Smartphone based structural health monitoring using deep neural networks

Abstract

In order to create a low-cost and easy-to-implement structural health monitoring method, this paper uses smartphones to collect data on the movement of a building to determine whether it has suffered any damage. Smartphones have easy handling, sensors embedded, and data processing properties. The smartphone-based structural health monitoring system have many advantage over the classical methods. Recurrent fuzzy neural network of long- and short-term memory is applied to determine the damage of the structure. Comparisons with classical neural networks and fuzzy systems are given to show the proposed deep neural networks have better performances. Finally, several experiments are applied to evaluate the proposed methods for the determination of structural damage. • A smartphone-based building structural health monitoring system is developed. • This system uses fuzzy long and short term memory (LSTM) to deal the data obtained from the smartphone. • This system is implemented in a prototype of structural health monitoring.