论文部分内容阅读
热值是反映农作物剩余物作为燃料利用潜力的重要参数。运用氧弹方法测试热值费时费力。基于可见-近红外光谱分析技术,分别对采用了不同预处理方法处理的光谱建立了5种农作物秸秆的偏最小二乘法(PLR)和主成分回归方法 (PCR)热值模型,分析了预测模型的准确性与稳定性。结果显示,10点平滑的PLR模型效果最好,预测相关系数R2和预测标准差RMSEP分别为0.853 7和0.443 4。过多的平滑点处理产生了过平滑现象,导致模型性能变坏,采用了微分处理和多元散射校正(MSC)处理的光谱预测模型性能未见明显提高。研究结果可为热值快速测试设备的研发提供基础数据和数学模型优化支持。
Calorific value is an important parameter that reflects the potential of crop residues as fuel utilization potential. It is time consuming and laborious to test the calorific value by the oxygen bomb method. Based on visible-near-infrared spectroscopy (NIRS), partial least squares (PLR) and principal component regression (PCR) calorific value models of crop straw were established for the spectra treated by different pretreatment methods. The prediction model The accuracy and stability. The results show that the 10-point PLR model is the best, the prediction correlation coefficient R2 and prediction standard deviation RMSEP are 0.853 7 and 0.443 4, respectively. Too smooth smoothing caused by over-smoothing resulted in poor performance of the model. The performance of spectral prediction model using differential processing and multivariate scatter correction (MSC) did not improve obviously. The results provide basic data and mathematic model optimization support for R & D of rapid heating value test equipment.