The use of near infrared hyperspectral imaging for the prediction of processing parameters associated with the pelleting of biomass feedstocks

Hyperspectral Imaging to Determine the Properties and Homogeneity of Renewable Carbon Materials

“…[21] As no estimate of replicate error within the reference values was available, RDP was calculated as the ratio of population standard deviation to RMSECV.P reviously models with RER values greater than 10 have been considered very good, as RPD values greater than 3w ere found suitable for screening purposes. [5,21] Good prediction performance based on RER and RPD was previously reported for the properties of biocoal [8] and bioenergy crops [22] based on NIRs pectroscopy and the moisture content of biomass pellet feeds [11] based on hyperspectralimaging within the NIR region.…”

“…[9] NIR-based modelsf or char and liquid properties from hydrothermalt reatment have been reported, [10] indicating that char and liquid components can be predicted even based on as maller amount of calibrations amples. In addition, hyperspectral imaging has recently been used for studying pellets made from energy crops, [11] visualizing the extractivec ontents of wood, [12] and characterizing spent mushroom substrate. [13] Herein, we determine the performance of NIR-based hyperspectral imaging in predictingt he properties of hydrothermally prepared carbon on the material and pixel levels.…”

NIR hyperspectral imaging for animal feed ingredient applications

“…Pelleting is one of the processes done in feed manufacturing that directly affects feed quality, aside from contributing to energy consumption and throughput. In a study on biomass feedstocks, hyperspectral imaging (880-1720 nm spectral range) system was used to predict the specific energy required in pelleting, the moisture content of the agricultural feedstocks, and the feeding rate of the feedstocks into the pellet die (Gillespie et al, 2016). For moisture content, the PLSR model developed obtained R 2 of 0.94, RPD of 4.14, RMSECV of 1.11% and latent variables of 7 and these statistics were considered by Gillespie et al (2016) as indication of a robust developed model which could be applied in any purpose.…”

“…In a study on biomass feedstocks, hyperspectral imaging (880-1720 nm spectral range) system was used to predict the specific energy required in pelleting, the moisture content of the agricultural feedstocks, and the feeding rate of the feedstocks into the pellet die (Gillespie et al, 2016). For moisture content, the PLSR model developed obtained R 2 of 0.94, RPD of 4.14, RMSECV of 1.11% and latent variables of 7 and these statistics were considered by Gillespie et al (2016) as indication of a robust developed model which could be applied in any purpose. In prediction of electric consumption, the model, which produced R 2 of 0.64, RPD of 0.91 and RMSECV of 0.12 kWh/kg, was reported as suitable to use for screening purposes only (Gillespie et al, 2016).…”

“…For moisture content, the PLSR model developed obtained R 2 of 0.94, RPD of 4.14, RMSECV of 1.11% and latent variables of 7 and these statistics were considered by Gillespie et al (2016) as indication of a robust developed model which could be applied in any purpose. In prediction of electric consumption, the model, which produced R 2 of 0.64, RPD of 0.91 and RMSECV of 0.12 kWh/kg, was reported as suitable to use for screening purposes only (Gillespie et al, 2016). Then for the prediction of feed rate to pellet die, the developed model had R 2 of 0.70, RMSCEV of 0.20 kg/min and RPD of 1.81 (Gillespie et al, 2016).…”