The Software Complex for the Selection and Analysis of Algorithms Predicting Key Quality Indicators of Polymer Film Materials of Industrial Production
Tamara Balabekovna Chistyakova, Anton Damrin, Nikita D. Grishchenkov · 2023
The following algorithms for constructing predictive models of key quality indicators of polymer film materials are considered and implemented: adaptive boosting of decision trees (AdaBoost), recurrent neural network of long short-term memory (LSTM). The characteristics of production data and the stages of their preparation for the construction of predictive models are presented. The comparison of trained predictive models for the defect by the number of black dots per m2 of the film web is carried out by various algorithms according to the main metrics: precision, recall, f1-score, to identify the best depending on the production data.