Research On Active Disturbance Rejection Control Of The Servo System Of The Hand-held Stabilized Pan Tilt

The hand-held stabilized pan tilt is a kind of photographic auxiliary equipment to keep the camera's Los stable,and the servo system control is the basis of the pan tilt.There are many uncertain factors in the pan tilt servo system.When the system is affected by external air flow,fuselage jitter,friction moment,load change and other internal and external disturbances,the system has time-varying and nonlinear characteristics.These uncertain factors will restrict the performance of the pan tilt.Therefore,it is particularly important to design a good control system.At present,most of the pan tilt control systems are based on the classical PID control,but in the complex working environment,the PID control has the defects of large overshoot and poor immunity.Therefore,this paper mainly studies several active disturbance rejection control(ADRC)algorithms to estimate and compensate the uncertain factors.The main research work is as follows:1.If the extended state observer of ADRC fails to accurately estimate the disturbance,there will be a big gap between the system compensated by disturbance and the "series integral standard" system,and the performance of the conventional nonlinear state error feedback control will be poor.Therefore,this paper proposes a self disturbance rejection control method based on RBF neural network with additional inertia term.The nonlinear state error feedback control law in ADRC is improved by using single neuron and RBF neural network with additional inertia term.The self-learning ability of neural network is used to improve the self-adaptive ability of ADRC,and the performance stability control of PTZ servo system is realized.2.In view of the time-consuming and laborious problem in the process of parameter adjustment of ADRC,this paper proposes an ADRC method based on BP neural network.In this paper,a single BP neural network is designed to automatically adjust five key parameters of ADRC on-line at the same time,and a method based on sampling step length is proposed to select the gain coefficient of the output layer of the neural network.The key parameters of the extended state observer and the nonlinear state error feedback in ADRC are self-tuningoptimized at the same time,and are successfully applied to the PTZ servo system Under control.3.In order to overcome the influence of uncertain factors on the pan tilt servo system,this paper proposes a sliding mode control method based on tracking differentiator and RBF neural network approximation disturbance.Firstly,the uncertainty of the system is regarded as the total disturbance,and a sliding mode controller is designed.Then,based on Lyapunov stability analysis theory,RBF neural network is designed to approximate the total disturbance,and a sliding mode control law with feedforward and disturbance compensation is designed.By improving the time limited convergence third-order Tracking differentiator,the problem of feedforward signal extraction in the control law is solved,and the servo system of PTZ is realized Unified control.The simulation results show that the above control methods achieve the stable and precise control of the pan tilt servo system.Compared with the traditional PID control,conventional active disturbance rejection control,sliding mode control and other control methods,the above improved algorithms have better adaptive ability,better response to the impact of uncertain factors.It has the advantages of good rapidity,high control accuracy,strong robustness and immunity.Therefore,several control algorithms studied in this paper have better theoretical reference value and practical application value.