Research and Simulation of Boiler Combustion System Based on Convolution Neural Network. Issue 6 (March 2019)

• Record Type: Journal Article
• Authors: Ouyang, Chunming; Xiao, Li; Xu, Zhibin; Zhang, Weidong
• Is Part Of: Journal of physics. Volume 1176:Issue 6(2019)
• Publication Date: 2019-03
• Subjects: Physics -- Congresses; 530.5
• DOI: 10.1088/1742-6596/1176/6/062017
• Languages: English
• ISSNs: 1742-6588

Abstract

Abstract: This paper developed a coal-fired boiler modeling method based on CNN (convolutional Neural Network). In recent years, the deep learning using CNN has made remarkable achievements in image research field. This paper combined the image convolution method and the signal system convolution theory in coal-fired boiler modeling, which regards the DCS records as time series signal. This network is consisted of two convolutional layers which map the input to the feature map, followed by two fully connected layers to model the feature to the output. All the parameters in the network, including the convolution kernel parameter are optimized by minimizing the cost function. This model provided a good solution of modeling coal-fired boiler system with characters of time delay, highly nonlinear and multi-variable coupling.