HIERARCHICAL INTELLIGENT CONTROL OF INDUSTRIAL PROCESSES

TECHNIQUES ARE DEVELOPED FOR THE EFFECTIVE HIERARCHICAL CONTROL OF LARGE-SCALE PROCESSES. INDUSTRIAL PROCESSES ARE USUALLY CHARACTERIZED BY SUCH ATTRIBUTES AS LARGE DIMENSION ALITY, THE INTERACTION OF MANY SUBSYSTEMS, MULTIPLE TIME- SCALES DYNAMIC BEHAVIOR, MODEL UNCERTAINTIES, AND CONFLICT ING PERFORMANCE CRITERIA. USING AS AN EXAMPLE THE CONTROL OFA TYPICAL THERMAL PROCESS, FUNCTIONAL REQUIREMENTS ARE IDENTIFIED AND OPERATIONAL MODES SPECIFIED IN ORDER TO DELINEATE SUBSYSTEM HIERARCHIES AND INTERACTIONS. EACH PROCESS SUBSYSTEM IS CONTROLLED LOCALLY VIA AN ADAPTIVE DYNAMIC CONTROLLER DESIGNED TO MEET LOCAL CONTROL OBJECTIVES. A QUASI STEADY-STATE APPROACH IS ADOPTED TO INTEGRATE THE LOCAL DYNAMIC CONTROLLERS INTO A HIERARCHICAL STRUCTURE SATISFYING MINIMUM ENERGY OBJECTIVES. THIS DECOUPLING OF SYSTEM LEVEL CONTROLS FROM THE SUBSYSTEM DYNAMIC BEHAVIOR IS MADE POSSIBLE THROUGH DIFFERENCES IN TIME RESPONSES. FINALLY, THE LAST CONTROL HIERARCHY IS INVOLVED WITH THE INTELLIGENT CONTROL OF THE PROCESS UNDER EMERGENCY CONDITIONS. A METHODOLOGY BASED UPON HEURISTICS MAY BE EMPLOYED IN ORDER TO INCORPORATE ARTIFICIAL INTELLIGENCE FOR DECISION- MAKING, CONTROL, AND LEARNING IN MULTIVARIABLE, NONLINEAR PROCESSES, SUCH AS THE THERMAL PROCESS CONSIDERED. HEURISTICS, MANIFESTED IN A SET OF CONTROL PROTOCOLS, YEILD THE NECESSARY CORRECTIVE ACTIONS FOR SUCH PROCESSES THE DECISION-MAKING AND CONTROL ENVIRONMENT OF A HUMAN OPERATOR. AN INTEGRATION APPROACH IS PROPOSED TO IMPLEMENT THE CONTROL ALGORITHMS. INTEGRATES EFFECTIVELY CONVENTIONAL CONTROL CONCEPTS WITH AI-BASED TECHNIQUES. SYSTEM STRUCTURAL AND FUNCTIONAL HIERARCHIES WILL BE IDENTIFIED AND ALGORITHMS DEVELOPED DESCRIPTIVE OF RELATIONSHIPS AMONG HIERARCHICAL ECHELONS. EXPERT SYSTEMS WILL ENHANCE OR REALIZE IN AN AUTONOMOUS MANNER THE OPERATING CHARACTERISTICS OF CONVENTIONAL CONTROLLERS. POTENTIAL APPLICATIONS OF THE ANTICIPATED RESULTS ENCOMPASS A WIDE RANGE OF LARGE-SCALE COMPLEX INDUSTRIAL