Optimizing combustion of coal fired boilers for reducing NOx emission using Gaussian Process

Abstract Since the mechanism of boiler combustion is extremely complicated and difficult to apply to model and optimize the combustion process directly, data-driven models attract increasingly attention from industry. This paper focuses on the application of Gaussian Process (GP) in optimizing combustion process for reducing NOx emission of a 330 MW boiler. GP is used to model the relationship between the NOx emission characteristic and boiler operation parameters. The hyperparameters of the GP model are optimized via Genetic Algorithm (GA). Based on 670 sets of production data from the 330 MW tangentially fired boiler, two GP models with 13 and 21-inputs are developed, respectively. The experimental result shows that the 21-inputs model provides better prediction performance than 13-inputs model does. The comparison between Support Vector Machines (SVM) and GP is also given under the 21-inputs circumstance. The influences of some inputs are investigated separately. Then, the predicted NOx emission is used as the objective of searching the optimal parameters for the boiler combustion. Under a given production combustion condition, the NOx decreases from 345 ppm to 238 ppm via optimizing the boiler operational parameters using the 21-inputs GP model, which is a reasonable achievement for the coal fired combustion process.

[1]  Saeid R. Dindarloo,et al.  Prediction of the Unburned Carbon Content of Fly Ash in Coal-Fired Power Plants , 2015 .

[2]  Hao Zhou,et al.  Modeling NOx emissions from coal-fired utility boilers using support vector regression with ant colony optimization , 2012, Eng. Appl. Artif. Intell..

[3]  Primož Potočnik,et al.  Multi-step-ahead prediction of NOx emissions for a coal-based boiler , 2013 .

[4]  Bojan Likar,et al.  Predictive control of a gas-liquid separation plant based on a Gaussian process model , 2007, Comput. Chem. Eng..

[5]  J Kocijan,et al.  Application of Gaussian processes for black-box modelling of biosystems. , 2007, ISA transactions.

[6]  Mika Liukkonen,et al.  Modeling of the fluidized bed combustion process and NOx emissions using self-organizing maps: An application to the diagnosis of process states , 2011, Environ. Model. Softw..

[7]  Wei Zhao,et al.  Coal chemical industry and its sustainable development in China , 2010 .

[8]  Yang Tingting,et al.  Modeling on SCR process of a coal-fired boiler using LSSVM method , 2015, The 27th Chinese Control and Decision Conference (2015 CCDC).

[9]  Hao Zhou,et al.  Modeling and optimization of the NOx emission characteristics of a tangentially fired boiler with artificial neural networks , 2004 .

[10]  Iain Murray,et al.  A framework for evaluating approximation methods for Gaussian process regression , 2012, J. Mach. Learn. Res..

[11]  Hsuan-Tien Lin.  A Study on Sigmoid Kernels for SVM and the Training of non-PSD Kernels by SMO-type Methods , 2005 .

[12]  Cen Ke-fa.  SUPPORT VECTOR MACHINE MODELING ON THE UNBURNED CARBON IN FLY ASH , 2005 .

[13]  Chih-Jen Lin,et al.  Asymptotic Behaviors of Support Vector Machines with Gaussian Kernel , 2003, Neural Computation.

[14]  Preeti Manke,et al.  ARTIFICIAL NEURAL NETWORK BASED NITROGEN OXIDES EMISSION PREDICTION AND OPTIMIZATION IN THERMAL POWER PLANT , 2013 .

[15]  Matthias W. Seeger,et al.  Bayesian Gaussian process models : PAC-Bayesian generalisation error bounds and sparse approximations , 2003 .

[16]  Jaroslaw Krzywanski,et al.  Neurocomputing approach for the prediction of NOx emissions from CFBC in air-fired and oxygen-enriched atmospheres , 2017 .

[17]  Mika Liukkonen,et al.  Artificial neural networks for analysis of process states in fluidized bed combustion , 2011 .

[18]  Zbigniew Michalewicz,et al.  Genetic Algorithms + Data Structures = Evolution Programs , 1996, Springer Berlin Heidelberg.

[19]  V. T. Sathyanathan,et al.  Prediction of unburnt carbon in tangentially fired boiler using Indian coals , 2004 .