Prediksi Kandungan Kimia Buah Kelapa Sawit Secara Nondestruktif Berdasarkan Karakteristik Admitansi Listrik dan Metode Kalibrasi Principal Component Regression

Abstract

Industri minyak sawit Indonesia berkembang pesat dan berkontribusi terhadap industri minyak nabati dunia. Penentuan kematangan buah kelapa sawit secara konvensional kurang akurat, metode destruktif membutuhkan waktu dan biaya yang besar sehingga diperlukan metode nondestruktif untuk mengurangi waktu dan biaya. Penelitian ini bertujuan memprediksi kadar air, minyak, dan asam lemak bebas (ALB) buah kelapa sawit berdasarkan admitansi listriknya menggunakan principal component regression (PCR). Sampel diukur sifat admitansi listriknya pada frekuensi 50 Hz-5 MHz menggunakan LCR meter, selanjutnya diukur kadar air, minyak, dan ALB menggunakan metode kimia. Data admitansi diolah menggunakan dua pre-treatment dan kemudian hasilnya dikalibrasi dengan data kimianya. Prediksi terbaik kadar air adalah admitansi tanpa menggunakan pre- treatment (PC-13) (r =0,91, SEC=8,87%, SEP= 9,63%, CV = 15,20%, RPD = 2,49, konsistensi = 102,80%). Prediksi kadar minyak terbaik diperoleh menggunakan admitansi dengan pre-treatment deresolve (PC-18) (r = 0,92, SEC = 7,26%, SEP = 7,21% CV = 44,84%, RPD = 2,49, konsistensi = 100,69%). Sedangkan prediksi kadar ALB terbaik didapat menggunakan admitansi dengan pre-treatment normalization (PC-14) (r = 0,70, SEC = 1,46%, SEP = 1,46%, CV = 39,67%, RPD = 1,29, konsistensi = 100,20%). Metoda impendansi dan PCR yang dikembangkan dapat digunakan untuk memprediksi kadar minyak dan kadar air buah sawit secara nondestruktif, sedangkan untuk prediksi kadar asam lemak bebas belum dapat diterapkan.

The Indonesian palm oil industry is rapidly developing and contributes significantly to the global vegetable oil industry. Conventional methods for determining oil palm fruit ripeness are often inaccurate, and destructive methods are time-consuming and costly. Therefore, a nondestructive method is needed to reduce both time and expense. This research aims to predict the moisture, oil, and free fatty acid (FFA) content of oil palm fruit based on its electrical admittance using Principal Component Regression (PCR). Sample’s electrical admittance properties were measured at frequencies from 50 Hz to 5 MHz using an LCR meter, and their moisture, oil, and FFA content were subsequently determined using chemical methods. Admittance data were processed using two pre-treatments, and the results were then calibrated with the chemical data. The best prediction for moisture content was achieved using raw admittance data without pre-treatment (PC-13), yielding an (r =0.91, SEC=8.87%, SEP= 9.63%, CV = 15.20%, RPD = 2.49, consistency = 102.80%). The best oil content prediction was obtained using admittance data with deresolve pre-treatment (PC-18), resulting in an (r = 0.92, SEC = 7.26%, SEP = 7.21%, CV = 44.84%, RPD = 2.49, consistency = 100.69%). Meanwhile, the best FFA content prediction was achieved using admittance data with normalization pre-treatment (PC-14), yielding an (r = 0.70, SEC = 1.46%, SEP = 1.46%, CV = 39.67%, RPD = 1.29, consistency = 100.20%). The developed impedance and PCR method can be used for nondestructive prediction of oil and moisture content in oil palm fruit, but it is not yet applicable for predicting free fatty acid content.