MEASUREMENT AND PREDICTION OF STRESS-STRAIN FOR EXTRUDED OILSEED USING NEURAL NETWORKS UNDER UNIAXIAL COLD PRESSING

摘要

A visualization of testing apparatus was developed to measure property of oilseeds relevant to physical mechanics during mechanical pressing for oil extraction. Stress-strain relationships were measured for extruded peanut, soybean, sesame and linseed compressed at thirteen pressures under uniaxial cold pressing. The prediction model of the stress-strain relationship was developed based on BP neural network. Results indicated that the stress-strain relationships were nonlinear. Over 50% strains for extruded soybean, sesame and linseed occurred at stress below 20MPa. Over 60% strain for extruded peanut occurred at stress below 10MPa. No more than 13% strain occurred at stress over 20MPa for extruded soybean sesame and linseed, and no more than 13% strain occurred at stress over 10MPa for extruded peanut. The maximum error between prediction and measurement for the stress-strain relationship was less than 0.0084 and the maximum training times was less than 88.