A method for generating images of abnormal combustion state in MSWI process based on DCGAN

This article is to provide qualified images of abnormal combustion state for the research of machine vision in municipal solid waste incineration (MSWI) process. Owing to the scarcity of the images of abnormal combustion state and the high cost of labeling, it is difficult to obtain sufficient images of abnormal combustion state. Aim at the problem, this paper proposes a method for generating images of abnormal combustion state based on a deep convolutional generative adversarial network (DCGAN). First, the real image data of abnormal combustion state is preprocessed. Second, the abnormal combustion state image generation generates false combustion images. Third, the real images and the generated images are fed into the discrimination network. The loss values are used to train the discrimination and generation. Finally, whether to update the parameters of the generation and discrimination network is determined by the error and epoch. The qualified generated abnormal combustion state images are obtained after the epoch setting met. The evaluation result of the generated image quality based on the Fréchet Inception Distance (FID) shows that DCGAN can realize the generation of abnormal combustion state images.