Method and system for detecting fungaltoxin in feed

Abstract

The invention relates to a method and a system for detecting fungaltoxin in feed. The method comprises the following steps: acquiring a data sequence which is measured by a detector of HPLC (High Performance Liquid Chromatography) equipment and contains the content of fungaltoxin in the feed; the method comprises the following steps: acquiring an appearance time range when each mycotoxin is subjected to HPLC (High Performance Liquid Chromatography), determining an average value of data sequence fragments not in the time range, determining the number of EMD decomposition according to permutation entropy, determining a variance of white noise added during each EMD decomposition based on the average value, adding the white noise to the data sequence, and then performing EMD decomposition to obtain a corresponding EMD decomposition value; removing abnormal values of all first modal components obtained by EMD decomposition to obtain a first modal component mean value; obtaining noise according to the first modal component mean value, subtracting the noise from the data sequence to obtain a denoised data sequence, calculating the permutation entropy of a denoised data sequence fragment, and if the reduction of the permutation entropy is smaller than a preset value, repeating the above process on the denoised data sequence to obtain a denoised data sequence; otherwise, the content of the fungaltoxin is obtained based on the denoised data sequence.

Description

Technical Field

[0001] The present invention relates to the field of feed, and in particular to a method and system for detecting mycotoxins in feed. Background Art

[0002] With the development of modern animal husbandry, the quality and safety of feed have received more and more attention. Mycotoxins, as a potential feed contaminant, have a particularly significant impact on the health of livestock and poultry. Mycotoxins include aflatoxin, vomitoxin, zearalenone and other toxic substances. Long-term intake of feed containing mycotoxins may cause livestock and poultry poisoning, and even death in severe cases, which will cause losses to farmers.

[0003] High performance liquid chromatography (HPLC) is one of the most widely used methods for mycotoxin detection, with the characteristics of high sensitivity and high resolution. However, HPLC analysis is inevitably affected by noise, which may come from instruments, environmental factors or the sample itself, resulting in more interference in the chromatogram signal, which in turn affects the quantitative analysis results of mycotoxins. In order to improve the accuracy of detection, it is usually necessary to perform noise reduction on the chromatogram obtained by HPLC. Currently, common noise reduction methods include filtering and denoising algorithms, but these traditional methods often find it difficult to balance noise reduction effects and signal fidelity in practical applications, and there are problems such as information loss or signal distortion. Therefore, there is an urgent need for an efficient chromatogram noise reduction method to further improve the detection accuracy while maintaining the sensitivity of mycotoxin detection and ensure the reliability of feed safety detection. Summary of the invention

[0004] In order to solve the above problems, the present invention provides a method for detecting mycotoxins in feed, which comprises the following steps: Acquiring a data sequence containing the content of mycotoxins in feed measured by a detector of an HPLC device; Obtaining the time range in which each fungal toxin appears when subjected to HPLC, determining the average value of data sequence fragments that are not within the time range, and determining the permutation entropy of the data sequence fragments, determining the number of EMD decompositions according to the permutation entropy, and determining the variance of white noise added during each EMD decomposition based on the average value, performing EMD decomposition after adding white noise to the data sequence, and obtaining the first modal component mean after removing outliers from all first modal components obtained by EMD decomposition; The noise is obtained according to the mean value of the first modal component, and the noise is subtracted from the data sequence to obtain a denoised data sequence. The permutation entropy of the denoised data sequence fragments is calculated. If the decrease in the permutation entropy is less than a preset value, the above process is repeated for the denoised data sequence. Otherwise, the content of fungal toxins is obtained based on the denoised data sequence.

[0005] Preferably, the determining of the permutation entropy of the data sequence fragments is specifically: Calculate the upper envelope and the lower envelope after smoothing the data sequence segments not within the time range; The average of the permutation entropy of the upper envelope and the permutation entropy of the lower envelope is taken as the permutation entropy of the data sequence segment.

