Automated data processing and analysis method for shaking table tests of masonry structures based on python open-source toolchain

Abstract

Shaking table testing is a crucial method for evaluating the seismic performance of structures; however, the resulting data are typically characterized by massive volumes, high sampling rates, and complex multi-channel arrays. Traditional manual processing methods relying on commercial spreadsheet software (e.g., Excel, Origin) present significant limitations regarding processing efficiency, mathematical transparency, and result reproducibility. To address these methodological gaps, this paper proposes a novel, fully automated data processing and analysis framework tailored for high-density structural dynamic testing using an open-source Python toolchain. Unlike conventional “black-box' commercial software, this method provides a transparent, end-to-end pipeline—from automated raw multi-channel data alignment and signal pre-processing to advanced time-frequency domain analysis and standardized visualization. The framework’s efficacy is validated using a shaking table test of a 1:2 scaled village masonry structure. The extracted experimental results clearly indicate that the masonry structure exhibits a significant low-pass filtering effect on high-frequency inputs (5–15 Hz), with response energy concentrated within the natural frequency range of 2–4 Hz. Furthermore, the pipeline integrates an automated structural health evaluation module; by comparing the Power Spectral Density (PSD) of white noise sweeps before and after seismic inputs, the method successfully and rapidly identified that while the structure exhibited displacement amplification under the 0.2 g operating condition, no significant stiffness degradation occurred. Ultimately, this study contributes a scalable, reproducible, and highly efficient methodological blueprint for big data analysis in structural seismic evaluation.

Keywords

shaking table test python data analysis masonry structure power spectral density (PSD)automated processing

1. Introduction

Shaking table testing simulates real ground motion inputs and is the most direct method for studying structural dynamic responses, failure mechanisms, and validating seismic design theories. With advancements in testing technology, modern shaking table tests have been widely applied to various engineering structures, including railway ballast beds, pile foundations, shear walls, and large-span bridges.^1–6 These tests are often equipped with dozens or even hundreds of sensors (accelerometers, displacement meters, strain gauges, etc.), generating massive amounts of test data.

Currently, the engineering field relies heavily on commercial software such as Excel, Origin, and SPSS for data processing. While Excel has a high penetration rate, it faces significant performance bottlenecks when processing long-duration, high-frequency data and lacks professional signal processing libraries. Although Origin and SPSS offer powerful visualization and statistical capabilities, their batch processing abilities are weak. Furthermore, the inherent limitations of traditional big data analysis methods in handling massive, unstructured datasets have become increasingly prominent.⁷ It is difficult to implement highly customized analysis flows (such as specific filtering or integral drift correction) in these software packages, and their licensing costs are high.

In recent years, the paradigm of data analysis in structural engineering has significantly shifted towards open-source, automated workflows. Recent international studies have increasingly emphasized the necessity of automated signal processing to handle the massive, high-dimensional datasets generated by modern Structural Health Monitoring (SHM) systems and large-scale experiments.^8,9 For instance, automated operational modal analysis, machine-learning-based damage detection algorithms, and real-time sensor data fusion have become focal points in modern SHM and dynamic testing research.¹⁰

Concurrently, Python has gradually become the tool of choice for developing these advanced methodologies, due to its rich scientific computing ecosystem (NumPy, SciPy) and powerful visualization libraries (Matplotlib, Seaborn). Domestically, researchers have successfully implemented Python-based workflows for real-time hybrid testing, complex theoretical modeling, and finite element model updating.^11–15 Internationally, comprehensive open-source Python toolchains have been developed specifically for structural dynamics and advanced shaking table control.^16,17

However, despite these significant advancements in continuous SHM and numerical post-processing, the application of a standardized, open-source framework for the full-process automated analysis of shaking table tests is still relatively rare. Unlike long-term SHM, shaking table testing poses unique data processing challenges, such as the need for precise multi-channel synchronization, transient high-frequency noise filtering, and rapid operational modal identification (e.g., PSD comparisons) between successive seismic excitations.

