【作者】 吕宁； 于晓洋； 刘少波；

【Author】 Lv Ning Yu Xiaoyang Liu Shaobo(School of Automation Harbin University of Science and Technology Harbin 150080)

【机构】 哈尔滨理工大学自动化学院；

【摘要】 针对递推偏最小二乘(RPLS)方法对间歇生产过程故障监控的不足,提出了一种改进的跟踪递推最小二乘(TRPLS)算法,TRPLS建模采用非线性核函数对数据的拟合能力,解决了PLS非线性建模问题,通过引入跟踪因子,不但解决了数据的饱和现象,而且不用再考虑数据的填补问题。仿真结果表明TRPLS不仅可以实时监控系统的运行,而且也能够准确的识别故障,减少了误报、漏报现象,提高了监控和诊断的准确率。更多还原

【Abstract】 An improved track recursive partial least squares (TRPLS) model is presented in order to detecting deficiencies of fault diagnosis MPCA in batch process.TRPLS models can detect nonlinear model of PLS,data saturation phenomenon and data Packing question with fitting ability nonlinear kernel function for data and track factor.Simulation results show that the TRPLS model not only detect the known faults,but also can recognize the unknown faults,the validity of batch process monitoring and fault diagnosis.更多还原