[0006] Preferably, the number of EMD decompositions to be performed is determined according to the permutation entropy, specifically: Get the permutation entropy threshold and the upper and lower limits of the number of decompositions; If the permutation entropy is not greater than the permutation entropy threshold, the number of EMD decompositions is the lower limit; If the permutation entropy is greater than the permutation entropy threshold, the difference between the permutation entropy and the permutation entropy threshold is calculated, and a proportional factor is multiplied by the difference to obtain the increased number. The sum of the increased number and the lower limit is used as the number of EMD decompositions. If the calculated number of EMD decompositions is greater than the upper limit, the upper limit is taken as the number of EMD decompositions.

[0007] Preferably, the variance of adding white noise when determining each EMD decomposition based on the average value is specifically: Determine a plurality of coefficients according to the number of EMD decompositions performed, each EMD decomposition corresponding to a coefficient; The square of the product of the coefficient and the mean value is taken as the variance of white noise added during EMD decomposition.

[0008] Preferably, the first modal component mean is obtained by removing outliers from the first modal components obtained by decomposing all EMDs, specifically: Obtain a feature matrix according to the first modal components obtained by all EMD classifications, wherein each row of the feature matrix is ​​a first modal component obtained by EMD decomposition; The feature matrix is ​​smoothed by using Gaussian filtering or mean filtering; A column average operation is performed on the smoothed feature matrix to obtain a mean value of the first modal component after removing outliers.

[0009] In another aspect, the present invention provides a system for detecting mycotoxins in feed, the system comprising the following modules: A data acquisition module for acquiring a data sequence containing the mycotoxin content in the feed measured by a detector of the HPLC device; A decomposition module, used for obtaining the time range in which each fungal toxin appears when performing HPLC, determining the average value of data sequence fragments not within the time range, and determining the permutation entropy of the data sequence fragments, determining the number of EMD decompositions according to the permutation entropy, and determining the variance of white noise added during each EMD decomposition based on the average value, performing EMD decomposition after adding white noise to the data sequence, and obtaining the first modal component mean after removing outliers from all first modal components obtained by EMD decomposition; The content determination module is used to obtain noise according to the mean value of the first modal component, subtract the noise from the data sequence to obtain a denoised data sequence, calculate the permutation entropy of the denoised data sequence fragments, and repeat the above process for the denoised data sequence if the decrease in the permutation entropy is less than a preset value, otherwise obtain the content of fungal toxins based on the denoised data sequence.

[0010] Preferably, the determining of the permutation entropy of the data sequence fragments is specifically: Calculate the upper envelope and the lower envelope after smoothing the data sequence segments not within the time range; The average of the permutation entropy of the upper envelope and the permutation entropy of the lower envelope is taken as the permutation entropy of the data sequence segment.

[0011] Preferably, the number of EMD decompositions to be performed is determined according to the permutation entropy, specifically: Get the permutation entropy threshold and the upper and lower limits of the number of decompositions; If the permutation entropy is not greater than the permutation entropy threshold, the number of EMD decompositions is the lower limit; If the permutation entropy is greater than the permutation entropy threshold, the difference between the permutation entropy and the permutation entropy threshold is calculated, and a proportional factor is multiplied by the difference to obtain the increased number. The sum of the increased number and the lower limit is used as the number of EMD decompositions. If the calculated number of EMD decompositions is greater than the upper limit, the upper limit is taken as the number of EMD decompositions.

[0012] Preferably, the variance of adding white noise when determining each EMD decomposition based on the average value is specifically: Determine a plurality of coefficients according to the number of EMD decompositions performed, each EMD decomposition corresponding to a coefficient; The square of the product of the coefficient and the mean value is taken as the variance of white noise added during EMD decomposition.

[0013] Preferably, the first modal component mean is obtained by removing outliers from the first modal components obtained by decomposing all EMDs, specifically: Obtain a feature matrix according to the first modal components obtained by all EMD classifications, wherein each row of the feature matrix is ​​a first modal component obtained by EMD decomposition; The feature matrix is ​​smoothed by using Gaussian filtering or mean filtering; A column average operation is performed on the smoothed feature matrix to obtain a mean value of the first modal component after removing outliers.

[0014] In a final aspect, the present invention further provides a computer-readable storage medium, on which a computer program is stored, and the computer program implements the above-mentioned method when executed by a processor.

