Study On Inertial Platform Position Control Based On ADRC

Inertial platform is a high technology which concerns multiple disciplines such as theelectronics, electromechanical, computer, self-control and so on. In addition to solving thepose control and space isolation in the complex environment, the inertial platform alsoplays a crucial role in the areas of the national defense and aerospace and is a frontiertechnology which all countries are competing to developBased on three axis inertial platform as the research object, Using the principle of therotation coordinate system. the dynamics formulas of the three-axis platform arecalculated from the energy perspective. Based upon the calculated dynamics formula, it isrealized that disturbances are important issues that need to be resolved in the design ofthree-axis inertial platform control algorithm. By studying the Active DisturbanceRejection Controller(ADRC) algorithm, this article addresses the problem of platformdisturbance, specific works as follows:Fristly, for the disturbance factors such as coupling factors in the dynamics formulas,as well as uncertainties in the mathematical models and outside disturbances, ADRCcontrol algorithm is proposed. ADRC controller algorithm is to use a trackingdifferentiator(TD)tracking controller input signal, the differential signal of the input isextracted reasonnbly; By using extended state observer(ESO), the proportion signal anddifferential signal of the controlled object input are estimated, And the total perturbation ofthe system are estimated on real-time; Finally, the nonlinear state error feedback(NLSEF)is used, the nonlinear feedback of the error signal, By using the nonlinear feedback of theerror signal, the quality of the controller loop is Improved, and the output of the controlleris compensated, Simulation results show that the proposed control method can achievesatisfactory results.Secondly, In order to improve the nonlinear tracking ability and adaptive ability ofADRC controller, this paper designed an online parameter tuning controller. based onquasi-diagonal recurrent neural network. The key parameters, which are similar to theproportion coefficient and the differential coefficient in the PID algorithm, determines the output signal of the controller. The quasi-diagonal recurrent neural network (QDRNN) hasthe advantages of simple structures and less weighted connections and the connectionsbetween the hidden layer neurons ensure the nonlinear tracking ability of the network. Byusing the online identification function of the QDRNN, the key parameters are set on line.Finally, the simulation results verify the effectiveness of the designed controller.