A knowledge-based Monitoring and Diagnostic System for Emulsion Polymerisation
A knowledge-based system (KBS) has been developed for an emulsion polymerisation process. The reactor operation is carried out in batch and semi-batch modes. The system incorporates both quantitative and qualitative modelling framework. The KBS enables extraction of knowledge from incomplete or uncertain information, allows representation of process behaviour at varied levels of detail and integrates different tasks and problem-solving approaches. Mathematical models developed for the polymerisation process includes partial and integro-partial differential equations. The model predictive control algorithm allows manipulation of temperature, the flow rates of monomers (styrene and MMA), the surfactant and the initiator to control the particle size distribution (PSD) and molecular weight distribution (MWD) over the entire operating regime. The expert system architecture provides continuous support and rectification of product off-specs, monitoring and maintaining safety functionality, as well as retaining process continuity. Process events such as the particle growth and secondary nucleation could be picked up by the rule-based expert system. This expert system was built using G2 (Gensym Corp). The expert system provides diagnostic and decision support to the operator. The decision support system (DSS) incorporates several features such as intelligent polymer recipe input interface to pre-select optimal pump controls. The expert system was put to test on the actual facility and has been aiding the operator running the facility. Multi-way Principal Components Analysis (MPCA) provides intelligent statistical monitoring which will be integrated into our expert system. Herein the successful application of our system is described.