[0015] The present invention determines the number of EMD decompositions and the variance of added white noise based on the average value and permutation entropy of data sequence fragments. If the signal is more complex or has more noise, more EMD decompositions are performed to ensure that the result after adding different noises multiple times and averaging is more accurate, and can ensure that the variance of the white noise matches the actual noise level of the signal, thereby reducing the aliasing phenomenon between modes during the decomposition process. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a flow chart of Embodiment 1; Figure 2 It is the time range of appearance of different mycotoxins in HPLC; Figure 3 Schematic diagram of EMD decomposition; Figure 4 Schematic diagram for noise determination. DETAILED DESCRIPTION

[0017] In this article, relational terms such as first and second, etc. are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0018] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only 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.

[0019] In the first embodiment of the present invention, the present invention provides a method for detecting mycotoxins in feed, such as Figure 1 As shown, the method comprises the following steps: S1, obtaining a data sequence containing the mycotoxin content in the feed measured by a detector of an HPLC device; High-Performance Liquid Chromatography (HPLC) can quantitatively calculate the content of mycotoxins. In HPLC, the sample solution is pumped into the liquid chromatography column by a high-pressure pump. Different compounds in the sample are separated according to their retention time (RT) in the chromatographic column, and their response signals are finally measured by the detector. Among them, mycotoxins include aflatoxins, fumonisins, etc. The feed is ground to make a sample solution, which is injected into the HPLC system, and the sample enters the chromatographic column with the mobile phase. In the chromatographic column, different compounds are separated according to their properties and the interaction force with the column packing. Each compound will reach the detector at a different retention time. For example, aflatoxin may be detected at 5 minutes, while fumonisins may be detected at 7 minutes. When the compounds in the sample pass through the detector, the detector (such as a UV detector, a fluorescence detector, or a mass spectrometer) will generate a response signal, the size of which is proportional to the concentration of the compound. Over time, the signals recorded by the detector will form a data sequence. The data in the data sequence represents the response intensity of the detector over time, that is, the concentration of different mycotoxins.

[0020] S2, obtaining the time range in which each fungal toxin appears when subjected to HPLC, determining the average value of the data sequence fragments that are not within the time range, and determining the permutation entropy of the data sequence fragments, determining the number of EMD decompositions according to the permutation entropy, and determining the variance of white noise added during each EMD decomposition based on the average value, performing EMD decomposition after adding white noise to the data sequence, and obtaining the first modal component mean after removing outliers from all first modal components obtained by EMD decomposition; In HPLC, different mycotoxins appear at different retention times, which represent the time period in which the mycotoxins appear in HPLC detection, that is, the time interval in which they produce signal peaks in the data sequence. For example, aflatoxin appears at 3-4 minutes. Each mycotoxin has another time range of appearance, such as Figure 2 As shown in the figure, the data in the non-occurring time range in the data sequence is mainly noise, and data segments are extracted from these non-said time ranges, and their average values ​​are calculated as the baseline of the background noise.

[0021] Permutation entropy is a method to measure the complexity of time series. The size of permutation entropy reflects the randomness and complexity of these noises. Since the data of the data sequence fragment is noise, it has great randomness. This randomness leads to little difference in the calculated permutation entropy. In order to highlight the difference of the data sequence fragments, in an optional embodiment, the permutation entropy of the data sequence fragment is determined as follows: Calculate the upper envelope and the lower envelope after smoothing the data sequence segments not within the time range; The average of the permutation entropy of the upper envelope and the permutation entropy of the lower envelope is taken as the permutation entropy of the data sequence segment.

[0022] The data sequence within the non-retention time range usually contains more noise and high-frequency components, resulting in greater signal randomness. In order to reduce the impact of this randomness on the calculation of permutation entropy, we first smooth the data sequence fragments. The smoothing methods shown include but are not limited to moving average, Gaussian filtering, etc. The upper envelope represents the curve composed of the local maximum value of the signal, and the lower envelope represents the curve composed of the local minimum value of the signal. Through these envelopes, more important trend information in the signal is obtained instead of directly processing the noise in the original data. The upper and lower envelopes are determined by detecting the local maximum and minimum values, and then these points are connected by spline interpolation to form the upper and lower envelopes.

