| The direct torque control (DTC) of three-phase induction machine has two typical control strategies: switching-table-based DTC (ST-DTC) and direct torque controlled space vector modulated (DTC-SVM). ST-DTC has received a lot of attention from scholars around the world, which is characterized by fast dynamic response, structural simplicity and strong robustness. However, it has several drawbacks, such as large torque ripple and difficulties in control at low speeds. Most problems can be solved by DTC-SVM under adequate decoupling. Nevertheless, it is hard to achieve strong robustness to time-variant disturbance and unmodeled dynamics in DTC-SVM.Based on the review of domestic and overseas literatures, this dissertation proposed a new solution of DTC. Considering the nonlinearity and strong coupling characteristics of induction machine, the effects of uncertainty owing to imperfect modeling and parameter variation will be treated as perturbances and handled by ADRC. Some key technologies, including control strategy, parameter optimization, flux observation, stator resistance perturbation estimator and digital control are researched in detail to improve the performance of DTC system.ADRC has been developed into two forms in recent years. Combining the characteristics of transmission system, the profound theoretical analysis and study on the two forms of ADRC has been made in this dissertation, and the necessary conditions for the stability of two-order ADRC are derived based on real shift theory and Routh criterion.The performance of control systems are closely related to the design of control strategies. This dissertation focuses on control strategy design and performance improvement of DTC-SVM systems. Firstly, the classical ESO is combined with discontinuous projection modificaiton so as to be applicable with systems under time-variant disturbance. Then, the speed/flux controllers are developed by using the improved ADRC, with the stability and stability conditions both addressed in the Lyapunov stability sense. Moreover, a modified adaptive genetic algorithm with linear constraints is proposed to simplify the parameter adjustment of ADRC. Accurate flux observation is critical to high performance torque control. In this dissertation, a new stator flux observer based on the model-optimized ADRC is developed to improve the observing performance at low speed.The sensitivity to machine parameters is analysed according to the amplitude-frequency and phase-frequency response. Simulation results show better performance under different speed and load situations.The accuracy of flux observation is sensitive to stator resistance, which is, however, hard to determine in advance due to temperature variation, skin effect and other effects. To deal with this problem, the equivalent control technique in sliding-mode control and discontinuous projection modification are incorporated to construct an improved equivalet-compensation type ESO (ECESO), such as to reduce the lumped uncertainty in flux observation. Furthermore, a closed-loop flux observer based on ECESO is proposed to reject the disturbance due to stator resistance variation in stator current measurement. The stability of the flux observer is proved in the Lyapunov sense. Analysis on its sensitivity to stator resistance is also established. Results show that good robustness to stator resistance variation can be achieved in full speed range.A fully digital experimental platform based on a TMS320F2812 digital signal processor is established. All the proposed schemes using ADRC are applied with several test conditions to further demonstrate the performance. Experimental results reveal the effectiveness of the proposed schemes.
|
PDF Full Text Request