机构地区: 东北大学材料与冶金学院轧制技术及连轧自动化国家重点实验室
出 处: 《东北大学学报(自然科学版)》 2009年第7期989-992,共4页
摘 要: 鉴于连铸过程中存在众多不确定性,考虑某中厚板厂的实际生产状况与中厚板热送热装的控制要求,以表面纵裂为例,按照模糊专家系统的开发思路,建立起连铸坯纵裂知识库,开发推理机制.采用纵裂指数预测和神经网络预测相结合的方法对连铸坯质量进行离线判断、推理和决策,并给出相应的缺陷产生原因解释.采用VC++6.0软件建立人机界面,数据库采用Oracle,调试结果表明该系统可以达到较高的判定准确率(92.8%),为在线应用奠定了基础. In view of so many uncertainties found in the continuous casting process, actual operating conditions and controlling requirements for hot-charging and hot-delivery in plant, a reasoning approach is proposed taking the longitudinal cracks on slabs surface as example to set up a relevant knowledge database in accordance to the thought to build fuzzy expert system. Then, the off-line judgement, reasoning and decision-making are available to the quality of continuous casting slabs by means of the prediction combining the longitudinal cracks index with artificial neural network, with the reason why a defect forms given. With the software VC++ 6.0 and Oracle database used, the man-machine interface is built. The debugged results showed that in this way the judgement accuracy can be up to 92.8%, thus laying a foundation for its online applications.
关 键 词: 连铸坯质量 模糊专家系统 纵裂 知识库 神经网络 人机界面
领 域: [自动化与计算机技术] [自动化与计算机技术]