Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 Oct 17;24(20):6675.
doi: 10.3390/s24206675.

Comparative Analysis of Machine Learning Techniques for Identifying Multiple Force Systems from Accelerometer Measurements

Giovanni de Souza Pinheiro  1 , Fábio Antônio do Nascimento Setúbal  1 , Sérgio de Souza Custódio Filho  2 , Alexandre Luiz Amarante Mesquita  1 , Marcus Vinicius Alves Nunes  1
Affiliations

Comparative Analysis of Machine Learning Techniques for Identifying Multiple Force Systems from Accelerometer Measurements

Giovanni de Souza Pinheiro et al. Sensors (Basel). .

Abstract

The knowledge of the forces acting on a structure enables, among many other factors, assessments of whether the component's useful life is compromised by the current machine condition. In many cases, a direct measurement of those forces becomes unfeasible, and an inverse problem must be solved. Among the solutions developed, machine learning techniques have stood out as powerful predictive tools increasingly applied to engineering problem-solving. This study evaluates the ability of different machine learning models to identify parameters of multi-force systems from accelerometer measurements. The models were assessed according to their prediction potential based on correlation coefficient (R2), mean relative error (MRE), and processing time. A computational numerical model using the finite element method was generated and validated by vibration measurements performed using accelerometers in the laboratory. A robust database created by the response surface methodology in conjunction with Design of Experiment (DOE) was used for the evaluation of the ability of machine learning models to predict the position, frequency, magnitude, and number of forces acting on a structure. Among the six machine learning models evaluated, k-NN was able to predict with a 0.013% error, and Random Forests showed a maximum error of 0.2%. The innovation of this study lies in the application of the proposed method for identifying parameters of multi-force systems.

Keywords: finite element method; force identification; harmonic analysis; machine learning; vibration measurements.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflicts of interest.

Figures

Figure 1
Figure 1
Flowchart of the fundamental steps taken for achieving the study objective.
Figure 2
Figure 2
Basic measurement system used in experimental modal analysis with vibration exciter (Shaker), force transducer, and accelerometer.
Figure 3
Figure 3
Experimental system: suspended sheet with vibration exciter at point 49.
Figure 4
Figure 4
Computational geometry: (a) measurement point locations; (b) finite element mesh.
Figure 5
Figure 5
Numerical harmonic analysis with fixed force positioned at point 33 and variable force positioned at point 13.
Figure 6
Figure 6
Frequency response function of points 01, 26, and 49 on the plate highlighting the first five natural frequencies.
Figure 7
Figure 7
Mode shape associated with the first bending mode, 60 Hz: (a) experimental model (b) numerical model.
Figure 8
Figure 8
Modal shape associated with the second bending mode, 81 Hz: (a) experimental model (b) numerical model.
Figure 9
Figure 9
Comparison of natural frequencies and modal forms of experimental and numerical models: (a) graph with MAC values; (b) graph with COMAC values.
Figure 10
Figure 10
Response surface for acceleration at point 49 as a function of the force component at point 33 and frequency.
Figure 11
Figure 11
Comparison between the predicted response surface values and the points generated by the harmonic analysis for the force at point 40.
Figure 12
Figure 12
Importance of vibration measurement locations (features).
See this image and copyright information in PMC

References

    1. Rezayat A., Nassiri V., de Pauw B., Ertveldt J., Vanlanduit S., Guillaume P. Identification of Dynamic Forces Using Group-Sparsity in Frequency Domain. Mech. Syst. Signal Process. 2016;70–71:756–768. doi: 10.1016/j.ymssp.2015.09.015. - DOI
    1. Feng W., Li Q., Lu Q. Force Localization and Reconstruction Based on a Novel Sparse Kalman Filter. Mech. Syst. Signal Process. 2020;144:106890. doi: 10.1016/j.ymssp.2020.106890. - DOI
    1. Lin T.-K., Liang J.-C., Zhou G.-D., Yi T.-H., Xie M.-X., Ciang C.C., Lee J.-R., Bang H.-J. Structural Health Monitoring for a Wind Turbine System: A Review of Damage Detection Methods. Meas. Sci. Technol. 2008;19:122001. doi: 10.1088/0957-0233/19/12/122001. - DOI
    1. Zhang F., Ji S., Shi Y., Zhan C., Zhu L. Investigation on Vibration Source and Transmission Characteristics in Power Transformers. Appl. Acoust. 2019;151:99–112. doi: 10.1016/j.apacoust.2019.03.011. - DOI
    1. Altstadt E., Scheffler M., Weiss F.P. Component Vibration of VVER-Reactors—Diagnostics and Modelling. Prog. Nucl. Energy. 1995;29:129–138. doi: 10.1016/0149-1970(95)00002-2. - DOI

LinkOut - more resources