真空注型智能质量控制方法及系统实现

Intelligent quality control method and system implemention for vacuum casting

针对传统真空注型工艺参数的设定大多依靠人为经验而导致浇注件的质量难以控制的问题,考虑到真空注型工艺过程为典型的多变量非线性的间隙工业过程,结合实例推理、神经网络和模糊推理技术,构建真空注型智能质量控制系统。系统采用模块化结构,其中实例推理模块主要利用实例推理技术实现类似浇注实例工艺参数的自动检索;神经网络模块主要利用神经网络技术建立浇注件几何特征和工艺参数的关系模型,从而实现对新浇注实例工艺参数的智能推荐;模糊推理模块主要采用模糊推理技术实现工艺参数的智能修正。通过将该系统用于自制的真空注型物理样机验证了所研究的理论方法的可行性,以及所开发系统的高可靠性。

The Vacuum Casting （VC） process is a typical interval process of multi-variable and non-linear, and most of the traditional process parameters setting mainly depend on experience of operators, which is difficult to control the quality of VC products. For this problem, the intelligent quality controlling system for VC was constructed by combining with Case-Based Reasoning （CBR）, Neural Network-Based Reasoning （NNR） and Fuzzy-Based Reason- ing （FBR）. The module structure was adopted by this system, in which the automated retrieval of CBR module＇s ca- ses process parameters was realized, the relation model between geometry feature and process parameters for NNR module＇s cases process parameters was established, and the intelligent correction of FBR module＇s cases process parameters was realized. The system was applied to physical VC machine, and the result showed the feasibility and re- liability of the proposed system.