Application of Boundary Local Feature Scale Adaptive Matching Extension EMD Endpoint Effect Suppression Method in Blasting Seismic Wave Signal Processing

Wu, Jing; Wu, Li; Sun, Miao; Lu, Ya-ni; Han, Yan-hua

doi:https://doi.org/10.1155/2021/2804539

Shock and Vibration

On this page

Abstract Introduction Conclusions Data Availability Conflicts of Interest Acknowledgments References Copyright Related Articles

Special Issue

Advances in Vibration Signal Time-Frequency Analysis for Defect Detection 2021

View this Special Issue

Research Article | Open Access

Volume 2021 | Article ID 2804539 | https://doi.org/10.1155/2021/2804539

Application of Boundary Local Feature Scale Adaptive Matching Extension EMD Endpoint Effect Suppression Method in Blasting Seismic Wave Signal Processing

Jing Wu,¹Li Wu,²Miao Sun,³Ya-ni Lu,¹and Yan-hua Han¹

Academic Editor: Chaoqun Duan

Received23 Jun 2021

Accepted07 Aug 2021

Published14 Aug 2021

Abstract

The intrinsic endpoint effect of empirical mode decomposition (EMD) will lead to serious divergence of the intrinsic mode function (IMF) at the endpoint, which will lead to the distortion of IMF and affect the decomposition accuracy of EMD. In view of this phenomenon, an EMD endpoint effect suppression method based on boundary local feature scale adaptive matching extension was proposed. This method can consider both the change trend of the signal at the endpoint and the change rule of the signal inside. The simulation results showed that the proposed method had better suppression effect on the intrinsic endpoint effect of EMD than the traditional EMD endpoint effect suppression method and achieved high-precision IMF. The endpoint effect suppression method of EMD based on boundary local feature scale adaptive matching extension was used to process the actual blasting seismic signal. The decomposition results showed that the method can effectively suppress the endpoint effect of EMD of blasting seismic signal and are helpful to extract the detailed characteristic parameters of blasting seismic signal.

1. Introduction

The endpoint effect is an unavoidable problem in most signal-processing methods. Empirical mode decomposition (EMD) [1] has been widely used as an adaptive algorithm that decomposes according to the characteristics of the data itself, so it is extremely important to solve the problem of endpoint effect of EMD [2–4].

Some methods have been put forward to solve the problem of endpoint effect of EMD, such as extremum extension method [5] and polynomial fitting extension method [6], which focus on the local change trend of signal endpoint but ignore the global signal feature. There are also a few methods that consider the global signal feature but ignore the change trend of signal endpoint [7].

In view of the above research status, an EMD endpoint effect suppression method based on boundary local feature scale adaptive matching extension is proposed, which can combine the local change trend of the signal endpoint with the feature of the original global signal itself. Compared with the traditional methods, this method has better effect and higher signal decomposition accuracy after it is processed by this method. Through the analysis of the simulation signal and the actual signal, it is proved that this method can not only effectively suppress the endpoint effect but also accurately extract signal feature parameters [8–11].

2. The Principle of Endpoint Effect

The intrinsic mode function (IMF) [12, 13] of the signal obtained by EMD needs to be screened many times. The essence of screening is to calculate the local mean value of the signal according to the upper envelope determined by all maximum points of the signal and the lower envelope determined by all minimum points of signal [14, 15]. However, the endpoint of the signal cannot be at the maximum or minimum at the same time, and it is not necessarily the extreme point. Therefore, the upper and lower envelope may diverge at the endpoint, which distorts the EMD result [16, 17].

3. The Principle of Boundary Local Feature Scale Adaptive Matching Extension EMD Endpoint Effect Suppression Method

The boundary local feature scale adaptive matching extension EMD endpoint effect suppression method consists of two parts. The first part is boundary local feature scale extension (BLFSE), which considering the variation trend of the signal endpoint amplitude and the internal relationship between the global time of the signal and the interval relationship between the global time of the signal and the time interval the endpoint. The second part is adaptive matching, finding the time series with the highest matching degree in the global signal according to the extended local feature scale.

