Method for testing moisture content of soil by using piezoelectric material

Abstract

The invention discloses a method for testing soil moisture content by using a piezoelectric material, and particularly relates to the technical field of soil moisture content measurement, signals are captured through a data acquisition system on the basis of a piezoelectric ceramic sensor, wavelet analysis and manifold learning are combined by inputting captured to-be-processed signal data, and the soil moisture content is measured. According to the method, the monitored signals can be processed and analyzed more deeply and comprehensively, so that the signal characteristics of the soil moisture content are extracted more accurately, and the accuracy of the monitoring result is improved.

Description

Technical Field

[0001] The invention relates to the technical field of soil moisture content measurement, and in particular to a method for testing soil moisture content by using a piezoelectric material. Background Art

[0002] In recent years, measuring the moisture content in the soil has played a vital role in agricultural irrigation, ecological protection, and climate and hydrological research. In order to accurately measure soil moisture, scientists have proposed a variety of measurement methods. At present, the mainstream methods for measuring soil moisture include drying method, tensiometer method, resistance method, FDR frequency domain reflectometry, and TDR time domain reflectometry. The above methods for determining soil moisture content are either time-consuming and require the destruction of soil structure; or the measurement range is limited, and the measurement results are greatly affected by factors such as soil texture, soil temperature, and organic matter content; or the instrument price is high, the operation steps are complicated, and professional personnel are required to operate and maintain it. Most soil moisture test methods are not yet widely popularized. Therefore, a more accurate, efficient and cost-effective soil moisture content monitoring method is needed. Summary of the invention

[0003] In order to solve the problems of operational complexity and environmental factors in the current soil moisture measurement methods, as well as the long measurement process and high cost, the present invention provides a method for testing soil moisture content using piezoelectric materials. The method not only improves the accuracy of the test results, reduces the test time, and is applicable to different types of soils, but also reduces the monitoring cost through reasonable system design and technical combination.

[0004] The present invention provides a specific technical solution as follows: a method for testing soil moisture content using a piezoelectric material, wherein a waveform generator is precisely connected to a voltage amplifier, and then the amplified signal output end is connected to a driving end piezoelectric sensor; the driving end piezoelectric sensor is completely inserted into a test soil sample in a vertical posture, and then the receiving end piezoelectric sensor is also inserted into the soil in a vertical manner, and a parallel spacing of 1 cm is maintained with the driving end piezoelectric sensor; then the receiving end piezoelectric sensor is connected to a data acquisition card, and finally the data acquisition card is connected to an experimental computer through a USB interface to achieve data communication;

[0005] The initial sine wave signal is set and emitted through the waveform generator and the parameters are set. The initial sine wave signal is amplified by the voltage amplifier. The piezoelectric signal emitted by the driving end piezoelectric sensor propagates inside the test soil and is captured by the receiving end piezoelectric sensor. The captured signal is recorded and stored in real time by the data acquisition device and then uploaded to the computer through the USB interface.

[0006] Input the captured signal data to be processed and draw the signal time domain diagram of the current window; then perform continuous complex wavelet transform and draw the time-frequency diagram to capture the time-frequency characteristics of the signal from the two dimensions of time and frequency; set the threshold to distinguish the signal from the noise, and perform multi-level wavelet decomposition on the signal, and then split the signal into high frequency and low frequency and draw the wavelet time domain diagram after noise reduction, divide the signal into equal intervals, draw contour maps to extract the time-frequency points at the centroid of the wavelet ridge band; perform polynomial fitting on the extracted time-frequency points and draw the structural time-frequency change curve;

[0007] The LLE algorithm represents each sample through the linear combination of the local neighbor points of the structure time-frequency change curve, and reconstructs the data after fixing the neighbor weights to minimize the mapping error. The algorithm flow is as follows:

[0008] (1) Input sample set X = {x1, x2, …, x n ∈R N}, the sample true dimension is d and d=2, the ε neighborhood parameter is k and k=8;

[0009] (2) Calculate the weight w ij , use the least squares method to minimize the reconstruction cost function:

[0010]

[0011] In the formula, x i , x j is the sample point; w ij is the weight, which is the point x i The k nearest neighbors around x are i The weight of; W is the weight matrix, data X i It can only be described linearly by its immediate neighborhood. If x i , x j is not a k-neighbor point, then w ij =0 and

[0012] (3) Weight w ij The original sample is restored by the true dimension d (d < < N) of the manifold structure sample, where N is the dimension of the sample set X, i.e., an N×1 column vector; assuming that the sample data of the known mapping manifold structure is Y i , at this time the minimum objective cost function is calculated as:

