Qualifying scientific work on the rights of the manuscript. The dissertation on competition of a scientific degree of the candidate of technical sciences on a specialty 05.09.03 – electrotechnical complexes and systems – Dnipro University of Technology, Ministry of Education and Science of Ukraine, Dnipro, 2021.
The dissertation deals with the actual scientific problem of the development of energy-efficient control laws of coordinates of asynchronous electric drive operating in modes with changing loads and speed setpoint.
The author proposes two modifications of conventional vector-controlled induction motor drive. For the first time, it is proposed to use a modified nonlinear gradient-based model predictive control toolbox GRAMPC instead of a flux controller in a field-oriented control scheme. This modification ensures a control law for transferring the electromechanical system from one operating point (torque, speed) to the desired point following the minimum energy loss criteria in real-time, taking into account possible changes in the desired future state of the electric drive as a result of the influence of the input reference signal and as a result of the influence of change in load (including simultaneously). The optimal control problem is defined as the minimization of the time integral of the energy losses with constraints. To this end, the expression according to Hamilton's method is defined and first-order optimality conditions are determined based on Pontryagin’s Maximum Principle. The effect of the model algorithmic parameters: prediction horizon, the maximum number of iterations, numerical integration method, and the number of data points is considered and default values in terms of real-time demands are determined. Utilizing the multiple-criteria decision-making approach, it is shown that by combining the number of gradient iterations with the number of discrete horizon intervals, the desired accuracy and speed of precalculating the loss-minimizing trajectory of field-producing current is achieved.
The second proposed modification is a simple technique for sub-optimal online loss minimization achieved through the use of the law of the rotor flux generation augmented with adaptive low-pass filtering of the flux reference at each sampling step. It is shown that by appropriately choosing the filter time constant as a fraction of the rotor time constant the instantaneous power losses after a load torque step can be significantly reduced compared to the standard case. The analysis for the appropriate choice of the filter time constant is based on a numerical study and modeling for three different induction motors with different rated powers.
A laboratory testbench was created for experimental research of the closed-cycle operation of a 370-W field-orientation induction machine in dynamic behavior when load conditions are changing, considering the nonlinearities of the main inductance. Both the steady-state and dynamic performance of the proposed methods is investigated. The developed control laws were implemented in hardware using the code generation technology directly from MATLAB/Simulink environment for execution in the dSPACE real-time controller. Handling real-time applications are made in ControlDesk experiment software for seamless ECU development. The results obtained during experimental studies confirm the reliability and high accuracy of the results of analytical calculation and modeling, as well as the adequacy of the proposed mathematical model of asynchronous machine and control systems. Additionally, the comparison of measurement results with conventional methods is provided, which validates the advantages and performance of the control schemes.
Keywords: induction motor, field-oriented control, power losses, energy-efficient control, dynamic mode, adaptive filtering, model predictive control, real-time implementation, multiple-criteria decision-making