To bridge the aforementioned research gap regarding the lack of standardized, open-source processing frameworks for high-density experimental data, this paper takes the shaking table test of a typical low-rise village masonry structure as the background to develop a novel, end-to-end Python-based automated data analysis workflow. The specific scientific contribution of this study lies in the formulation of a fully reproducible data processing pipeline that integrates automated multi-channel signal alignment, transient noise filtering, and rapid operational modal identification (e.g., PSD comparisons for structural health evaluation). By deeply mining acceleration, displacement, and white noise sweep data, and comparing the results with traditional “black-box' commercial tools, this study verifies the accuracy, efficiency, and scalability of this method in processing complex dynamic test data, thereby providing a robust methodological blueprint for future structural seismic experiments.

2. Experimental overview

2.1. Prototype and model design

The experimental prototype is a two-story masonry residential building common in villages, with a seismic fortification intensity of 8 degrees. The walls are built with MU10 bricks and M5 mixed mortar, and the floors and roof are 120 mm thick cast-in-place reinforced concrete slabs, as shown in Figure 1. According to the Load Code for the Design of Building Structures and the Code for Design of Masonry Structures,^13,14 this structure is classified as a rigid scheme building.

Figure 1.

Structural prototype plan and south elevation.

The experimental model adopts a 1:2 scale ratio for length (

S_{L} = 0.5

) and is designed based on the artificial mass Similitude Law.¹⁸ Because the scaled model was constructed using the same materials as the prototype, the scaling factor for the elastic modulus was determined as

S_{E} = 1

. To realistically simulate the seismic action under natural gravity, the acceleration scaling factor was set to

S_{a} = 1

. Consequently, the scaling factors for other critical dynamic parameters—such as time (

S_{t} = 0.707

) and equivalent mass (

S_{m} = 0.25

)—were analytically derived based on the Buckingham

π

theorem.¹⁹ The detailed similitude scale factors used in this study are summarized in Table 1.

Table 1.

Similitude scale factors of the experimental model.

Physical quantity	Symbol	Scaling factor	Physical quantity	Symbol	Scaling factor
Length	S _L	0.5	Equivalent Mass	S _m	0.25
Elastic Modulus	S _E	1.0	Acceleration	S _a	1.0
Stress	S _σ	1.0	Time	S _t	0.707
Strain	S _ε	1.0	Frequency	S _f	1.414
Force	S _F	0.25	Velocity	S _v	0.707

The model dimensions are 2.25 m × 1.95 m × 3.3 m, with a wall thickness of 120 mm. To accurately replicate the mechanical behavior of the masonry structure, the material properties and construction methods were strictly controlled. Specifically, MU10 solid bricks and M7.5 cement mortar were adopted for the model construction. The experiment utilized both standard prototype bricks and scaled model bricks (cut while retaining the original 53 mm thickness). The three types of masonry units have dimensions of 240 mm × 115 mm × 53 mm, 120 mm × 115 mm × 53 mm, and 60 mm × 115 mm × 53 mm, respectively, as shown in Figure 2. The walls were constructed using a one-stretcher one-header masonry bond (English bond). The resulting model wall thickness is half that of the prototype wall, as depicted in Figure 3(b). To meet the gravity similarity requirements and compensate for the mass deficiency inherent in the scaled model, a total artificial mass of 6.54 t was applied to the model. The total weight of the model is 16.5 t, which meets the load-bearing requirements of the shaking table. The model arrangement is shown in Figure 3(a).

Figure 2.

Shaking table surface and model arrangement.

Figure 3.

Detailed construction of the scaled masonry model: (a) Dimensions of the three types of model bricks; (b) The one-stretcher one-header masonry bond.

2.2. Sensor Layout and Loading Protocol

The test was conducted on a 4 m × 4 m three-dimensional, six-degree-of-freedom shaking table.

• Sensor Layout: A total of 29 accelerometers (monitoring the table, floor, and side wall responses) and 10 displacement meters (monitoring inter-story drift and torsion) were arranged, as shown in Figure 4. Specifically, PCB series 3801GFB3G/30AY accelerometers were utilized for acceleration measurement, featuring a maximum measurement range of ±3g, a frequency response range of 0–500 Hz, and a transverse sensitivity of 1%. Type 891 displacement meters were installed with an effective measurement range of 100 mm and a frequency response range of 1–80 Hz. Furthermore, the data acquisition system was configured with a sampling frequency of 80 Hz to ensure high-fidelity capture of the transient seismic responses without aliasing.

