Condition Identification of MSW Incineration Based on Multimodal Fusion and Large Language Models
Yang Wang, Jian Chen, Xin Wei · 2025
With the acceleration of urbanization and the steady growth of industrial economy in China, the annual generation of Municipal Solid Waste (MSW) is increasing rapidly. Accurate identification of MSW incineration conditions is crucial for optimizing the incineration process, reducing pollutant emissions, and ensuring the safe operation of equipment. However, traditional identification methods, such as monitoring by Distributed Control Systems (DCS), observation with industrial cameras, and manual inspection, have limitations and uncertainties. Since the 1990s, although artificial intelligence technology has been introduced into this field, its adaptability and recognition ability still need to be improved under the significant changes in MSW composition and complex operating conditions. To address the above issues, this paper proposes an MSW incineration condition identification method based on multimodal fusion and large language models. This method integrates multimodal data, such as images from industrial cameras and time-series data from DCS, utilizes the Retrieval-augmented Generation (RAG) technology of large language models to achieve efficient recognition. Comparative experiments show that this method significantly improves the accuracy and robustness of MSW incineration condition identification and can provide real-time operational suggestions for operators, offering a new approach for the intelligent management of MSW incineration processes.