[0013]

[0014] Where, M = (IW) T (IW), where I is the unit matrix; Y is the corresponding projection of the N-dimensional sample set X in the low-dimensional d dimension; Y i , Y jCorresponding to the i-th and j-th columns of matrix Y; in order to remove the shortcomings of Y's translation invariance and reduce the generation of degenerate solutions, it is necessary and Remove the eigenvectors corresponding to 0, and finally get the manifold embedding represented as the 2nd to d+1th vectors in the feature matrix;

[0015] LE maintains the proximity of neighboring samples in high-dimensional data in the low-dimensional manifold and uses weighted distance as the loss function to explore the low-dimensional embedding characteristics of the data. The algorithm flow is as follows:

[0016] (1) Input sample X = {x1, x2, …, x n ∈R N}, the sample true dimension is d and d=2, the ε neighborhood parameter is k and k=15;

[0017] (2) Construct a neighbor graph G; if two points x i , x j Not adjacent, the weight is 0, otherwise w ij =1;

[0018] (3) Constructing low-dimensional embedding:

[0019]

[0020] Where L = DW is the Laplacian matrix, D is the diagonal matrix, and And use the conditional constraints YDY T =I, removing the deficiency caused by the scale factor;

[0021] Where (3) is transformed into the generalized eigenvalue decomposition problem

[0022] Ly=λDy (4)

[0023] Take the eigenvectors corresponding to the smallest d eigenvalues, excluding the eigenvector corresponding to the smallest eigenvalue 0, and obtain the global low-dimensional coordinates Y = {y1, y2, ..., y n ∈R d}.

[0024] Preferably, a 0.5 cm×5 cm rectangular PZT piezoelectric ceramic coated with a pure acrylic waterproof coating is dried for 24 hours and used as the driving end piezoelectric sensor; a 1.63 cm×4.14 cm rectangular PVDF piezoelectric film sensor LDT1-028K is used as the receiving end sensor; and the test soil sample is screened with a sieve with a pore size of 0.45 mm.

[0025] Preferably, the continuous complex wavelet transform introduces the complex wavelet basis function on the basis of the continuous wavelet transform, so as to more comprehensively capture the time-frequency characteristics of the signal from the two dimensions of time and frequency, wherein the continuous wavelet transform expression is:

[0026]

[0027] Where: W x (a, b) are wavelet coefficients, which represent the intensity of the frequency component of the signal at a specific scale and position); a is the scale parameter to control the expansion and contraction of the wavelet; b is the translation parameter to control the position of the wavelet on the time axis; ψ a,b is the wavelet basis function; ψ * (·) is the complex conjugate of ψ(·); if continuous complex wavelet transform is used, the wavelet basis function is transformed into an adjustable complex Morlet wavelet basis function:

[0028]

[0029] Where: F b is the wavelet bandwidth parameter; F c is the center frequency of the wavelet; in the specific analysis process, by setting the parameter F b and F c The value of is used to select the target analysis frequency band.

[0030] Preferably, before distinguishing the signal from the noise, the signal in the time domain is converted into a signal in the frequency domain by Fourier transform for some places that are difficult to process in the time domain. The Fourier transform formula is as follows:

[0031]

[0032] In the formula, w is the frequency, t is the time, e -iwt is a complex function.

[0033] Preferably, the piezoelectric signal data displayed on the computer is monitored in real time using Loggernet data management software, and when the data tends to be stable, the measurement is continued for 30 minutes.

[0034] Preferably, the specific steps of wavelet decomposition include:

[0035] (1) Select cmor3-3 wavelet basis and 3 decomposition levels for wavelet decomposition;

[0036] (2) The wavelet coefficients are denoised by setting the threshold thr and thr=560 to obtain the denoised wavelet coefficients;

[0037] (3) Reconstruct the signal using the coefficients after threshold processing to complete denoising, as follows:

[0038] The wavelet ridge band is extracted by using the contour method, the amplitude c of the contour line is calculated and determined, and then the upper and lower edges of the wavelet ridge band are extracted, and the time-frequency point at the centroid of the ridge band is identified;

[0039] A0=(m+c) / 2(5)

[0040] Among them, A0 is the amplitude at the isosurface;

[0041] Where m is the maximum value of the wavelet coefficient modulus, m divides the signal into equal intervals, and the contour line method is used to draw the contour line to obtain the wavelet ridge band, and the time-frequency point at the centroid of the wavelet ridge band is saved.

