In this research, an automatic grading machine for FFB is built
and tested through a series of field tests. All four machine subsystems
have satisfactorily passed the tests with minor adjustments.
The machine automatically graded the FFBs without causing injuries
to the bunch. Using Stepwise discriminate analysis, machine
correctly classified the FFBs into two classes, namely rejected (class
1), and accepted (class 2) with success rate of 93.53%. For fraction
classification, IOPRI standard was used. In the algorithm K-means
clustering and squared Euclidean distance analysis were applied,
which showed the success rate of 88.7%. Eight models were used
to estimate FFBs weight, one for each fraction, and showed average
R2 of 0.9603. Development of this automatic grading machine for
oil palm FFB using real-time and non-destructive method provides
the oil palm industries in Indonesia with the first FFB automatic
grading machine that works on-site and grades the FFBs based
on IOPRI grading standard.