一种改进的间歇过程故障诊断方法

An Improved Fault Diagnosis Method for Batch Process

针对单一故障诊断方法对间歇过程故障诊断效率和准确率低的缺点,提出将快速独立主元分析(FastICA)与递推最小二乘支持向量机(RLSSVM)相结合的集合型故障诊断方法 FastICA-RLSSVM。利用FastICA对非高斯间歇过程数据快速提取特征分量,通过RLSSVM对该时变过程进行快速分类。为验证该方法的有效性,将该方法应用于青霉素发酵过程故障诊断,并与提升小波—递推最小二乘支持向量机(LW-RLSSVM)方法进行对比分析,实验结果证明FastICA-RLSSVM诊断间歇过程故障准确率高,适应性好,分类效果稳定。

In order to overcome the low efficiency and accuracy problem of conventional single fault diagnosis methods for batch process,a novel ensemble approach based on fast independent component analysis（FastICA） and recursive least squares support vector machines（RLSSVM） is proposed.Firstly,FICA is applied to extract fastly the information of the non-Gaussian batch process.Then,the time-varying process is classified fastly by RLSSVM.To certify the characteristic of the method,FastICA-RLSSVM method is applied to diagnose the faults in penicillin fermentation process,and to comparatively analyze with the method,whose name is RLSSVM based on lifting wavelet.Simulation results show that,FastICA-RLSSVM is more accurate and adaptive,and the classification results have higher stability.