Taking the “boundary local feature scale” as the reference, a time series with the highest matching degree with “boundary local feature scale” will be found in the global signal, which is the result of the “boundary local feature scale adaptive matching extension EMD endpoint effect suppression method.” Figure 1 shows the specific operation flow of the EMD endpoint effect suppression method based on the boundary local feature scale adaptive matching extension.

3.1. The Principle of Boundary Local Feature Scale Extension

The change trend of the signal is not only reflected at the endpoints but also reflected inside the signal [18, 19]. Therefore, there is an inherent connection between the global time of the signal and the interval time of the endpoints. According to this connection, time parameters corresponding to a maximum value point and a minimum value point can be calculated at the left and right endpoint of the signal. By importing the time parameters into the polynomial established by the maximum (minimum) value point of the endpoint, the amplitude parameters corresponding to the maximum (minimum) value point can be obtained.

3.1.1. The Time Parameter of the Boundary Local Feature Scale Extension

Take the left endpoint of the signal as an example, find the occurrence time of all the maximum points of the signal, and record them as t_max1, t_max2, …, t_maxi (i = 1,2,3, …, M). In the same way, find the occurrence time of all the minimum points of the signal and record them as t_min1, t_min2, …, t_mini (i = 1,2,3, …, N). The maximum and minimum points which need to be extended are recorded as t_max0 and t_min0, respectively. The calculation of t_max0 and t_min0 are divided into the following four cases. Case 1: t_max1 < t_min1∩t_maxM < t_minN, M = N; t_min0 and t_max0 are solved with equations (1) and (2), respectively: Case 2: t_max1 < t_min1∩t_maxM > t_minN, M = N + 1; t_max0 is solved with the same equation (2) and t_min0 is solved with Case 3: t_max1 > t_min1∩t_maxM > t_minN, M = N; t_min0 and t_max0 are solved with equation (4) and equation (5), respectively: Case 4: t_max1 > t_min1∩t_maxM < t_minN, M = N − 1; t_max0 is solved with the same equation (5), and t_min0 is solved with

3.1.2. The Amplitude Parameter of the Boundary Local Feature Scale Extension

Take the left endpoint of the signal as an example, find all the amplitudes corresponding to the occurrence time of the maximum value points of the signal, and record them as x_max1, x_max2, …, x_maxi (i = 1,2,3, …, M). In the same way, find all the amplitudes corresponding to the occurrence time of the minimum value points of the signal and record them as x_min1, x_min2,…, x_mini (i = 1,2,3, …, N). According to the amplitude variation trend near the endpoint, a maximum value point and a minimum value point near the left endpoint are extended.

The specific steps of the boundary local feature scale extension are as follows: Step 1: take maximum points closest to the left endpoint of the signal, i.e., x_max1, x_max2, …, x_maxa. The value of a is related to the sample size. Step 2: polynomial fit is x_max1, x_max2, …, x_maxa. Then, (t_max0, x_max0) can be obtained by taking t_max0 obtained in Section 3.1.1 into the fitting formula in “Step 1.”

Similarly, (t_min0, x_min0) can be obtained.

3.2. The Principle of Adaptive Matching

The trend of the signal is reflected in the endpoint and in the whole signal. Therefore, it is possible to find a curve within the signal that has the highest matching degree with the “boundary local feature scale extension” obtained in Section 3.1.

Specific steps of adaptive matching are as follows: Step 1: take a time series that includes 6 extreme points, which contains the left point continuation results (t_max0, x_max0) and (t_min0, x_min0) and the two closest maximum and minimum value points (t_max1, x_max1), (t_min1, x_min1), (t_max2, x_max2), and (t_min2, x_min2). Record this time series as l₀, which contains k sampling points. Step 2: divide the original signal into n time series with the number of sampling points k and record them as l₁, l₂, l₃, ...l_i ... l_n (i = 0, 1, 2, ..., n). Step 3: calculate the adaptive matching coefficient (adaptive matching, ξ_am) [7] and find out l_i (i = 0 1 2...n) with the highest matching degree with l₀. The adaptive matching coefficient is calculated in equation (7), where A = max {x_max0, x_max1, x_max2} and B = max {x_min0, x_min1, x_min2}. Find the minimum value of the adaptive matching coefficient, ξ_ammin = min {ξ_am1, ξ_am2, ξ_am3, …, ξ_amn}; the corresponding l_i is the desired time series, which has the highest matching degree with l₀: Step 4: shifting l_i satisfying the minimum value of ξ_am to the position of l₀, then carry out EMD.

