Raw meal fineness is the percentage content of 80 µm sieving residue after the cement raw material is ground. The accurate prediction of raw meal fineness in the vertical mill system is very helpful for the operator to control the vertical mill. However, due to the complexity of the industrial environment, the process variables have coupling, time-varying delay and nonlinear characteristics in the grinding process of cement raw material. At present, few people pay attention to the coupling characteristics among variables, thus solving this problem is particularly important in raw meal fineness prediction. In this article, we propose a two-dimensional convolutional neural network method that is used to predict raw meal fineness during the grinding process of raw material. Convolutional neural network has strong feature extraction capabilities and does not require manual feature selection. The two-dimensional convolution kernels are used to extract the coupling, time-varying delay and nonlinear features among variables, especially the coupling features. In addition, two important parameters P and L of two-dimensional convolutional neural network model are optimized, respectively. The optimized model solves the problems of coupling, time-varying delay and nonlinearity among variables. Our two-dimensional convolutional neural network model is proved to be very effective by comparing with the state-of-the-art methods.
MulderEde JongTPFeenstraL.Closed cycle construction: an integrated process for the separation and reuse of C&D waste. Waste Manag2007; 27: 1408–1415.
2.
BaoYZhuYDuW, et al. A distributed PCA-TSS based soft sensor for raw meal fineness in VRM system. Control Eng Pract2019; 90: 38–49.
3.
ChengYHuangYPangB, et al. ThermalNet: a deep reinforcement learning-based combustion optimization system for coal-fired boiler. Eng Appl Artif Intell2018; 74: 303–311.
4.
ShuYHuangYLiB.Design of deep learning accelerated algorithm for online recognition of industrial products defects. Neural Comput Appl2019; 31: 4527–4540.
5.
LopsYChoiYEslamiE, et al. Real-time 7-day forecast of pollen counts using a deep convolutional neural network. Neural Comput Appl2020; 32: 11827–11836.
6.
HarbolaSCoorsV.One dimensional convolutional neural network architectures for wind prediction. Energy Convers Manag2019; 195: 70–75.
7.
Carolin MabelMFernandezE. Analysis of wind power generation and prediction using ANN: a case study. Renew Energy2008; 33: 986–992.
8.
YiJWangQZhaoD, et al. BP neural network prediction-based variable-period sampling approach for networked control systems. Appl Math Comput2007; 185: 976–988.
9.
ChauchardFCogdillRRousselS, et al. Application of LS-SVM to non-linear phenomena in NIR spectroscopy: development of a robust and portable sensor for acidity prediction in grapes. Chemom Intell Lab Syst2004; 71: 141–150.
10.
HuangGZhouHDingX, et al. Extreme learning machine for regression and multiclass classification. IEEE Trans Syst Man Cybern Part B2012; 42: 513–529.
11.
LinXYangZSongY.Short-term stock price prediction based on echo state networks. Expert Syst Appl2009; 36: 7313–7317.
12.
XuLLiuS.Study of short-term water quality prediction model based on wavelet neural network. Math Comput Model2013; 58: 807–813.
13.
WuYGaoRYangJ.Prediction of coal and gas outburst: a method based on the BP neural network optimized by GASA. Process Saf Environ Prot2020; 133: 64–72.
14.
MohandesMAHalawaniTORehmanS, et al. Support vector machines for wind speed prediction. Renew Energy2004; 29: 939–947.
15.
HeFZhangL.Prediction model of end-point phosphorus content in BOF steelmaking process based on PCA and BP neural network. J Process Control2018; 66: 51–58.
16.
LiuWGuoGChenF, et al. Meteorological pattern analysis assisted daily PM2.5 grades prediction using SVM optimized by PSO algorithm. Atmos Pollut Res2019; 10: 1482–1491.
17.
LinXQianZ.Modeling of vertical mill raw meal grinding process and optimal setting of operating parameters based on wavelet neural network. In: 2014 international joint conference on neural networks (IJCNN), Beijing, China, 6–11 July 2014. New York: IEEE.
18.
