模糊神经网络在线辨识定子电阻的直接转矩控制技术

Direct Torque Control Technology of Fuzzy NN Network Online Distinguishing Stator Resistance

提出用 FNNS(模糊神经网络)在线辨识异步电机的定子电阻,把电机运行时间、频率、电流作为输入变量,得到输出变量、定子电阻值,其中利用模糊规则的增减优化 BP 网络,节约处理时间,并按规则数确定学习率(具有可变性),保证了系统的高效性、稳定性及收敛性。用 DSP 进行运算处理,对定子转矩和磁通等进行快速运算,减小了由于运算时间而带来的误差。由实验得,整个系统可以减弱电机低速下运行不平稳和指令突变时所带来的影响。

Stator resistance is identified on -line by FNNS(fuzzy -neural network system)in this paper.Working time,rotating speed of the engine and the current are used as fuzzy state variables and the value of the stator resistance is obtained;during this process,BP network can be optimized by additions and deletions of the fuzzy rules.process time can be saved;the learning rate is determined by the number of the rules(change ability),the hign-efficiency,stabilization and convergence of the system can be assured;DSP is used for calculating process in stator torque and flux quickly in that it can reduce the error which is caused by calculating time;from the simulation result,the system can weaken the effect caused by rotating placity in low speed and the wave of the order.