[0042] The present invention has the following advantages:

[0043] The present invention is based on a piezoelectric ceramic sensor, does not require expensive experimental equipment or instruments, has a low measurement cost, and has significant advantages in the field of soil moisture monitoring due to the fast response characteristics of piezoelectric ceramics.

[0044] The present invention combines wavelet analysis and manifold learning to enable the present invention to process and analyze the monitored signals more deeply and comprehensively, thereby more accurately extracting the signal characteristics of soil moisture content and improving the accuracy of the monitoring results. Through this precise signal feature extraction, misjudgment caused by external interference or signal complexity is effectively avoided, greatly improving the accuracy of the monitoring results, and providing reliable technical support for the accurate monitoring of soil moisture content. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 A flow chart of the method for testing soil moisture content using piezoelectric materials provided by the present invention;

[0046] Figure 2 A configuration diagram of the data acquisition system provided by the present invention;

[0047] Figure 3 A data analysis flow chart provided by the present invention;

[0048] Figure 4 The spectrum diagram provided by the present invention;

[0049] Figure 5 The wavelet time-frequency diagram provided by the present invention;

[0050] Figure 6 The contour lines and wavelet ridge band diagrams provided by the present invention;

[0051] Figure 7 The time-frequency variation curve diagram of the structure provided by the present invention;

[0052] Figure 8 The LLE algorithm provided by the present invention extracts the manifold structure graph;

[0053] Fig. 9 The LE algorithm provided by the present invention extracts a manifold structure graph. DETAILED DESCRIPTION

[0054] The following is a description of the implementation of the present invention by specific embodiments. People familiar with the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0055] This embodiment provides a method for testing soil moisture content using a piezoelectric material, and the specific steps are as follows:

[0056] In step S1, a pure acrylic emulsion coating is evenly applied on a 0.5 cm×5 cm rectangular PZT piezoelectric ceramic surface and air-dried for 24 hours; the test soil sample is screened with a sieve with an aperture of 0.45 mm.

[0057] In step S2, first accurately connect the DG1022U arbitrary waveform generator to the YK_HA170D voltage amplifier, and then closely connect the amplified signal output end to the driving end piezoelectric sensor. Secondly, insert the 0.5cm×5cm rectangular PZT piezoelectric ceramic coated with a pure acrylic waterproof coating, which serves as the driving end sensor, completely into the test soil sample in a vertical posture. Then, insert the 1.63cm×4.14cm rectangular PVDF piezoelectric film sensor LDT1-028K, which serves as the receiving end sensor, into the soil in a vertical manner, and maintain a parallel distance of 1cm with the driving end sensor. Then reliably connect the receiving end piezoelectric sensor to the NI data acquisition card. Finally, use the USB interface to connect the NI data acquisition card to the experimental computer to achieve data communication. Figure 2 shown.

[0058] In step S3, the initial sine wave signal is accurately set and emitted through an arbitrary waveform generator, and its parameters are set as follows: low potential 0V, high potential 2V, and frequency stable at 1KHz. The signal is amplified by the voltage amplifier and adjusted to 20V. The piezoelectric signal emitted by the piezoelectric sensor at the driving end propagates inside the test soil and is captured by the piezoelectric sensor at the receiving end. The captured signal is recorded and stored in real time by the data acquisition card CR1000 every 0.2s, and then uploaded to the computer through the USB interface. During the experiment, the piezoelectric signal data displayed on the computer was monitored in real time using the Loggernet data management software. When the data tends to be stable, the measurement is continued for 30min.

[0059] Step S4 processes the original data, including denoising, continuous complex Morlet wavelet transform, manifold learning, and wavelet analysis, and draws the manifold structure diagram (such as Figure 8-9 As shown in the figure), spectrum diagram (as shown in the figure), Figure 4 As shown in the figure), time-frequency diagram and structure time-frequency change curve diagram. Figure 3 shown.

