Seismic observation prediction system

Seismic Test Structure Behavior Prediction System Based on NARX Neural Network Model

This technology is a seismic observation and prediction system that places a multi-story test structure equipped with an accelerometer on the top floor onto a vibration table to apply seismic signals. It compares measured behavior values with predictions from a neural network-based NARX model to select the model with the lowest error. The structure is also equipped with a Tuned Mass Damper (TMD) to reduce vibrations.

Seismic design is essential for high-rise buildings and long-span bridges, so behavior data has traditionally been collected by applying artificial vibrations to test structures before construction. However, relying solely on physical measurements for behavior analysis is costly and time-consuming, and there has been a lack of technology to easily predict the behavior of test structures and actual buildings.

This technology enables fast and accurate mathematical prediction of structural behavior by determining the NARX model with the minimum MAE or RMSE during the simulation phase and verifying it by increasing the TMD counterweight during the testing phase. It can be applied to seismic design verification for high-rise buildings and long-span bridges, as well as seismic performance evaluation of existing structures, allowing for the reduction of physical test cycles while pre-evaluating various seismic scenarios.

‍

‍

Key Features:

• A test structure including multiple floors with an accelerometer on the top floor and a Tuned Mass Damper (TMD) to attenuate vibrations
• A vibration table that supports the test structure and triggers seismic signals, and a control unit that manages the vibration table
• A configuration that selects the prediction model with the minimum mean absolute error or root mean square error between the accelerometer measurements and predicted values
• A configuration that uses a neural network NARX function with seismic signals and delayed measured values as inputs, and increases the TMD counterweight during the testing phase after simulation

‍

This invention was developed through research support from the Korea Agency for Infrastructure Technology Advancement (KAIA) for the development and performance verification of a source system for seismic reinforcement bracing damper devices using new materials.

‍