AI predicts physics of future fault slip in laboratory earthquakes

An artificial intelligence approach borrowed from natural language processing can predict future faultfriction and the next failure time with high resolution in laboratory earthquakes. The technique, applying artificial intelligence to the fault's acoustic signals, advances previous work and predicts the future state of the fault's physical system.

Predicting futurefriction is what we do." Chris Johnson, co-lead author of a paper on the findings, said that it provides a potential path to near term forecasting of earthquake timing in Earth.

A team led by Paul Johnson has made steady advances in applying machine learning techniques to the challenge of forecasting earthquakes in the laboratory and in the field.

The future fundamental physics of the system can be seen in the acoustic signals emitted by the laboratory fault. That is the first thing that's ever been seen.

In a novel approach, the Los Alamos team applied a transformer model to acoustic emissions from the laboratory fault.

Chris Johnson said that the deep-learning transformer model is synonymous with a language translation model such as Google Translate. "Imagine writing an email in English and having the artificial intelligence translate the English to Japanese while also anticipating your words and filling in the blanks."

Chris Johnson said that the artificial intelligence takes data and says what's happening next on the fault.

The Los Alamos team used machine learning techniques to forecast fault failure timing in lab earthquakes and historical slow-slip Earth data. Machine learning was applied to data from laboratory shear experiments to show that fault emissions are imprinted with information about where it is in the slip cycle.

The Los Alamos researchers were able to predict the evolution of the fault with the help of statistical features and machine learning.

The data is input to a model to predict the current state of the fault system. A description of the current state of the system is what's included in that prediction.

Future predictions are beyond describing the instantaneous state of the system. Chris Johnson said that the model is learning from the past to predict the future and when the next slip event will occur.

Chris Johnson said that the model is not constrained with physics but predicts the physics.

The next challenge is whether we can do this in Earth to predict future fault displacement. We don't have long data sets for model training like we do in the laboratory.

The method could be applied to other disciplines, such as non-destructive materials testing, where it could give information about damage to a bridge.

More information: Kun Wang et al, Predicting future laboratory fault friction through deep learning transformer models, Geophysical Research Letters (2022). DOI: 10.1029/2022GL098233 Journal information: Geophysical Research Letters Citation: AI predicts physics of future fault slip in laboratory earthquakes (2022, October 10) retrieved 10 October 2022 from https://phys.org/news/2022-10-ai-physics-future-fault-laboratory.html This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.