Deep learning-based soft-sensing method for operation optimization of coke dry quenching process

In the modern industrial process control, the development of distributed control system (DCS) makes the application of data-driven soft-sensing methods available. Deep learning (DL), as a novel training strategy of deep neural networks, has large potential for soft sensor modeling. In comparison with shallow neural network, because DL can make full use of massive process by greedy layer-wise training approach, deep structure of neural network has better representation and generalization ability. Compared with the traditional coke wet quenching, coke dry quenching (CDQ) has the advantage of waste heat recovery, which is advanced, energy saving and environmentally friendly, and is the main coke quenching method adopted in iron and steel plant. A deep learning-based soft-sensing method for operation optimization of coke dry quenching process is put forward in this paper. By defining the economic efficiency, the data with high economic efficiency is used for modeling and optimizing the CDQ operating variable, i.e. supplementary air flow rate (FSA). The experimental results show that, adopting the adjusted optimal operation, a remarkable raise (1.58%) of economic efficiency can be acquired on average. Thus, the presented deep learning-based soft-sensing method for operation optimization is effective and feasible for improving the waste heat recovery in CDQ system.

[1]  Pierre Baldi,et al.  Deep autoencoder neural networks for gene ontology annotation predictions , 2014, BCB.

[2]  Dexian Huang,et al.  Data-driven soft sensor development based on deep learning technique , 2014 .

[3]  Xin-xin Zhang,et al.  Mathematical model for fluid flow and heat transfer in the cooling shaft of Coke Dry Quenching unit , 2002 .

[4]  M. Errera,et al.  Thermodynamic analysis of a coke dry quenching unit , 2000 .

[5]  Meng Wang,et al.  Multimodal Deep Autoencoder for Human Pose Recovery , 2015, IEEE Transactions on Image Processing.

[6]  Pierre Baldi,et al.  Autoencoders, Unsupervised Learning, and Deep Architectures , 2011, ICML Unsupervised and Transfer Learning.

[7]  Xin-xin Zhang,et al.  Computational and experimental study of cooling process in Coke Dry Quenching experimental shaft , 2002 .

[8]  Yoshua. Bengio,et al.  Learning Deep Architectures for AI , 2007, Found. Trends Mach. Learn..

[9]  Jürgen Schmidhuber,et al.  Deep learning in neural networks: An overview , 2014, Neural Networks.

[10]  Simon J. Doran,et al.  Stacked Autoencoders for Unsupervised Feature Learning and Multiple Organ Detection in a Pilot Study Using 4D Patient Data , 2013, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[11]  Pei-Hsun Lin,et al.  Exergy Analysis of a Coke Dry Quenching System , 2009 .

[12]  Geoffrey E. Hinton,et al.  Reducing the Dimensionality of Data with Neural Networks , 2006, Science.

[13]  Marc'Aurelio Ranzato,et al.  A Sparse and Locally Shift Invariant Feature Extractor Applied to Document Images , 2007, Ninth International Conference on Document Analysis and Recognition (ICDAR 2007).

[14]  Kai Sun,et al.  Model predictive control for improving waste heat recovery in coke dry quenching processes , 2015 .

[15]  Yee Whye Teh,et al.  A Fast Learning Algorithm for Deep Belief Nets , 2006, Neural Computation.

[16]  Marc'Aurelio Ranzato,et al.  A Sparse and Locally Shift Invariant Feature Extractor Applied to Document Images , 2007 .

[17]  Wei Wang,et al.  Generalized Autoencoder: A Neural Network Framework for Dimensionality Reduction , 2014, 2014 IEEE Conference on Computer Vision and Pattern Recognition Workshops.

[18]  Qinghua Hu,et al.  Transfer learning for short-term wind speed prediction with deep neural networks , 2016 .