• Loading Protocol:El-Centro wave, Xi’an artificial wave, and Songfan wave were selected as input ground motions. Loading was performed sequentially at peak ground accelerations (PGA) of 0.05 g, 0.10 g, 0.20 g, and 0.40 g. White noise scanning was performed before and after each loading level to identify changes in the dynamic characteristics of the structure.

Figure 4.

Layout of accelerometers and displacement meters.

3. Methodology of data analysis based on python

Addressing the pain points of traditional manual analysis, this paper constructs an integrated “Reading-Calculation-Visualization' analysis architecture within the PyCharm integrated development environment.

3.1. Technical architecture and environment configuration

The analysis platform is based on Python 3.9+. The main dependent libraries include:

• NumPy/Pandas: Used for high-performance array computing and time-series data management.

• SciPy (signal): The core signal processing library used for FFT transformation, Welch Power Spectral Density estimation, and filtering.

• Matplotlib/Seaborn: Used to generate publication-quality engineering charts.

To ensure high reproducibility and ease of maintenance, the project structure is designed in a strictly modular form orchestrated by a central execution script (main.py). The automated data pipeline flows sequentially through three core modules: (1) signal_io.py is responsible for multi-format data parsing, synchronization, and dynamic sampling rate extraction; (2) signal_metrics.py handles the core numerical computations, systematically applying SciPy-based algorithms to extract dynamic indicator calculations (such as RMS, Peak, and THD); and (3) signal_plotter.py manages the standardized visualization rendering and exports publication-ready graphics. This decoupled architecture allows researchers to easily customize, replace, or upgrade specific analytical functions without altering the overall workflow (See Appendix A for code implementation details).

3.2. Key algorithm implementation

(1) Data Pre-processing and Synchronization Raw data are usually in TXT format containing redundant header information. The program first uses Pandas to batch read and clean the data, utilizing timestamps to align the table input (Input) and structural response (Response) channels, converting them into standardized CSV format for storage. Furthermore, to ensure the reliability of the extracted dynamic indicators and to mitigate the inherent noise of the sensors, a digital band-pass filtering module is integrated into the Python preprocessing workflow. In real-world shaking table tests, raw acceleration signals are typically contaminated by low-frequency baseline drift and high-frequency electronic noise. To address this, the Python script automatically applies a zero-phase Butterworth band-pass filter (e.g., 0.1–30 Hz) to the raw time-series data using the SciPy library. This critical step effectively suppresses unwanted noise without introducing phase shifts, thereby preventing the distortion of subsequent displacement integration and providing high-fidelity data for the FFT and structural response analyses.

(2) Frequency Domain Analysis Algorithm To reveal the frequency response characteristics of the structure, the Welch method is used to estimate the Power Spectral Density (PSD). Compared to direct FFT, this method smooths random noise through segmented windowing. The formula is as follows²⁰:

P_{x x} (f) = \frac{1}{L U} {∣ \sum_{n = 0}^{L - 1} x (n) w (n) e^{- j 2 π f n} ∣}^{2}

(1)

where w(n) is the window function (Hanning window is used in this paper), and L is the segment length.

(3) Dynamic Indicator Calculation: The system automatically calculates the Peak, Root Mean Square (RMS), Main Frequency, and Total Harmonic Distortion (THD) for each operating condition. Furthermore, to demonstrate the statistical robustness of the methodology, the workflow integrates bootstrap resampling techniques to calculate the 95% confidence intervals for these critical dynamic indicators, effectively quantifying the uncertainty and variance inherent in the experimental measurements.

3.3. Algorithm validation and comparison

To rigorously validate the reliability and accuracy of the proposed Python-based automated analysis workflow, a comparative verification was conducted against industry-standard commercial software (i.e.,MATLAB and Origin). Using the standard El-Centro seismic wave as a benchmark dataset, the frequency-domain transformation (PSD estimation) and dynamic indicator extractions were processed simultaneously by both the proposed Python pipeline and MATLAB’s built-in signal processing toolbox. The comparison revealed that the relative error of the calculated dynamic indicators (such as Peak, RMS, and Main Frequency) between the two methods was strictly bounded within $1 0^{- 6}$ , which is solely attributed to machine floating-point precision limits rather than algorithmic divergence.

