摘要
针对传统神经网络专家系统解释机制不健全及无法提供推理过程的不足,采用结合质量分析及追溯的专家系统,在进行热轧型钢产品各项参数检测后,生成质量分析结果并进行质量追溯,进一步对检测结果进行详细解释说明。通过纵向追溯和横向追溯分别为专家系统提供解释结果和推理过程。此外,根据钢铁行业的产品特点,对物理性能检测部分应用检测-预测集成一体的复合神经网络,简化了实际生产过程中物理性能参数检测工作环节并提高了成品质量检测工作的自动化程度。
Traditional neural network expert system has deficiency in imperfect explanation mechanism and not able to provide the reasoning process, in light of this, we adopt an expert system which combines the quality analysis and tracing to generate quality analysis result and to trace back the quality after every parameter df hot rolled steel products is detected, and give further expatiation on the detection results. Through longitudinal and horizontal tracing the explanation results and the reasoning process are provided to the expert system respectively. Besides, according to the characteristics of the products in iron ＆ steel industry, the use of the integrated-to-one detection and forecast composite neural network in physical properties detection part simplify the aspect of detection work in regard to physical property parameters in practical production process and enhance the automation degree of products quality detection.
出处
《计算机应用与软件》
CSCD
北大核心
2013年第7期185-187,227,共4页
Computer Applications and Software
基金
河北省重点基础研究项目(09963536D)
天津师范大学博士基金项目(52XB1001
52X0 9013)
关键词
专家系统
径向基神经网络
质量分析
质量追溯
Expert system Radial basis function neural network Quality analysis Quality tracing
