Development of a quality management system based on predictive analytics to prevent the risks of product nonconformity

The article is devoted to the current problem of improving the quality management system for products of industrial enterprises based on mathematical statistics and digital technologies. Expanding the ability to solve production problems using digital twins and cyber-physical systems also includes intelligent forecasting of quality problems and defect risks. The purpose of the article is to develop organizational and technical solutions for the development of a quality management system based on predictive analytics for preventing the risks of product non-conformity. The main research methods are the structuring of production factors by categories and functional characteristics, the formalization of the sequence of operations of predictive analytics, and the mathematical description of the relationships between the objects of the quality management system. The development of a predictive analytics model was carried out using regulatory documents in the field of product quality management, equipment maintenance, production automation and data management, and artificial intelligence. The article highlights the problems of implementing predictive quality analytics, which consist in the technical readiness of production: the availability of reading devices, digitalization and automation of processes, synchronization of databases and hardware and software. A conceptual structural model of predictive analytics of product quality and nonconformity risk management has been developed, which provides for the formation of three databases of defect risks and a database of ready-made predictive models. The result of predictive analytics is the prediction of product quality discrepancies and the ability to change production process settings to prevent defects. A mathematical description of the predictive analytics process has been made with the formalization of a formulaic recording of the approximation of the predictive function and the dependent variable. A diagram of the organization of the quality control system at the stages of the technological process for the production of powdered milk is visualized, indicating control points for data collection and regulatory documentation. The situation of production of low-quality milk powder of reduced solubility was modeled and the production reasons for defects were substantiated. The research materials can be used in the development and implementation of programs for the development of quality management systems at industrial enterprises, modernization of automated systems and software products for production management, planning projects to save resources and reduce losses from defects.