Neural Networks in Material Humidity Sensors
Svetlana V. Artemova, Maria A. Kamenskaia, Evgenii Sergeevich Mityakov, Chien Vu, Andrey Ivanovich Ladynin, Dmitry N. Lapaev · 2022
The article discusses creating sensors technique for moisture assessment suitable for use in information and control systems regarding drying both bulk and pasty materials. Roller-belt and drum-type dryers developed sensors examples are given. Neural networks training for such sensors is organized according to technological parameters exemplary measures in dryers with a fixed accuracy. Trained neural network inputs are technological parameters normalized measured values that affect drying process directly. From neural network output, material’s normalized moisture content estimate is obtained. Based on the data received from the sensor regarding current humidity inside the drying plant, information system makes a control decision, minimizing the specified drying parameter criteria. The use of such systems in industry makes it possible to increase high-quality material output without drying process productivity losses.