论文部分内容阅读
尽管基于解析冗余的故障诊断方法有许多突出的优点而越来越多地得到研究和应用， 但它依赖于系统的模型， 当系统存在非线性或不确定性时， 存在难以建模的困难， 模糊神经网络可以通过学习建立系统的模型，且模型参数有明确的物理意义，初始参数易于选择，成为解决这一问题的优选途径， 作者通过把模糊神经网络的学习转化为竞争聚类和线性优化问题， 基于竞争聚类和最小二乘原理， 提出了一种模糊神经网络学习算法， 并在某伺服机构上进行了学习和故障诊断的实验， 获得了良好的实验结果
Although the fault diagnosis method based on analytic redundancy has been studied and applied more and more prominently, it relies on the system model. When the system has nonlinearity or uncertainty, it is difficult to model. , The fuzzy neural network can learn the model of the system through modeling, and the parameters of the model have explicit physical meaning. The initial parameters are easy to choose and become the preferred way to solve this problem. By transforming the learning of fuzzy neural network into competitive clustering and linear Based on the principle of competition clustering and least squares, a fuzzy neural network learning algorithm is proposed. Experiments of learning and fault diagnosis are carried out on a servo mechanism, and good experimental results are obtained