Synthesis of a parametrically invariant servo drive using the model parameters recovery method

   RELEVANCE. Servo drives operate with a law of change of the setting action unknown in advance and provide reproduction of this law by the output coordinate. Servo drives find application in robotic and mechatronic systems, machine tools, systems of automatic control and remote transmission of information, radar stations, guidance units, etc. The operation of servodrives often proceeds in conditions of instability parameters and characteristics elements of the electric drive. Corrective devices synthesized by classical methods of automatic control theory cannot cope with providing the specified accuracy of reproduction of the input signal and the required quality of transients.   THE PURPOSE. In this regard, an important and urgent task is the synthesis of an active correction system with a non-stationary controller that provides the required quality and accuracy of the control process due to the coefficient self-tuning algorithm.   METHODS. When solving this problem, methods for identifying parameters based on the gradient algorithm and numerical integration of the object of study dynamics equations, implemented by means of the MatLab software environment, were used.   RESULTS. The paper solves the problem of synthesizing the self-tuning algorithm for the coefficients of the servo drive corrective device based on the identification approach. The parameters are identified by a searchless gradient algorithm while minimizing the discrepancy between the object of study and its inverse model, as well as restoring the coefficients of differential equations using integration and the corresponding computational procedures. An servo drive with negative position feedback is tuned to the modular optimum with a proportional controller whose coefficients are completely determined by the parameters to be identified. The self-tuning algorithm consists in calculating the correction factor of the non-stationary P-controller and forming a multiplicative channel of the active correction closed loop.   CONCLUSION. The simulation of the electric drive in the MatLab software environment showed high accuracy and quickness of the process identifying parameters in a wide range of their change. When forming an active correction contour, a necessary requirement is to distinguish between the identification cycle and the self-tuning cycle. This makes it possible to avoid singular perturbations and reduce resonant facts during the operation of a parametrically invariant electric drive. The developed method of active correction with a priori known and unchanged structure of the object model of study makes it possible to maintain the required accuracy and quality of the operation of the electric drive under conditions of parametric disturbances with permissible deviations of accuracy and quality indicators. Implementation of the method does not require additional equipment, organization of special test signals, significant computational costs. The method of synthesizing a parametrically invariant electric drive can be used to develop robust control systems for non-stationary objects, including when the hypothesis of quasi-stationarity is not fulfilled.