基于核独立元分析的非线性工业过程故障诊断

复杂工业过程的数据具有非高斯、非线性特性,在进行故障检测时,利用核独立元分析(kernel independent component analysis, KICA)方法能有效解决这一问题。然而,由于在处理数据时使用了核函数,无法将线性的贡献图方法直接用于故障诊断,因此采用一种基于改进KICA结合非线性贡献图的方法,对非线性工业过程进行故障检测与诊断。该方法利用基于超松弛因子的FastKICA方法建立监控模型,得到检测故障信息。在发生故障后,通过非线性贡献图法诊断故障变量。最后,选用带钢热连轧工业过程实测数据进行仿真,通过与传统贡献图方法比较,结果表明此方法能够对非线性数据进行有效可靠的故障检测和故障诊断,验证了非线性贡献图的有效性。

Fault Diagnosis of Nonlinear Industrial Process Based on Kernel Independent Component Analysis

Because kernel independent component analysis (KICA) can process nonlinear data, it has been widely used in the field of complex industrial process monitoring. However, since the kernel function is used when processing data, in order to solve the problem that the traditional contribution plot cannot be directly used for fault diagnosis. Therefore, a diagnostic method based on the improved KICA method for fault detection and nonlinear contribution plot is proposed. Then the fault variables can be clearly defined. The method uses the FastKICA method based on the over-relaxation factor to establish a monitoring model and obtains fault information. After a fault occurs, the fault variable is diagnosed by a nonlinear contribution plot method. Finally, the actual measured data of hot strip mill industrial process is selected for simulation. Compared with the linear contribution plot method, the results show that this method can effectively and reliably detect and diagnose faults in nonlinear data, and verify the validity of nonlinear contribution plot.