中文摘要：

针对位置传感器的引入使得开关磁阻电机（SRM）结构变得复杂,可靠性降低这一问题,将RBF神经网络和终端滑模控制（TSMC）相结合建立了自适应神经终端滑模观测器,用RBF神经网络逼近观测器的控制输入,无需知道扰动项的上界,通过终端滑模控制使电流偏差为零,实现对组合变量iω L/ θ的观测。依据电感特性将电机的工作区间分为电感近似线性区和电感非线性区,建立全周期电感数学模型,并将电感和电流代入观测变量达到对开关磁阻电机转子位置准确跟踪的目的。仿真结果表明,在电机稳态运行和负载转矩突变时都能实现对转子位置的精确估计,为开关磁阻电机无位置传感器控制提供了基础。

英文摘要：

As switched reluctance motor （SRM） with position sensor has complex structure and low reliability, radial basis function （RBF） neural network and terminal sliding mode control （TSMC）arecombined to build adaptive neural terminal sliding mode observerin this paper.RBF neuralnetworkis used toapproach observerinput.TSMC can make current deviation approach zeroregardlessupper disturbance boundary. Thissystem canrealizeobservation ofcombined variable iω？L/？θ.The paper divides motor’s working regionintolinear and nonlinear inductance regionsdependingon inductance characteristic.Complete period inductancemodelis builtand current and inductanceissubstituted into observation variable, thustrackingrotor position accurately.Simulation results verifyvalidityofestimationresult when motor works in steady state and loading torque variessuddenly.The methodthus lays foundation for sensorless SRM control.