一种新的状态估计混合智能模型

为实现发电机状态维修，研究一种新的状态估计混合智能模型。设计了一种基于径向基信息神经网络和模糊推理的模型，阐述径向基信息神经网络的结构和训练算法，介绍一阶Sugeno模糊模型的5层结构，根据专家经验知识选择合适的模糊输入隶属函数和模糊逻辑推理规则，而结论参数采用递归最小二乘法辨识。该模型充分利用各种在线监测装置的监测数据，特别是可利用领域专家的实践经验和专业知识，实现发电机状态估计。

A New Hybrid Intelligent Model for Condition Estimation

In order to carry out condition-based maintenance for generators, a new hybrid intelligent model of condition estimation is proposed. The model is established with fuzzy inference and neural network using a radial basis function. The structure and training algorithm of the neural network are described. The first order Sugeno fuzzy model has five layers. The input membership functions and fuzzy logic inference rules are selected according to expert experience and knowledge. Consequent parameters are calculated with a recursive least squares estimator. The model makes full use of the data of various online monitors. The expert's practical experience and professional knowledge are input to the Sugeno fuzzy model. The generator's condition is estimated after training of the hybrid intelligent model.