The traditional corn deep-processing plant for corn starch to sugar has a complex process,which generates a large amount of industrial data with complex structure,and the production process can be affected by a variety of factors,resulting in large fluctuation in product quality.In order to solve this problem,a method is proposed to construct an agent model considering the uncertainty of data and optimize the operation parameters.First of all,using the actual industrial production data as a data source,artificial neural network is used as an agent model to fit the input and output data,the uncertainty of the data is analyzed through the variance and confidence intervals,and finally the operating parameters are optimized by using genetic algorithm and particle swarm optimization algorithm,respectively with the highest fructose content as the goal.It is found that the fructose content obtained by genetic algorithm optimization is 1.45% higher than that by particle swarm optimization algorithm.The optimization model proposed can be applied to assist industrial production,thus improving product quality.
BieglerL T, LangY D, LinW. Multi-scale optimization for process systems engineering[J]. Computers & Chemical Engineering, 2014, 60(2):17-30.
[4]
ArablooM, ZiaeeH, LeeM, et al. Prediction of the properties of brines using least squares support vector machine (LS-SVM) computational strategy[J]. Journal of the Taiwan Institute of Chemical Engineers, 2015, 50:123-130.
[5]
Hiromi, Baba, Jun-ichi, et al. In Silico predictions of human skin permeability using nonlinear quantitative structure-property relationship models[J]. Pharmaceutical Research, 2015, 32(7):2360-2371.
[6]
ZhouT, JhambS, LiangX, et al. Prediction of acid dissociation constants of organic compounds using group contribution methods[J]. Chemical Engineering Science, 2018, 183:95-105.
[7]
WangK J, WangP S, NguyenH P. A data-driven optimization model for coagulant dosage decision in industrial wastewater treatment[J]. Computers & Chemical Engineering, 2021, 152:107383.
[8]
KwonY, LeeD, ChoiY S, et al. Uncertainty-aware prediction of chemical reaction yields with graph neural networks[J]. Journal of Cheminformatics, 2022, 14(1):1-10.
LoweD, ZapartC. Point-wise confidence interval estimation by neural networks:A comparative study based on automotive engine calibration[J]. Neural Computing and Applications, 1999, 8(1):77-85.
[11]
WuL L, ChangW X, GuanG F. Extractants design based on an improved genetic algorithm[J]. Industrial & Engineering Chemistry Research, 2007, 46(4):1254-1258.
[12]
LiY Q, YangX G, LiuY J. An improved PSO algorithm for solving non-convex NLP/MINLP problems with equality constraints[J]. Computers & Chemical Engineering, 2007, 31(3):153-162.
[13]
JezowskiJ, PoplewskiG, JezowskaA. Simulated annealing optimization in chemical and process engineering Ⅰ.Optimisation method with the use of a simplex and simulated annealing[J]. Inzynieria Chemiczna i Procesowa, 2003, 24(1):47-62.
TongY, ShuM, LiM, et al. A neural network-based production process modeling and variable importance analysis approach in corn to sugar factory[J]. Frontiers of Chemical Science and Engineering, 2023, 17:358-371.
[17]
MengJ, LiC, TaoJ, et al. RNN-LSTM-based model predictive control for a corn-to-sugar process[J]. Processes, 2023, 11:1080.
[18]
MaplesM P, ReichartD E, KonzN C, et al. Robust chauvenet outlier rejection[J]. The Astrophysical Journal Supplement Series, 2018, 238(1):2-21.
[19]
ElkoG W, SondhiM M, WestJ E. Noise reduction processing arrangement for microphone arrays[J]. Journal of the Acoustical Society of America, 1998, 88(6):134-140.
[20]
LinJ, LiS. Sparse recovery with coherent tight frames via analysis Dantzig selector and analysis LASSO[J]. Applied and Computational Harmonic Analysis, 2014, 37(1):126-139.
[21]
ZhangX, LiuL, LongG, et al. Episodic memory governs choices:An RNN-based reinforcement learning model for decision-making task[J]. Neural Networks, 2021, 134:1-10.
PDF
(2265KB)
