科研之友
微信
新浪微博
Facebook
分享链接
摘要
In complex production processes, there is a complex, nonlinear relationship between product quality and process parameters. In this paper, a product quality prediction method for complex production processes based on model fusion is proposed to improve the accuracy and stability of the final product quality prediction. For the complex production process, the overall prediction model and staged prediction model based on the improved random forest algorithm are first established. Among them, a feature selection method combining correlation analysis and redundancy removal is proposed for the overall prediction model, which is susceptible to low correlation between input and output. An error correction mechanism is proposed for the error accumulation problem of the staged prediction model. Second, the stacking integrated learning algorithm is used to realize the fusion of the overall prediction model and the phased prediction model, and the prediction results of the final product quality are obtained by using the prediction advantages of the two. Finally, taking the prediction of cigarette moisture content at the inlet of the cut tobacco dryer in the tobacco production process as an example, the accuracy and stability of the model fusion prediction method in this paper are verified by comparing the prediction methods of the traditional single model. ? 2023 Chinese Academy of Sciences.
- 单位
