Data Driven Research On Fault Diagnosis For SBR Wastewater Treatment Process In Paper Mill And Its Application

SBR process is widely used and occupied an important place in wastewater treatment.A fault diagnosis system for SBR process can timely detect the faults,and improve operational stability.It has special significance for high efficiency and low costs.This dissertation research on fault diagnosis for SBR process in paper mill and its application based on the thesis of multivariate statistical analysis.First,a fault diagnosis model based on multivariate principal component analysis(MPCA)was developed.The data samples of normal batch and faulty batch were used for model verification.The study results shown that:(1)for normal batches,training set has better diagnosis effect with lower false alarm rate than testing set;(2)for faulty batches,MPCA model could detect fault sources timely and show good monitoring ability,but lack of ability to identify abnormal situation caused by process conditions change.SBR process is a typical batch process with multi-operating characteristics.This research propose a period subdivision MPCA method,called sub-MCPA.Three division ways,namely,craft process,variable trends,similarity of load matrix were used for subdividing SBR process.After model verification,sub-MPCA model reduced false alarm rate during normal batch diagnostics.Sub-MPCA model based on division by similarity of load matrix had good ability to locate the fault source.It can be used as a fault diagnosis method for SBR process.Once the fault source has been found,then,the sensor fault type in SBR process need to be identified.It is hard to distinguish between Precision Degradation and Drifting or Bias.So this research combined wavelet analysis with contribution plots to solve this problem and proposed a step-by-step identification strategy.Through verification by four simulated faults shown that the strategy was accurate and feasible.Combined with the previous research,we finally provided a way how to design fault diagnosis and monitoring system for SBR process in paper mills.The system completed fault diagnosis and monitoring in three steps,respectively,diagnostic model,real-time monitoring,batch monitoring.The application program was developed with Python and PyQt.This application is easy to use and strong practicality.