Search

• Propose a novel deep learning model based on Runge-Kutta recurrent neural network (RKRNN) with prior knowledge to realize structural system identification and seismic response prediction. • Formulate a partition training strategy to train the proposed neural network to improve the efficiency of training. • Utilize three numerical examples to valid the feasibility of RKRNN model including a linear three degrees of freedom system, a nonlinear single degree of freedom system with Bouc-Wen hysteresis and a simply supported bridge. • Site monitoring data from a bridge located in California has been used to further validate the proposed approach. In the seismic analysis of structural systems, dynamic response prediction is an essential problem and is significant in every stage during the structural life cycle. Conventionally, response analysis is carried out by numerical analysis. However, when the structural parameter is unknown, the establishment of a numerical model will be difficult. Enlightened by the Runge-Kutta (RK) numerical algorithm, this paper proposes a novel recurrent neural network named Runge-Kutta recurrent neural network (RKRNN) to realize the seismic response prediction. A partition training strategy is formulated to train the proposed neural network and to improve the efficiency of training. The proposed model can be trained by using a limited number of samples. Three numerical examples are utilized to validate the feasibility of RKRNN model including a linear three degrees of freedom (DOFs) system, a nonlinear single DOF system with Bouc-Wen hysteresis, and a numerical reinforced concrete bridge model. Additionally, the site monitoring data from a real-world bridge is utilized to further validate the proposed network. The results show that the proposed RKRNN model can effectively and efficiently predict the structural response under seismic load and exhibits robustness to noise, with good potential for applications in engineering practice. [ABSTRACT FROM AUTHOR]

[Display omitted] • Proposes a novel deep learning model that utilizes both 1DCNN and RNN variants. • Uses the attention mechanism to improve the performance of the 1DCNN-RNN variant. • Uses raw acceleration time-series type of sequential data as the input data. • Compares the proposed model with equivalent dl -based model architectures. • Moreover, examines the environmental variable which affects structural response. In recent years, by emerging deep learning (DL) based algorithms, researchers have been exploring dl -based models to identify structural damage through data-driven approaches. dl -based data-driven techniques using autonomous feature extraction from raw sequential data are more robust under environmental variations. Extraction of robust features while considering the sequential dependencies will significantly improve the accuracy of damage identification by these techniques. In this regard, various architectures of dl -based models have been proposed. This study presents a novel dl -based model that utilizes both one-dimensional convolutional neural network (1DCNN) and recurrent neural network (RNN) variants using an attention mechanism. Attention mechanism improves the performance of the 1DCNN-RNN variants model precisely when its input data is raw acceleration time-history, as a kind of sequential data. The IASC-ASCE phase II and Qatar University grandstand simulator benchmarks are used to evaluate the proposed model by comparing its performance with dl -based neural network architectures that could be equivalent to this combination. Moreover, the environmental variable which affects structural response is also examined. Results demonstrate that the CNN-ATT-biGRU model architecture has the best accuracy and appropriate training time and model size among nine compared architectures. [ABSTRACT FROM AUTHOR]