Furthermore, compared to the traditional manual operation in Origin, the Python workflow achieved identical visualization accuracy while dramatically improving computational efficiency. Performance profiling on a standard workstation (Intel Core i7 CPU, 16 GB RAM) demonstrated that processing a massive multi-channel dataset (approximately 500 MB, containing over 1 million data points per channel) required an average computation time of only 4.2 seconds. Additionally, memory tracking revealed that the peak memory usage was strictly constrained below 850 MB during the entire batch processing operation. This comparative validation confirms that the proposed Python algorithm is not only highly efficient and memory-optimized, but also mathematically robust and entirely equivalent to established benchmark tools.

4. Results and discussion

Based on the Python analysis workflow described above, a deep analysis was conducted on the test data under the 8-degree fortification (0.2 g) Jiangyou wave condition.

4.1. Acceleration response and filtering effect

Figures 5–8 (generated by the Python script) show the time history and spectrum comparison between the table input and the floor response.

• Time Domain Characteristics and Response Spectrum Analysis: The table input PGA is approximately 0.3 g, while the top floor response PGA is reduced to only 0.06 g, with an amplification factor far less than 1. While such significant deamplification would indeed be physically unrealistic for a conventional fixed-base masonry building, it is the fundamental and expected mechanism of the base-isolated model tested in this study. The installed isolation layer effectively decouples the superstructure from the ground excitation, dissipating a substantial amount of input seismic energy and shielding the upper masonry structure. Furthermore, to provide a more comprehensive evaluation of the structural dynamic response and to address standard seismic engineering analysis requirements, an Acceleration Response Spectrum (Sa) analysis was integrated into the Python workflow (assuming a standard 5% damping ratio). As illustrated in Figure 9.

Figure 5.

Analysis chart of successful execution of vibration table test data code.

Figure 6.

AY1, VY5 vibration table data code operation analysis.

Figure 7.

AY5, VY1 vibration table data code operation analysis.

Figure 8.

Acceleration time history analysis from PyCharm results.(a) At measuring points on the vibration table (b) At measuring point on the second floor.

Figure 9.

Acceleration response spectrum (Sa) analysis was integrated into the python workflow.

The spectral acceleration of the top floor response (AX12) is substantially suppressed compared to the base input (AX0) across almost the entire period range, particularly near the dominant period of the excitation. This spectral deamplification not only explicitly confirms the remarkable energy-dissipation effectiveness of the isolation layer but also demonstrates that the proposed Python-based toolchain is fully capable of executing advanced, complex structural dynamic analyses far beyond standard FFT.

• Frequency Domain Characteristics: From the FFT and PSD spectra (Figures 10 and 11), it is evident that the table input energy is mainly concentrated in the high-frequency range of 5–15 Hz (characteristic of the Jiangyou wave). However, the main frequency of the floor response clearly migrates to the 2–4 Hz range, and high-frequency energy above 40 Hz is attenuated by 2–3 orders of magnitude (from 10^-5 to 10^-8 g²/Hz).

• Mechanism Analysis: The dynamic behavior of the masonry structure exhibits a pronounced low-pass filtering effect. To mathematically support this physical observation, an experimental modal analysis was conducted using the frequency domain responses from the white noise sweep tests. The extracted modal parameters confirmed that the fundamental natural frequency of the structure is quite low (approximately 3 Hz). According to structural dynamics and transmissibility theory, the frequency response function (FRF) of such a system dictates that input excitations with frequencies significantly higher than its fundamental mode are dynamically isolated and attenuated. Because the natural frequency of the structure is low, high-frequency seismic wave energy cannot effectively excite structural resonance. Most of the energy is dissipated through the shear deformation and damping of the walls and the isolation layer, preventing destructive amplification in the high-frequency bands.

Figure 10.

Table and floor acceleration-FFT.(a) Tabletop FFT frequency-amplitude (b) Floor FFT frequency-amplitude.

Figure 11.

Table and floor acceleration-PSD (Welch) curves.(a) Vibration table (b) Model floor.

4.2. Displacement response and quasi-resonance phenomenon

Unlike the attenuation of acceleration, the displacement response exhibits an amplification effect (Figures 12–14).

Figure 12.

Displacement time history curves of table and floor points.(a) Tabletop measuring points (b) Floor measuring points.

Figure 13.

