特种起重机的蚁群寻优智能控制方法研究

发布时间：2018-01-19 01:10

本文关键词： 特种起重机模糊神经网络智能控制自适应蚁群算法　出处：《哈尔滨工程大学》2012年硕士论文　论文类型：学位论文

【摘要】：论文的研究对象特种起重机是一种工业机器人和船载起重机相结合的产物，它兼具了二者的模型结构和工作特点，是一种新型的船载智能起重设备。它由一组起重臂、旋转支承底座、回转机构以及多传感器通过依次铰接组成，完成起重机的俯仰、回转等动作。该特种起重机在工作中依靠关节产生的驱动来推动臂梁运动，，将起重臂末端所持物品准确地转移到指定地点。鉴于该特种起重机这种类似于多关节机械臂系统运动的特点，论文借鉴三关节机械臂的模型推理方法，建立了该特种起重机简化模型的运动学和动力学模型，即给出了系统的数学模型。由于该特种起重机作业于船舶上，因此它在工作时无法避免由于海浪的运动对其造成的不利影响以及其他诸多不确定因素的影响，且其非线性强，动力学行为复杂。若使用传统的基于模型的控制方法来对这样一个复杂且耦合性很强的系统实施控制，难度无疑会很大。随着模糊控制和神经网络等智能控制方法的飞速发展，这类复杂机械系统的控制问题获得了更多的解决办法。由于单纯的模糊控制和神经网络控制各有其不足之处，将这两种方法相互结合，形成一种模糊神经网络(FNN)控制器，该控制器不仅可以对信息进行推理归纳，而且能够并行处理信息，同时该控制系统还具有自学习能力和优秀的联想泛化能力，是一种可以较好的对复杂机械系统进行控制的智能控制器。本文使用FNN控制器对特种起重机的数学模型进行了动力学仿真研究，仿真结果验证了该控制器的有效性。论文最后研究将基本蚁群算法和自适应蚁群算法分别应用到模糊神经网络智能控制系统的参数学习训练中。并使用该训练好的模糊神经网络控制器对特种起重机在MATLAB环境下进行控制仿真研究。仿真结果显示，基于BP算法训练的FNN控制器要比传统的PD控制和模糊控制的控制效果好，而基于基本蚁群算法学习训练的FNN控制器参数训练的收敛速度要比BP算法快，对特种起重机的控制效果也要比基于BP算法训练的FNN控制器效果好。同时相对比于基本蚁群优化算法，改进的自适应蚁群算法具有更快的优化训练收敛速度以及更好的轨迹跟踪误差效果。论文通过仿真研究讨论了该智能控制系统的可行性以及有效性。
[Abstract]:The research object of this paper is the special crane which is the product of the combination of industrial robot and shipborne crane. It has both model structure and working characteristics. It is a new type of shipborne intelligent lifting equipment. It is composed of a group of lifting arm, rotating support base, rotary mechanism and multi-sensor by hinge in turn to complete the pitching of the crane. Rotation and other actions. The special crane in work depends on the joint of the drive to promote the movement of the arm beam. The objects held at the end of the lifting arm are transferred to the designated location accurately. In view of the characteristics of the special crane which is similar to the motion of the multi-joint manipulator system, the paper uses the model reasoning method of the three-joint manipulator for reference. The kinematics and dynamics model of the simplified model of the special crane is established, that is, the mathematical model of the system is given. Because the special crane works on the ship, it can not avoid the adverse effect of the wave motion on it and many other uncertain factors, and its nonlinear is strong. The dynamic behavior is complex. If the traditional model-based control method is used to control such a complex and highly coupled system. With the rapid development of intelligent control methods such as fuzzy control and neural network. The control problem of this kind of complex mechanical system has been solved more and more. Because the pure fuzzy control and the neural network control have their own shortcomings, the two methods are combined with each other. A fuzzy neural network (FNN) controller is formed, which can not only infer the information, but also process the information in parallel. At the same time, the control system also has self-learning ability and excellent associative generalization ability. It is a kind of intelligent controller which can control the complex mechanical system. In this paper, FNN controller is used to simulate the mathematical model of special crane. Simulation results verify the effectiveness of the controller. Finally, the basic ant colony algorithm and adaptive ant colony algorithm are applied to the parameter learning training of fuzzy neural network intelligent control system, and the trained fuzzy neural network controller is used for special lifting. The control simulation is carried out in MATLAB environment. The simulation results show that. The FNN controller based on BP algorithm is better than the traditional PD control and fuzzy control. The convergence speed of FNN controller parameter training based on basic ant colony algorithm is faster than that of BP algorithm. The control effect of the special crane is better than that of the FNN controller based on BP algorithm, and compared with the basic ant colony optimization algorithm. The improved adaptive ant colony algorithm has faster training convergence speed and better trajectory tracking error effect. The feasibility and effectiveness of the intelligent control system are discussed by simulation in this paper.
【学位授予单位】：哈尔滨工程大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TH21

【相似文献】