摘要
针对批量生产过程的配方作为一种综合性的知识表达,在批量生产过程的建模、软测量、控制和优化方面难以进行定量分析的问题,利用概率神经网络模型提出了一种根据样本间的相似性程度,对数据进行处理的同时对文字之间的顺序和相关性进行测度的新型预测配方的分类方法。仿真结果表明,该配方预测分析方法能够准确地预测配方的类别。
Aimed at the problems that recipe as an integrated knowledge representation was difficult to analyze in batch process modeling,software measurement,control and optimization,a new prediction analysis method for batch process recipe by using of probabilistic neural network(PNN) model was presented.A recipe distance measurement method was built based on the degree of similarity between samples.This approach processed data and took into account the correlation between the texts.Simulation results show that the approach of recipe prediction analysis can accurately predict the recipe categories.
出处
《化工自动化及仪表》
CAS
北大核心
2010年第7期24-27,共4页
Control and Instruments in Chemical Industry
基金
国家自然科学基金资助项目(70671035)