Comparison of Prediction Models for Determining the Degree of Damage to Korla Fragrant Pears

For a fast and accurate evaluation of the values damaged fragrant pears, prediction method damage degree Korla pears was proposed. To study variation laws damages under different volumes squeezing deformation, partial least squares regression (PLSR), generalised neural network (GRNN) adaptive fuzzy inference system (ANFIS) were chosen to predict establish optimal model. The results demonstrated that with increase ripeness or deformation value, increases gradually. performance comparison models based on PLSR, GRNN ANFIS, it found trained ANFIS can all pears. which inputs membership function dsigmf (R2 = 0.9979, RMSE 46.6) psigmf 46.6), achieves best performance. Research provide theoretical references commodity value quality grading design picking machine.