Capacitive cotton seed monitoring signal noise reduction method and system based on CEEMDAN

Abstract

The invention discloses a noise reduction method and system for capacitive cotton seed monitoring signals, and relates to the technical field of signal noise reduction processing. Comprising the following steps: performing signal decomposition on a capacitive cotton seed monitoring signal sequence by using a CEEMDAN method to obtain IMF components with different frequencies, calculating a correlation coefficient R corresponding to each component, and designing a correlation coefficient threshold formula for distinguishing noisy IMF components needing to be further filtered and directly reserved effective IMF components; performing wavelet threshold denoising on the noisy IMF component to obtain a filtered IMF component; and superposing the effective IMF component and the denoised IMF component, and reconstructing to obtain a denoised signal. The method and the system can be applied to the fluted disc type dibbler, and the monitoring precision of the cotton seed sowing condition can be effectively improved.

Description

Technical Field

[0001] The invention belongs to the field of capacitive sensor signal processing, and in particular provides a noise reduction method and system for a capacitive seed taking state detection signal of a cotton toothed disc seed drill. Background Art

[0002] The mainstream mode of mechanized cotton planting in Xinjiang is based on a toothed disc seed drill, which adopts a planting method of spot seeding on the membrane. At present, the use of the capacitance method to achieve accurate detection of crop sowing amount has made breakthroughs in stability and resistance to high dust pollution in the field. When using the capacitance method for detection, the working site environment is complex, and the raw data obtained is often mixed with useless signals, such as baseline drift caused by impedance changes between cotton seeds and sensors, power frequency interference caused by capacitance and inductive coupling of power lines during signal acquisition, etc., and cotton seeds are small and medium-sized seeds, and the capacitance changes produced are small, which are difficult to identify and process. If the noise is not removed in time, it will affect the determination of the number of cotton seeds. Therefore, a cotton seed drill seeding state detection sensor is developed to detect abnormal seeding conditions in time. Summary of the invention

[0003] In order to solve the above problems, the present invention provides a method and system for reducing noise of the capacitive seeding state detection signal of a cotton toothed disc seed drill. The present invention obtains the shape characteristics of the signal by analyzing the installation method and working principle of the capacitive sensor, constructs an ideal simulation signal for such characteristics, performs CEEMDAN-wavelet threshold joint denoising on the constructed simulation signal, and analyzes the noise reduction effect through four parameters: signal-to-noise ratio (SNR), root mean square error (RMSE), curve similarity parameter (NCC), and smoothness index (SR).

[0004] In order to achieve the above technical effects, the technical solution of the present invention is specifically as follows:

[0005] A capacitive cotton seed monitoring signal noise reduction method based on CEEMDAN, comprising:

[0006] S1 decomposes the original signal by CEEMDAN to obtain n intrinsic mode functions (IMFs), which are named IMF1, IMF2, ···, IMFn, ···, where n=1, 2, ···, N;

[0007] S2 designs a correlation coefficient threshold formula, determines the demarcation point of the IMF component according to the correlation coefficient value of the IMF component, and identifies the useful component and the noise component.

[0008] S3 retains the useful signal and performs wavelet denoising on the noise component.

[0009] S4 combines the denoised noise IMF component with the IMF component dominated by the useful signal to reconstruct the denoised signal.

[0010] The method for reducing noise of capacitive cotton seed monitoring signals of CEEMDAN according to claim 1 is characterized in that S1 is specifically as follows:

[0011] S11 first performs EMD decomposition on the signal and decomposes it into the sum of several regular signals, namely:

[0012]

[0013] Where is the intrinsic mode function. C i (t) must meet two conditions: one is that the signal is locally symmetrical about the time axis, and the other is that the number of zeros and poles is equal or differs by at most 1. r1(t)=x(t)-C1(t). If r1(t) does not meet the following conditions: one is that r1(t) is a monotonic signal, and the other is that r1(t) is less than a predetermined value, then r1(t) is regarded as a new x(t) and decomposed to obtain new C2(t) and r2(t), and r1(t)=C2(t)+r2(t) until a satisfying r is obtained. n (t), then:

[0014]

[0015] S12 adds normally distributed Gaussian white noise. The specific steps are as follows: Determine the number N of added noise to obtain N new original signals:

[0016] The EMD algorithm is used to obtain the first-order residual component for N initial quantities, and the first-order average residual component is calculated by averaging:

[0017]

[0018] By the formula: The first-order average IMF component can be obtained, and then Repeat the above steps to obtain the second order , the third-order average IMF component ,……,until Until it can no longer be decomposed, the signal after CEEMDAN decomposition is obtained.

