This article investigates the effect of processing parameters (conditioning temperature and binder content), on final quality of produced agro-pellets for heat energy generation, obtained from four different olive cultivars using different technological parameters. Technological, physical and chemical properties of pellets (carbon, hydrogen, nitrogen and sulphur content, particle density, abrasion length, moisture, ash content, higher and lower heating values, fixed carbon and volatile matter content) have been determined to assess their quality. The performance of Artificial Neural Network (ANN) was compared with the performance of second order polynomial (SOP) model, as well as with the obtained experimental data in order to develop rapid and accurate mathematical model for prediction of final quality parameters of agro-pellets. SOP model showed high coefficients of determination (r2), between 0.692 and 0.955, while ANN model showed high prediction accuracy with r2 between 0.544 and 0.994. [Projekat Ministarstva nauke Republike Srbije, br. III 46005 i br. TR-31055]
展开▼
机译：本文研究了加工参数（调节温度和粘合剂含量）对生产的用于生产热能的农用颗粒的最终质量的影响，该颗粒是从四个使用不同工艺参数的不同橄榄品种获得的。已确定颗粒的技术，物理和化学性质（碳，氢，氮和硫含量，颗粒密度，磨蚀长度，水分，灰分含量，较高和较低的热值，固定碳和挥发性物质含量）以评估其质量。将人工神经网络（ANN）的性能与二阶多项式（SOP）模型的性能以及获得的实验数据进行了比较，以便开发出快速，准确的数学模型来预测农业颗粒的最终质量参数。 SOP模型显示出较高的测定系数（r2），在0.692至0.955之间，而ANN模型显示出较高的预测准确性，r2在0.544至0.994之间。 [Projekat Ministarstva nauke Republike Srbije，br。 III 46005 i br。 TR-31055]
展开▼
