M3RTNet: Combustion State Recognition Model of MSWI Process Based on Res-Transformer and Three Feature Enhancement Strategies

Sustainability, Год журнала: 2025, Номер 17(8), С. 3412 - 3412

Опубликована: Апрель 11, 2025

The accurate identification of combustion status can effectively improve the efficiency municipal solid waste incineration and reduce risk secondary pollution, which plays a key role in promoting sustainable development treatment industry. Due to low accuracy incinerator flame state recognition current process, this paper proposes Res-Transformer model based on three feature enhancement strategies. In paper, is used as backbone network integrate local features global features. Firstly, an efficient multi-scale attention module introduced into Resnet, uses parallel sub-network establish long short dependencies. Then, deformable multi-head designed Transformer layer, self-attention extract long-term Finally, we design context fusion efficiently aggregate spatial information shallow channel deep network, enhance cross-layer extracted by network. order verify effectiveness proposed comparative experiments ablation were conducted image dataset. results showed that Acc, Pre, Rec F1 score indices 96.16%, 96.15%, 96.07% 96.11%, respectively. Experiments demonstrate robustness method.

Язык: Английский

Effects of different types of municipal solid waste incineration slag on landfill stabilization and related microbiological mechanism

Waste Management, Год журнала: 2025, Номер 202, С. 114833 - 114833

Опубликована: Апрель 26, 2025

Язык: Английский