Modeling, safety verification and optimization of operating procedures in process systems using hybrid Petri nets
IEEE SMC'99 Conference Proceedings. 1999 IEEE International Conference on Systems, Man, and Cybernetics (Cat. No.99CH37028)
In this paper, a hybrid Petri net-baaed hierarchical control and modeling framework for real-time process procedural control is introduced. The new modeling tool, called as hybrid predicate Petri net, is an extension of hybrid Petri net and Predicate/Transition net, in which the ftig speeds of its continuous transitions are changeable and controlled by upper level and continuous mathematical model. Predicates are introduced to discrete transitions to represent the various constraints possibly
... straints possibly existing in process systems. Deftition of net structure and running rules of the hybrid predicate Petri net are provided. On the basis of supervisory model, the issues of safety verification, optimization of control parameters and scheduling of operating sequences are discussed. At last, a distillation column is used as an example for modeling and analysis. 1. INTRODUCTION Process industries involve both continuous and batch processes. In process industry, process control layer under normal operation has achieved high automation degree, yet the startup, shutdown and emergency handling processes are still lying in manual operation level. Because of degraded operator efficiency and the ever-increasing degree in coupling of the system components and response time requirements, it is necessary to develop the automation of safe process startup and shutdown under irregular conditions to improve the production efficiency and the safety. Startup and shutdown of processes are equivalent situations where the model of the plant must evolve with time in order to be in phase with the reality. At the same time, procedures are applied to start and stop the production, unit by unit, obeying to sequential control actions. Consequently, in such procedural control systems, both continuous and discrete aspects are present. They are typically hybrid dynamic systems. The synthesis of operating procedures involves the determination of sequence of primitive operations, such as, opening or closing valves, turning on or off motors, and changing the values of set points, which can be implemented online and automatically through distribute control system. Procedures and sequences must be established considering a variety of aspects including technical process requirements, safety and economics, that is, objectives and constraint conditions. Due to the operating complexity of process industry, a diverse set of constraints, such as (1) Temporal constraints; (2) Quantitative inequality constraints; (3) Physical constraints; and (4) Logical, mixing constraints, must be satisfied while scheduling and optimizing the operating procedures. It's clear that a comprehensive hybrid system model that can represent all these kinds of constraints is needed before safety verification and optimization. Different approaches to the representation of hybrid systems have been proposed. Some authors define a homogeneous model that links the discrete-event part and the continuous part in a single formalism [l-3]. Others use specific formalisms for each of the two parts or defme a model based on the interface between the two parts [4,5]. Naturally, most of the efforts are based on the discrete-event part and involve models for discrete event dynamic system. In which, Petri nets [6,7] are one of the most popular model due to its powerful modeling mechanism of concurrency, synchronization, asynchronization and its uniform representation for system modeling and analysis. Especially, ordinary Petri nets can be extended to integrate as much as possible continuous aspects in them, such as timed Petri nets[S], continuous Petri nets[9,10], hybrid Petri nets[3,7]. It is O-7803-5731-0,99/$10.0001999 IEEE I-854 advantageous to be able to represent both continuous and discrete parts of a hybrid system in the same context as in hybrid Petri nets. Due to the complexity of procedural control in process systems, a hierarchical approach allows to decompose the whole system into smaller sub-problems according to different competence levels. From the point of control, it involves 3 levels, that is, local control of the plant, supervisory control, optimization and decision. From the point of modeling, the process recipes can also be hierarchically modeled by describing the unit operation details in lower level. This can reduce the scale of the model, and also facilitate the analysis and control.