Refineries often need to find similar crude oil to replace the scarce crude oil for stabilizing the feedstock property. We introduced the method for calculation of crude blended properties firstly, and then created a crude oil selection and blending optimization model based on the data of crude oil property. The model is a mixed-integer nonlinear programming(MINLP) with constraints, and the target is to maximize the similarity between the blended crude oil and the objective crude oil. Furthermore, the model takes into account the selection of crude oils and their blending ratios simultaneously, and transforms the problem of looking for similar crude oil into the crude oil selection and blending optimization problem. We applied the Improved Cuckoo Search(ICS) algorithm to solving the model. Through the simulations, ICS was compared with the genetic algorithm, the particle swarm optimization algorithm and the CPLEX solver. The results show that ICS has very good optimization efficiency. The blending solution can provide a reference for refineries to find the similar crude oil. And the method proposed can also give some references to selection and blending optimization of other materials.
为增强近红外光谱模型通用性,解决直接正交信号校正算法在光谱处理过程中可能出现过拟合、模型不稳定的现象,提出一种将随机森林与直接正交信号校正算法相结合的模型传递方法(Random Forest-Direct Orthogonal Signal Correction,RF-DOSC)。该方法首先利用随机森林算法进行近红外光谱波长点筛选,然后采用直接正交信号校正方法进行光谱处理并建立回归方程,由PLS计算回归系数求得模型传递矩阵。实验使用三台光谱仪(S,S1,S2)测得的玉米近红外光谱数据集建立传递模型,数据集1(D1)水分、油分、蛋白质、淀粉成分预测标准偏差(SEP)分别为0.1267、0.0982、0.1569和0.4051,数据集2(D2)四种成分的SEP分别为0.1548、0.0819、0.1366和0.3836,均小于传统方法。实验结果表明本文所提模型传递方法能有效消除光谱噪声,减小主仪器和从仪器光谱之间的差异,提高模型的稳定性和准确性,实现不同仪器之间模型的共享。
The drug supervision methods based on near-infrared spectroscopy analysis are heavily dependent on the chemometrics model which characterizes the relationship between spectral data and drug categories.The preliminary application of convolution neural network in spectral analysis demonstrates excellent end-to-end prediction ability,but it is sensitive to the hyper-parameters of the network.The transformer is a deep-learning model based on self-attention mechanism that compares convolutional neural networks(CNNs)in predictive performance and has an easy-todesign model structure.Hence,a novel calibration model named SpectraTr,based on the transformer structure,is proposed and used for the qualitative analysis of drug spectrum.The experimental results of seven classes of drug and 18 classes of drug show that the proposed SpectraTr model can automatically extract features from a huge number of spectra,is not dependent on pre-processing algorithms,and is insensitive to model hyperparameters.When the ratio of the training set to test set is 8:2,the prediction accuracy of the SpectraTr model reaches 100%and 99.52%,respectively,which outperforms PLS DA,SVM,SAE,and CNN.The model is also tested on a public drug data set,and achieved classification accuracy of 96.97%without preprocessing algorithm,which is 34.85%,28.28%,5.05%,and 2.73%higher than PLS DA,SVM,SAE,and CNN,respectively.The research shows that the SpectraTr model performs exceptionally well in spectral analysis and is expected to be a novel deep calibration model after Autoencoder networks(AEs)and CNN.
Pengyou FuYue WenYuke ZhangLingqiao LiYanchun FengLihui YinHuihua Yang