[0019] The method for reducing noise of capacitive cotton seed monitoring signals of CEEMDAN according to claim 1 is characterized in that S2 is specifically as follows:

[0020] After the S21 signal is decomposed, the IMF component with the largest upward jump in the correlation coefficient value of adjacent IMF components and the components after it are the components with the characteristics of the original signal. Based on this, a threshold formula for the correlation coefficient is designed to determine the noise-dominated component and the useful signal-dominated component:

[0021]

[0022] where y i is the correlation coefficient between the i-th order decomposition IMF component and the original signal, λ is the difference between the adjacent correlation coefficient values, and N is the total decomposition order. The i when λ takes the maximum value is recorded as , This is the threshold value of the correlation coefficient.

[0023] S22 calculates n intrinsic mode functions IMF1, IMF2, ···, IMF n The correlation coefficients y1, y2, ···, y n ,According to the above formula, the judgment method is obtained:

[0024] The method for reducing noise of capacitive cotton seed monitoring signals of CEEMDAN according to claim 1 is characterized in that S3 is specifically as follows:

[0025] S31 is greater than or equal to The useful components are denoted as imf1, imf2⋯imf m , less than The noise component is denoted as imf m+1 、imf m+2 ⋯IMF n . Keep the useful signal component and perform wavelet threshold denoising on the noise component

[0026] S32 selects soft threshold denoising and uses sym4 wavelet as the wavelet basis. The formula of soft threshold method is:

[0027]

[0028] Where:

[0029] —wavelet coefficient; λ—the set threshold.

[0030] Assuming the signal to be denoised is S(n), the denoising process steps are as follows: (1) Select a suitable wavelet function and then perform N-layer wavelet decomposition on S(n); (2) Select a suitable threshold and use the threshold value function to quantize the coefficients of each layer; (3) Based on the low-frequency coefficients of the Nth layer and the high-frequency coefficients of each layer of wavelet decomposition, realize wavelet reconstruction of the signal.

[0031] S5 A capacitive cotton seed monitoring signal noise reduction system based on CEEMDAN, the system comprising:

[0032] The signal acquisition module composed of the interdigital capacitance sensor and the micro capacitance detection circuit is used to obtain the signal. The lower computer is used to transmit and process the signal to obtain the capacitance signal after noise reduction. The upper computer is used to display and store data. The micro capacitance detection chip uses the PCAP04 chip and the single chip microcontroller uses the STM32F103C8T6. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 The present invention provides a flow chart of the capacitive cotton seed monitoring signal noise reduction method.

[0034] Figure 2 A diagram showing the relationship between the noise-reduced signal and the original signal provided in an embodiment of the present invention.

[0035] Figure 3 This is a diagram showing the change in capacitance values ​​for different cotton species provided by an embodiment of the present invention.

[0036] Figure 4 This is a structural diagram of a capacitance detection system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0037] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0038] Example 1

[0039] This embodiment provides a method for reducing noise of laser gyro signals based on CEEMDAN. Figure 1 As shown, the following steps are included:

[0040] S1 decomposes the original signal by CEEMDAN to obtain n intrinsic mode functions (IMFs), which are named IMF1, IMF2, ···, IMF n ,···,where n=1,2,···,N;

[0041] S2 designs a correlation coefficient threshold formula, determines the demarcation point of the IMF component according to the correlation coefficient value of the IMF component, and identifies the useful component and the noise component.

[0042] S3 retains the useful signal and performs wavelet denoising on the noise component.

[0043] S4 combines the denoised noise IMF component with the IMF component dominated by the useful signal to reconstruct the denoised signal.

[0044] On the whole, multiple intrinsic modal components are first decomposed by CEEMDAN, and the modal components containing noise are screened out using the correlation coefficient. Then, the modal components containing noise are processed using the wavelet threshold denoising method. Finally, the unprocessed modal components and the denoised components are reconstructed to obtain the final pure cotton seed monitoring signal.

