|
| Model optimization for determination of moisture, fat, protein and ash in chia seed by near infrared spectroscopy |
| |
| DOI：10.12166/j.zgyz.1003-7969/2020.07.029 |
| KeyWord:chia seed near infrared spectroscopy partial least-squares quantitative model |
| FundProject:国家重点研发计划（2018YFC1602300） |
|
| Hits: 1543 |
| Download times: 540 |
| Abstract: |
| Determination of moisture, fat, protein and ash, four important quality indexes of chia seed, are based on chemical method, which is tedious, time-consuming, laborious, and can’t monitor multiple indexes simultaneously. 103 chia seed samples from various producing areas were collected. A total of 19 kinds of spectral pretreatment and selection of optimal spectral range, the partial least-squares models of moisture, fat, protein and ash in chia seed were established. The results showed that the spectra were processed by Savitzky-Golay filter (SG) for moisture; the combination of first derivative (1st) and multiplicative signal correction (MSC) was found to be the best preprocessing method for fat model; protein was pretreated with 1st, standard normal variate (SNV) and SG; for the detection of ash, 1st, SNV and Norris derivative filter (ND) spectral pretreatment method was better. Furthermore, new models were developed with the selected optimal pretreatments by artificial process. For moisture, fat, protein and ash, the correlation coefficients of prediction set were 0.993, 0.972, 0.925 and 0.923, respectively. Accordingly, near infrared spectroscopy could achieve the simultaneous, rapid and nondestructive detection of moisture, fat, protein and ash in chia seed to improve the detection efficiency for massive samples. |
| View full text View/Add Comment Download reader |
| Close |
|
|
|
