| Industrial intelligent robot is a representative product of intelligent manufacturing,which integrates many kinds of modern technologies such as modern manufacturing,new material,intelligent control and artificial intelligence.The development,manufacturing and application of industrial intelligent robot have become an important symbol of the innovation and manufacturing level of a country.For this reason,this dissertation will take the latest version robot operating system as the reference research frame,the Delta parallel robot as application object,and use the component-based method of robot system to study the key technologies of the industrial intelligent robot.The main contents includes: kinematics and dynamics modelling,optimization of mechanism parameters,fuzzy logic system based intelligent control method,reusable motion planning method,and experiments on the intelligent control and motion planning method,in order to build an industrial intelligent robot system with high performance,low cost and easy to integration.The kinematics and dynamics model of the Delta parallel robot are built.On this basis,a method for optimizing the mechanism parameters of the robot is proposed considering control task requirements.The method contains a dynamic anisotropic optimization algorithm and a global dynamic optimization algorithm.In terms of the mechanism parameter optimization algorithm considering the dynamic anisotropic property,two objective functions are designed,which are actuating joints torque index of the whole workspace and actuating joints torque fluctuation index of the whole workspace.PSO algorithm is adopted to solve this optimization problem.The optimization results show that the two objective functions can improve the dynamic performance of the robot effectively.In order to improve the dynamic performance of the robot in the whole workspace,a multi-objective optimization algorithm is designed based on global dynamic performance index.An objective function for quantifying the robot global dynamic sensitivity is designed.Optimization results show that the optimized mechanism parameters can not only improve the global dynamic performance of the robot but also improve the global kinematic performance.These models and algorithms are programmed using component-based method,the modules can be recombined to create new mechanism parameter optimization models to meet the needs of new tasks.From the perspective of motion control of the industrial intelligent robot,the partially dependent design method for interval type-2 fuzzy logic control system is proposed and applied to the trajectory tracking control of parallel robot with internal and external uncertainties.An initial type-1 fuzzy logic controller is constructed first and then extended into the interval type-2 fuzzy logic controller via a blurring process.In the view of how to blur the type-1 fuzzy membership functions,three kinds of blurring method are studied,and the control performance of the blurred interval type-2 fuzzy logic controllers is analyzed.By analyzing the relationship between burring degree,burring method and output control surface,the best blurring method for making the controller to have the optimal nonlinear characteristics is obtained.In order to improve the control quality,the output signal enhance coefficient is introduced.The effectiveness of the design and optimization method for interval type-2 fuzzy logic control system is verified by simulations.In order to make the controller of the industrial intelligent robot have good environmental adaptability,an intelligent controller based on self-learning interval type-2 fuzzy neural network is proposed.The controller has a parallel structure: an interval type-2 fuzzy neural network and traditional PD controller.In the aspect of designing the interval type-2 fuzzy neural network,the partially dependent design approach is used to design the interval type-2 fuzzy sets.In the design process of the antecedent fuzzy sets,a double sequence symmetric trapezoid membership function arrangement is proposed.This arrangement makes the self-learning laws and the stability analysis to have an analytical form,which benefits the hardware realization of the algorithm.In the aspect of designing the self-learning laws,a learning algorithm based on sliding mode control theory is established to tune the structure parameters of the interval type-2 neural network online.The stability of the proposed control system is demonstrated using Lyapunov stability theorem.Three simulations are carried out according to the trajectory tracking control problem of the Delta robot.Simulation results show that the self-learning interval type-2 fuzzy neural network controller achieves better control performance and strong adaptability in terms of improving the tracking accuracy and the robustness under unknown system uncertainties.In order to solve the robot motion planning problem,a reusable particle swarm optimization motion planning method is proposed.This method combines the motion planning problem with the problem of reusing optimal trajectory parameters.The motion planning problem of a robot with complex constrains is studied.A trajectory characterization method based on trajectory feature point is adopted,which makes different trajectories to have a unified time and spatial scale by normalized calculation,and increases the generality of this method.A global motion planning algorithm based on PSO is proposed.Fitness functions with multiple constrains while considering the obstacle avoidance are designed.In order to improve the convergence speed,a particle initialization method based on trajectory feature point fitness is designed.In the aspect of reusing the optimized trajectories,an optimized trajectories reuse method based on optimized trajectories modification database is proposed.In order to improve the retrieval precision,a retrieval method based on enhanced feature point fitness vector is designed,which makes the retrieval vector contain more information about the trajectory feature points and the obstacles.A new data storage form of the optimized trajectory database is built using the primary key for data connection.The enhanced feature point fitness database and the optimized trajectories modification database is separated by this data storage form.A database indexing method based on vocabulary tree is established,which could improve the retrieval speed of the local database.Simulation results show that,the proposed motion planning method is fast in computation speed and stable in planning results,which can effectively handle motion planning problems with complex constrains.The industrial intelligent robot control and motion planning experimental platform is built on the basis of the Delta parallel robot platform.The experiments of intelligent control system based on self-learning interval type-2 fuzzy neural network and reusable particle swarm optimization motion planning method are designed,and the effectiveness of the intelligent control system and the motion planning method are verified on the experimental platform.The experimental results show that the intelligent control system based on self-learning interval type-2 fuzzy neural network can effectively improve the trajectory tracking precision and the system stability of the robot.The reusable particle swarm optimization motion planning method can solve the robot motion planning problem with complex constrains efficiently online. |
PDF Full Text Request