摘要
研究质量监测在部分最小二乘(PLS)框架下的可检测性问题,并给出一种新的主成分确定方法.根据故障子空间和变量的显著性测度确定不同故障对质量的影响力,定义最优检测函数,进而建立PLS最优检测模型(ODM),使模型对严重影响产品质量的重要过程故障具有最优检测能力.以TennesseeEastman(TE)过程为案例进行统计质量控制研究,结果表明此方法对重要过程故障的敏感性很高,有助于及时采取补救措施,稳定产品质量.
The fundamental conditions for faults detectability under partial least square (PLS) framework were analyzed. A new component number selection method, the PLS optimum detection model (PLSODM), was proposed to optimize the faults detection capability of the PLS method. This model was defined according to the fault subspace and the index of variable's importance in projection, which reflected the importance of the process variables to quality. The application of PLS-ODM in the Tennessee Eastman (TE) Benchmark statistical quality control confirms that the weak important faults can be detected as rapidly as possible, so that the product quality can be controlled to be more stable, while operation cost can be decreased greatly.
出处
《浙江大学学报(工学版)》
EI
CAS
CSCD
北大核心
2005年第5期657-662,共6页
Journal of Zhejiang University:Engineering Science
基金
国家自然科学基金资助项目(20206028).