Research On Inverse Kinematics Problems Of Transmission Line De-icing Robot

The power grid hazards of ice storm are that an increase in snow or ice loads may result in some harm phenomenons,such as the power line dance and contamination flashover.The loads exceed the specified design level may lead to the power tower deformation,even collapse,and finally cause the structural damage of the entire power grid.During all power grid accidents,the power line accidents have the most serious impact on poser grid.At present,the main de-icing methods for transmission lines include hot melting method and mechanical ice breaking method.The former removes the ice by increasing the line current.These hot melting methods are very efficient.However,its energy consumption and operating costs are also very considerable.The disadvantages of latter are high-risk,labor-intensive,low efficiency.In this paper,a novel method of de-icing robot is proposed for overhead high-voltage transmission line de-icing.The advantages of de-icing robot are low power comsumption,high efficiency,low-risk,continuous online operation.Compared with the traditional de-icing methos,this new de-icing method has the obvious advantages.To ensure the normal work of de-icing robot,mainly depends on the stability of robot.The transformation of the position and orientation of a manipulator end-effector from Cartesian coordinates to joint coordinates is known as the inverse kinematics problem.The problem is important for the robot mechanical design and optimaization,robot trajectory planning and motion control,virtual reality and robot visual servo control technologys.So,this dissertation focuses on the methods for solving the inverse kinematics of de-icing robot.Main results and contributions of this dissertation are as follows:Firstly,the power grid hazards of ice storm,the common de-icing methods and its shortcomings are introduced.A new method of de-icing robot is proposed.Then,the research progress of the high-voltage transmission line robots and the inverse kinematics problems of series robot are generalized.As the de-icing robot structure with three manipulators is complex,it is very hard to build an accurate kinematic model.In charpter 2,a time-sharing control strategy is proposed.That is,the individual manipulator moves sequentially.Now,the kinematic model of de-icing robot can be simplified as a series robot model.On this basis,a link coordinate system for de-icing robot is established and the link parameters are available.At the same time,an optimization problem which is described by the errors between the current and the desired locations of the end-effector is built for solving the original inverse kinematics problem.In this charpter,the analytical of the gradient vector and an effective method to approximate the optimum search step size are developed.As a result,the computation load of the gradient-based optimal methods is significantly reduced.In summary,the general formula can be applied to solve the inverse kinematics problem of serial manipulators with any degree of freedom.The derivative-based optimal methods,such as Newton iterative method,are one of the most important methods for solving the inverse kinematics problem.Although the analytical form of the gradient vector and an approximate the optimum search step size had been developed in charpter 2.However,since the Jacobian matrix and many vector operations are used,the computational process is more complex.To reduce the computational complexity,a modified DFP algorithm was developed in chartper 3.Unlike the traditional DFP algorithm,this modified DFP algorithm randomly chooses the search step size between 0 and 1.Thus the computational complexity is greatly reduced,and more importantly,this method is not only computationally efficient but also numerically stable.Moreover,in order to improve the search efficieny of the initial value,this charpter presented the application of electromagnetism-like(EM)algorithm to solve inverse kinematics of robots.The original inverse kinematics problem can be transformed into a function optimization problem.And accurate numerical solutions to this problem can be achieved by using optimial methods.However,these methods do not apply for the robot with unknown structural parameters.The artificial neural network,which has significant flexibility and learning ability,has been used in the inverse kinematics problem independently of the robot structre.Two issues of concern for solving the inverse kinematics with neural networks are:1)how to collect training set without knowing inverse kinematic expression.2)The slow training process,especially for a complex configuration manipulator,or a large set of training data.In charpter 4,there are two new methods that are developed for solving the shortcoming of the neural networks.The first proposed method train network under joint subspace instead of whole robot joint space.The second method uses the pre-classified samples to train network.Furthermore,a new learning algorithm called extreme learning machine(ELM)is apply to train the neural network.The results show that the proposed method has not only greatly reduced the computation time but also improved the precision.In the above charpters,the inverse kinematics problems without constraints are discussed.However,many real-world optimization problems are often constrained.For example,the de-icing robot studied in this dissertation has an additional rotational degree of freedom at the suspension point,the amout of which are decided by robot structural parameters and the amout of other joints.The former methods are usually not applicable to solve these problems.In charpter 5,a modified electromagnetism like mechanism method based on the feasibility and dominance rules is proposed for solving the inverse kinematics problems with constrains.Furthermore,these methods can also be applied for other constrained global optimization problems.In charpter 6,the key and difficult technical problems are summaried.Followly,the mechanical structure and control system components of the de-icing robot are introuduced.Finally,the main innovations of the thesis are summarized,and the fields for further research are expected.