[0023] The permutation entropy is calculated for the calculated upper and lower envelopes respectively. The specific calculation method of the permutation entropy includes segmenting and sorting the data, counting the frequency of each subsequence pattern, and then calculating the entropy value. In order to obtain the final permutation entropy of the entire data sequence segment, we take the average of the upper envelope permutation entropy and the lower envelope permutation entropy as the permutation entropy of the entire segment, and eliminate the influence of local random fluctuations in the original noise signal on the permutation entropy calculation, so as to more effectively distinguish the complexity of different data segments.

[0024] When the permutation entropy is high, it means that the noise is more complex and more modes are needed for decomposition. In one embodiment, the number of EMD decompositions is determined according to the permutation entropy, specifically: Get the permutation entropy threshold and the upper and lower limits of the number of decompositions; If the permutation entropy is not greater than the permutation entropy threshold, the number of EMD decompositions is the lower limit; If the permutation entropy is greater than the permutation entropy threshold, the difference between the permutation entropy and the permutation entropy threshold is calculated, and a proportional factor is multiplied by the difference to obtain the increased number. The sum of the increased number and the lower limit is used as the number of EMD decompositions. If the calculated number of EMD decompositions is greater than the upper limit, the upper limit is taken as the number of EMD decompositions.

[0025] If the permutation entropy of the signal is less than or equal to the set threshold, it means that the complexity of the signal is low, the noise is not obvious, or the signal is relatively simple. The number of EMD (empirical mode decomposition) decompositions is directly set as the lower limit; when the permutation entropy of the signal is greater than the threshold, it means that the signal is more complex and may contain more noise or irregular components. More EMD decompositions are used. Assuming the permutation entropy threshold is 0.7, the lower limit of the decomposition number is 3, and the upper limit of the decomposition number is 8. If the permutation entropy is 0.9 and the scaling factor is 10, the number of EMD decompositions is 5. If it is not an integer, it is rounded down or rounded up. Adding white noise to the data sequence and performing EMD decomposition is an EMD decomposition. The number of EMD decompositions refers to the number of times the above process is performed and the number of repeated EMD operations, that is, the number of EMD decompositions after adding white noise, such as Figure 3 shown.

[0026] Before each EMD, white noise is added to the data sequence. The intensity (i.e., variance) of the white noise will interfere with the original data. If it is too small, the modal aliasing problem cannot be solved. In one embodiment, the variance of the white noise added to each EMD decomposition is determined based on the average value, specifically: Determine a plurality of coefficients according to the number of EMD decompositions performed, each EMD decomposition corresponding to a coefficient; The square of the product of the coefficient and the mean value is taken as the variance of white noise added during EMD decomposition.

[0027] In this embodiment, a coefficient is set for each EMD decomposition, so that the white noise of different EMD decompositions is different. The modal aliasing problem is reduced by white noise with different variances. If the average value of the data sequence is large, it means that the overall strength of the signal is large. Accordingly, the variance of the added white noise should also be increased appropriately to ensure that the interference of noise on the signal is moderate; if the average value of the data sequence is small, it means that the signal is weak, and the variance of the white noise should also be small to prevent the noise from excessively interfering with the signal.

[0028] Suppose that N times EMD is decomposed, so each EMD decomposition will correspond to a coefficient, and these coefficients are the parameters that add white noise intensity when being used to control different EMD decompositions.By distributing different coefficients for each EMD decomposition, the variance of the white noise added when each EMD decomposition can be dynamically adjusted.When calculating variance, calculate the product of coefficient and mean value first, then square operation is performed to the product, so that the variance of adding white noise when different EMD decompositions is obtained.In an optional embodiment, the coefficient is the numerical value between 0.2-0.5.

[0029] Before EMD decomposition, adding white noise to the signal can help better separate the different frequency components in the signal. White noise is a kind of random noise that can help the EMD algorithm to more robustly separate the modal functions of different frequencies in the signal. The data after adding noise is subjected to EMD decomposition to obtain multiple intrinsic mode functions (IMFs), which represent the different frequency components of the signal. Usually, the first modal function (IMF1) obtained by decomposition contains the high-frequency components and noise in the signal. The first modal component usually contains the highest frequency part of the signal, and may also be mixed with more noise components. Outlier processing is to remove some overly discrete or extreme values ​​in this modal component. Outliers can be identified and removed by a variety of methods, such as using the standard deviation method or the interquartile range method, to ensure that extreme values ​​do not affect subsequent analysis. The specific method can be to remove values ​​outside the range of three times the standard deviation of the mean, or to remove data points with large deviations according to the threshold set according to business needs. Of course, clustering and other methods can also be used to find the abnormal first modal component and delete the abnormal first modal component.

