Research on noise reduction technology of blade-icing signal based on acoustic emission technology

Размещено на http://www.allbest.ru/

“Automation and system engineering”, (Pacific National Univer¬sity)

the Sino-Russian Institute,

“Automation and system engineering”, (Pacific National Univer¬sity)

Changchun University, Changchun, China

“Automation and system engineering”

Research on noise reduction technology of blade-icing signal based on acoustic emission technology

Qin Hongwu - PhD, Professor,

Li Guangbin - Master, School of Electronic and Information Engi¬neering

Chye En Un - Doctor of Engineering, Professor, Head of The Department

Voronin V.V. - Doctor of Engineering, Professor

Ivanov V.E. - Cand. Sci. (Eng.), Associate Professor

Abstract

Aiming at the noise interference problem of the fan blade icing signal detected based on the acoustic emission technology, this paper adopts a multi-resolution decomposition algorithm of wavelet analysis, and finally selects db5 wavelet as the wavelet base and performs 5 layers of decomposition. For the wavelet decomposition, Data corresponding to each wavelet decomposition parameter in the energy spectrum coefficient table of each level. It is found that the energy occupied by the signal on the d2 layer is the vast majority of the main energy, and this layer can be extracted and decomposed, and it is found that the main information of the leaf ice covered by the signal can be obtained.

Keywords: acoustic emission technology, wavelet analysis, noise, blade -icing.

Аннотация

Авторы:

Цинь Хуну - Чанчуньский университет, Чанчунь, КНР

Ли Гуанбинь - Чанчуньский университет, Чанчунь, КНР

Чье Ен Ун - Тихоокеанский государственный университет, Хабаровск, Россия

Воронин В.В. - Тихоокеанский государственный университет, Хабаровск, Россия,

Иванов В. Э. -Тихоокеанский государственный университет, Хабаровск, Россия.

Для подавления помех в сигнале обледенения лопастей ветрогенератора, сформированном с использованием акустической эмиссии используется вейвлет- анализ с пятиуровневым разложением. Данные, соответствующие каждому параметру вейвлет-разложения приводятся в таблице коэффициентов энергетического спектра каждого уровня. Показано, что энергия, сигнала на втором уровне разложения, составляет основную долю от полной энергии сигнала, и на этом уровне разложения можно извлечь информацию о наличии обледенения.

Ключевые слова: ветрогенератор, акустическая эмиссия, вейвлет-анализ, подавление шумов, обледенение.

Introduction

Wind energy, as a clean and renewable energy source, is receiving increasing attention from countries around the world. Wind power equipment is increasing day by day. During the operation of the wind turbine, the blades often cause relatively serious accidents due to the ice coating on the surface, which affects the overall operation of the wind turbine. Therefore, it is possible to monitor the icing status of the fan blades in real time, especially the timely and effective monitoring of the position and size of the icing of the fan blades has important practical significance for ensuring the reliable operation of the fan equipment. This paper analyzes the blade icing signals collected by acoustic emission technology [1-4]. It is found that the collected icing signals of fan blades have a lot of noise interference problems, and the results cannot be accurately and effectively analyzed. Therefore, multi-resolution based on wavelet analysis is selected. The decomposition algorithm was performed, and the simulation analysis was performed using MATLAB simulation software. It was found that the reconstructed signal contains most of the information of the original signal. It was experimentally verified that the wavelet analysis based on wavelet analysis can effectively reduce the ice -covered signal of the blades [5]. Noise processing improves the accuracy of icing signal acquisition of fan blades by acoustic emission technology.

Theoretical method

At present, wind turbine blades are basically thin shell structures made of composite materials, and the content of composite materials usually exceeds 90%. Therefore, wavelet analysis is one of the most effective methods for analyzing signals of such composite materials [6]. It can decompose the acoustic emission signal into a series of sub-signals in different frequency ranges, and obtain the characteristics of the acoustic emission source by analyzing each sub-signal wavelet decomposition Therefore, the wavelet can better describe the local characteristics of the signal.

As far as the current research and analysis status is concerned, the research work of wavelet analysis on acoustic emission technology mainly focuses on the characteristic's analysis of acoustic emission signals, the analysis of the location of acoustic emission sources, and the study of the propagation characteristics and attenuation characteristics of acoustic emission signals. Among them, the extraction of signal feature quantities is the basis of positioning analysis and propagation attenuation characteristics analysis, and it is also a key point for breakthrough in the application of acoustic emission engineering [7]. Therefore, this paper adopts a multi-resolution decomposition algorithm based on wavelet analysis. The signal in different regions during ice time are decomposed into time-domain signal components in different frequency ranges. The multi-resolution analysis idea that decomposes ice-covered signals into different frequency ranges is an important theoretical basis for the feature positioning of acoustic emission signals in this paper.