[0060] In the denoising stage, wavelet threshold denoising is used to reduce noise interference. The key principle is to set a threshold to distinguish between signals and noise, and then perform multi-level wavelet decomposition on the signal to split the signal into high-frequency components and low-frequency components. In terms of manifold learning, a feature extraction method based on the LLE (local linear embedding) algorithm and the LE (Laplacian eigenmap) algorithm is used to process and reduce the dimension of the data. The continuous complex wavelet transform introduces complex wavelet basis functions on the basis of the continuous wavelet transform, so as to more comprehensively capture the time-frequency characteristics of the signal from the two dimensions of time and frequency. The continuous wavelet transform expression is:

[0061]

[0062] Where: W x (a, b) are wavelet coefficients (indicating the intensity of the frequency component of the signal at a specific scale and position); a is the scale parameter (controlling the expansion and contraction of the wavelet); b is the translation parameter (controlling the position of the wavelet on the time axis); ψ a,b is the wavelet basis function; ψ * (·) is the complex conjugate of ψ(·). If continuous complex wavelet transform is used, the wavelet basis function is transformed into an adjustable complex Morlet wavelet basis function:

[0063]

[0064] Where: F b is the wavelet bandwidth parameter; F c is the center frequency of the wavelet. In the specific analysis process, this embodiment sets the parameter F b and F c The value of is used to select the target analysis frequency band.

[0065] like Figure 1 As shown, the method provided by the invention is based on the fast response characteristics of piezoelectric ceramic sensors, providing a method that can monitor the changes in soil moisture content in real time, and through the combination of wavelet analysis and manifold learning, more accurately extract signal features related to soil moisture content, thereby improving the accuracy of monitoring results.

[0066] In step S1-2, the performance of piezoelectric ceramics is protected by a waterproof coating to improve the durability of piezoelectric ceramics; the consistency and purity of soil samples are ensured by screening, reducing the interference of impurities on the test results.

[0067] In step S3, stable moisture content data is obtained by real-time monitoring of the image data displayed on the computer.

[0068] Step S4: The purpose of Fourier transform is to transform the signal in the time domain (i.e., time domain) into the signal in the frequency domain (i.e., frequency domain). As the domains change, the angle of understanding of the same thing also changes. Therefore, the frequency distribution of the signal cannot be observed in the time domain, but it can be observed more conveniently in the frequency domain. The Fourier transform formula is as follows:

[0069]

[0070] In the formula, w is the frequency, t is the time, e -iwt is a complex function.

[0071] The Fourier transform result is shown in the figure Figure 3 shown.

[0072] By setting a threshold to distinguish between signal and noise, the signal is decomposed into high-frequency and low-frequency components by multi-level wavelet decomposition. The high-frequency part mainly contains noise, and the low-frequency part reflects the overall trend of the signal. The specific steps include:

[0073] (1) Perform wavelet decomposition and select the appropriate wavelet basis and decomposition level;

[0074] (2) Setting a threshold to process the wavelet coefficients, usually using a hard threshold or a soft threshold;

[0075] (3) Reconstruct the signal using the coefficients after threshold processing to complete denoising.

[0076] The effect after denoising is as follows Figure 5 shown.

[0077] The wavelet ridge band is extracted using the contour method, the amplitude c of the contour line is calculated and determined, and then the upper and lower edges of the wavelet ridge band are extracted, and the time-frequency point at the centroid of the ridge band is identified.

[0078] A o =(m+c) / 2 (4)

[0079] In the formula, m is the maximum value of the wavelet coefficient modulus. Divide the signal into equal intervals, draw contour lines using the contour method, obtain the wavelet ridge band, save the time-frequency point at the centroid of the wavelet ridge band, and draw Figure 6 .

[0080] Finally, the time-frequency curve of the structure is obtained by fitting the fifth-order polynomial of each window time-frequency point through the least squares method. Figure 7 shown.

[0081] The LLE algorithm represents each sample through a linear combination of local neighbor points, and reconstructs the data after fixing the neighbor weights to minimize the mapping error. The algorithm flow is as follows:

[0082] (1) Input sample X = {x1, x2, …, x n ∈R N}, the sample true dimension is d and d=2, the neighborhood parameter is k and k=8 (ε neighborhood).

[0083] (2) Calculate the weight w ij . Use the least squares method to minimize the reconstruction cost function:

[0084]

[0085] DataX i It can only be described linearly by its immediate neighborhood. If x i , x j If it is not a neighbor point (k neighborhood), then w ij =0 and

[0086] (3) Weight w ij Fixed, the manifold structure d (d < < D) is used to restore the initial sample. Assume that the sample data of the known mapping manifold structure is Y i , at this time, the minimum objective cost function can be calculated as:

[0087]

[0088] Where M = (IW) T (IW). In order to remove the shortcomings of Y translation invariance and reduce the generation of degenerate solutions, it is necessary and Remove the eigenvectors corresponding to 0, and finally get the manifold embedding represented as the 2nd to d+1th vectors in the feature matrix. The result is as follows Figure 8 shown.