LinXFZhangMQ.Modelling of the vertical raw cement mill grinding process based on the echo state network. In: 2016 12th world congress on intelligent control and automation (WCICA), Guilin, China, 12–15 June 2016. New York: IEEE.
19.
YeTZhuX.The bridge relating process neural networks and traditional neural networks. Neurocomputing2011; 74: 906–915.
20.
HuangG-BZhuQ-YSiewC-K.Extreme learning machine: theory and applications. Neurocomputing2006; 70: 489–501.
21.
WangJ-GWangYYaoY, et al. Stacked autoencoder for operation prediction of coke dry quenching process. Control Eng Pract2019; 88: 110–118.
22.
EslamiEChoiYLopsY, et al. A real-time hourly ozone prediction system using deep convolutional neural network. Neural Comput Appl2019; 32: 8783–8797.
23.
HintonGEOsinderoSTehY-W.A fast learning algorithm for deep Belief nets. Neural Comput2006; 18: 1527–1554.
24.
HintonGESalakhutdinovRR.Reducing the dimensionality of data with neural networks. Science2006; 313: 504–507.
ZangDWeiZBaoM, et al. Deep learning–based traffic sign recognition for unmanned autonomous vehicles. Proc IMechE, Part I: J Systems and Control Engineering2018; 232: 497–505.
28.
SladojevicSArsenovicMAnderlaA, et al. Deep neural networks based recognition of plant diseases by leaf image classification. Comput Intell Neurosci2016; 2016: 3289801.
29.
FuYZhangYGaoY, et al. Machining vibration states monitoring based on image representation using convolutional neural networks. Eng Appl Artif Intell2017; 65: 240–251.
30.
ElleuchMTagouguiNKherallahM.Optimization of DBN using regularization methods applied for recognizing Arabic handwritten script. Procedia Comput Sci2017; 108: 2292–2297.
31.
YangJHanYWangY, et al. Optimization of real-time traffic network assignment based on IoT data using DBN and clustering model in smart city. Futur Gener Comput Syst2020; 108: 976–986.
32.
QingXNiuY.Hourly day-ahead solar irradiance prediction using weather forecasts by LSTM. Energy2018; 148: 461–468.
33.
JiangNHuXLiN.Graphical temporal semi-supervised deep learning–based principal fault localization in wind turbine systems. Proc IMechE, Part I: J Systems and Control Engineering2020; 234: 985–999.
34.
HaoXWangZShanZ, et al. Prediction of electricity consumption in cement production: a time-varying delay deep belief network prediction method. Neural Comput Appl2019; 31: 7165–7179.
35.
LiJChengKWangS, et al. Feature selection: a data perspective. ACM Comput Surv2017; 50: Article 94.
36.
SunYLiJLiuJ, et al. Using causal discovery for feature selection in multivariate numerical time series. Mach Learn2015; 101: 377–395.
37.
SchmidhuberJ.Deep learning in neural networks: an overview. Neural Networks2015; 61: 85–117.
38.
HintonGESrivastavaNKrizhevskyA, et al. Improving neural networks by preventing co-adaptation of feature detectors. arXiv Preprint arXiv12070580.
39.
KingmaDPBaJ.Adam: a method for stochastic optimization. arXiv Preprint arXiv14126980.
40.
SrivastavaNHintonGKrizhevskyA, et al. Dropout: a simple way to prevent neural networks from overfitting. J Mach Learn Res2014; 15: 1929–1958.
41.
HaraKSaitoDShounoH.Analysis of function of rectified linear unit used in deep learning. In: 2015 international joint conference on neural networks (IJCNN), Killarney, 12–17 July 2015, pp.1–8. New York: IEEE.
42.
AngermuellerCPärnamaaTPartsL, et al. Deep learning for computational biology. Mol Syst Biol2016; 12: 878.
43.
LiQMengQCaiJ, et al. Applying support vector machine to predict hourly cooling load in the building. Appl Energy2009; 86: 2249–2256.
44.
GersFAEckDSchmidhuberJ, et al. Applying LSTM to time series predictable through time-window approaches. London: Springer, 2002, pp.193–200.
