Scholarlink

<p>The earthquake warning system (EWS) is a critical technological advancement that prevents significant loss of life and infrastructure, particularly considering the current inability to predict earthquakes. Few developed nations, like the USA, Japan, etc. have implemented EWS. India has also made efforts to implement EWS for metro rails and Uttarakhand. Expanding EWS in developing nations necessitates the development of cost-effective and reliable solutions. Seismic sensors based on geophones and MEMS accelerometers have emerged as promising, cost-effective alternatives to expensive conventional sensors due to their performance. However, challenges persist related to the suboptimal acceleration response of these sensors. Therefore, experiments have been conducted on a tri-axial shake table to analyse the acceleration response of MEMS-based Raspberry Shake 4D (RS4D), geophone-based Raspberry Shake 3D (RS3D), with the conventional strong-motion sensor Guralp CMG-5TC by capturing simulated seismic waveforms. Based on the observations, machine learning (ML) techniques have been leveraged to establish the correlations between sensor responses in predicting acceleration using RS4D and RS3D sensors. Out of 41 regressors of Lazy Predict API, stochastic gradient descent (SGD) yields the best performance with a root-mean-square error (RMSE) of 0.2 and an R<sup>2</sup> score of 97.46. This predictive capability is significant and vital for multilevel earthquake warnings, significantly when acceleration profoundly impacts the operational management of critical installations. An Internet of Things (IoT)-based framework utilising ML-based acceleration predictions for EWS realisation has also been discussed. Integrating heterogeneous sensors with IoT technology provides an innovative framework to develop a cost-effective, reliable EWS.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An IoT-Based Architectural Framework for Earthquake Warning System Using Low-cost Heterogeneous Seismic Sensors

  • 期刊论文

  • 发表日期：

    2025年5月24日

查看全文
  • Amarendra Goap,
  • Siddhartha Sarkar,
  • Anubrata Roy,
  • C Rama Krishna,
  • Satish Kumar

摘要

The earthquake warning system (EWS) is a critical technological advancement that prevents significant loss of life and infrastructure, particularly considering the current inability to predict earthquakes. Few developed nations, like the USA, Japan, etc. have implemented EWS. India has also made efforts to implement EWS for metro rails and Uttarakhand. Expanding EWS in developing nations necessitates the development of cost-effective and reliable solutions. Seismic sensors based on geophones and MEMS accelerometers have emerged as promising, cost-effective alternatives to expensive conventional sensors due to their performance. However, challenges persist related to the suboptimal acceleration response of these sensors. Therefore, experiments have been conducted on a tri-axial shake table to analyse the acceleration response of MEMS-based Raspberry Shake 4D (RS4D), geophone-based Raspberry Shake 3D (RS3D), with the conventional strong-motion sensor Guralp CMG-5TC by capturing simulated seismic waveforms. Based on the observations, machine learning (ML) techniques have been leveraged to establish the correlations between sensor responses in predicting acceleration using RS4D and RS3D sensors. Out of 41 regressors of Lazy Predict API, stochastic gradient descent (SGD) yields the best performance with a root-mean-square error (RMSE) of 0.2 and an R2 score of 97.46. This predictive capability is significant and vital for multilevel earthquake warnings, significantly when acceleration profoundly impacts the operational management of critical installations. An Internet of Things (IoT)-based framework utilising ML-based acceleration predictions for EWS realisation has also been discussed. Integrating heterogeneous sensors with IoT technology provides an innovative framework to develop a cost-effective, reliable EWS.