Coated in ice due to snow-fall and/or freezing rain in winter,power transmission lines sag and often break,resulting in insulator flashover,pole collapsing,communication interruption,power outages,and so on.It not only causes inconvenience to people’s life but also has harmful effect on both agricultural and industrial production.Thus,it is necessary and significant to research and develop the de-icing robot on power transmission lines,an efficient and safe deicing technique.The essence of deicing robot control is to how to attain the precise control for obstacle-crossing of deicing robot arms.However,t he de-icing robot is a complex electromechanics system and is also time-varying nonlinear system.Furthermore,the de-icing robot generally has to face external disturbances and various uncertainties in its nonlinear dynamics.It is difficult to establish an appropriate mathematical model for the design of a model-based control system.Therefore,the general claim of intelligent control methods such as fuzzy control and neural network control is that they can attenuate the effects of unknown disturbances and unstructured uncertainties by using the rule reasoning and the powerful learning capability without a detailed knowledge of the controlled plant.On the other hand,for improving the control performance of robot manipulator,the fuzzy rules and the network parameters need to be optimized in advance by optimization algorithms.In this dissertation,the fuzzy systems,the neural networks,and the optimization algorithms are integrated with traditional cont rol methods including adaptive control,robust control,proportional-integral-differential control,and sliding mode control to obtain the desired trajectory tracking performance for transmission lines de-icing robot manipulator.The main contributions of this dissertation are shown as follows:1.The first contribution is to represent an adaptive fuzzy-neural control utilizing sliding mode-based learning algorithm for robot manipulator to track the desired trajectory.A traditional sliding mode controller is applied to ensure the asymptotic stability of the system,and fuzzy rule-based wavelet neural networks are employed as the feedback controllers.Additionally,a novel adaptation of the parameters of the fuzzy rule-based wavelet neural networks is derived from the sliding mode-based learning algorithm in the Lyapunov stability theore m.Hence,the adaptive fuzzy-neural control approximates parameter variation,unmodeled dynamics,and unknown disturbances without the detailed knowledge of robot manipulator,while resulting in an improved tracking performanc e.2.This contribution designs an artificial chemical reaction optimization algorithm and neural network based adaptive control scheme for robot manipulator to attain the desired trajectory tracking.Radial basis function neural network is applied to approximate the uncertainties in robo t dynamics.The network parameters in initial stage are optimized by utilizing artificial chemical reaction optimization algorithm.The radial basis function neural network weights are determined based on adaptive tuning strategy in Lyapunov theory.Thus,the convergence and stability of whole system are guaranteed,and the tracking performance of robot manipulator is improved.3.In this contribution,a sliding mode control system based on combining chemical reaction optimization algorithm with radial basis f unctional link net for an n-link robot manipulator is proposed to achieve the high-precision position tracking.In the proposed scheme,a three-layer radial basis functional link net with powerful approximation ability is employed to approximate the uncert ainties,such as parameter variations,friction forces,and external disturbances,and to eliminate chattering phenomenon of the sliding mode control.In order to achieve the expected performance in the initial phase as well as the improved convergence rate,the parameters of radial basis functional link net need to be optimized in advance.Therefore,the initial parameters of the radial basis functional link net are optimized offline by chemical reaction optimization algorithm instead of random selection.Furthermore,the weights of radial basis functional link net are determined online according to adaptive tuning laws in the sense of a projection algorithm and the Lyapunov stability theorem to guarante e the stability and convergence of the system.4.This contribution suggests a hybrid algorithm based optimized fuzzy proportional-integral-differential(PID)control scheme for deicing robot manipulator.In this proposed method,a PID controller,whose input and output variables are the error signals and the PID controller gains,is represented in terms of fuzzy rules.A hybrid algorithm,which combines chemical reaction optimization with particle swarm optimization,is suggested to concurrently determine all of the parameters such as the PID controller gains,the widths and centers of Gaussian membership functions,and the number of fuzzy rules.Moreover,a fitness function including different performance criteria is defined based on the concept of multi-objective optimization.Thereby,the fuzzy PID controller with adaptive gains is more capable and flexible than traditional PID controller with fixed gains.5.This contribution presents an adaptive robust sliding tracking control scheme by using recurrent fuzzy wavelet functional link net for power line deicing robot manipulator.In the adaptive robust sliding tracking control system,the highlighted features of recurrent technique,fuzzy logic system,wavelet transform,and radial basis functional link net are incorporated into sliding mode control technique.A newly generalized recurrent fuzzy wavelet functional link net with improved learning capability is suggested to approximate unmodeled dynamics,parameter variations,and friction forces.An adaptive robust term is added into the adaptive robust sliding tracking control law for eliminating unstructured parts,functional reconstruction errors,and inescapable disturbances.Additionally,the weights and parameters of recurrent fuzzy wavelet functional link net are tuned by a new adaptation strategy based on the Lyapunov stability theory to ensure the boundedness of both the tracking errors and the parameter estimation errors.Thus,the whole control system is stable and robust while can result in the high-accuracy position tracking without the chattering.Lastly,the numerical simulation and experimental results of two-link robot manipulator and three-link power line deicing robot manipulator are provided to verify the effectiveness and robustness of the proposed methodologies. |