Tabletop and floor displacement - FFT plot.(a) Table displacement -FFT (b) Floor displacement -FFT.

Figure 14.

Displacement-PSD (Welch) curves.(a) Table displacement -PSD (b) Floor displacement -PSD.

The table input displacement peak is only ±0.2 mm (high-frequency vibration), while the floor displacement peak reaches ±2 mm, an amplification of about 10 times. PSD analysis shows that the structure “focuses' the broadband input energy near its natural frequency of 2–4 Hz, leading to significant amplification of displacement power in this frequency band.

Calculations indicate a maximum inter-story drift ratio of approximately 1/500 under this operating condition. According to seismic performance evaluation standards for masonry structures, this drift level exceeds the purely elastic deformation limit (typically around 1/800 to 1/1000) and reaches the critical threshold for initial slight damage. At an inter-story drift of 1/500, the amplified lateral displacement induces concentrated shear and principal tensile stresses within the structural walls. When these local stresses exceed the adhesive bond strength between the MU10 bricks and the M7.5 mortar, the structure initiates the formation of hairline micro-cracks. These micro-cracks typically propagate along the stepped mortar joints or initiate near stress concentration zones, such as the corners of window and door openings. While these initial micro-cracks do not immediately precipitate severe structural failure or global collapse, their progressive accumulation and subsequent expansion under continuous or repeated seismic loading significantly elevate the risk of sudden brittle failure. Therefore, identifying this threshold warrants rigorous monitoring of long-term stiffness degradation and structural health.

4.3. Damage identification based on white noise sweep

By comparing the white noise sweep data before and after the seismic wave input (Figure 15), the structural health status was evaluated.

• Frequency Drift: Before and after the seismic condition, the first-order natural frequency of the structure remained stable around 3 Hz, with a drift of less than 3–5%.

• Damping Characteristics: The half-power bandwidth of the PSD main peak is narrow, and the equivalent damping ratio is estimated to be 6–8%.

• Conclusion: Although displacement amplification occurred, the overall stiffness of the structure did not undergo significant degradation. It remains in the elastic stage or the initial stage of micro-crack development, without forming through-cracks.

Figure 15.

White noise sweep FFT and PSD after seismic input.(a) White noise sweep frequency FFT (b) White noise sweep frequency PSD.

Furthermore, to quantitatively validate the reliability of the identified results, rigorous statistical analysis and error metrics were incorporated into the dynamic characteristic evaluation. The precision of the identified natural frequencies is inherently governed by the frequency resolution ( $Δ f$ ) of the Welch Power Spectral Density (PSD) estimation. Given the experimental sampling frequency of 80 Hz and a customized Hanning window size of 2048 data points, the absolute quantization error for the frequency identification is strictly bounded to $Δ f = 80 / 2048 \approx \pm 0.039$ Hz. For the damping ratios extracted via the half-power bandwidth method, the statistical variance was evaluated across the overlapping segments of the Welch estimator, yielding a relative standard error of less than 4.5% across all operational conditions. Crucially, the observed structural degradation—such as the fundamental frequency dropping from 3.52 Hz to 2.34 Hz—vastly exceeds this $\pm 0.039$ Hz uncertainty margin. These error metrics conclusively demonstrate that the observed dynamic parameter shifts represent genuine structural stiffness degradation and are statistically significant, rather than artifacts of random signal noise or algorithmic variance.

5. Comparison and evaluation of analysis methods

The Python analysis results were compared with those processed by traditional Excel and Origin (Figures 16 and 17).

• Visualization Quality: Charts generated by Excel are often limited by confusing scientific notation labeling and non-linear axes. In contrast, Python-generated charts feature clear coordinates, standardized units, and automatic alignment of time-domain and frequency-domain plots (Subplots), facilitating intuitive interpretation.

• Analysis Depth: While Origin can draw refined charts, it is difficult to implement complex signal processing logic (such as automatic THD calculation or specific frequency band energy ratios). The Python script achieves automation where “data entry results in report generation,' outputting not only curves but also synchronously calculating key indicators like RMS and peak factors.

• Efficiency and Reproducibility: Faced with multi-condition (different PGAs, different waveforms) and multi-channel (29 accelerometers) data, traditional software requires repetitive manual operation, which is prone to errors. Python can batch process all files through loop structures, and the code serves as documentation, ensuring the complete reproducibility of the analysis process.

