| Summary: | 博士 === 國立臺北科技大學 === 電機工程系所 === 99 === In this dissertation, an adaptive supervisory Gaussian cerebellar model articulation controller (ASGCMAC) is developed for speed control of direct torque control (DTC) system of induction motor. The ASGCMAC comprises a supervisory controller and an adaptive Gaussian-CMAC (GCMAC). The supervisory controller confines the tracking error within a predefined bounded range to upgrade system behavior during transient; the adaptive GCMAC learns and approximates system dynamics to smoothly dominate steady-state response. The weight memories of the ASGCMAC are adjusted on-line by the adaptation laws, which are derived by Lyapunov stability theory, to promote system dynamics while operating condition varies.
In this research, an adaptive rotor speed estimator (ARSE) and an adaptive stator resistance estimator (ASRE) are designed under the structure of adaptive stator flux estimator (ASFE), for the realization of speed-sensorless induction motor drive with real-time stator resistance identification.
For the realization of the proposed system, the ASGCMAC, ARSE and ASRE are integrated and implemented in a DTC induction motor drive. The experimental results show that, in contrast to PI control, the ASGCMAC not only demonstrates superior performance in a wide speed range (36 rpm to 1800 rpm), but also exhibits better robustness in the face of external load disturbance. For the verification of ASRE, the experimental results prove that the ASRE is capable of estimating the actual value of stator resistance while the mismatch of stator resistance occurs, and that the quality of speed estimation is improved and the robustness of system is promoted accordingly.
|
|---|
