Modeling And Scheduling Of Hybrid Processes

The research on production optimization and scheduling is a very active and popular subject to which more and more attention has been paid by both academia and industry. The hybrid processes consisted of both continuous and discrete production processes, is a typical hybrid dynamic system and is very difficult to analyze and control. The dissertation focuses on the modeling, simulation, evaluation and schedule optimization of hybrid processes.The main content and contribution of the dissertation are as follows:1. In order to model hybrid processes, a new kind of extended Petri net, hybrid object-oriented Petri nets (HOPN) which integrates high-level Petri nets, differential equations and object-oriented concepts into one modeling frame, is put forward. And the definition, the evolutionary pattern and the conflict mode of HOPN are studied in detail in the dissertation.2. Object-oriented HOPN (OOHOPN), which cooperates object-oriented design method into modeling process, is put forward. The models formed by OOHOPN are reusable, expandable and emendable.3. A method about how to detecting deadlocks in an OOHOPN is presented. Deadlocks in OOHOPN are divided into structural deadlock, restrictive deadlock and mark deadlock. To detect non-restrictive deadlock, an equivalent model of OOHOPN is constructed. Then a method that adopts ideas of decomposition and coordination of large-scale system is put forward. It divides detecting procedure into two parts, one is to detect deadlocks in every equivalent sub-model and the other is to detect deadlocks between equivalent sub-models. The method can decrease the complexity of deadlock detecting of OOHOPN.4. In order to solve simulation problem of complex Petri nets, a new method is puts forward, which includes Petri nets realization and simulation algorithms. Detail procedure that uses the model to simulate is offered in the part of simulation algorithms. As an example, simulation results of production processes of some enterprise are offered.5. A kind of new architecture of scheduling system, which takes simulation as center and integrates processes model, information model and decision model into one framework, is proposed.6. In order to solve multi-target decision-making question, an analytic hierarchy process (AHP) optimization method combined with fuzzy evaluating is figure out to select scheduling rules.7. One method which can obtain best scheduling rule set in specific manufacture environment is proposed, genetic algorithm and process simulation is integrated in this method, process simulation is used to get adaptive function value and genetic algorithm is used to search optimum solution. And, for lessening calculation time, serial genetic algorithm is replaced by parallel genetic algorithm.8. Since the existing production plan of discrete processes or continuous processes can't meet the demand of hybrid processes, a four-staged planning method that may be used to generate operation plan of hybrid process is brought forward.9. Taken Northwest Zircotube Co. Ltd as an example, the schedule system of this hybrid process is researched and implemented successfully.