设计总说明：本论文研究的涉及到的是一个自动物料分拣机器人中的一个环节来做重点研究。在设计的全过程中主要分成四个重要的环节分别是1自动物料分拣机器人的本体设计和建模；2自动物料机器人的控制系统设计；3自动物料机器人的视觉定位技术；4自动物料机器人的视觉定位系统设计和实验。本课题将对自动物料机器人的分拣控制系统进行设计。利用虚拟样机技术对控制系统进行设计，介绍了控制系统和智能控制算法。传统的分拣系统大部分都是基于计算机的技术和控制理论的基础上建立起来的。自动物料分拣机器人的主要组成是由三个直线驱动单元和控制系统再加上末端执行器。对于伺服电机的控制采用PID的参数直接控制。而本设计的精度要求和智能化程度都比较高，顾采用基于单神经元整定PID 控制器参数来控制 PID 对伺服电机驱动控制。改善系统的动态特性，对动作变化趋势能够提前感应到与告知，得以让系统对于复杂的环境可以做到自适应性。该设计不仅做到高精度控制的同时也做到了智能化的捡取。论文主要研究的课题方向是自动物料分拣机器人的控制系统设计，目的是为了控制机器人的情却运动使用虚拟样机技术对机器人物理样机来做实验和性能测试，随时都可以对直言用的虚拟样机进行修改，等得出最好的方案在进行物理样机的制造首先设计好整套控制系统的具体分工模块。就是所谓的采用智能控制方式分工环节规划模块选择控制方式所需要的硬件种类完成控制系统的设计。选择控制方式为运动控制卡和工控机相互配的控制方式。将物理样机的模型导入到软件中建立虚拟样机本设计中用到的软件是Matlab2014a。将本课题中本体设计出的三文模型导入为虚拟模型建立虚拟样机的bitmap。控制物理样机的移动精确要求很高达到0.02mm的精度要求。在输入量的时候需要考虑到精度的要求与系统稳态的要求采用传统的PLC的精度要求很难达到所需要的精度所以排除了这一个方案选择了高计算能力的Matlab/Simulink来作为本虚拟样机的模拟软件。借用Matlab/Simulink这个软件来生成自动物料分捡器的虚拟样机模型；再利用这个虚拟样机来设计控制系统；在最后建立起整个自动物料分拣机器人的虚拟样机模型进行仿真分析。在物料分拣的控制过程中需要准确控制运动所以采用了PID控制运动控制卡然而在仿真的过程中却发现如果只是选择设定好的参数来进行控制会导致电机不能平稳精确的运行而传统的PID控制确实又和控制对象有关。对象改了PID控制器自身并不能做出调整所以需要对PID参数进行整定和寻优。神经网络的控制中自学习能力很强有能提高系统的控制精度加入了神经网络对PID参数的实时控制才每一次控制对象的更新后神经网络学习模块就可以不断地根据闭环收回的误差来自己学习调整函数参数再由PID控制器来输出新的控制率对机器人实施控制。本设计中的神经网络将会采用有导师学习规则的单神经元自学习控制PID参数的整定寻优。在设计中各个极限位置都会设置有传感器来作为边界的报警信号。以此来防止将会因为智能控制的错误带来的危险的事故。而本课题带来的设计是简而言之就是使用虚拟样机技术在虚拟仿真中利用单神经元来整定PID参数控制伺服电机的运行。37002
毕业论文关键词： 虚拟样机、神经网络、PID参数控制、导师学习规则
Design General Information：It is the focus of research related to the material to do an automatic sorting robot in a part of this thesis. In the whole design process is pided into four main important aspects are the main design and modeling an automated material sorting robot; 2 automatic control system design materials robot; 3 automatic material robot vision positioning technology; 4 automatic material Robot visual positioning system design and test.This topic will be to automatic material sorting of the robot control system design.The use of virtual prototyping technology to design the control system.It describes the control system and the intelligent control algorithm.Most of the traditional sorting system based on computer technology and established on the basis of control theory. Automatic material sorting of the robot is mainly composed by three linear drive unit plus the end of the actuator and control system. And the precision requirement of the design and intelligent degree is high .so based on single neuron setting parameters of the PID controller to control the whole system and improve the system dynamic characteristics ,Change trend of action will sensing and let us know in advance and make the system adaptability for complex environment can do. This design not only achieve high precision control and accomplish the intelligent pick up。The main research topics direction is to design the control system of automated material sorting robot, in order to control the situation, but the movement of the robot using a virtual prototype technology robot physical prototypes to experiment and performance testing, will always be able to use the virtual respect prototype modification, etc. come to the best solution during the first physical prototype fabrication designed the entire control system specific pision modules. The so-called intelligent control part of the planning pision of the control mode module hardware needed to complete the type of control system design. Select the control mode for the motion controller and IPC control mode with each other.The physical prototype model into the software to create a virtual prototype of the software used in the design is Matlab2014a. The main issue in the design of three-dimensional model into a virtual model to create a virtual prototype of the bitmap. Precision motion control demanding physical prototypes reach 0.02mm accuracy. Enter the amount of time to take into account the requirements of precision and accuracy requirements of the system homeostasis traditional PLC accuracy requirements so difficult to achieve the desired ruled out a program of selected high computing power Matlab / Simulink as this virtual prototype simulation software. Borrowing Matlab / Simulink software to generate automated material sorting the virtual prototype model; then use this virtual prototyping to design control systems; virtual prototype model of the entire automated material sorting robot simulation analysis in the final build.In the control process of sorting materials require accurate control of movement so using PID control motion control card, however in the course of the simulation was found that if just choose to set a good parameter to be controlled can cause the motor not running smoothly precise and traditional PID control does turn and control objects related. Objects changed PID controller itself does not need to make adjustments so PID tuning parameters and optimization. Control neural networks have a strong self-learning ability to improve the control precision real-time control of neural network to join the PID control parameters only once per object After updating the neural network learning module can continue to recover according to the closed-loop error to own Learning to adjust function parameters and then by the PID controller to control the output of the new rate for the robot to exercise control.The design of the neural network will use the supervised learning rule of a single neuron self-learning control PID tuning parameters optimization.Various extreme positions are provided with a sensor in the design as a boundary alarm signal. In order to prevent wrong it will be because of intelligent control the risks of accidents.And this project is designed to bring in a nutshell is the use of virtual prototyping technology uses a single neuron in a virtual simulation to tuning PID control servo motor operation. 自动物料分拣机器人控制系统设计+Matlab源程序+图纸:http://www.751com.cn/jixie/lunwen_35649.html