[0045] Example 2

[0046] In this embodiment, the noise reduction method of the capacitance signal based on CEEMDAN proposed in Embodiment 1 is further described. First, the signal to be processed is decomposed by CEEMDAN to obtain signal components and reconstruct the process as follows:

[0047] S11 uses a filtering noise reduction algorithm to perform noise reduction on the measured missed broadcast and rebroadcast signals. The noise reduction algorithm used in this embodiment is CEEMDAN-wavelet threshold. Among them, the original signal is the measured signal obtained using the built capacitor seeding detection platform. First, the signal is decomposed by EMD into the sum of several regular signals, namely:

[0048]

[0049] S12 adds normally distributed Gaussian white noise. The specific steps are as follows: Determine the number N of added noise to obtain N new original signals:

[0050] The EMD algorithm is used to obtain the first-order residual component for N initial quantities, and the first-order average residual component is calculated by averaging:

[0051]

[0052] By the formula: The first-order average IMF component can be obtained, and then Repeat the above steps to obtain the second order , the third-order average IMF component ,……,until Until it can no longer be decomposed, the signal after CEEMDAN decomposition is obtained.

[0053] The CEEMDAN algorithm parameters are as follows: the noise weight Nstd is 0.4, the number of noise additions NR is 7, the maximum number of decomposed components MaxIter is 500, and 8 IMF components are obtained.

[0054] In this embodiment, the correlation coefficient threshold formula described in step S2 is designed, and the process of identifying the useful component and the noise component is:

[0055] After the S21 signal is decomposed, the IMF component with the largest upward jump in the correlation coefficient value of adjacent IMF components and the components after it are the components with the characteristics of the original signal. Based on this, a threshold formula for the correlation coefficient is designed to determine the noise-dominated component and the useful signal-dominated component:

[0056]

[0057] Where yi is the correlation coefficient between the i-th order decomposition IMF component and the original signal, λ is the difference between the adjacent correlation coefficient values, and N is the total decomposition order. The value i when λ takes the maximum value is denoted as , which is the threshold value of the correlation coefficient.

[0058] Calculate n intrinsic mode functions IMF1, IMF2, ···, IMF n The correlation coefficients y1, y2, ···, y n ,According to the above formula, the judgment method is obtained:

[0059]

[0060] According to the threshold formula , determine the noise-dominated component and the useful signal-dominated component. i is the correlation coefficient between the i-th order decomposition IMF component and the original signal, λ is the difference between the adjacent correlation coefficient values, and N is the total decomposition order. The value i when λ takes the maximum value is recorded as , which is the threshold value of the correlation coefficient, which is greater than or equal to The useful amount is less than is the noise component. In this example, the formula can be used to calculate .

[0061] In this embodiment, the process of performing wavelet denoising on the noise component in step S3 is as follows:

[0062] S31 is greater than or equal to The useful components are denoted as imf1, imf2⋯imf m , less than The noise component is denoted as imf m+1 、imf m+2 ⋯IMF n . Useful signal components are retained, and wavelet threshold denoising is performed on the noise components. Therefore, the decomposed IMF1~IMF4 components are subjected to wavelet threshold denoising. When the IMF1~IMF4 components are subjected to wavelet threshold denoising, the sym4 wavelet is used as the wavelet basis, the threshold function selects the soft threshold function, and the decomposition layer number N is 5 layers. The denoised IMF1~IMF4 components are added to IMF5~IMF8 for reconstruction as the denoised signal.

[0063] S32 takes the capacitance output wave peak value under different seeding conditions from the noise-reduced signal to obtain the response curve between cotton seed quantity and capacitance value. The results are as follows: Figure 3 shown.

[0064] Based on this, a capacitive cotton seed monitoring signal noise reduction method based on CEEMDAN provided by the present invention has scientific design and good effect. The actual signal noise reduction effect shows that the processed capacitance value can be obviously stratified according to the number of cotton seeds, which is convenient for subsequent threshold division and realizes real-time monitoring of seeding quality.

[0065] Example 3

[0066] In this embodiment, a capacitive cotton seed monitoring signal noise reduction system for a toothed disc cotton seed drill CEEMDAN is provided, the system comprising: a toothed disc cotton seed drill test model machine, a forked finger capacitance sensor, a micro capacitance detection circuit, a host computer, and a slave computer.

