摘要
多牌号产品生产过程经常涉及到牌号切换,而切换后新牌号生产过程的变量关系可能随之发生变化,故采用单一的故障检测和诊断方法,无法对多牌号产品连续生产过程出现的异常做出有效的判断。这就需要及时准确地识别出新牌号,并对每个牌号有相应的故障检测和诊断模型。为此,本文引入人工神经网络(ANN),将其用于牌号识别,提出了牌号识别和主成分分析(PCA)相结合的方法,即利用历史数据建立各个牌号的BP神经网络(BPNN)模型和PCA模型,在线数据经过BPNN识别确认牌号类型后,调用对应牌号的PCA模型进行故障检测和诊断。结果表明,BPNN不仅可以准确识别牌号,识别率较规格界限法更高,而且可以对牌号过渡过程进行判断。另外,与不进行牌号识别仅采用单一牌号正常样本或者所有牌号正常样本混合建立的PCA模型相比较,采用牌号识别后进行故障检测时的精度更高,证明了该方法的有效性。
Multi-grade production process often involves the grade transition, and the relationship between variables in the process of new grade production may be changed along with it after transition, so a single approach of fault detection and diagnosis can not make an effective judgments for abnormal condition in the continuous multi-grade production process. Therefore, new grade first need to be identified timely and accurately, and a corresponding fault detection and diagnosis model also need to be established for each grade. To this end, artificial neural network （ANN） is introduced for grade recognition in this paper, an approach which combines grade recognition with principal component analysis （PCA） is proposed, namely, historical data is used to establish BP neural network（BPNN） model and PCA model for various grades, on-line data is recognized to confirm the current grade type by using BPNN, and then PCA models are called corresponding to the grade to perform fault detection and diagnosis. The results show that, BPNN not only can accurately recognize the grade type, and recognition rate is higher than the specification limits method, but also can be used to determine the grade transition process. In addition, compared to PCA-based fault detection only using PCA model established by employing normal samples of a single grade or mixed normal samples of all grades but without grade recognition, PCA-based fault detection after grade recognition is more accurate, and effectiveness of the approach is proved.
出处
《计算机与应用化学》
CAS
CSCD
北大核心
2010年第5期703-708,共6页
Computers and Applied Chemistry
关键词
神经网络
PCA
牌号过渡
牌号识别
故障诊断
neural networks, PCA, grade transition, grade recognition, fault diagnosis
