IWO与BPNN混合智能算法在 转子故障诊断中的应用

在经典IWO杂草算法的基础上提出一种适用于神经网络优化的新算法。该算法将多种结构的神经网络权值阈值编码为不同维度的杂草种子，以神经网络均方误差作为种子适应度的统一评价标准，同时对多个维度的杂草种子进行排序筛选，实现了神经网络权值阈值与结构同时优化的目的。应用该方法于转子系统故障分类问题，实验结果表明该方法可以在结合BP算法优势的同时有效优化神经网络各参数，可以得到分类精度高、结构最简且泛化能力强的神经网络故障分类器。

Application of Hybrid Intelligent Algorithm of Invasive Weed Optimization and Neural Network in Rotor System Fault Diagnosis

Based on the classical ecologically inspired meta-heuristic Invasive Weed Optimization (IWO) algorithm, propose an improved variant. This work combines various structures of neural networks, whose parameters are coded as different dimensions of weed seed, and using the MSE of networks as uniform fitness of every weed. Weeds of multiple uniforms are optimized at the same time. The application of this method to fault classification problems in rotor system and the results of the experiment shows that this new method can use the advantages of the BP algorithm while optimizing neural network.