Process Monitoring For Multiphase Batch Processes With Uneven Operation Durations

Since the 90s of the 20th century,with the urgent market demands of the personalized,diversified and high-quality products,the batch process has become an important industrial production models and many engineers have focus on the safety and reliability of batch processes.Compared to the continuous process,the characteristics such as multi-stage,repeated operation and.the change of product led the monitoring for batch process to a complicated problem.Gaussian model,as a method of dividing and modeling,was successfully applied to various fields owing to its solid theoretical foundation and good practical performance.The process monitoring method based on Gaussian model is proposed for multi-phase batch processes with uneven operation durations.(1)The step-wise sequential phase partition based on Gaussian Single Model(GSM)is proposed.The sequential of batch process operation is considered and monitoring variables within a certain period of time are evaluated.The phase is divided into many initial phase according to whether or not they obey Gaussian Single Model.Then according to the specific characteristics of the actual production process,the initial phase is divided into the stable period and the transition period.(2)After the division of stable period and transition period,The data characteristics are considered and the monitoring model is established respectively according to the characteristics of their own data.And then the appropriate statistics are selected as the monitoring index.(3)Considering the characteristics of unequal length in process,the identification of the period is not finished according to the sampling time.Combined with the several situations that may occur the next time,the identification of the online period problem may be effectively solved.After the completion of the identification of period,the corresponding online monitoring is selected.(4)The simulation results of the penicillin fermentation process shows the effectiveness of the proposed method.