摘要
提出了一种基于神经网络与专家系统,具有一定自学习能力的铸造质量控制方法,具体描述了其工作原理。该系统由工艺参数优化模块和缺陷诊断模块组成。工艺参数优化模块以神经网络为推理机,以过去生产数据和有限元数值模拟数据作为训练样本,建立工艺参数与铸件性能之间的非线性关系。缺陷诊断模块以基于产生式规则的推理方式,诊断缺陷类型、产生原因及防治措施,且诊断结果反馈到工艺参数优化模块,用于神经网络再学习。试验结果表明,该系统提高了工艺参数准确率,增强了缺陷诊断能力,减少了铸件次品率。
A quality control approach based on neural networks and expert system is presented, and the working principle is described in detail. The system is composed of technical parameter optimization module and defect diagnosis module. Artificial neural networks（ANN） has been used in technical parameter optimization module, which forms the nonlinear relation between technical parameter and casting performance, based on the experience data and numerical simulation data. Defect diagnosis module is a rulebased expert system, which has the ability of diagnosing and analyzing the defect. At the same time, it will transfer the results to the technical parameter optimization module as feedback. Experimentation results demonstrate that the proposed approach is effective for casting quality control.
出处
《测控技术》
CSCD
2007年第2期38-40,共3页
Measurement & Control Technology
关键词
专家系统
神经网络
学习系统
铸造
expert system
neural network
learning system
casting