This analysis method constructs a low-pass filter H(w) and a high-pass filter G(w) according to the basis function and scale function of the orthogonal wavelet, and the corresponding impulse response functions are h (n) and g (n). And meet the formula (1):

If the fan blade is in the ice-covered state, the sampling sequence of the acoustic emission signal is f (n), and it is an approximate signal on scale 0. The decomposition algorithm can be expressed as shown in equation group (2):

In the above formula, each parameter is shown in Table 1.

Table 1

In short, on each wavelet decomposition scale j, a convolution operation is performed on the approximate signal of the previous scale through a low-pass filter and a high-pass filter respectively [8]. Because the signal multi-resolution decomposition of the wavelet transform is a complete non-redundant decomposition, the result is obtained {Aj(n)UDj(n)|1<j<J}to reconstruct the low-frequency component and high-frequency component signals respectively on the decomposition scale j, and the algorithm expression is as shown in (3) Show:

Therefore, the signal f (n) can be expressed as:

Therefore, in the analysis of blade ice-covered acoustic emission, for a given acoustic emission signal f(n), the wavelet transform multi-resolution decomposition algorithm is used to describe features in a relatively small dimension. In combination with ice-covered analysis, the acoustic emission signal needs to be analyzed. For analysis requirements, the selection of a suitable wavelet base should meet the fast processing requirements for a large amount of data, and wavelets are extremely sensitive to defect signals and insensitive to structural noise. After transforming on the scale, it should include and characterize defect information, and at the same time it can effectively Enhance useful information and suppress useless information [9]. The equation set (1) is used to perform wavelet decomposition of the signal f (n) on the scale J, and according to (2), the J + 1 frequency range components after the wavelet decomposition are reconstructed, and the reconstructed signals are subjected to Fourier transform

Leaf transformation to obtain detailed spectral information on each decomposition scale. Because the components of the wavelet decomposition contain the signal characteristics of the icing of the leaves, the time domain signals and spectral signals on each wavelet decomposition scale of the signals are compared, and the characteristic values reflecting the signal characteristics can be extracted. The wavelet analysis is aimed at the acoustic emission waves of different composite materials such as blades

It extracts a single frequency or a waveform formed in a narrow frequency band, selects the peak value of the formed waveform, and effectively compensates the attenuation signal, so as to achieve blade coverage. Accurate positioning of acoustic emission sources when ice conditions occur, to confirm the target position of the blade. signal wind turbine acoustic

Experimental scheme

This paper adopts the multi-resolution decomposition algorithm in wavelet analysis, which is achieved by decomposing the signals of different areas when the leaves are covered with ice into time-domain signal components in different frequency ranges. Through laboratory collection, we find the collected signals. All are noisy signals, which need to be processed for noise reduction [12]. They cannot be directly used for the analysis of the characteristic parameters and waveforms of the acoustic emission source [13]. This will have a greater impact on the determination of the icing of the leaves. The ratio is called the ideal processing method, which is based on the multi-resolution decomposition algorithm in wavelet analysis. According to the different characteristics of the signal and noise on the wavelet scale, the modulus maximum points generated by the noise are selected and kept at their positions, and then the wavelet coefficients are reconstructed. The remaining modulus maximum points are used in the reconstruction [14]. The signal reconstructed is the recovered signal of real information. The multi-resolution wavelet decomposition structure is shown in Figure 1.

Fig. 1. Multi-resolution wavelet decomposition structure diagram

According to the foregoing description, MATLAB software is used for simulation analysis, and a rich wavelet function is provided in the MATLAB wavelet toolbox, including as shown in Table 2.

Table 2 Wavelet function classification

When selecting the wavelet base, we must consider the multiple standard requirements of wavelet. The wavelet base selected in this paper should have the following capabilities: Enough for fast processing of large amounts of data and discrete wavelet transform; Wavelets can exhibit characteristics of transientity and diversity; Sensitive to defect signals and can accurately represent defect information; Wavelet bases should have linearly correlated rows.

Considering the selection criteria of wavelet and the review of related data in the past, db5 wavelet was finally selected as the wavelet base and it was decomposed into 5 layers, as shown in Figure 2 for the db5 wavelet effect.

Fig. 2. db5 wavelet effect diagram

Decompose the wavelet shown in Figure 2 above in 5 layers to get the effect diagram shown in Figure 3.

Fig. 3. db5 wavelet 5-layer decomposition effect diagram The energy spectrum coefficients of each level are:

According to the figure above, the db5 wavelet can quickly decay in the frequency domain, and its waveform is relatively symmetrical, with an exponential function-like attenuation signal. Observe the data corresponding to each wavelet decomposition parameter in the energy spectrum coefficient table of each level after the wavelet decomposition in Table 3. It can be found that the energy of the signal on the d2 layer is about 52 % of the main energy, and it has more than half of the energy of the signal. Therefore, this layer can be extracted and decomposed, and the acoustic emission contained in the signal can be obtained. The main information of the source.

