论文部分内容阅读
故障诊断问题实质上是一个模式识别问题,即多分类问题。采用Morlet小波来构造支持向量机(Sup-port Vector Machine,SVM)的核函数,Morlet小波核SVM比普通SVM具有更好的鲁棒性和更强的泛化能力。在一对一算法的基础上实现Morlet小波核多类支持向量机的故障诊断,并将此方法成功应用于电厂汽轮发电机组的故障诊断。实验仿真结果表明Morlet小波核多类SVM故障分类器比BP神经网络训练和测试速度快,且其分类精度在高斯噪声干扰下还保持100%,比BP神经网络高出11.8%。因此该方法能够快速而准确地对电厂汽轮发电机组的故障进行诊断,满足电力系统实时操作的要求。
The problem of fault diagnosis is essentially a pattern recognition problem, that is, a multi-classification problem. Morlet wavelet is used to construct kernel function of Sup-port Vector Machine (SVM). Morlet wavelet kernel SVM has better robustness and generalization ability than ordinary SVM. On the basis of one-to-one algorithm, the fault diagnosis of Morlet wavelet multi-class SVM is implemented and the method is successfully applied to the fault diagnosis of power plant turbo-generator. Experimental results show that the Morlet wavelet multi-class SVM fault classifier is faster than BP neural network in training and testing, and its classification accuracy remains 100% under the Gaussian noise disturbance, which is 11.8% higher than BP neural network. Therefore, the method can quickly and accurately diagnose the fault of power plant turbo-generator and meet the requirements of real-time operation of power system.