摘要
乙烯裂解炉管在石化行业中至关重要,其炉管质量问题对整个化工过程安全生产也有很大影响,它由离心铸造过程获得。离心铸造过程属于间歇过程,但它在每个批次中也包含离散数据信息,而非时间序列。为了实现对炉管质量的监测,本文利用离心铸造过程中可以获得的参数,提取时间序列中的关键统计信息,形成每个批次的多变量样本信息,用主元分析方法建立炉管质量监测模型,所得到质量监测模型的目前正确率为80%,误报率为10%。通过贡献图的方法对预报结果进行了分析,剖析了引起炉管质量问题的原因,为工业炉管的正常生产提供了指导。
Ethylene cracking furnace tube plays an important role in petrochemical industry. It is manufactured by centrifugal casting. Centrifugal casting process is a batch process, but it also contains information of discrete data in each batch. This paper uses the measurements that obtained from the centrifugal casting process to extract the key statistical information in time sequence and form a set of multivariate variate information for each batch. The principle component analysis is employed to build the model for industrial centrifugal casing process monitoring. The correct identification rate and the false alarm rate of this model are 80% and 10%, respectively. The contribution plots are used to analysis forecasting results and identify root cause for failure batch, which provide a reference for the improvement of tube manufacturing.
出处
《计算机与应用化学》
CAS
2015年第12期1475-1478,共4页
Computers and Applied Chemistry
关键词
主元分析
离心铸造
质量监测
principle component analysis
centrifugal casting process
quality monitoring