[0067] The toothed disc cotton seed drill test model is used as the research carrier to verify the method proposed in this paper, in which sensors and their signal acquisition and processing equipment are installed.

[0068] The signal acquisition module 301 composed of the interdigital capacitance sensor and the micro capacitance detection circuit is used to obtain the capacitance signal generated when the cotton seeds pass through the detection area and convert it into a digital signal.

[0069] The lower computer in the signal transmission and processing module 302 is a single chip microcomputer, which is used to receive and transmit the collected digital signals and perform noise reduction processing on the digital signals.

[0070] The host computer in the data display module 303 is a computer, which visualizes the data and stores it in an Excel table.

[0071] The tiny capacitance detection circuit includes the PCAP04 chip and its peripheral circuits, the RT9166 linear regulator, etc.

[0072] The single chip microcomputer is STM32F103C8T6, which communicates with Pcap04 through SPI protocol. After the capacitance value is measured, the display module is controlled through GPIO to display the measurement result.

[0073] In addition, when the noise reduction processing program in the above-mentioned single-chip microcomputer is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention.

[0074] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0075] This article uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only used to help understand the method and core ideas of the present invention. At the same time, for those skilled in the art, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for reducing noise of capacitive cotton seed monitoring signals based on CEEMDAN is implemented in the following steps: S1: Decompose the original signal by CEEMDAN to obtain n intrinsic mode functions (IMFs), which are named IMF1, IMF2, ···, IMF n ,···,where n=1,2,···,N; S2: Design a correlation coefficient threshold formula, determine the cutoff point of the IMF component according to the correlation coefficient value of each IMF component, and distinguish between the useful component and the noise component; S3: retain the useful signal and perform wavelet denoising on the noise component; S4: The denoised noise IMF component is combined with the IMF component dominated by the useful signal to reconstruct the denoised signal.

2. The method for reducing noise of a capacitive cotton seed monitoring signal of CEEMDAN according to claim 1, characterized in that: The S1 is specifically as follows: S11 performs EMD decomposition on the signal and decomposes it into the sum of several regular signals; S12 adds normally distributed Gaussian white noise. The specific steps are as follows: First, determine the number of noises to be added, N, and get N new original signals x N (t); Secondly, the EMD algorithm is used to obtain the first-order residual component for N initial quantities, and the first-order average residual component is calculated by averaging. . Find the first-order average IMF component , and then , repeat the above steps to obtain the second order , the third-order average IMF component ,……,until Until the signal cannot be decomposed any further, the signal has achieved CEEMDAN decomposition.

3. The method for reducing noise of a capacitive cotton seed monitoring signal of CEEMDAN according to claim 1, characterized in that: The S2 is specifically as follows: After the S21 signal is decomposed, the IMF component with the largest upward jump in the correlation coefficient value of adjacent IMF components and the components after it are the components with the characteristics of the original signal. Based on this, a threshold formula for the correlation coefficient is designed to determine the noise-dominated component and the useful signal-dominated component:

4. Where y i is the correlation coefficient between the i-th order decomposition IMF component and the original signal, λ is the difference between the adjacent correlation coefficient values, and N is the total decomposition order. The i when λ takes the maximum value is recorded as , That is the threshold of the correlation coefficient; S22 calculates n intrinsic mode functions IMF1, IMF2, ···, IMF n The correlation coefficients y1, y2, ···, y n ,According to formula (1), the judgment method is obtained: 。 5. The method for reducing noise of a capacitive cotton seed monitoring signal of CEEMDAN according to claim 1, characterized in that: The S3 is as follows: The useful components of S31 greater than or equal to are denoted as imf1, imf2⋯imf m , the noise component less than is recorded as imf m+1 、imf m+2 ⋯IMF n . Useful signal components are retained, and the noise components are denoised by wavelet threshold; S32 selects sym4 wavelet as the wavelet basis and soft threshold denoising.

6. A capacitive cotton seed monitoring signal noise reduction system based on CEEMDAN, characterized in that: include: The toothed disc cotton seed drill test model is used as a research carrier to verify the method proposed in this paper; the signal acquisition module composed of the interdigital capacitance sensor and the micro capacitance detection circuit is used to obtain the signal; the lower computer is used to transmit and process the signal to obtain the capacitance signal after noise reduction; the upper computer is used to display and store data.

Images