In summary, the boundary local feature scale adaptive matching extension EMD endpoint effect suppression method can be realized by completing the above four steps.

4. Comparative Study of Multiple EMD Endpoint Effect Suppression Methods for Simulated Signals

The simulation signal is used to conduct a comparative study of multiple EMD endpoint effect suppression methods, and the correlation coefficient and standard deviation of error between the IMF and the original signal are analyzed to evaluate the EMD endpoint effect suppression.

Simulation signal S(t) is composed of three sinusoidal signals with frequencies of 10 Hz, 30 Hz, and 60 Hz, that is, S(t) = x₁(t) + x₂(t) + x₃(t), where x₁(t) = sin(2 × π × 10 × t), x₂(t) = sin(2 × π × 30 × t), and x₃(t) = sin(2 × π × 60 × t). Sampling point N = 200, sampling time t is 0, π/100, 2π/100, 3π/100....... 2π, that is, on [0, 2π], the middle point is taken with an interval of π/100.

EMD is performed on S(t) directly, and the IMF is obtained as shown in Figure 2(a). It can be found that the two ends of IMF1, IMF2, and IMF3 have different degrees of divergence. With the decomposition, the divergence of IMF3 is the most serious and tends to develop into the data. The decomposition results obtained by boundary local feature scale adaptive matching extension EMD endpoint effect suppression method, extremum extension method, and polynomial fitting method are shown in Figures 2(b)–2(d), respectively. The correlation coefficient (r_xy) [14] and standard deviation of error (D_sde) [20] between IMF component and corresponding sinusoidal signal are calculated one by one. The equations of correlation coefficient and standard deviation of error are shown in equations (8) and (9), respectively, where N is the number of sampling points, x corresponds to x₁(t), x₂(t), and x₃(t), and y corresponds to IMF1, IMF2, and IMF3. The calculation results are shown in Table 1.

(a)

(b)

(c)

(d)

The following conclusions can be drawn from Figure 2 and Table 1:(1)The three IMF components obtained by the boundary local feature scale adaptive matching extension EMD endpoint effect suppression method reflects the three sinusoidal signals contained in S(t) and has high correlation and small error with the corresponding sinusoidal signals(2)The IMF obtained by the boundary local feature scale adaptive matching extension EMD endpoint effect suppression method has the highest accuracy(3)The effect of extremum extension method is slightly better than polynomial fitting method, especially, for the suppression of low frequency components

By further analysis, the IMF obtained in Figures 2(b)–2(d) are transformed by Hilbert transform, and the marginal spectrum-based boundary local feature scale adaptive matching extension EMD endpoint effect suppression method, extremum extension method, and polynomial fitting method are obtained, respectively, as shown in Figures 3(b)–3(d). Figure 3(a) is the marginal spectrum obtained by direct Hilbert transform of x₁(t), x₂(t), and x₃(t). In Figure 3, the energy spectral density (ESD) is the ordinate.

