摘要
设计了神经网络自适应滑模控制器。用RBF神经网络自动调整滑模控制器的切换项增益,无需建立包含参数摄动和干扰在内的整个系统的精确数学模型,有效提高了系统的稳定性和鲁棒性。采用Lyapunov稳定性理论证明了系统稳定性,并针对常值干扰、时变干扰和参数摄动情况分别进行了仿真与实验。与传统的PI控制相比,神经网络自适应滑模控制器具有更好的稳定性和抗干扰能力。
Considering the sensitivity to parameter variation and load disturbance of Permanent magnet synchronous motor (PMSM) , this paper proposed a neural network based adaptive sliding mode control (NNASMC) for higher stability and robustness. RBF neural network was used to adjust the gain of the switch part of sliding mode control input. So the accurate mathematic model of the whole system including uncertain parameters and disturbance was not required. The stability of the system was proved by Lyapunov theory. Simulations and experiments are done under the situation of constant disturbance, time - raring disturbance and parameter variation. The proposed NNASMC has a better stability and noise reduction compared with PI control.
出处
《电机与控制学报》
EI
CSCD
北大核心
2009年第2期290-295,共6页
Electric Machines and Control
基金
“985”工程学科建设投资项目(107008200400020)
关键词
永磁同步电机
RBF神经网络
滑模控制器
参数摄动
负载扰动
permanent magnet synchronous motors
RBF neural network
sliding mode control
parameter variation
load disturbance