On-line firmness sensing of dates using a non-destructive impact testing device

• Non-destructive falling impact to develop date firmness predictive models.
• PLS models with the entire impact history had a strong potential.
• Conveyor belt speed played an important role in the models power.
• Potential of the method was revealed in developing date sorting lines.

Based upon a non-destructive falling impact test, a rapid on-line device was developed for assessing date fruits firmness. The device consisted of a conveyor unit used for carrying and throwing fruits onto a flat plate connected to a load cell and a data acquisition unit for acquiring the impact force time history. All the tests were carried out at three speeds (1, 1.5 and 2 m s−1) of the conveyor belt. The firmness predictive models were initially developed using simple and multiple regressions with individual impact indices derived from the first half-wave. These indices were also computed for the first twenty positive half-waves of the impact signal and used for further regression analysis. Moreover, all data points of the first half-wave and the entire impact signal were used as independent variables in developing partial least squares regression models. Generally, a substantial increase in prediction power appeared when using the impact indices of the twenty positive half-waves (coefficient of determination in prediction (Rp2) of 0.756 and standard deviation ratio (SDR) of 2.46) instead of the impact indices of the first half-wave (Rp2 = 0.527, SDR = 1.74). Partial least squares models with the entire impact history led to slightly better results (Rp2 = 0.800, SDR = 2.77) as compared to those developed from the first half-wave information (Rp2 = 0.776, SDR = 2.61). The intermediate forward speed of 1.5 m s−1 indicated the best predictions in almost all different methods.