摘要
本文将在化工领域广泛使用的软测量技术应用来进行异步电机转矩测量。根据感应电动机的数学模型,提出了仅用葶于神经网络的电机定子电流的电磁转矩辨识方法,实现异步电机转矩的软测量。用改进的BP算法对神经网络进行学习和训练,构建了适合电机转矩观测的多层前馈神经网络,仿真研究表明:基于神经网络的电机转矩辨识模型具有良好的性能。
This paper uses soft-sensing technology, which has been used broadly in chemical industry, to measure the torque of asynchronous motor. According to the fundamental equations of induction motor, a electromagnetic torque identification method that only uses stator current based on neural networks is set up, and the torque of asynchronous machine is measured by soft-sensing technology. The structure of multi-layer feed-forward neural networks that is fit for electromagnetic torque measure is founded and trained with Back Propagation Levenberg-Marquardt's method. The simulation results show that the neural network - based identification model of motor electromagnetic torque is of better Performance.
出处
《自动化信息》
2006年第7期44-45,36,共3页
Automation Information
关键词
软测量技术
软测量模型
神经网络
仿真
Soft-sensing Technology
Soft-sensing Model
Neural Networks
Simulation