Pengembangan Model Prediksi Kandungan Kimia Buah Kelapa Sawit Secara Non-Destruktif berbasis Electrical Impedance Spectroscopy (EIS)

Abstract

Minyak sawit merupakan komoditas strategis Indonesia yang mutu bahan bakunya ditentukan oleh kandungan minyak, kadar asam lemak bebas, dan kadar air. Analisis parameter tersebut umumnya masih menggunakan metode destruktif yang memerlukan waktu lama sehingga kurang sesuai untuk pengukuran cepat. Penelitian ini bertujuan menentukan karakteristik listrik buah kelapa sawit pada rentang frekuensi 50 Hz–5 MHz, menganalisis hubungannya dengan kandungan minyak, kadar asam lemak bebas, dan kadar air, serta mengembangkan model prediksi non-destruktif berbasis Electrical Impedance Spectroscopy (EIS). Pengukuran meliputi impedansi, resistansi, admitansi, induktansi, kapasitansi, konduktansi, dan reaktansi. Data dimodelkan menggunakan regresi linier, Partial Least Squares (PLS), Multiple Linear Regression (MLR), dan Artificial Neural Network (ANN). Hasil menunjukkan bahwa impedansi dan resistansi menurun, sedangkan admitansi dan konduktansi meningkat seiring kenaikan frekuensi dan kematangan buah. Admitansi memberikan hubungan terbaik dengan kandungan minyak dan kadar asam lemak bebas, sedangkan impedansi terbaik untuk kadar air. Model terbaik menghasilkan R² 0,98 (RPD 7,99) untuk minyak, R² 0,96 (RPD 5,27) untuk kadar asam lemak bebas, dan R² 0,98 (RPD 7,02) untuk kadar air. Hasil ini menunjukkan bahwa EIS multi-frekuensi berpotensi menjadi metode cepat, akurat, dan non-destruktif untuk penentuan mutu buah kelapa sawit.

Palm oil is a strategic commodity that significantly contributes to Indonesia’s economy, with the quality of fresh oil palm fruit being primarily determined by its oil content, free fatty acid (FFA) content, and moisture content. Conventional analytical methods for these parameters are destructive, time-consuming, and unsuitable for rapid quality assessment. This study aimed to characterize the electrical properties of oil palm fruit over a frequency range of 50 Hz–5 MHz, investigate their relationships with oil, FFA, and moisture contents, and develop non-destructive prediction models based on Electrical Impedance Spectroscopy (EIS). Electrical parameters, including impedance, resistance, admittance, inductance, capacitance, conductance, and reactance, were measured and modeled using linear regression, Partial Least Squares (PLS), Multiple Linear Regression (MLR), and Artificial Neural Networks (ANN). The results showed that impedance and resistance decreased, whereas admittance and conductance increased with increasing frequency and fruit maturity. Admittance exhibited the strongest relationship with oil and FFA contents, while impedance showed the highest correlation with moisture content. The best prediction models achieved R² values of 0.98 (RPD = 7.99) for oil, 0.96 (RPD = 5.27) for FFA, and 0.98 (RPD = 7.02) for moisture. These findings demonstrate that multi-frequency EIS is a rapid, accurate, and non-destructive approach for assessing the quality of oil palm fruit.