Flame combustion state recognition in municipal solid waste incineration processes based on image multi-threshold segmentation and DQN-PL model

Ce Cao, Qiang Zhang, Menghan Li and Shaoyang Wang

Energy, 2025, vol. 331, issue C

Abstract: Flame combustion state identification is critical for control systems in power plants. In this paper, a Deep Q Network-Pseudo Label (DQN-PL) model combining DQN with a pseudo-label generation strategy is proposed to enhance the accuracy of flame state identification and classification during municipal solid waste incineration. Multi-threshold segmentation using a genetic algorithm-simulated annealing (GA-SA) distinguishes the flame areas more accurately through a combination of global search and local fine-tuning, resulting in segmentation accuracy improvements of 3.96 % and 6.57 % compared to GA and SA alone, respectively. L1 regularization, recursive feature elimination and random forest screening are successively used, followed by principal component analysis (PCA) for dimensionality reduction to further reduce the complexity of the data. Finally, the flame combustion state is classified using the DQN-PL model, which dynamically expands the samples through a high-confidence pseudo-labeling mechanism and optimizes the decision strategy by combining DQN reinforcement learning. The results show that the accuracy of DQN-PL is 8.15 % and 16.94 % higher than that of DQN and DNN, respectively. On small and medium-sized datasets with missing labels, DQN-PL achieves higher accuracy than ViT and Swin Transformer with lower computational cost.

Keywords: Combustion state recognition; Flame image segmentation; Deep reinforcement learning; Pseudo label generation (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S036054422502609X
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:energy:v:331:y:2025:i:c:s036054422502609x

DOI: 10.1016/j.energy.2025.136967

Access Statistics for this article

Energy is currently edited by Henrik Lund and Mark J. Kaiser

More articles in Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().