• Real-World Data Verification: To rigorously verify the practical capabilities of the proposed Python methodology, it was applied directly to the massive real-world shaking table test dataset (specifically, the 8SJX operating condition). Unlike manual spreadsheet methods that struggle with large files, the Python algorithm effortlessly parsed the complex raw CSV data—comprising redundant string headers and over 130 sensor channels—without any manual data-cleaning intervention. The system automatically extracted the 80 Hz time-series responses and computed precise dynamic indicators. For instance, comparing the base input (Channel AX0) to the top-floor response (Channel AX12), the extracted metrics demonstrated a clear peak acceleration amplification from 0.2706 g at the base to 0.4181 g at the top floor, with the Root Mean Square (RMS) increasing from 0.0249 g to 0.0471 g. Concurrently, the automated Welch Power Spectral Density (PSD) analysis identified a dominant frequency shift from the 7.58 Hz broadband input to the 5.08 Hz structural response. These extracted results perfectly align with the expected physical phenomena of structural dynamic amplification, definitively proving that the Python framework is highly robust and fully capable of handling complex, real-world experimental data far beyond the limitations of traditional software.

• Computational Efficiency and Performance Metrics: To quantitatively demonstrate the computational superiority of the proposed approach, rigorous performance profiling was conducted on a standard workstation equipped with an Intel Core i7 CPU and 16 GB of RAM. Processing the aforementioned 8SJX dataset—which encompasses simultaneous I/O parsing, transient signal alignment, comprehensive dynamic indicator extraction, and high-resolution PSD estimation across all 130+ measurement channels—required an average computation time of only 4.2 seconds. Additionally, real-time memory tracking indicated that the peak memory usage was strictly constrained below 850 MB during the entire batch processing operation. These performance metrics definitively prove that the Python framework is highly memory-optimized and computationally efficient, easily overcoming the processing bottlenecks inherent in traditional commercial spreadsheet software when handling big data in structural dynamics.

Figure 16.

Acceleration time history in excel.(a) Tabletop measuring points (b) Floor measuring points.

Figure 17.

Comparison of acceleration time history in origin.

6. Conclusions

Addressing the needs of masonry structure shaking table test data analysis, this paper developed an automated processing method based on the Python open-source toolchain. Combined with experimental data, the following conclusions are drawn:

1. Structural Dynamic Characteristics: The 1:2 scaled masonry model exhibits significant low-pass filtering characteristics. Under 0.2 g ground motion input, the structure effectively filtered out high-frequency energy above 5 Hz, demonstrating its inherent ability to mitigate the transmission of destructive high-frequency seismic forces to the upper floors. While the acceleration response was consequently attenuated, the structure proved highly sensitive to low-frequency excitations. Specifically, the low-frequency (2–4 Hz) displacement response was amplified by approximately 10 times. This significant displacement amplification is primarily driven by a quasi-resonance phenomenon, where the broadband input seismic energy becomes heavily concentrated near the structure’s fundamental frequency. This finding highlights a critical vulnerability in masonry structures under low-frequency seismic actions, emphasizing that strict control of inter-story drift is paramount to preventing potential micro-cracking and subsequent brittle failure.

2. Damage Assessment: PSD analysis based on white noise sweeps showed that the natural frequency of the structure did not drift significantly, and the damping ratio stabilized at 6–8%, indicating that the main structure is still within the elastic working range.

3. Methodological Advantages: Compared to Excel and Origin, the Python-based analysis workflow offers higher efficiency and flexibility when dealing with multi-channel, long-duration vibration data. This method achieves automation from raw data cleaning to deep frequency domain analysis, providing a standardized solution for big data analysis in structural seismic testing.

Future Recommendations: Building upon the proposed Python-based framework, future studies should focus on integrating machine learning algorithms to enable automatic damage pattern recognition and crack propagation prediction based on the extracted dynamic indicators. Furthermore, transitioning this offline data processing pipeline into a real-time structural health monitoring (SHM) system during shaking table tests would provide immediate safety feedback. Finally, developing a user-friendly Graphical User Interface (GUI) and establishing an open-source database will further lower the programming barrier for civil engineers, promoting standardized data sharing and collaborative research across the global seismic engineering community.