(a)

(b)

(c)

(d)

It can be seen from Figures 3(b)–3(d) that the marginal spectral frequencies obtained by the boundary local feature scale adaptive matching extension EMD endpoint effect suppression method are 10.15 Hz, 30.11 Hz, and 60.12 Hz, respectively; the marginal spectral frequencies obtained by the extremum extension method are 8.91 Hz, 28.15 Hz, and 59.03 Hz, respectively; the marginal spectral frequencies obtained by the polynomial fitting method are 7.47 Hz, 33.38 Hz, and 60.12 Hz, respectively. The marginal spectrum shows the frequency of simple harmonic wave contained in IMF by different endpoint effect suppression methods. The error between the marginal spectrum obtained by the boundary local feature scale adaptive matching extension EMD endpoint effect suppression method and the marginal spectrum obtained by x₁(t), x₂(t), and x₃(t) direct Hilbert transform is the smallest, which indicates that the boundary local feature scale adaptive matching extension EMD endpoint effect suppression method can accurately detect the characteristic frequency contained in S(t). The results are consistent with those in Figure 2 and Table 1, which further shows that the boundary local feature scale adaptive matching extension EMD endpoint effect suppression method can effectively suppress the EMD endpoint effect, realize the extraction of the original signal detail feature parameters, and obtain high-precision IMF.

5. Application of Boundary Local Feature Scale Adaptive Matching Extension EMD Endpoint Effect Suppression Method of Blasting Seismic Wave Signal

The development of engineering blasting has greatly improved work efficiency and brought great convenience to the country’s infrastructure construction. However, the impact of its seismic effects and air shock waves on the surrounding environment has become increasingly prominent. The main manifestations are destruction and cracking of existing buildings, slope instability and collapse, and fear of humans and animals. Among them, blasting seismic effect is considered to be the primary hazard of engineering blasting [21].

Especially, in the construction urban blasting engineering, the impact of blasting seismic effects on surrounding buildings is more prominent. Based on the blasting excavation project of the water intake tank of Huanghuayuan Bridge in Chongqing, the length of the water intake tank is 135 m, the upper and bottom width is 69 m, and the lower and lower width is 24 m. The layout plan of the blasting site is shown in Figure 4. It can be seen in Figure 4 that there is a residential building 82 meters away from the water intake tank. The residential building is the key monitoring object of this blasting construction, with 7 floors above the ground. The TC-4850 intelligent blasting vibration instrument is used to monitor the building. The layout of the measuring points is shown in Figure 5. Only the layout of the measuring points of the first floor is shown here, and the other floors are the same as the first floor.

The impact of the blasting seismic effect on surrounding buildings is studied through the EMD method. The boundary local feature scale adaptive matching extension method is studied to suppress the EMD endpoint effect of the blasting seismic wave signal.

In the monitoring results, a typical seismic wave signal is selected as the research object, as shown in Figure 6. The sampling frequency of the signal is 4000 sps. According to the Nyquist sampling theorem [22], the Nyquist frequency of the measured blasting seismic wave signal is 2000 Hz, including 4096 sampling points.

The signal in Figure 6 is decomposed by the boundary local feature scale adaptive matching extension EMD endpoint effect suppression method. The decomposition results are shown in Figure 7(a). It can be found from Figure 7(a) that there is only slight divergence at the right end of IMF4, and the endpoint effect suppression of other components is well controlled. Further analysis is carried out to calculate the marginal spectrum of each IMF. The calculation results are shown in Figure 7(b). It can be found that each IMF carries a set of specific frequency signals of the blasting seismic wave signal, which once again shows that the boundary local feature scale adaptive matching extension EMD endpoint effect suppression method can realize the accurate extraction of signal feature parameters. The total marginal spectrum of the signal, as shown in Figure 8, is further obtained. Figure 8 shows that the energy of the underwater drilling blasting seismic wave is mainly concentrated in 0∼50 Hz, which is consistent with the conclusion drawn by [23]. The dominant frequency of the signal is the frequency corresponding to the maximum energy density [24], so the dominant frequency of this signal is 14.286 Hz. When the frequency of blasting seismic wave is the same as the natural frequency of the residential building, the amplitude of the structure will reach the maximum, thus inducing resonance harm.

(a)

(b)

Through the finite element analysis software of YJK, the three-dimensional model of the residential building is obtained, as shown in Figure 9, and the natural vibration frequency of the house is calculated. The natural vibration frequency of the first eight-order formation of the residential building is shown in Table 2. It can be found from the analysis of Table 2 that the second-order formation of the residential building is 13.806 Hz, and the dominant frequency of this blasting is 14.286 Hz, which are very close to each other. Therefore, the seismic wave generated by the blasting is very likely to cause the resonance of the residential building.

