论文部分内容阅读
在冷轧过程中,断带故障是冷轧工序的主要生产故障之一。针对冷轧过程断带故障的特点,提出一种基于核主元分析(KPCA)非线性特征提取和最小二乘支持向量机(LSSVM)分类的故障诊断方法。此方法采用KPCA理论将冷轧过程原始空间数据映射到高维空间,并在高维空间进行主元分析,从而降维、去相关性,得到冷轧过程非线性特征向量。将降维后的特征主元作为LSSVM输入进行训练和识别,根据LSSVM的输出结果判断冷轧过程工作状态与故障类型。仿真结果表明:基于KPCA非线性特征提取和LSSVM分类的故障诊断方法计算速度快,能有效地提取冷轧过程断带故障特征,识别断带故障类型。
In the cold-rolling process, the break-tape failure is one of the main production failures in the cold-rolling process. Aiming at the characteristics of fault in cold rolling process, a fault diagnosis method based on kernel principal component analysis (KPCA) nonlinear feature extraction and least square support vector machine (LSSVM) is proposed. In this method, the KPCA theory is used to map the original spatial data of the cold rolling process into high-dimensional space and carry out principal component analysis in high-dimensional space to reduce the dimensionality and decorrelation and get the non-linear eigenvectors of the cold-rolling process. The dimensionality-reduced feature PCU is trained and identified as LSSVM input, and the working status and fault type of the cold rolling process are judged according to the output of LSSVM. The simulation results show that the fault diagnosis method based on KPCA non-linear feature extraction and LSSVM classification is fast in calculation, which can effectively extract the fault features of the fault during the cold rolling process and identify the fault types of the fault.