APPLICATION OF ARTIFICIAL NEURAL NETWORKS FOR PREDICTING TEMPERATURE IN A SOLAR DRYER

Abstract

Medicinal plants play an important role in healthcare and the treatment of various diseases. They are widely used in traditional and modern medicine, contain biologically active compounds with various therapeutic properties. Changes in the composition of plants occur mainly as a result of microbiological degradation of the crop, which is associated with an increase in the number of microorganisms depending on storage conditions and humidity levels. To minimize losses in the composition, such basic storage methods as freezing, vacuum packaging, canning, irradiation and drying are used. The drying process is considered the most environmentally friendly method of minimizing losses in the composition, allowing you to preserve the quality, structure and color of the product during long-term storage. Dried products take up significantly less space compared to frozen and canned ones. The drying process takes place due to heat and mass transfer in the dried product. To improve the efficiency of dryers, analyze and predict their operation, their modeling is necessary. Modeling allows taking into account the complex dynamics of the drying process and predict control parameters. Predicting the temperature in the drying chamber increases the ef�iciency of the process and improves the quality of the product. Temperature control allows to effectively reduce the moisture level of the product while maintaining its nutritional properties. Therefore, preliminary forecasting of temperature values during the drying process is a prerequisite for improving product quality and increasing the ef�iciency of the process. In this paper, a series of experiments with a solar dryer were conducted, and its main characteristics were studied. Based on the results of experiments using the arti�icial neural network model, it was predicted that with solar radiation of 673 W/m² and an ambient temperature of 44.3 °C, the temperature in the drying chamber will be 49 °C, and with solar radiation of 700 W/m² and an ambient temperature of 45.7 °C, the temperature in the drying chamber will be 50.8 °C. These results are fully consistent with the experimental data.
The accuracy of the proposed model was the root mean square error RMSE equal to 0.36  C, and the percentage of root mean square error was 0.83%. This method allows predicting the efficiency of the dryer and using it in future scientific and practical research. The proposed method is not limited to solar dryers, it also allows evaluating other types of solar drying technologies.