際際滷

際際滷Share a Scribd company logo

AI Artificial Intelligence. Landslide velocity prediction applied to the Cassas Landslide, Piedmont, Italy

Download as PPT, PDF
Claudio Angelino
Claudio Angelino

This document describes an artificial intelligence system called "AI LANDSLIDE" that analyzes monitoring data to predict landslide velocity based on past inclinometric data and rainfall. It discusses applying the system to the Cassas Landslide in Italy, including input data used in the learning phase and accurate predictions of velocity data from 1998-2002 and displacement from 1991-2002. The document also mentions using the predictions to trigger civil protection alerts more reliably and studying mitigative measures like drainage tunnels. It concludes by noting the system could potentially be applied to predicting other natural phenomena beyond landslides.

1 of 15
Downloaded 27 times
AI Artificial Intelligence Landslide Velocity Prediction Applied to the Cassas Landslide, Italy
Table of contents Introduction Cassas Landslide Input data Velocity data prediction Past displacement prediction Mitigative measures Other applications End
Introduction "AI LANDSLIDE" analyses monitoring data and allows reliable predictions of landslide velocity based on past inclino- metric data and rainfall. Civil Protection Alerts can be triggered with more reliability, thus avoiding many costlycostly crisis planning errorscrisis planning errors.
PARIS - LYON GENEVA TURIN - MILAN GENOA - ROME
Cassas Landslide, Italy - Volume 20 - 30 million m3 - Lenght appx. 1.8 Km - Average slope 50 to 55 % - 際際滷 depth appx. 60 m
Cassas Landslide Rainfall 1998-2002
The learning phase encompassed 40 points. A.I. LANDSLIDE can already effectively predict velocities after a rather short learning phase! Prediction: Velocity data 98-02
1999 1999 2000 2000 2001 2001 2002 2002
Rainfall 1990-2002 Rainfall data from 1991 to 2002 are available. Thus it is possible to estimate the displacement during this period
AI Artificial Intelligence Landslide Prediction Past displacement
?
Displacement Real displacement Nov-98 to Ago-02 (46 months) Inclinometer n属3 20.5 cm 20.5 cm Calculated displacement Jan-91 to Nov-98 (94 months) Inclinometer n属3 21.1 cm 21.1cm
According to the 束 AI Landslide 損 model the need of implementing mitigative measures has clearly emerged. The main scope being the depression of the ground water table Three alternatives have been studied with a risk management approach: Deep drainage by vertical shafts and submerged pumps 600 m long tunnel in stable ground with ascending drainage boreholes 150 m long 3 x 3 m drainage tunnel in the sliding mass with subhorizontal drains 束AI Landslide損 engineering application
The system 束 AI Landslide 損 can be custom tailored to study other events and phenomena than landslides. Practically, any activity in which the driving parameters can be measured and identified can be studied with 束 AI Landslide 損. However the model has to be custom tailored for each phenomenon: - Levels of water in a river or creek - Deformations and settlements - Tunnels convergence - Rock mass movements - Dispersion of contaminants - Etc. Other 束AI Landslide損 applications
END More information in: www.ailandslides.com www.riskope.com

More Related Content

AI Artificial Intelligence. Landslide velocity prediction applied to the Cassas Landslide, Piedmont, Italy

  • 1. AI Artificial Intelligence Landslide Velocity Prediction Applied to the Cassas Landslide, Italy
  • 2. Table of contents Introduction Cassas Landslide Input data Velocity data prediction Past displacement prediction Mitigative measures Other applications End
  • 3. Introduction "AI LANDSLIDE" analyses monitoring data and allows reliable predictions of landslide velocity based on past inclino- metric data and rainfall. Civil Protection Alerts can be triggered with more reliability, thus avoiding many costlycostly crisis planning errorscrisis planning errors.
  • 4. PARIS - LYON GENEVA TURIN - MILAN GENOA - ROME
  • 5. Cassas Landslide, Italy - Volume 20 - 30 million m3 - Lenght appx. 1.8 Km - Average slope 50 to 55 % - 際際滷 depth appx. 60 m
  • 7. The learning phase encompassed 40 points. A.I. LANDSLIDE can already effectively predict velocities after a rather short learning phase! Prediction: Velocity data 98-02
  • 9. Rainfall 1990-2002 Rainfall data from 1991 to 2002 are available. Thus it is possible to estimate the displacement during this period
  • 10. AI Artificial Intelligence Landslide Prediction Past displacement
  • 11. ?
  • 12. Displacement Real displacement Nov-98 to Ago-02 (46 months) Inclinometer n属3 20.5 cm 20.5 cm Calculated displacement Jan-91 to Nov-98 (94 months) Inclinometer n属3 21.1 cm 21.1cm
  • 13. According to the 束 AI Landslide 損 model the need of implementing mitigative measures has clearly emerged. The main scope being the depression of the ground water table Three alternatives have been studied with a risk management approach: Deep drainage by vertical shafts and submerged pumps 600 m long tunnel in stable ground with ascending drainage boreholes 150 m long 3 x 3 m drainage tunnel in the sliding mass with subhorizontal drains 束AI Landslide損 engineering application
  • 14. The system 束 AI Landslide 損 can be custom tailored to study other events and phenomena than landslides. Practically, any activity in which the driving parameters can be measured and identified can be studied with 束 AI Landslide 損. However the model has to be custom tailored for each phenomenon: - Levels of water in a river or creek - Deformations and settlements - Tunnels convergence - Rock mass movements - Dispersion of contaminants - Etc. Other 束AI Landslide損 applications
  • 15. END More information in: www.ailandslides.com www.riskope.com