【作者】 胡世广；

【导师】 王太勇；

【作者基本信息】 天津大学 ， 机械制造及其自动化， 2008， 博士

【摘要】 在现代制造领域,数控系统的广泛应用极大推动了制造技术的进步。随着市场需求的不断提升以及相关支撑技术的发展,数控系统正在经历一次以开放性与功能复合化为突出特征的技术变革,由此必将引起系统体系架构与控制策略的变革性发展。针对于此,本文在数控系统开放式体系、结构建模、动态分析以及系统智能策略四个方面进行了深入探讨。基于对数控系统功能模块的综合规划,提出了一种四层结构的数控系统开放体系参考模型。给出了一种层级链式模块注册策略及注册单元体的形式化结构定义,可实现系统局部结构调整。完成了基于功能组件的用户专有技术配置机制的分析。分析表明该参考模型可支持三级程度的系统配置。提出了基于面向对象技术的数控系统建模方案及其标准化建模流程。在对系统用例深入分析的基础上,重点建立了系统的核心用例包—运动控制包对应的子系统的类图模型。基于此对类对象行为进行了分析,构建了通道伺服控制序列图及通道状态图模型,进而通过活动图对系统操作逻辑进行了建模分析。引入网论进行数控系统运行时动态性能分析。提出了决策库所的形式化定义以解决环境信息外延不完整带来的冲突问题,进而完成了对系统总体结构的仿真验证。引入形式化数学分析方法对DNC过程进行了分析。提出了带抑制弧赋时Petri网的等效模式以降低模型维度与分析复杂度,基于此进行了实例仿真分析并给出了相应的系统结构调整规划。基于着色Petri理论对多轴伺服控制的分析模型进行了压缩,表明该方法可有效精简具有相似结构的系统分析模型。对系统智能策略的信息作用范围及加工作业周期内各阶段智能策略的实施目标进行了探讨,提出了系统运行时分类智能控制策略。基于自适应遗传算法给出了加工参数预优化方案及算例。提出了基于电流与电压综合监测的加工过程模糊自适应控制方案并建立了控制模型,进而引入径向基网络进行了加工自适应控制过程的仿真分析。结果表明所建立的模糊自适应控制系统能够以较高的辨识与控制精度实现数控加工的智能自适应控制。基于上述研究成果,已开发出五联动数控系统功能样机,并完成了与五轴加工中心的配套工作,初步实现了对相关理论成果的实例化验证。更多还原

【Abstract】 In the field of modern manufacturing, promotion of manufacturing technology is significantly encouraged by the wide application of CNC system. With the increase of market requirements and the development of relevant supporting technologies, CNC system is now undergoing a technological innovation that is characterized mostly by openness and functional composition, which will certainly result in the innovation of structure and control strategy of the CNC system. According to this situation, this dissertation makes deep research on open architecture, structure modeling, dynamic analysis and intelligent strategy.Based on the comprehensive planning of function module of CNC system, a four-layer open architecture reference model is proposed. A hierarchical chain-typed registration strategy of function module is given, and the definition of registration unit structure is presented. With this strategy, local structure of the CNC system can be adjusted. User-defined configuration of specialized technology can be archived by the presentation of function component. Further analysis shows that the reference model supports three deferent degrees of system configuration.A modeling strategy of the CNC system based on object-orient technology is proposed, and a standard modeling flow is discussed. On the basis of deep research on use-case of the system, a class diagram model of the kernel use-case package, in other words, the motion control package of the system is built. Behavior of each class is discussed, on the basis of which, a sequence diagram of channel servo-control system under auto-mode is built, and a state machine diagram is presented. Finally, operation logic of the system is analyzed with an activity diagram.Net theory is introduced to the analysis of dynamic performance of the CNC system. A decision place is defined to get rid of conflicts caused by the imperfection of extension of environment information. With this definition, simulation is carried out to verify the general structure of the system. Analysis method based on formalized mathematics is introduced to analyze DNC process. An equal transformation of timed transition Petri net with inhibitor arc is proposed to descend its dimension and to decrease analyzing difficulty. On the basis of this, a simulation example is carried out and a strategy of structure adjusting is given according to analysis conclusions. Further simulation illustrates the effect of the adjustment. Theory of colored Petri net is introduced to condense the Petri net model of multi-axis servo control progress. Analysis shows that with this technology, the system’s Petri net models with similar structure units can be simplified effectively.By discussing the influencing range of intelligent strategy and control target of each stage of machining cycle, a runtime intelligent strategy based on information classification is proposed. Adaptive GA is introduced to find optimization cutting parameters at the earlier stage of machining cycle. A fuzzy adaptive machining control strategy based on the synthetic monitoring on current and voltage of spindle motor is put forward, and the relevant control model is given. On the basis of that fuzzy control model, RBF neural network is introduced to build a simulation model of the adaptive control system. Simulation analysis proves that the fuzzy adaptive control system is able to realize intelligent control of NC machining with higher identification as well as control precision.A CNC system prototype with five-axis simultaneous control power is developed based on the foregoing research achievements, and is integrated to a five-axis machining center, which presents a preliminary practical verification of relevant theoretical achievements.更多还原