Refactoring and Optimization of Bridge Dynamic Displacement Based on Ensemble Empirical Mode Decomposition

Yingquan Zou, Yunpeng Chen, Peng LiuSensors 2019 · DOI

TL;DR

This paper introduces a novel method for accurately reconstructing a bridge's dynamic displacement from accelerometer data. By combining Ensemble Empirical Mode Decomposition (EEMD) with time-domain integration, the approach effectively filters out noise and drift, leading to significantly improved accuracy compared to existing methods. This advancement is crucial for reliable bridge health monitoring.

What Problem Does This Paper Address?

Accurately measuring the dynamic displacement of bridges is vital for structural health monitoring (SHM) and ensuring safety. While accelerometers are cost-effective and easy to install, converting their raw output into precise displacement data is challenging. Direct integration of acceleration signals suffers from amplification of low-frequency drift and high-frequency noise, leading to significant errors. Existing methods, such as frequency-domain integration or basic Empirical Mode Decomposition (EMD), have limitations like truncation errors, modal mixing, and end effects, which compromise accuracy, especially for long-term monitoring. This paper aims to overcome these limitations by proposing a more robust and accurate displacement reconstruction method.

Methods

The core of the proposed method lies in the combination of Ensemble Empirical Mode Decomposition (EEMD) and time-domain integration.

  1. Acceleration Signal Acquisition: Raw vibration data is collected using MEMS accelerometers.
  2. EEMD Decomposition: The accelerometer signal is decomposed into multiple Intrinsic Mode Functions (IMFs) using EEMD. EEMD is an enhancement over EMD that addresses modal mixing and end effects by adding white noise and averaging results over multiple trials. This process effectively separates the signal into components representing different oscillatory modes.
  3. IMF Selection and Reconstruction: The "effective" IMFs, which represent the true dynamic response of the bridge, are selected. These selected IMFs are then used to reconstruct a cleaner acceleration signal.
  4. Velocity Reconstruction: The reconstructed acceleration signal is integrated once in the time domain to obtain the velocity signal.
  5. Displacement Reconstruction: The reconstructed velocity signal is integrated again in the time domain to obtain the final dynamic displacement signal.

This approach leverages EEMD's ability to adaptively filter out noise and drift inherent in the acceleration signal before the integration steps, thus preventing error amplification.

Datasets

The paper utilizes the following datasets for evaluation:

  • Simulated Bridge Analog Signals: Ideal displacement and acceleration signals were generated using mathematical functions (e.g., s = 10 sin(5πt) + 10 sin(10πt)). Noise was added to the acceleration signal to simulate real-world conditions.
  • Vibration Test Bench Data: Data was collected from a vibrating table simulating bridge vibrations at a specific frequency (8.7 Hz) and amplitude (0.13 mm). This allowed for comparison with a laser displacement sensor.
  • Field Data from Highway Elevated Bridge: Real-world acceleration data was collected from an actual highway elevated arch bridge under traffic loads.

Key Insights and Ideas

  • EEMD's Superiority for Noise and Drift: The paper strongly demonstrates that EEMD is significantly better than traditional frequency-domain filtering or basic EMD at handling the low-frequency drift and high-frequency noise present in accelerometer data, especially when preparing for integration.
  • Preventing Error Amplification: By filtering the signal before the double integration process, EEMD effectively prevents the amplification of errors that plague direct integration methods.
  • Adaptive Filtering: EEMD's adaptive nature, decomposing the signal based on its intrinsic time scales, makes it highly suitable for complex, non-stationary vibration signals from bridges.
  • Robustness to Excitation: While the paper acknowledges limitations with broadband excitation, the method shows high accuracy for typical bridge vibration frequencies.

Key Results

The proposed EEMD-based method consistently outperformed other tested methods (Frequency Domain Bandpass Filtering, EMD Adaptive Filtering, FFT+EMD Filtering) across various evaluations:

  • Simulation: The EEMD method achieved the lowest Root-Mean-Square Error (RMSe), Kurtosis Error (Ke), and Phase Error (Pe) compared to other methods, with RMSe and Ke below 1.5% and Pe less than 2°.
  • Vibration Table Test: The EEMD method produced displacement signals that were highly coincident with the ideal signal and the laser displacement sensor measurements, outperforming other methods in terms of accuracy and avoiding significant offsets.
  • Field Test: The EEMD method provided accurate dynamic displacement reconstruction for a highway elevated bridge, capturing significant displacement peaks (up to 1.4 mm) caused by heavy trucks, with results closely matching expected values.

Limitations and Future Work

  • Broadband Excitation: The authors acknowledge that the proposed method is not suitable for broadband excitation signals.
  • Future Work: The team plans to improve the method to handle broadband excitation signals, aiming to reconstruct acceleration signals from such complex scenarios.

Summary updated

Abstract

Considering the lack of precision in transforming measured micro-electro-mechanical system (MEMS) accelerometer output signals into elevation signals, this paper proposes a bridge dynamic displacement reconstruction method based on the combination of ensemble empirical mode decomposition (EEMD) and time domain integration, according to the vibration signal traits of a bridge. Through simulating bridge analog signals and verifying a vibration test bench, four bridge dynamic displacement monitoring methods were analyzed and compared. The proposed method can effectively eliminate the influence of low-frequency integral drift and high-frequency ambient noise on the integration process. Furthermore, this algorithm has better adaptability and robustness. The effectiveness of the method was verified by field experiments on highway elevated bridges.
BibTeX
@inproceedings{zou2019refactoring,
  author = {Yingquan Zou and Yunpeng Chen and Peng Liu},
  title = {{Refactoring and Optimization of Bridge Dynamic Displacement Based on Ensemble Empirical Mode Decomposition}},
  booktitle = {Sensors},
  year = {2019},
  doi = {10.3390/s19143125},
}
Figure