Research On Rolling-Process-Oriented RP-agent Distributed Cooperative Control System

With the increasing demands of the plate product quality, specifications andcost-effective and the increasingly fierce market competition, and the newtechnologies and processes such as medium and heavy plate hydraulicstraightening machine, full-hydraulic rolling shear and bilateral scissors,stainless steel composite carbon-base, have been widely applied in recent years.At the same time, the means of achieving modern rolling process actions havegradually evolved into multi-cylinder cooperative control or process intelligentdecision-making control. All the above technologies put forward higherrequirements of the rolling process automation control research anddevelopment. In order to solve the new situations of production process andadapt to these new technology developments, a process control system is eagerlyneeded to integrate information, so that field detection sensor, basic automation,drive automation and process automation can be combined into ahighly-integrated information platform. On the basis of cooperative control offour servo cylinder with multi-degree of freedoms on medium and heavy platefull-hydraulic straightening machine, two servo cylinder collaborative control onthe full-hydraulic rolling shear, and how to integrate into the full range ofprocess automation and factory automation equipment as a single largeengineering practice, the dissertation contents and findings are as follows:(1) The development of rolling process computer control system has beensummarized, and the three architectures of rolling process automation computercontrol system and two domestic architectures have been detailed illustrated.The digitization, miniaturization, network-based, low-cost, strong computationwill be the future directions of development. (2) Agent theory and methods were used to demonstrate the Agent motioncoherence. This article first proposed the RP-Agent distributed cooperativecontrol architecture for the rolling process, and introduced the definition of theRP-Agent model for rolling process, the communication layer, coordinationlayer and control layer of the model’s constitutions; together the definition of thecore layer of the RP-Agent model form by coordination layer ontology model,the collaborative model, the computational model and self-learning. Thecoordinative layer was pointed out to be the different reactions of RP-Agentsystem to external environment or their own activities. FDTC was made as anexample to introduce the concept of single and multiple RP-Agent controlgroups. The control framework program, during which a big problem wasbroken down into a number of small issues and two RP-Agents were shared forsolving complex problems, were brought forward. Communication, coordinationand control as the process resources of computer control system, when two ormore RP-Agents are existing in the system, a RP-Agent was defined as the corecontroller of Host RP-Agent to be responsible for the performances and logics ofoverall system and the strategies on task activation and redistribution, otherRP-Agents were defined as Slave RP-Agent only for extending the calculatingability of Host RP-Agent so as to ensure the coordination and orderly control ofthe entire control system.(3) A detailed requirement analysis was made on RP-Agent by Gaiadevelopment method, the five kinds of problems to be solved in rolling processautomation and control system were summarized, and five system roles werefounded on this basis: CommC role, CoorC role, ContC role, EnviOb role,BlackBoard role. Gaia graphical language was used to build asystematically-interactive model by the detailed analysis on the informationinteractive relationships among the various actors. RP-Agent body model,service model and acquaintance model, which were oriented for the rollingprocess, were designed based on the role model and perfect interaction model.Moreover, the whole Gaia-based and RP-Agent-oriented modeling process wassummarized. (4) The targeted classification of "events" in the rolling process was made.The architecture of RP-Agent event-driven was built, and the task allocationcalculation model based on the overall load determining system was defined byHost RP-Agent. The whole process of the environmental events and self events,i.e., the RP-Agent event-driven model of behaviors, was established fromlistening, capture and distribution to processing. The static knowledgerepresentation of ontology was used to describe the compositions of all thecontrol objects of RP-Agent control system and their interconnections. Partialobjects of the main console of full-hydraulic medium and heavy plate rollingshear were used as samples to introduce on how to build system ontology. Thedecision-making process of RP-Agent-based event-driven internal behaviors andthe implementation of the algorithm were analyzed and described in detail. Thecontrollability and reliability of the proposed algorithm have been enhanced bymodifying the RP-Agent matching index weights for improving the system’sexecution ability according to the designed plan.(5) Oracle has been proved to have huge advantages and many meritsthrough the applications of OntologyList and RelatedProperties tables in rollingprocess automation control system. Secondly, through the implementationprocess of "blackboard" role, the establishment of the "blackboard" wasconcluded to be an important and effective means to solve parallel processing,independent design and debugging, as well as the solution of the large anduncertain complex issues.(6) The multiplex hydraulic servo in the rolling process was introduced indetail, and the RP-Agent-based control system structure was put forward. Basedon the command value dynamic hyperbolic and dynamic tracking strategy, theintelligence, collaborative and effectiveness of RP-Agent architecture have beenvalidated.