针对异步电机转子电阻识别结果偏差较大的问题,文中提出了一种基于融合神经网络的异步电机转子电阻在线识别技术.该技术方案以模型参考自适应识别技术作为基础框架,采用电压模型磁链观测系统作为参考单元,用电流模型磁链观测系统作为调控单元,设计了基电压和电流模型的转子电阻识别方法.并进一步通过卷积神经网络对参考单元系统模型进行训练优化,实现对异步电机转子电阻在线的精确识别.实验测试结果表明,该技术方案的可行性较高,比同类方法的转子电阻识别响应速度提升了约25%,识别精度也达到了较高水平,且具有较强的抗扰动稳定性,为异步电机矢量控制的策略设计提供了 一种新思路.
In response to the problem of significant deviation in the identification results of asyn-chronous motor rotor resistance,this paper proposes an online identification technology for asynchronous motor rotor resistance based on fusion neural networks.This technical solution uses Model Reference Adaptive System(MRAS)as the basic framework,voltage model flux observation system as the reference unit,and current model flux observation system as the control unit.A rotor resistance identification meth-od based on voltage and current models is designed.Furthermore,the reference unit system model is trained and optimized through convolutional neural network to realize the accurate identification of asyn-chronous motor rotor resistance online.The experimental test results show that this technical solution is highly feasible,with a response speed improvement of about 25%compared to similar methods for rotor resistance identification.The recognition accuracy has also reached a high level,and it has strong anti-interference stability.This provides a new idea for the design of vector control strategies for asynchronous motors.