Cyber-Physical Twins for Environmental Process Modeling and Forecasting

Background

Environmental process modeling is a key component of understanding how environmental systems behave through the use of mathematical and statistical models. Environmental modeling can be used to address a wide range of systems including agricultural disease spread, impact of natural disasters on specific ecosystems, water and air quality management, and habitat conservation. However, there are issues with the scalability and transferability of current models, particularly those developed for interdisciplinary applications.

Invention Description

Researchers at Arizona State University have developed a new cyber-physical twin paradigm for learning environmental process dynamics by leveraging both digital and physical twins working in tandem. This technology can help to address current limitations of environmental process modeling, including scalability and transferability, through the use of a physical twin to close the loop on model learning and improvement. This cyber-physical twin paradigm enables deep reinforcement and self-supervised learning algorithms to validate the models learned for the digital twin. 

Potential Applications:

• Agricultural disease hotspot spread analysis
• Post-wildfire transition hydrology (e.g., debris flows)
• Ground seismicity impact assessment
• Habitat conservation

Benefits and Advantages:

• Scalable & transferable – uses both a physical and digital twin to close the loop on model learning and improvement
• Improved model validation – enables direct human engagement and the execution of real-world experiments
• AI integration capability – can enable timely integration of dynamic, high-resolution data through data fusion and dimensionality reduction