Design of Adaptive Sliding Recurrent Cerebellar Model Articulation Controller for Switched Reluctance Motor Drive Systems

• Main Authors: Si-Shin Jian, 簡士睎
• Other Authors: Shun-Yuan Wang
• Format: Others
• Language: zh-TW
• Online Access: http://ndltd.ncl.edu.tw/handle/pvu78n

碩士 === 國立臺北科技大學 === 電機工程研究所 === 103 === Switched reluctance motors (SRMs) have several advantages such as high torque output, high efficiency, nonrotor windings loss, and low cost. However, the doubly salient pole structure of both the stator and rotor leads to high nonlinearity of the torque output and renders the control of the SRMs difficult. Recently, there are many intelligent control methods have been proposed by researchers, including the cerebellar model articulation controller (CMAC).The output value of the associated memory of the CMAC differs from the input state. When the memory is activated, the memory value is equal to 1 or 0. This phenomenon reduces the network resolution and renders the learning performance insufficient for achieving the desired control performance. This study proposes an adaptive sliding recurrent CMAC (ASRCMAC) system comprising a recurrent CMAC, sliding surface, and modified compensator. Furthermore, the recurrent neural network concept was used to capture the system dynamics and convert the static CMAC into a dynamic one. To ensure the stability of the system, the weights and recurrent weights of the recurrent CMAC were adjusted online according to adaptive rules derived from the Lyapunov stability theory. To verify the performance and effectiveness of the proposed ASRCMAC system, we used the proposed ASRCMAC scheme for controlling a direct torque control system of an SRM speed controller drive system. The simulation and experimental results showed that when the system was operated at low, medium, high, and positive-inversion speeds, the load torque of the motor was 1 Nm. The error in the transient response speed was less than 10 rpm (in the steady state, it was within ±2 rpm), and the SRM exhibited a desirable torque response in a wide speed range for the 1 Nm torque load. Compared with a conventional CMAC, the root mean square error of the ASRCMAC system was lower. Moreover, the ASRCMAC system was more robust and offered more effective control, which can be used for enhancing the dynamic response of SRM speed controller drive systems.