基于补偿模糊神经网络的智能诊断系统

Compensatory Fuzzy Wavelet Network Based Intelligent Fault Diagnosis

提出了一种基于补偿模糊神经网络的智能诊断系统,该系统将神经网络和补偿模糊逻辑相结合,采用动态、全局优化的运算,充分利用了相互间的优点。在神经网络的学习算法中,动态优化补偿模糊运算,使网络更适用、更优化。网络不仅能适当调整输入输出模糊隶属函数,也能借助于补偿逻辑算法动态优化相应的模糊推理,由于补偿模糊逻辑神经网络引入了补偿模糊神经元,能使网络从初始定义的模糊规则开始训练,使网络容错率更高,系统更稳定。仿真实验证明该模型在智能诊断中具有收敛速度快,诊断精度高,而且适应性强等优点。

A compensatory fuzzy neural network (CFNN) using compensatory fuzzy operators is proposed for intelligent fault diagnosis.The CFNN combines compensative fuzzy logic and neural network,and is composed of control oriented cell and decision-making cell.The fuzzy computation is dynamic and global optimized,therefore its speed is fast.Compared to BP neural networks,the convergence speed and the error precision are improved a lot.Practice has proved that the method is worth further extending.