Monitoring and Forecasting Quality Indicators of Predough Production in the Bakery Industry

The work demonstrates the process of developing a machine learning model for a system for monitoring and controlling the quality of predough at a bakery. This system will provide continuous analysis of the predough parameters, such as temperature, humidity, lifting force and titranic acidity, which will ensure timely detection of deviations in the quality of the predough and avoid production problems. In addition, it will help automate the quality control process and ensure the stability of production processes. Analysis and forecasting of dough parameters will improve the quality of produced bread or other bakery products. Management based also on the predicted values of technological indicators of the predough will allow the enterprise to avoid unnecessary costs of materials and energy and increase economic efficiency. As part of the study, seven machine learning models were created and reviewed, including three linear and four varieties of binary decision trees. An analysis of the best model - a regression decision tree - was carried out, confirming its effectiveness and the validity of the resulting hierarchical structure, in particular, the prediction of important indicators of predough with a high degree of reliability, which indicates its importance in practical applications in production. The created model can be used to monitor and control product quality in real time, allowing operators to quickly identify and correct any deviations in process performance. Also, by analyzing changes in the variables of the predough production process, it is possible to identify precursors of possible production accidents or deviations in product quality, which will allow timely measures to be taken to avoid them.