Research On Open Intelligent CNC Architecture And Key Technology Based On STEP-NC

At present, a lot of researches have been done to improve the openness of the CNC and the CNC openness theories are getting mature gradually. However, there is still a bottle neck between CNC and the high level manufacturing system CAx in aspect of sharing information. Therefore, to achieve the integration of the whole manufacturing system, a new machining data model, STEP-NC (STEP for numerical controller) was proposed. The utilization of STEP-NC as the machining data model, instead of traditional G&M codes, brings abundant information for designing, process planning and machining. Besides, STEP-NC provides a new opportunity for the intelligent CNC. Meanwhile, the use of STEP-NC causes some greate changes in the architecture of CNC system. For this reason, this paper promotes the research on the architecture of STEP-NC compliant CNC systems and the related key technologies, especially on the realization of intelligence of STEP-NC compliant CNC system.The basic structure of traditional CNC systems is reviewed and the software and hardware platforms are selected based on the requirements of intelligent CNC systems. The detailed structure and properties of STEP-NC data model are analyzed and the new characteristics of STEP-NC compliant CNC are studied. Then an architecture of open and intelligent STEP-NC compliant CNC system is presented. This CNC framework contains 5 main parts:a local process planning module, an intelligent decision making module, an open soft NC core, an expert knowledge base system and a real-time Ethernet fieldbus platform. This architecture is the basis for the following researches.The realization of some basic functions is also studied. The Java libraries provided by ST-Developer are used to develop the interpreter that is used to extract and explain STEP-NC data. The Java3D engine is used for display and simulation in 3D. The local process planning module is divided into three levels. An interlinkage method between the process planning and the NC kernel is given. The above three STEP-NC function modules are tested by developing the validation programes. The ontology conception is introduced to construct a three layered ontology expert model. The related knowledge of machining in CNC area is modeled. The importance of constructing the CNC machining ontology is analyzed. The developing language and tools are specified. The main concepts in CNC domain are classified and defined, and the SWRL is used to design the semantic rules. Meanwhile, the mapping mechanism between Express data model and OWL ontology model is studied. The integration method between CNC machining expert knowledge base and intelligent CNC system is also presented. The experiment indicates that the CNC ontology expert knowledge base can describe the knowledge in CNC area and possess the reasoning capability to support the intelligent behavior of STEP-NC compliant CNC system.The adoption of real-time industrial Ethernet as low level hardware platform can satisfy the requirement of openness, and thus can guarantee the integration with upper level units of designing and management. A real-time Ethernet which is called as EtherMAC is developed by our research group and adopted by this paper. The topology structure, working mechanism and slave scheme are detailed in this paper. The protocol stack for the master node under Linux RTAI is developed. The real-time and synchronization performance is almost equal to the prevailing real-time Ethernet even without proprietary hardware and high performance RTOS, so it totally satisfies the requirement of isochronous real-time motion control. In addition, the master protocol stack also provides a flexible integrating interface for other application programs.The soft NC kernel of EMC2 under Linux RTAI is studied. An integrating scheme for STEP-NC compliant open and intelligent CNC is proposed. The EMC2 is regarded as the core, the STEP-NC process planning module is integrated on upper level and the EtherMAC platform is integrated on lower level. The HAL conception of EMC2 is used here, where the EtherMAC protocol stack is integrated. The canon language is adopted as the middle language for CNC machining, where the STEP-NC process planning module is integrated. In this integrated architecture, the capability of driving machining with the data model of STEP-NC and the capability of motion controlling with EtherMAC platform are validated. With this architecture, the intelligent modules and function modules are easy to extend. These indicate that the system proposed in this thesis is feasible, and can provide a basic platform for future researches on intelligent STEP-NC compliant CNC systems.