论文部分内容阅读
提出了一种基于支持向量机的机械故障诊断模型,该模型建立在VC维理论和结构风险最小原理基础上,根据有限的样本信息在模型的复杂性和学习能力之间寻求最佳折中,以期获得最好的推广能力。在选取诊断模型输入向量时,对故障信号功率谱进行小波分解,简化了故障特征向量的提取。仿真结果验证了该模型的有效性。
A model of mechanical fault diagnosis based on support vector machine is proposed. The model is based on VC dimension theory and the principle of least risk of structure. Based on the limited sample information, the best compromise between the complexity and learning ability of the model is sought. In order to obtain the best promotion ability. When the input vector of the diagnostic model is selected, the power spectrum of the fault signal is decomposed by wavelet to simplify the extraction of fault eigenvectors. The simulation results verify the effectiveness of the model.