GUI program,the posture transformation logic program and the BP neural network-based ultrasonic inspection optimal pose angle prediction program design are emphasized.Aiming at the data training needs of neural networks,an offline 3D scanning body data acquisition method is proposed.First,based on MATLAB,through analyzing,processing,and ANN fitting of body position data,it reveals the mapping relationship between the basic data of patient’s gender,height,weight,etc.and the angle of roll,hip flexion and leg bending in the process of ultrasound inspection.The fitting effect of the tested ANN model is good,the minimum mean square error of the test sample is 157.6014,and the R value is 0.922.Subsequently,the C++ implementation of the BP neural network pose and angle prediction program was studied,and the above research was verified at the functional level in the form of a prototype.The ultrasonic body position transformation robot realizes the mutual switching of up to 7 body positions by assisting the human body in the rational design of the mechanism and the bed surface.Innovatively designed a non-forced squeeze rollover scheme,which realizes the passive rollover and rollover of disabled patients without assistance.Compared with other medical auxiliary devices with rollover function,this research designed rollover Process comfort and safety are improved.For the first time,the intelligent body position parameter prediction based on neural network is proposed.With the help of the nonlinear fitting characteristics of neural network,the accurate prediction of body position transformation parameters in the ultrasound inspection process is realized.Through longterm clinical application,the neural network control program of the robot can continuously learn,thereby improving the accuracy of posture transformation.The mechanical system,hardware system and software system of the ultrasonic inspection body transformation robot have been tested by a third party and can operate in coordination to meet the technical requirements of safety,rapid response and precise body position transformation.更多还原