Therefore, the corresponding control measures must be taken in the actual construction to ensure the safety of the residential building. The conclusion also shows that the method proposed in this paper is not only helpful to suppress the EMD endpoint effect and obtain higher precision IMF but also can accurately extract the frequency parameters contained in the blasting seismic wave, which is helpful to control blasting vibration and provide basis for formulating scientific antiseismic measures.

6. Conclusions

(1)The boundary local feature scale adaptive matching extension method not only considers the local change trend of the signal at the endpoint but also retains the unique internal attributes of the signal through the global search ability of “adaptive matching,” so as to retain the authenticity of the original signal to the greatest extent(2)By comparing the decomposition results of simulation signals, it is found that the boundary local feature scale adaptive matching extension EMD endpoint effect suppression method can effectively suppress the EMD endpoint effect and obtain higher accuracy IMF(3)The frequency energy information contained in the blasting seismic wave can be effectively extracted from the marginal spectrum of the IMF obtained by the boundary local feature scale adaptive matching extension EMD endpoint effect suppression method, which is helpful to identify the characteristic parameters of blasting seismic wave signal and control blasting vibration

Data Availability

The data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest

The authors declare no conflicts of interest.

Acknowledgments

The authors would like to thank Ma Chengyang for his advice on the composition of the article. This work was supported by the National Natural Science Foundation of China (41672260 and 41907259) and the Scientific Research Program of Hubei Provincial Education Department (Q20202701).

