Timezone: »

 
Cross-Modal Virtual Sensing for Combustion Instability Monitoring
Tryambak Gangopadhyay · Vikram Ramanan · Chakravarthy S.R. · Soumik Sarkar

@
in Machine Learning and the Physical Sciences »

In many cyber-physical systems, imaging can be an important but expensive or 'difficult to deploy' sensing modality. One such example is detecting combustion instability using flame images, where deep learning frameworks have demonstrated state-of-the-art performance. The proposed frameworks are also shown to be quite trustworthy such that domain experts can have sufficient confidence to use these models in real systems to prevent unwanted incidents. However, flame imaging is not a common sensing modality in engine combustors today. Therefore, the current roadblock exists on the hardware side regarding the acquisition and processing of high-volume flame images. On the other hand, the acoustic pressure time series is a more feasible modality for data collection in real combustors. To utilize acoustic time series as a sensing modality, we propose a novel cross-modal encoder-decoder architecture that can reconstruct cross-modal visual features from acoustic pressure time series in combustion systems. With the "distillation" of cross-modal features, the results demonstrate that the detection accuracy can be enhanced using the virtual visual sensing modality. By providing the benefit of cross-modal reconstruction, our framework can prove to be useful in different domains well beyond the power generation and transportation industries.

Author Information

Tryambak Gangopadhyay (Iowa State University)

As a Ph.D. student at Iowa State University, I am working as a Research Assistant at Self-aware Complex Systems Lab on Machine Learning and Deep Learning for healthcare, agriculture, and different cyber-physical systems. I am doing a concurrent Masters in Computer Science. In 2019, I did my summer internship at the Machine Learning group of Lawrence Livermore National Laboratory. I developed deep learning fusion models for clinical prediction tasks using Electronic Health Record datasets. In my Ph.D. thesis, I am focusing on implementing state-of-the-art architectures and designing new algorithms for large-scale image/video data, multivariate time series data and text data with applications focusing on cyber-physical systems. My interests include Autoencoders (For 3D and 2D Data), LSTM Models, Attention Models, Explainability for Spatiotemporal Data, Action Recognition, Anomaly Detection using ML, 3D, and 2D CNN for large-scale video datasets, Natural Language Processing, Multimodal Learning using text, time-series and images, Spatiotemporal Interpretability for Multivariate Time-Series Data. Recently, I have started developing an interest in Reinforcement Learning. Please contact me at tryambak@iastate.edu, tryambak95@gmail.com for more details about my research.

Vikram Ramanan (Iit madras)
Chakravarthy S.R. (Indian Institute of Technology Madras)
Soumik Sarkar (Iowa State University)

More from the Same Authors