A Convolution-Circulation Neural Network Based Method for Identification of Rotary Kiln Sequence Working Conditions

Abstract

The invention provides a rotary kiln sequence working condition identification method based on a convolution-circulation neural network, relating to the technical field of image classification and pattern recognition. Firstly, the video sequence information of a rotary kiln firing zone is preprocessed under different working conditions; PCA principal component analysis is used to extract the features of a region of interest and to reduce the dimension of the region of interest; then a CNN-RNN convolution loop neural network is designed, and the dynamic information between image features and image sequences are further extracted; a random search super-parameter optimization method is adopted to select the optimal super-parameters of the loop neural network, and an optimal CNN-RNN neural network classifier model is obtained, to achieve the rotary kiln image sequence of the working condition recognition. The rotary kiln sequence working condition identification method based on the convolution-circulation neural network can make use of not only the image space characteristics but also the correlation information and dynamic characteristics between the image sequences, so the method canachieve better classification effect on the recognition of rotary kiln image sequence working conditions.

Description

technical field [0001] The invention relates to the technical field of image classification and pattern recognition, in particular to a method for recognizing sequential operating conditions of a rotary kiln based on a convolutional-cyclic neural network. Background technique [0002] The rotary kiln is up to 100 meters long and is in continuous rotation. The particularity of its structure and the complexity of the process make the sintering process of the rotary kiln complex, including the physical and chemical reactions of materials, fuel combustion, heat transfer between gases, materials, and linings. Multiple coupled processes such as gas and material movement. Due to the difficulties in the sintering process of the rotary kiln, such as the difficulty of online measurement of the clinker quality index and the difficulty of accurately identifying the firing state of key process parameters closely related to the quality of the clinker, the existing sintering process of the...

AI Technical Summary

Problems solved by technology

However, the research object of this method is only the flame grayscale image, and the multi-channel encoding problem of the image is not considered, nor is the timing relationship of the clinker sintering condition of the rotary kiln considered.

Images