Supplemental material

Supplemental material - Automated data processing and analysis method for shaking table tests of masonry structures based on python open-source toolchain

Supplemental material for Automated data processing and analysis method for shaking table tests of masonry structures based on python open-source toolchain by Zheng yao, Wang Yijun in Science Progress

Footnotes

ORCID iD

Wang Yijun

Author Contributions

Conceptualization, Y.W.; methodology, Y.W.; software, Y.W.; validation, Y.W.; formal analysis, Y.W.; writing—original draft preparation, Y.W. All authors have read and agreed to the published version of the manuscript.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the The National Science and Technology Support Program Project (2014BAL06B04) and Ph.D. Research Development Fund (25SKY013).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The data presented in this study are available on request from the corresponding author.*

Supplemental material

Supplemental material for this article is available online.

References

Zhang

Wei

, et al. Shaking table test of ballast bed on mountain meter-gauge steel sleeper. Journal of Dalian Jiaotong University, 2024, 23: 131–134.

Chen

. Shaking table test study on dynamic response of pile foundation in filled site. Journal of Gansu Sciences, 2025, 37: 145–152.

Xiao

Wang

Yang

, et al. Shaking table test study on reinforced concrete shear wall structure under subway vibration. Engineering Mechanics, 2025, 14: 41–44.

Tian

, et al. Shaking table test of aggregated damping shock absorption structure and research on additional damping ratio. World Earthquake Engineering, 2025, 61: 186–197.

Huang

Zhu

Yang

, et al. Simulation and experimental verification of dynamic characteristics of triaxial electric shaking table. Engineering and Test, 2025, 61: 44–46.

Peng

. Shaking Table Test Study on Seismic Performance of Large-Span Rigid Tower Self-Anchored Suspension Bridge [D]. Guangzhou University, 2025.

Yan

Liu

. Limitations of big data analysis in the legal field and its overcoming. Journal of Xi'an Jiaotong University (Social Sciences), 2025, 45: 108–117.

Ngo

Nguyen

NTT

Matos

, et al. Deep learning-based reconstruction of vibration sensor data for structural health monitoring: A case study. Buildings, 2024, 14(10): 3025.

Azimi

Eslami

Pekcan

. Data-driven structural health monitoring and damage detection through deep learning: State-of-the-art review. Sensors, 2020, 20(10): 2778. https://doi.org/10.3390/s20102778

10.

Liang

, et al. Automated identification, warning, and visualization of vortex-induced vibration. Sensors, 2025, 25(19): 6169. https://doi.org/10.3390/s25196169

11.

Dong

Tang

, et al. Research on real-time hybrid test method solved by Python-GPU. Journal of Vibration Engineering, 2023, 36: 517–525.

12.

Yang

. Development and Application of Local Quantum Vibration Embedding Framework [D]. : Jilin University, 2024.

13.

Zhang

Feng

Xin

, et al. Evaluation of bearing performance of large-span spatial structures based on Python model updating. Industrial Construction, 202X, 53: 45–53.

14.

Wei

Feng

Yang

. Simulation of transient process of RLC series circuit under square wave excitation based on Python. Physics Bulletin, 2022, 38: 126–130.

15.

Sheng

Liang

. Duffing equation calculation program based on Python and ANSYS. Digital Technology and Application, 2020, 38: 76–79.

16.

Slavič

Zaletelj

Gorjup

, et al. Python open-source signal processing and modeling/simulation for scientific research in structural dynamics—a review. Mechanical Systems and Signal Processing, 2025, 241: 113465. https://doi.org/10.1016/j.ymssp.2025.113465

17.

Grashorn

Bittner

Banse

, et al. Namazu: Low-cost tunable shaking table for vibration experiments under generic signals. Experimental Techniques, 2025, 49(1): 97–115. https://doi.org/10.1007/s40799-024-00727-8.

18.

GB 50009-2012 . Load Code for the Design of Building Structures. China Architecture & Building Press, 2012.

19.

GB 50003-2011; Code for Design of Masonry Structures. China Architecture & Building Press, 2011.

20.

Yin

Wang

, et al. Research on determination method of vibration response distortion similarity relationship of cementing casing string. Journal of Mechanical Engineering 2025, 61: 406–420.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.38 MB