Methods and Technologies for Primary Processing, Modeling and Visualization of Industrial Data for Quality Control of Multi-assortment Polymeric Films

The methods of primary processing, modeling, visualization of data of large-capacity production of polymeric films of a wide range are described. The methods are implemented in the product quality control system. Industrial data is characterized by a big volume, high accumulation rate, a variety of sources, incompleteness, encoded values, unstructured nature. Therefore, their primary processing consists in converting them into real values, filtering, structuring by types of production parameters. The lack of measurement of a number of important quality indices of the polymeric material at the production stages, including the quality indices of the extrudate, causes incompleteness of data about the control object. This incompleteness led to the creation of a library of adjustable mathematical models for calculating quality indices depending on the control actions measured in production, with different configurations of aggregates and types of films. Complete arrays of measured and calculated parameter values are stored in a database for subsequent analysis, processing (by methods of mathematical statistics and machine learning) and visualization. The data are visualized in the form of parameter trends and histograms for distributions of the specific number of defects of various types across the film rolls. Visualization allows you to study the influence of control actions and technological additives (for example, stabilizers) on the quality of the film to select a rational recipe and production mode of the film. The proposed methods are implemented in the form of a flexible software package. It is adjusted to the range of films produced, configurations of production lines and is designed to support decision-making on the control of consumer characteristics of polymeric films. Testing of the software package according to the data of industrial extrusion-calendering production of rigid packaging films based on polyvinyl chloride has confirmed its operability. The effectiveness of the package is achieved by reducing the preparation time of big data and ensuring the completeness of big data, saving expensive raw materials.