24 July 2024 in Journal of Process Control
14 August 2023 in IEEE Transactions on Instrumentation and Measurement
In the cement clinker burning process, the soft sensing of free calcium oxide (f-CaO) content has been a challenging task due to the dynamic time delay between f-CaO content and process variables, the different time scale between process variables and f-CaO content and the strong nonlinearity of the process data. With the development of data-driven modeling techniques, numerous soft sensing methods for f-CaO content based on process data have emerged. However, under this circumstance, the monotony of soft sensor input may become a bottleneck that limits further improvement of the f-CaO prediction performance. To address this issue, this paper proposes a novel end-to-end multi-source information fusion framework (MSIFF) for soft sensing the f-CaO content within the cement clinker. The MSIFF takes process data and flame images as inputs, and utilizes mechanistic knowledge by generating mechanistic features from process data using a first-principle rotary kiln model. The explainable dynamic features are extracted from the matched process data and flame image sequences with a multi-source dynamic feature extraction network (MSDFE), which further participates in the end-to-end modeling of f-CaO content together with the mechanistic features. The proposed MSIFF method is validated on a real cement production line. While providing valuable operating information for the cement clinker burning process, the MSIFF exhibits an improved f-CaO soft sensing performance compared to existing f-CaO estimation methods.
24 May 2025 in Engineering Applications of Artificial Intelligence
25 December 2024 in IEEE Transactions on Instrumentation and Measurement
As the key quality indicator in the iron ore sintering process, the ferrous oxide (FeO) content in the finished sinter directly reflects its thermal state and denotes sintering quality. Currently, data-driven soft sensor modeling technology has become the mainstream method for predicting this quality indicator. However, existing approaches either fail to leverage the multimodal properties of the sintering dataset or simply neglect to explore the interrelationships between multisource heterogeneous data, thus limiting the improvement of prediction accuracy. To address this issue, we propose a multilevel deep fusion framework (MDFF) for real-time FeO content detection, which combines data fusion, feature fusion, and decision fusion methods together to form an end-to-end framework for soft sensor modeling. Firstly, shallow features are extracted from cross-section images based on expert knowledge for data-level fusion. Then, two multiscale encoding networks and a shared representation learning network are employed to realize feature-level fusion. Subsequently, the final predicted FeO content is obtained by fusing the prediction results of shared representations and modality-specific features at the decision level. By integrating multi-source data and designing multi-level information fusion, the proposed MDFF method can make full exploitation of the intrinsic interactions among multisource information at various levels. The experimental results of the actual sintering process show that MDFF is superior to the existing models in terms of prediction accuracy and stability.
19 April 2025 in Measurement
6 May 2026 in Control Engineering Practice