Table 3 Table of energy spectrum coefficients of various levels after db5 wavelet decomposition

Select the collected acoustic emission signals with a lot of noise, as shown in Figure 4.

Fig. 4. Noise signal of acoustic emission

According to the foregoing description, the resolution analysis algorithm in the wavelet transform is used to analyze and process the graph, and the two graphs are compared. As shown in Figure 5, the noise emission signal is shown below:

Fig. 5. Acoustic emission signal diagram after noise reduction

It can be found from the comparison of the above figure that the acoustic emission signal in Figure 4 after the noise reduction can maintain the basic characteristics of the original signal waveform in FIG. 5 and ensure that the information contained in the original signal is not lost due to noise reduction.

Conclusion

In this paper, a multi-resolution decomposition algorithm based on wavelet analysis is adopted to solve the noise interference problem of the fan blade icing signal detected by the acoustic emission technology. The db5 wavelet is selected as the wavelet base and 5 layers of decomposition are used. Corresponding to each wavelet decomposition parameter data in the energy spectrum coefficient table of each level after wavelet decomposition, it is found that the energy of the signal on the d2 layer is about 52 % of the main energy. information. According to the simulation results, a part of the noisy acoustic emission signal is selected for verification. After comparison, it is found that the acoustic emission signal after noise reduction can maintain the basic characteristics of the original signal waveform, which can ensure that the information contained in the original signal will not Lost due to noise reduction. It shows that this scheme can be used in ice analysis of fan blades.

References

1. Sabatier J., Lanusse P, Feytout B. A solution for ice accretion detection on wind turbine blades // 14th International Conference on Informatics in Control, Automation and Robotics. ICINCO. NY : IEEE, 2017. Р. 414-421.

2. Чье Ен Ун, Куликов Д.А., Харитонов К.О. Обнаружение импульсов акустической эмиссии и обеспечение единого времени в системе сейсмоакустического контроля горного давления // Информатика и системы управления. 2007. № 2. С. 109-119.

3. Овчарук В.Н., Чье Ен Ун. Анализ амплитудно-частотных характеристик при изучении свойств керамических материалов методом акустической эмиссии // Вестник ТОГУ. 2007. № 4. С. 171-184.

4. Овчарук В.Н., Чье Ен Ун. Особенности построения систем регистрации и анализа сигналов акустической эмиссии // Приборы. 2014. № 1. С.37-43.

5. Yin Linhuan Li, Jie Hu Xiaodi. Application Research of Icing Sensor for Road Safety Early // Warning Journal of Wuhan Institute of Technology, 2016.

6. Yue Jianfei Shi Lei. Icing and Antiicing of Aircraft Science and Technology and Enterprises, 2014.

7. Wang D, Tsui K. L..Dynamic Bayesian wavelet transform: New methodology for extraction of repetitive transients // Mechanical Systems & Signal Processing. 2017. № 88. Р. 137-144.

8. Shu L, Li H, Hu Q. Study of ice accretion feature and power characteristics of wind turbines at natural icing environment // Cold Regions Science and Technology. 2018. № 147. Р. 45-54.

9. Jial In T., Sl Im S., Crist Inel M. An experimental study of acoustic emission methodology for in service condit ion moni toring of wind turbine blades // Renewable Energy. 2016. № 99. Р. 170-179.

10. Joosse P, Blanch M., Dut ton A. Acoust ic Emission Monitoring of Small Wind Turbine Blades // Journal of Solar Energy Engineering. 2013. № 124. Р. 401-411.

11. Christopher B., Gregory N., Morscher V. Transverse cracking in carbon fiber reinforced polymer composites: Modal acoustic emission and peak frequency analysis // Composites Science and Technology. 2015. № 116. Р. 26-32.

12. Baker C., Morscher G.N., Pujar V. V. Transverse cracking in carbon fiber reinforced polymer composites: Modal acoustic emission and peak frequency analysis // Composi tes Science and Technology. 2015. № 116. Р. 26-32.

13. Chao K., Liu J. Analysis of Technology Development Trend of the Wind Power Generation-Based on Patent Map Cluster Analysis System and Method // Advanced Materials Research. 2013. № 748. Р. 490-492.

14. Wang Y., Xu Y, Lei Y. An effect assessment and prediction method of ultrasonic deicing for composite wind turbine blades // Renewable Energy. 2018. № 118. Р. 1015-1023.

15. Заглавие: Исследование способа подавления помех в сигнале акустической эмиссии для обнаружения обледенения лопастей ветрогенератора

Размещено на Allbest.ru