Machine Learning Based Production Control And System Development For Tail Coil

The high rate of tail coil seriously affects the profitability of the cold rolling mill.It is always the goal that enterprise pursues to realize the accurate prediction of the tail coil and improve the technical parameters and the quality of the product in the production process through the large-scale production data.This thesis takes the steel coil process data produced by a certain steel company for 3 months as the analysis object,and based on the machine learning method,extracts the key process parameters that affect the product quality,and then builds the prediction model of whether the tail roll occurs.The main tasks as follows:(1)Aiming at the problems of data missing and data anomalies in actual industrial data,an effective data cleaning method is proposed,which can be efficiently deal with missing values,outliers,single values and repeated values in data sets so that the original data set can be normalized and modeled to build the model(2)Combining the statistical and machine learning analysis methods,the feature selection process of tail coil production control is proposed,and the feature subset which has the strongest correlation with the target variables in the data set is extracted to improve the interpretability and classification accuracy of the model.Based on the characteristics of screening,a Logistic prediction model for production control and prediction of tail coil was established.The classification accuracy of the model is verified through the test set,and the classification results are shown by combining the confusion matrix and the ROC curve.(3)Based on the forecasting model of tail coil production control established above,a data-driven optimization model aiming at minimizing the cost of quality control is established,and a particle swarm optimization algorithm is designed to solve the problem.The correctness of the model and the effectiveness of the algorithm are verified by computational experiments.(4)Based on the above research results,a data visualization analysis software is designed and developed,which can basically realize the data analysis function of iron and steel industry.In order to apply machine learning method to the field of product production control in iron and steel enterprises,a useful exploration is made.