[0089] LE maintains the proximity of neighboring samples in high-dimensional data in the low-dimensional manifold and uses weighted distance as the loss function to explore the low-dimensional embedding characteristics of the data. The algorithm flow is as follows:

[0090] (1) Input sample set X = {x1, x2, …, x n ∈R N}, the sample true dimension is d and d=2, the neighborhood parameter is k and k=15 (ε neighborhood).

[0091] (2) Construct the neighbor graph G. If two points x i , x j Not adjacent, the weight is 0, otherwise w ij =1.

[0092] (3) Construct low-dimensional embedding.

[0093]

[0094] Where L = DW is the Laplacian matrix, D is the diagonal matrix, and And use the conditional constraints YDY T =I, eliminating the deficiency caused by the scale factor.

[0095] Equation (7) can be transformed into the generalized eigenvalue decomposition problem

[0096] Ly=λDy (8)

[0097] Take the eigenvectors corresponding to the smallest d eigenvalues ​​(except the eigenvector corresponding to the smallest eigenvalue 0), and you get the global low-dimensional coordinates Y = {y1, y2, ..., y n ∈R d}. The result is as follows Fig. 9 shown.

[0098] In summary, the present invention has experimentally designed a method for testing soil moisture content using piezoelectric materials as drivers and sensors combined with wavelet analysis and manifold learning. This method has significant advantages in the field of soil moisture monitoring based on the rapid response characteristics of piezoelectric ceramics. The present invention combines wavelet analysis with manifold learning, so that the present invention can process and analyze the monitored signals more deeply and comprehensively, thereby more accurately extracting the signal characteristics of soil moisture content and improving the accuracy of monitoring results.

[0099] Although the present invention has been described in detail above by general description and specific embodiments, it is obvious to those skilled in the art that some modifications or improvements can be made to the present invention. Therefore, these modifications or improvements made without departing from the spirit of the present invention all belong to the scope of protection claimed by the present invention.

Claims

1. A method for testing soil moisture content using piezoelectric material, characterized in that: The waveform generator is precisely connected to the voltage amplifier, and then the amplified signal output end is connected to the driving end piezoelectric sensor; the driving end piezoelectric sensor is completely inserted into the test soil sample in a vertical posture, and then the receiving end piezoelectric sensor is also inserted into the soil in a vertical manner, and maintained parallel to the driving end piezoelectric sensor; then the receiving end piezoelectric sensor is connected to the data acquisition card, and finally the data acquisition card is connected to the experimental computer through the USB interface to achieve data communication; The initial sine wave signal is set and emitted through the waveform generator and the parameters are set. The initial sine wave signal is amplified by the voltage amplifier. The piezoelectric signal emitted by the driving end piezoelectric sensor propagates inside the test soil and is captured by the receiving end piezoelectric sensor. The captured signal is recorded and stored in real time by the data acquisition device and then uploaded to the computer through the USB interface. Input the captured signal data to be processed and draw the signal time domain diagram of the current window; then perform continuous complex wavelet transform and draw the time-frequency diagram to capture the time-frequency characteristics of the signal from the two dimensions of time and frequency; set the threshold to distinguish the signal from the noise, and perform multi-level wavelet decomposition on the signal, and then split the signal into high frequency and low frequency and draw the wavelet time domain diagram after noise reduction, divide the signal into equal intervals, draw contour maps to extract the time-frequency points at the centroid of the wavelet ridge band; perform polynomial fitting on the extracted time-frequency points and draw the structural time-frequency change curve; The LLE algorithm represents each sample through the linear combination of the local neighbor points of the structure time-frequency change curve, and reconstructs the data after fixing the neighbor weights to minimize the mapping error. The algorithm flow is as follows: (1) Input sample set X = {x1, x2, …, x n ∈R N }, the sample true dimension is d and d=2, the ε neighborhood parameter is k and k=8; (2) Calculate the weight w ij , use the least squares method to minimize the reconstruction cost function: In the formula, x i , x j is the sample point; w ij is the weight, which is the point x i The k nearest neighbors around x are i The weight of W is the weight matrix, data X i It can only be described linearly by its immediate neighborhood. If x i , x j is not a k-neighbor point, then w ij =0 and (3) Weight w ij The original sample is restored by the true dimension d (d < < N) of the manifold structure sample, where N is the dimension of the sample set X, i.e., an N×1 column vector; assuming that the sample data of the known mapping manifold structure is Y i , at this time the minimum objective cost function is calculated as: Where, M = (IW) T (IW), where I is the unit matrix; Y is the corresponding projection of the N-dimensional sample set X in the low-dimensional d dimension; Y i , Y j Corresponding to the i-th and j-th columns of matrix Y; in order to remove the shortcomings of Y's translation invariance and reduce the generation of degenerate solutions, it is necessary and Remove the eigenvectors corresponding to 0, and finally get the manifold embedding represented as the 2nd to d+1th vectors in the feature matrix; LE maintains the proximity of neighboring samples in high-dimensional data in the low-dimensional manifold and uses weighted distance as the loss function to explore the low-dimensional embedding characteristics of the data. The algorithm flow is as follows: (1) Input sample X = {x1, x2, …, x n ∈R N }, the sample true dimension is d and d=2, the ε neighborhood parameter is k and k=15; (2) Construct a neighbor graph G; if two points x i , x j Not adjacent, the weight is 0, otherwise w ij =1; (3) Constructing low-dimensional embedding: Where L = DW is the Laplacian matrix, D is the diagonal matrix, and And use the conditional constraints YDY T =I, removing the deficiency caused by the scale factor; Where (3) is transformed into the generalized eigenvalue decomposition problem Ly=λDy (4) Take the eigenvectors corresponding to the smallest d eigenvalues, excluding the eigenvector corresponding to the smallest eigenvalue 0, and obtain the global low-dimensional coordinates Y = {y1, y2, ..., y n ∈R d }.

