基于ELMAN神经网络的同步电机动态参数在线辨识

为提高同步电机参数在线辨识的速度和可靠性，减少辨识计算量，提出了一种基于神经网络的电机参数动态跟踪辨识方法。针对同步电机暂态、次暂态参数的非线性和动态特性，在多层前向BP网络中引入特殊关联层，形成有“记忆”能力的Elman神经网络，因而可以映射系统的非线性和动态特性。在网络训练算法中，提出一种自适应修正步长和矩量因子的算法，显著提高了训练的收敛速度。训练样本集以同步电机在各种典型运行模式下的检测数据经卡尔曼滤波、状态空间有限元等基于模型的辨识算法离线计算得到。文中还给出了由工控机、智能数据采集卡和传感器锁相环控制接口电路构成的在线辨识硬件电路设计。数字仿真和动模实验机组辨识算例证明，这种Elman神经网络模型能够实现同步电机动态参数的在线跟踪辨识。

ON-LINE DYNAMIC PARAMETER IDENTIFICATION OF SYNCHRONOUS MACHINES BASED ON ELMAN NEURAL NETWORKS

In order to improve the efficiency and the reliability of the on line parameter identification of synchronous machines, an artificial neural network (ANN) based approach for on line dynamic parameter identification of synchronous machines is proposed in this paper. For the nonlinearity and dynamic behavior of transient and subtransient parameters of synchronous machines, a special correlation layer is appended to hidden layer of BP network to form an Elman neural network with memorial ability, with which the nonlinearity and the dynamic behavior of the system can be mapped. In the training algorithm of the network an algorithm with adaptive step correction and momentum factor is put forward which obviously improves the convergence speed of training. The sample sets for training are generated in such a way that the data of synchronous machines detected under various typical operating mode are calculated by the model based identification algorithm, e.g., off line Carman filtration method and finite element method in state space, etc. The design of hardware for on line identification which is composed by industrial process controller, intelligent data acquisition card and control interface for phase locked loop of transducer is presented. The results of identification calculation example by digital simulation and dynamic simulation with experimental units show that the on line tracing identification of both static parameters and transient parameters of synchronous machines can be implemented by the presented Elman neural network model.