Multi-Stage Modeling and Monitoring for Penicillin Fermentation Process Based on Variable Sampling Period

Multi-principal component analysis(MPCA) is a more effective method in fault monitoring and diagnosis for the batch process.However,it is difficult to deal with complex nonlinear systems due to the linear feature of its own.According to the characteristics of fermentation process,a multi-stage MPCA method was proposed.The paper used multi-stage linear structure to replace the traditional linear single structure,overcomed the shortcomings of traditional MPCA method,and took a variable sampling period based on the relationship of variables among different stages,optimized the modeling data,improved the accuracy and timeliness of monitoring and diagnosis.