Fedin S. Quality maintenance of piston group details by on-line forecasting method

The object is statistical regulation of standard technological processes of production of responsible precision details of the cylindrical form; the aim is developmen of normative documentation and mechanism of operating accuracy and stability forecasting of the technological process of production of piston group details with usage of adaptive systems of an artifical intelligence; the methods are the product quality control, system approach, regression and correlation analysis, probability theory and mathematical statistics, theories of risk and artificial of neural networks; the novelty - the method of operating forecasting of accuracy and stability of the technological process which allows without the registration of the probability characteristics to predict a state of the process in limits or outside boundaries of regulation defined on a check card, and to carry out both continuous, and sampling of pistons quality is developed; results - the usage technique of neural networks for short-term prediction of single and complex metrics of production quality it is offered which includes: choice of the technical and economic factors of manufacture; realization of the correlation analysis of temporary numbers of the selected factors and of forecasting metrics; training, adjusting and maintenance of the neural network; a degree of implantation - the results of operation were approved and are used at manufacture of piston details group on firms of the machine-building profile of Ukraine as the guidelines on regulation of technological processes of processing details with application of neural networks; branch - quality management, standardization and certification.