The PI parameters set in advance under complex conditions can not meet the new operating status of the motor,then a speed estimator based on diagonal recurrent neural network(DRNN)PI parameter dynamic self-tuning was designed for this problem,and the PI parameters were dynamic self-tuning to improve system robustness and estimation accuracy. The principle and structure of speed estimation based on model reference adaptive system (MRAS)were analyzed,and the adaptive law of the estimation system was deduced by Popov hyperstability theory. On the basis of MRAS speed estimation,the PI parameters in MRAS speed estimation system were dynamically adjusted by DRNN,then the new speed estimation system based on DRNN-MRAS was obtained. Simulation and experimental results verify the effectiveness and practicability of the new DRNN-MRAS algorithm.%复杂工况下事先设定的PI参数不能满足电机新的运行状态,针对此问题设计一种基于对角递归神经网络PI参数动态自整定的转速估计器,对其中的PI参数进行动态整定以提高系统鲁棒性和估计精确度.分析了基于模型参考自适应系统转速估计的原理和结构,利用Popov超稳定理论推导出估计系统的自适应律.在MRAS转速估计的基础上,利用DRNN对MRAS转速估计系统中的PI参数进行动态在线整定,得到基于DRNN-MRAS的新型转速估计系统.仿真和实验结果验证了新型DRNN-MRAS算法的有效性和实用性.
展开▼
