Electronic Theses and Dissertation

Kelor (moringa oleifera) merupakan tanaman yang dikenal memiliki nilai gizi dan khasiat farmakologis yang tinggi. salah satu kandungan penting dalam daun kelor adalah zat besi (fe), yang berperan penting dalam pembentukan hemoglobin dan metabolisme tubuh. daun kelor, khususnya dalam bentuk serbuk, diketahui mengandung zat besi mencapai sekitar 60,5 mg/100 gram bahan. selama ini, analisis kadar fe pada daun kelor umumnya dilakukan menggunakan metode destruktif berbasis spektroskopi laboratorium. berdasarkan hal tersebut, penelitian ini dilakukan untuk mengembangkan alternatif sensor yang mampu memperkirakan kandungan fe pada tepung daun kelor secara non-destruktif. sensor yang digunakan adalah triad spectroscopy as7265x produksi sparkfun electronics, yang dalam penelitian ini dirancang menjadi bentuk portabel agar lebih praktis digunakan di lapangan. penelitian ini telah dilaksanakan di laboratorium instrumentasi dan energi program studi teknik pertanian fakultas pertanian serta di laboratorium kimia instrumen fakultas matematika dan ilmu pengetahuan alam universitas syiah kuala. sebanyak 16 sampel tepung daun kelor digunakan, yang berasal dari 8 kabupaten di indonesia. nilai aktual kandungan fe dari setiap sampel diperoleh melalui analisis menggunakan alat atomic absorption spectroscopy (aas). analisis data dilakukan meliputi uji outlier, perbandingan spektrum tanpa dan dengan pretreatment, serta perancangan model prediksi kadar fe menggunakan metode partial least square (pls). evaluasi model dilakukan dengan menggunakan parameter koefisien determinasi (r²), root mean square error of calibration (rmsec), mean absolute error (mae), dan residual predictive deviation (rpd). rentang panjang gelombang yang digunakan dalam analisis adalah 410–940 nm. hasil penelitian menunjukkan bahwa teknologi ini memiliki beberapa keunggulan, di antaranya proses pengukuran yang cepat, simultan, tidak merusak sampel, dan tidak memerlukan bahan kimia tambahan. berdasarkan hasil loading plot, panjang gelombang optimum diperoleh pada kisaran 670–690 nm. model pls menghasilkan nilai r² sebesar 0,850, rmsec sebesar 0,413, mae sebesar 0,341, dan rpd sebesar 2,7. setelah dilakukan pretreatment menggunakan standard normal variate (snv), kinerja model meningkat dengan nilai r² sebesar 0,880, rmsec sebesar 0,369, mae sebesar 0,275, dan rpd sebesar 3. berdasarkan nilai-nilai tersebut, model prediksi kandungan fe yang diperoleh termasuk dalam kategori sangat baik (excellent prediction accuracy).

Moringa (Moringa oleifera) is a plant known for its high nutritional value and pharmacological properties. One important component of moringa leaves is iron (Fe), which plays a vital role in hemoglobin formation and body metabolism. Moringa leaves, especially in powder form, are known to contain approximately 60.5 mg of iron per 100 grams of material. To date, analysis of Fe levels in moringa leaves has generally been conducted using destructive methods based on laboratory spectroscopy. Based on this, this study was conducted to develop an alternative sensor capable of non-destructively estimating the Fe content in moringa leaf powder. The sensor used is the Triad Spectroscopy AS7265x produced by Sparkfun Electronics, which in this study was designed in a portable form for more practical use in the field. This research was conducted at the Instrumentation and Energy Laboratory of the Agricultural Engineering Study Program, Faculty of Agriculture, and at the Instrumental Chemistry Laboratory of the Faculty of Mathematics and Natural Sciences, Syiah Kuala University. A total of 16 samples of moringa leaf flour were used, originating from 8 districts in Indonesia. The actual Fe content of each sample was obtained through analysis using an Atomic Absorption Spectroscopy (AAS) tool. Data analysis included outlier testing, comparison of spectra without and with pretreatment, and designing a prediction model for Fe content using the Partial Least Square (PLS) method. Model evaluation was carried out using the parameters of the coefficient of determination (R²), Root Mean Square Error of Calibration (RMSEC), Mean Absolute Error (MAE), and Residual Predictive Deviation (RPD). The wavelength range used in the analysis was 410–940 nm. The results of the study indicate that this technology has several advantages, including a fast, simultaneous measurement process, does not damage the sample, and does not require additional chemicals. Based on the loading plot results, the optimum wavelength was obtained in the range of 670–690 nm. The PLS model produced an R² value of 0.850, RMSEC of 0.413, MAE of 0.341, and RPD of 2.7. After pretreatment using the Standard Normal Variate (SNV), the model performance increased with an R² value of 0.880, RMSEC of 0.369, MAE of 0.275, and RPD of 3. Based on these values, the obtained Fe content prediction model is included in the excellent prediction accuracy category.