Process optimization in a supervision support system

We present details of the optimization task in a real-time (Thompson and Me rtz 1993) knowledge-based (Moore et al. 1991, Larsson 1992) supervision sup port system in the coal washing domain. The Ash Control Model (AshMod) assi sts operators in maximizing clean coal yield while keeping ash (impurity) c ontent within acceptable limits. AshMod assists the operator in plant situa tion assessment, fault diagnosis, and performance optimization. Situation a ssessment and fault diagnosis are mentioned briefly, since they have been d escribed elsewhere in detail (Villanueva and Lamba 1997, 1998). We focus on the optimization task, which employs a hybrid artificial intelligence and operations research approach. The process is modelled through a set of exte nded states associated with the entire process and with individual componen ts (circuits) within the plant. The scheduling optimizer continuously monit ors the process, assesses the process state, and dynamically plans and perf orms integer and real optimization of a sequence of actions. The supervisio n support system captures domain knowledge through multiview models (Terspr a et al. 1993) such as the goal tree success tree, plant schematic and faul t cause network. AshMod performs deep reasoning through the use of knowledg e models (Lind 1994) that capture purpose, function, structure, behaviour a nd heuristics. These knowledge models are mentioned briefly, since they hav e also been described elsewhere in detail (Villanueva and Lamba 1997, 1998) . The supervision support system (Lamba 1995) is currently undergoing onlin e validation at the B&C Coal Washing Plants operated by the Broken Hill Pro prietary Limited (BHP 1998) at Port Kembla, Australia. The system is expect ed to be fully operational by the end of 1998.