Bahaiev I. Іmprovement of qualimetric methods for assessing the quality risks of technological processes using information technology

The object of research is risk assessment of the quality of technological processes. The subject of the study is the use of information technology to improve the qualimetric methods of risk assessment in the operation of technological processes. The purpose of the study is to improve the methods of assessing the risks of poor quality of technological processes, taking into account the nonlinearity of the dependence of the measured values of individual indicators of product quality, as a result of the functioning of the technological process, and their estimates on a dimensionless scale and the use of effective mathematical apparatus of information technology. For the first time: a functional dependence between the actual values of the quality indicators of the technological process and their estimates on a dimensionless scale was obtained, which, unlike the existing ones, does not require the use of expert methods and is based on the geometric principle of dividing a segment in a given ratio; a method for assessing the risks of manufacturing low-quality products using the proposed functional dependence and Markov chain theory has been developed, which, unlike the existing ones, allows obtaining the probability of finding estimates of quality indicators in the specified control zones; a method for assessing the risks of manufacturing low-quality products based on the application of artificial intelligence methods and the use of the proposed functional dependence has been developed, which, unlike the existing ones, allows predicting the finding of quality indicators in the specified regulatory zones. The introduction substantiates the relevance of the topic, formulates the purpose and objectives of the research, defines the scientific novelty and practical value of the results obtained, and provides a general description of the work. The first section analyses 60 scientific publications. The development of social production in any country, its place in the world trade market, growth of the nation's welfare and living standards are closely linked to ensuring a high level of quality of products, services and functioning of socio-economic systems. The production of quality products is linked to the quality of technological processes at all stages of the production life cycle. Any technological process is associated with the risks of deviation of standardised product quality indicators from the ideal ones, i.e. those regulated by regulatory documents and technical regulations. In the second section, it is proposed to study and apply the logistic function to obtain estimates of unit quality indicators of any qualimetry objects on a dimensionless scale. The application of the logistic function allows obtaining a mathematical dependence that provides estimates of quality indicators on a dimensionless scale, regardless of the nature of the qualimetry object. To do this, it is necessary to know the minimum and maximum permissible values of qualimetry objects. Examples are given to test the methodology for applying the proposed functional dependence. It is proposed to evaluate products, processes and systems as qualimetry objects. The third section considers the possibility of applying one of the forecasting methods, for example, methods based on Markov chains. This method makes it possible to obtain high-quality forecasts for linear and nonlinear processes. The task of evaluating probabilistic forecasts is relevant, as the need for such forecasts is growing rapidly both in purely engineering applications and in financial and economic systems. A technological process is a complex system that needs to be evaluated, analysed, forecasted and, if necessary, adjusted. The article develops a methodology and software for predicting the production error for a particular technological process. It is shown that it is advisable to make a forecast only 4-5 steps ahead. If the Automatic Quality Management System is applied, this can lead to a 16% reduction in the variation of quality indicators. In the fourth section, it is proposed to apply one of the artificial intelligence tools to forecasting the quality and risks of the technological process. The author develops a methodology for using neural networks to assess the risks of process quality in the event of its non-stationarity. The corresponding methodology has been tested and the methodology has been tested in the form of a computational experiment. According to the results of computational experiments, it is proved that the use of the developed neural network models allows obtaining predictive estimates of stability and accuracy of technological processes of machining shaft-type parts with a reliability of 90%-96%. The developed neural network models can be used in automated systems to generate control influence and prevent deviations of part parameters from regulated values online when controlling the accuracy of the machining process.