M.M. Fozilov¹, T.B. Chistyakova², A.N. Polosin³

1,2,3 St. Petersburg State Institute of Technology (Technical University) (Saint Petersburg, Russia)

1 mukim.fozilov@yandex.ru; 2 nov@technolog.edu.ru; 3 polosin-1976@mail.ru

The efficiency of large-capacity, multi-assortment, continuous extrusion and extrusion-calendering productions of packing polymeric films in terms of throughput and product quality is directly related to the dependability (reliability, durability) of the operation of extruders preparing melts of polymeric compositions that are formed into films. The difficulty of ensuring the smooth operation of extruders is due to the many types of extruders, the constant change in the assortment of films produced, complex relationships between the parameters of the equipment, process state and dependability measures, the lack of a monitoring system for all parameters of the equipment state. Traditional approaches to information processing become ineffective for big production data (more than 100 thousand values of state parameters and quality indices). Therefore, the development of a computer tool to support decision making for operators to control extruders at productions of polymeric films taking into account actual state of extruders is relevant. The aim of this work is to develop an intelligence system that allows us, on the basis of big production data processing methods, mathematical models for calculating the extrudate quality indices and the extruder dependability measures that are not monitored at production and expert knowledge base of abnormal situations associated with equipment malfunctions, to solve the problems of resource-saving control of extruders and control under abnormal situations. The system includes a subsystem for making advices on resource-saving control of extruders and control under abnormal situations, a subsystem for calculating extrudate quality indices and dependability measures of extruders, an information subsystem, a data visualization subsystem for building trends of monitored and calculated parameters, interfaces for an extruder operator, an administrator, a knowledge engineer. The library of production data processing methods, which is the core of the subsystem for making advices on control, includes the principal component method (to reduce the dimensionality of the data), the linear multiple regression analysis (for synthesizing predictive empirical models), the discrete Fourier transform (for constructing spectra of vibration signals). The library of mathematical models, which is the core of the subsystem for calculating output parameters of extrudate preparation stage, contains models for calculating the average mixing degree, the solid fraction and the thermal destruction index of the extrudate, the reliability function and the failure probability of the extruder, the useful life of the insulation of the extruder screw drive motor. The information subsystem allows us to adjust the system to various production methods, film types and equipment configurations for extrudate preparation stage. It contains relational databases of production methods, films, extruder brands, equipment and process state parameters, extrudate and product quality indices, equipment dependability measures, production-frame knowledge base of abnormal situations, their reasons and recommendations for elimination. The novelty of the work lies in the complex use of big data processing methods, mathematical models and formalized expert knowledge for extruder control. The software package of the intelligence system has been developed using a stack of modern information technologies, which ensures the openness of the architecture and flexibility (customizability to the variable characteristics of the control object) of the system. Testing of the system according to the data of the extrusion-calendering production of the rigid pharmaceutical flat film with polyvinyl chloride as the core constituent at the plant in Russia and the extrusion production of the flexible blown film with low density polyethylene as the core constituent for packaging meat at the plant in Germany has been completed. The test results have confirmed that the system is operational and can be used as an advisor to operators for resource-saving control of extruders and control under abnormal situations. This allows us to reduce the time for making control decisions, increase the time periods between equipment stops, increase the equipment operating life, throughput and quality of the extrudate and film, reduce irretrievable waste (film defects).

Fozilov M.M., Chistyakova T.B., Polosin A.N. Intelligence system for resource-saving control of extrusion equipment
at multi-assortment productions of polymeric films. Dynamics of complex systems. 2020. T. 14. № 4. Р. 21-37.
DOI: 10.18127/j19997493-2