신경회로망을 이용한 영구자석 동기전동기의 센서리스 벡터제어

Abstract

신경회로망을 이용하여 PMSM의 속도제어를 기존의 속도 및 위치검출기 없이 행할수 있는 알고리즘을 제안 했다. PMSM을 새롭게 모델링하여 전동기의 추정속도를 신경회로망의 결합강도를 나타내는 가중치로 표현하여 PMSM에 신경회로망 이론을 적용할수 있도록 했다. 이렇게 하여 만들어진 2층 구조의 신경회로망에 역전파 알고리즘을 적용하여 신경회로망 출력이 PMSM의 출력과 같아지도록 가중치값을 조정하여 PMSM의 속도를 추정?杉?. 또한 이 추정속도를 이용하여 기존의 속도 및 위치검출기를 사용하지 않는 PMSM의 벡터제어가 별도의 훈련과정 없이 실시간으로 행해질수 있도록 했다. 제안된 알고리즘의 타당성을 컴퓨터 시뮬레이션과 실험을 통해 입증해 보였다.

The PMSM speed control technique without speed and position sensors is developed with the aid of the neural networks. The alternative mathematical model of PMSM is proposed and it is good for the application of the neural networks because the estimated motor sped is treated as the weights. The 2 layered neural network based on back propagation algorithm is used to estimate he rotor speed by adjusting the weights of the neural network. Using this estimated speed, the real time vector control of the PMSM is implemented without any other training process. The proposed control scheme is simulated digitally, and the experimental results agree very well with those of the simulated results

The PMSM speed control technique without speed and position sensors is developed with the aid of the neural networks. The alternative mathematical model of PMSM is proposed and it is good for the application of the neural networks because the estimated motor sped is treated as the weights. The 2 layered neural network based on back propagation algorithm is used to estimate he rotor speed by adjusting the weights of the neural network. Using this estimated speed, the real time vector control of the PMSM is implemented without any other training process. The proposed control scheme is simulated digitally, and the experimental results agree very well with those of the simulated results