Research On Hierarchical Modeling And Rule Description Method For Discrete Event Control Systems

Discrete event control systems (DECS) which widely exist in industrial control field are control systems of discrete event systems (DES). Just as DES, DECS has the characteristics of discreteness, asynchronism, concurrency and nondeterminacy. Furthermore both input and output of DECS are multi-dimension variables. Relations between input and output of DECS are set of logic relations between multiple variables and multiple state values. State space of EDCS is multi-dimension and multi-state. The characteristics of DECS lead to complexity of modeling and design of DECS. There are two primary problems: "curse of dimensionality" and "curse of modeling". However, researches on DECS model and its modeling method are not matured or perfect. It is not as perfect as that of continuous variable dynamic systems (CVDS) in oneness of form, conciseness of mathematical expression and feasibility of computation analysis. Meanwhile, with the rapid development of high technology, structures and behavior feathers of practical industrial control systems become more and more complex. So it needs imminently for new model and method which is simple, intuitionistic, convenient to grasp for domain engineers and easy to implementation.A new modeling and design method of DECS which combines object-based hierarchical model and rule description method (RDM) is put forward in the dissertation. The main work in the dissertation is as follows.(1) Aiming at the problem of "curse of dimensionality", object-based hierarchical model for DECS is put forward. The hierarchical model not only has the characteristics of the horizontal splitting structure such as parallelism, robustness and extendibility, but also has the characteristics of the vertical hierarchical structure such as wholeness and harmony. To construct object-based hierarchical model of DECS, objects of a system are divided into multiple parts and multiple levels according to their constitution and dependency. In hierarchical structure of DECS, the overall control task of a DECS is divided into multiple subtasks and distributed into multiple-layer objects. Rule is used to describe logic control relation between objects of high and low level. Message is used to transfer change of object state. The division of a system follows the principle of "highly cohesive and loosely coupled". Meanwhile the whole state space of a system is divided into multiple subspaces. So complexity of solving a problem is reduced and the problem of "curse of dimensionality" can be avoided.(2) On the basis of knowledge representation method with production rule and knowledge inference method of production system, rule description method for DECS design is put forward. Production rule description method is simple in form and easy to understand. Production rule has good independency and structural style, so it is propitious to formalization. Therefore rule description method is easy for domain engineers to understand, grasp and apply, it can solve the problem of "curse of modeling".(3) Formal syntax and semantics of logic rule description language (LRDL) are constructed. Formal syntax of LRDL is defined by EBNF. Axiomatic semantics of LRDL is presented and proved formally based on axiomatic system of Hoare logic, which offers theoretical basis for provement of the correctness of programs written with LRDL. Logic completeness of LRDL is proved. Describability of RDM for the three basic relations between events and activities of DECS is analyzed. The design and development procedure of DECS including six steps: requirement, analysis, design, implementation, verification and deployment is presented.(4) Several critical problems and solutions when using RDM to design and implement DECS are discussed. Rule simplification method based on Horn clause is presented. Aiming at the problem of real time and efficiency, a fast algorithm for rule match problem based on gene encoding and indexing is put forward. Aiming at the problem of correctness and reliability, a rule base verification and validation method of DECS is put forward. The method combines static analysis with dynamic emulation and test. A rule base verification method based on a kind of extended Petri net is presented.(5) A typical DECS, autobody control system based on bus is designed and implemented using object-based hierarchical model and RDM. Practical applications indicate that the method makes it easy to develop and maintain the software and hardware of autobody control system, and improves design quality and develop efficiency. The method is easy for technicians to understand, grasp and apply. So they can play an dominant role in system design and long-term maintenance.