An Unsupervised Classification Method for Flame Image of Pulverized Coal Combustion Based on Convolutional Auto-Encoder and Hidden Markov Model

Abstract

Combustion condition monitoring is a fundamental and critical issue that needs to be addressed in the wide-load operation of coal-fired boilers. In this paper, an unsupervised classification framework based on the convolutional auto-encoder (CAE), the principal component analysis (PCA), and the hidden Markov model (HMM) is proposed to monitor the combustion condition with the uniformly spaced flame images, which are collected from the furnace combustion monitoring system. First, CAE is adopted to extract the features from the flame images, which obtain the sparse representations in the images. Then, PCA is applied to project the feature vectors into the orthogonal space for robustness and computation efficiency. Finally, a HMM is built to calculate the corresponding optimal states by learning the temporal behaviors in the compressed representations. A coal combustion adjustment experiment was conducted in a 660 MW opposed-firing boiler, and the sequential 14,400 flame images with three different combustion states were obtained to evaluate the effectiveness of the proposed approach. We tested six different compression dimensions of the latent variable z in the CAE model and ensured that the appropriate compress parameter was 1024. The proposed framework is compared with five other methods: the CAE + Gaussian mixture model (GMM), CAE + Kmean, the CAE + fuzzy c-mean method, CAE + HMM, and the traditional handcraft feature extraction method (TH) + HMM. The results show that the proposed framework has the highest classification accuracy (95.25% for the training samples and 97.36% for the testing samples) and has the best performance in recognizing the semi-stable state (85.67% for the training samples and 77.60% for the testing samples), indicating that the proposed framework is capable of identifying the combustion condition, changing when the combustion deteriorates as the coal feed rate falls.

Combustion process monitoring based on flame intensity time series

“…One of the applications in which these networks perform particularly well is the detection of anomalies in the time series of different signals. 32–36 In the study, we utilised three types of DRNNs. They are as follows: LSTM recurrent neural network, one-dimensional (1D) convolutional neural network LSTM (CNN-LSTM) and 1D convolutional LSTM (ConvLSTM).…”

“…Gangopadhyay et al 14 leveraged images from hi-speed flame videos as well as the sound pressure data for detecting instability in an experimental combustion system. Qiu et al 15 used uniformly spaced flame images in recognition of the pulverised coal (PC) combustion status (stable, semi-stable and unstable) in the thermal power plant. Liu et al 16 used a deep belief network to predict the oxygen content at the combustion system outlet.…”

Hyperspectral Image Classification Based on Superpixel Pooling Convolutional Neural Network with Transfer Learning

“…A large number of existing studies have shown that deep learning-based classification method has good classification performance, and has achieved great success in computer vision, image processing and other fields. Some representative deep learning techniques include convolutional neural network (CNN) [9,10], recurrent neural network (RNN) [11,12], generative adversarial network [13,14] and convolutional auto-encoder [15,16]. Recently, a series of deep learning-based classification frameworks have been widely used in the field of remote sensing [17][18][19][20][21][22].…”

Flame Image Processing and Classification Using a Pre-Trained VGG16 Model in Combustion Diagnosis

“…Various architectures of such networks have been developed, and many of them achieve excellent performance in image recognition tasks, including flame images. Examples include deep convolutional neural networks [ 2 , 3 , 4 , 5 , 6 ], deep belief networks [ 7 , 8 ], deep convolutional auto-encoder [ 9 ], deep convolutional auto-encoder connected with the principal component analysis and the hidden Markov model [ 10 ], deep, fully connected neural networks [ 11 ], deep convolutional selective autoencoder [ 12 ] and various architectures followed by a symbolic time series analysis [ 13 , 14 ]. Each of the architectures mentioned above plays a vital role in a specific application area.…”