References

N. E. Huang, Z. Shen, S. R. Long, and M. L. C Wu, “The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis,” Proceedings of the Royal Society A: Mathematical, Physical & Engineering Sciences, vol. 454, no. 3, pp. 903–995, 1998.
View at: Publisher Site | Google Scholar
H. U. Aijun, “New process method for end effects of HILBERT-HUANG transform,” Chinese Journal of Mechanical Engineering, vol. 44, no. 4, pp. 154–158, 2008.
View at: Publisher Site | Google Scholar
C. H. Bai, X. C. Zhou, D. C. Lin et al., “PSO-SVM method based on elimination of end effects in EMD,” Xitong Gongcheng Lilun yu Shijian/System Engineering Theory and Practice, vol. 33, no. 5, pp. 1298–1306, 2013.
View at: Google Scholar
Z. Liang, S. P. Peng, and J. Zheng, “LEMD endpoint extension method and its application in microseismic signals’ denoising,” Zhendong yu Chongji/Journal of Vibration and Shock, vol. 33, no. 21, pp. 155–160, 2014.
View at: Publisher Site | Google Scholar
L. Shen, X. J. Zhou, Z. G. Zhang, and W. G. Zhang, “Boundary-extension method in hilbert-huang transform,” Journal of Vibration and Shock, vol. 28, no. 8, pp. 168–171, 2009.
View at: Google Scholar
H. Liu, “Dealing with the end issue of EMD based on polynomial fitting algorithm,” Computer Engineering and Applications, vol. 40, no. 16, pp. 84–86, 2004.
View at: Google Scholar
J. Yang, G. Shi, T. Zhou, and F. Gao, “Waveform extension method based on similarity sequential detection for the end effects reduction of EMD,” Journal of Vibration and Shock, vol. 37, no. 18, pp. 121–125, 2018.
View at: Publisher Site | Google Scholar
X. Cai, H. Zhao, S. Shang et al., “An improved quantum-inspired cooperative co-evolution algorithm with muli-strategy and its application,” Expert Systems with Applications, vol. 2021, Article ID 114629, 2021.
View at: Publisher Site | Google Scholar
H. Zhao, H. Liu, Y. Jin, X. Dang, and W. Deng, “Feature extraction for data-driven remaining useful life prediction of rolling bearings,” IEEE Transactions on Instrumentation and Measurement, vol. 70, no. 99, pp. 1–10, 2021.
View at: Publisher Site | Google Scholar
W. Deng and J. Xu, “An enhanced MSIQDE algorithm with novel multiple strategies for global optimization problems,” IEEE Transactions on Systems, Man, and Cybernetics: Systems, vol. 99, pp. 1–10, 2020.
View at: Publisher Site | Google Scholar
W. Deng, S. Shang, X. Cai, H. Zhao, Y. Song, and J. Xu, “An improved differential evolution algorithm and its application in optimization problem,” Soft Computing, vol. 25, no. 7, pp. 5277–5298, 2021.
View at: Publisher Site | Google Scholar
Xu Jia, S. Huang, and F. Ma, “The dynamic characteristics analysis for the large bridge based on the improved hilbert-huang transformation,” Geomatics and Information Science of Wuhan University, vol. 35, no. 7, pp. 801–805, 2010.
View at: Google Scholar
Z. Wu and N. E. Huang, “Ensemble empirical mode decomposition: a noise-assisted data analysis method,” Advances in Adaptive Data Analysis, vol. 1, no. 1, pp. 1–41, 2011.
View at: Publisher Site | Google Scholar
Z. He, Z. Zhu, and H. Xie, “Restrain boundary effect of EMD based on least square fitting,” Journal of System Simulation, vol. 30, no. 9, pp. 3377–3385+3398, 2018.
View at: Google Scholar
R. Hao and F. Li, “A new method to suppress the EMD endpoint effect,” Zhendong Ceshi Yu Zhenduan/Journal of Vibration, Measurement and Diagnosis, vol. 38, no. 2, pp. 341–345, 2018.
View at: Publisher Site | Google Scholar
Z. N. Han, J. X. Gao, Y. X. Li, and W. T. Sun, “Dealing with end effect of EMD based on gray theory,” Advanced Materials Research, vol. 228, pp. 1094–1099, 2011.
View at: Publisher Site | Google Scholar
D. Cao and X. Zhang, “Research on the comparison with methods of restraining ending effect of EMD and its application in fault diagnosis,” Journal of Mechanical Transmission, vol. 37, no. 3, pp. 83–87, 2013.
View at: Google Scholar
J. Qiu and J. J. Chen, “EMD in the research and application of deformation monitoring in embankment,” Applied Mechanics and Materials, vol. 501, pp. 1868–1872, 2014.
View at: Publisher Site | Google Scholar
L. Wu, Y. Zhang, Y. Zhao, G. Ren, and S. He, “Mode mixing suppression algorithm for empirical mode decomposition based on self-filtering method,” Radioelectronics and Communications Systems, vol. 62, no. 9, pp. 462–473, 2019.
View at: Publisher Site | Google Scholar
X. U. Ke, Z. H. Chen, and C. B. Zhang, “Rolling bearing fault diagnosis based on empirical mode decomposition and support vector machine,” Control Theory & Applications, vol. 36, no. 6, pp. 915–922, 2019.
View at: Google Scholar
S. Y. Qian, Application of HHT in Blasting Vibration Signal Processing, Central South University, Changsha, China, 2012.
P. P. Vaidyanathan, “Generalizations of the sampling theorem: seven decades after Nyquist,” IEEE Transactions on Circuits and Systems I: Fundamental Theory and Applications, vol. 48, no. 9, pp. 1094–1109, 2002.
View at: Publisher Site | Google Scholar
M. Sun, L. Wu, Q. Yuan, and C. Ma, “Time frequency analysis of blasting seismic wave signal based on CEEMDAN,” Journal of South China University of Technology, vol. 48, no. 3, pp. 76–81, 2020.
View at: Publisher Site | Google Scholar
X. Z. Shi, Time-Frequency Analysis of Blasting Vibration Signal and Research on Characteristic Parameters of Blasting Vibration and Hazard Prediction, Central South University, Changsha, China, 2007.

Copyright

Copyright © 2021 Jing Wu et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

PDF Download Citation

Download other formats

Order printed copies

Views

1615

Downloads

563

Citations