2. The method for testing soil moisture content using piezoelectric material according to claim 1, characterized in that: A 0.5 cm×5 cm rectangular PZT piezoelectric ceramic coated with a pure acrylic waterproof coating was used as the driving end piezoelectric sensor after being dried for 24 hours; a 1.63 cm×4.14 cm rectangular PVDF piezoelectric film sensor LDT1-028K was used as the receiving end sensor; and the test soil sample was screened with a sieve with an aperture of 0.45 mm.

3. The method for testing soil moisture content using piezoelectric material according to claim 1, characterized in that: Continuous complex wavelet transform introduces complex wavelet basis functions on the basis of continuous wavelet transform, so as to more comprehensively capture the time-frequency characteristics of the signal from the two dimensions of time and frequency. The expression of continuous wavelet transform is: Where: W x (a, b) are wavelet coefficients, which represent the intensity of the frequency component of the signal at a specific scale and position); a is the scale parameter to control the expansion and contraction of the wavelet; b is the translation parameter to control the position of the wavelet on the time axis; ψ a,b is the wavelet basis function; ψ * (·) is the complex conjugate of ψ(·); if continuous complex wavelet transform is used, the wavelet basis function is transformed into an adjustable complex Morlet wavelet basis function: Where: F b is the wavelet bandwidth parameter; F c is the center frequency of the wavelet; in the specific analysis process, by setting the parameter F b and F c The value of is used to select the target analysis frequency band.

4. The method for testing soil moisture content using piezoelectric material according to claim 1, characterized in that: Before distinguishing between signals and noise, in view of the fact that the frequency distribution of the signal cannot be observed in the time domain, the signal in the time domain is converted into a signal in the frequency domain through Fourier transform. The Fourier transform formula is as follows: In the formula, w is the frequency, t is the time, e -iwt is a complex function.

5. The method for testing soil moisture content using piezoelectric material according to claim 1, characterized in that: Loggernet data management software was used to monitor the piezoelectric signal data displayed on the computer in real time. When the data tended to be stable, the measurement was continued for 30 minutes.

6. The method for testing soil moisture content using piezoelectric material according to claim 1, characterized in that: The specific steps of wavelet decomposition include: (1) Select cmor3-3 wavelet basis and 3 decomposition levels for wavelet decomposition; (2) The wavelet coefficients are denoised by setting the threshold thr and thr=560 to obtain the denoised wavelet coefficients; (3) Reconstruct the signal using the coefficients after threshold processing to complete denoising, as follows: The wavelet ridge band is extracted by using the contour method, the amplitude c of the contour line is calculated and determined, and then the upper and lower edges of the wavelet ridge band are extracted, and the time-frequency point at the centroid of the ridge band is identified; A0=(m+c) / 2(5) Among them, A0 is the amplitude at the isosurface; Where m is the maximum value of the wavelet coefficient modulus, m divides the signal into equal intervals, and the contour line method is used to draw the contour line to obtain the wavelet ridge band, and the time-frequency point at the centroid of the wavelet ridge band is saved.

Images