[0030] In another specific embodiment, the first modal component mean is obtained by removing outliers from the first modal components obtained by decomposing all EMDs, specifically: Obtain a feature matrix according to the first modal components obtained by all EMD classifications, wherein each row of the feature matrix is ​​a first modal component obtained by EMD decomposition; The feature matrix is ​​smoothed by using Gaussian filtering or mean filtering; A column average operation is performed on the smoothed feature matrix to obtain a mean value of the first modal component after removing outliers.

[0031] S3, obtaining noise according to the mean value of the first modal component, subtracting the noise from the data sequence to obtain a denoised data sequence, calculating the permutation entropy of the denoised data sequence fragments, and repeating the above process for the denoised data sequence if the decrease in the permutation entropy is less than a preset value, otherwise obtaining the content of fungal toxins based on the denoised data sequence.

[0032] Since the noise is mainly concentrated in the first modal component, in one embodiment, the noise is obtained according to the first modal component mean value by using the first modal mean value as the noise. In another embodiment, the noise is obtained according to the first modal component mean value by using Gaussian filtering or high-pass filtering, for example Figure 4 shown.

[0033] The noise part in the original data sequence is removed to obtain the denoised data sequence, that is, the noise of the first modal component mean is subtracted from the original data sequence. After the signal is denoised, the permutation entropy of the denoised data segment is calculated to measure the complexity of the signal. By calculating the permutation entropy of the denoised data segment, it can be judged whether the noise of the signal has been effectively reduced. If the reduction in the permutation entropy of the denoised signal is not obvious, continue the denoising process, repeat the previous steps, and continue to repeat S2 to obtain the first modal component mean, and perform the process of obtaining the noise and denoising of the first modal component mean. Once the permutation entropy of the denoised signal is reduced to meet the requirements, it means that the denoising effect is sufficient. At this time, the denoised signal can be used for quantitative analysis of fungal toxins. The denoised signal can more accurately reflect the true content of fungal toxins and avoid misjudgment caused by noise interference.

[0034] In a second embodiment of the present invention, the present invention provides a system for detecting mycotoxins in feed based on HPLC, the system comprising the following modules: A data acquisition module for acquiring a data sequence containing the mycotoxin content in the feed measured by a detector of the HPLC device; A decomposition module, used for obtaining the time range in which each fungal toxin appears when performing HPLC, determining the average value of data sequence fragments not within the time range, and determining the permutation entropy of the data sequence fragments, determining the number of EMD decompositions according to the permutation entropy, and determining the variance of white noise added during each EMD decomposition based on the average value, performing EMD decomposition after adding white noise to the data sequence, and obtaining the first modal component mean after removing outliers from all first modal components obtained by EMD decomposition; The content determination module is used to obtain noise according to the mean value of the first modal component, subtract the noise from the data sequence to obtain a denoised data sequence, calculate the permutation entropy of the denoised data sequence fragments, and repeat the above process for the denoised data sequence if the decrease in the permutation entropy is less than a preset value, otherwise obtain the content of fungal toxins based on the denoised data sequence.

[0035] Preferably, the determining of the permutation entropy of the data sequence fragments is specifically: Calculate the upper envelope and the lower envelope after smoothing the data sequence segments not within the time range; The average of the permutation entropy of the upper envelope and the permutation entropy of the lower envelope is taken as the permutation entropy of the data sequence segment.

[0036] Preferably, the number of EMD decompositions to be performed is determined according to the permutation entropy, specifically: Get the permutation entropy threshold and the upper and lower limits of the number of decompositions; If the permutation entropy is not greater than the permutation entropy threshold, the number of EMD decompositions is the lower limit; If the permutation entropy is greater than the permutation entropy threshold, the difference between the permutation entropy and the permutation entropy threshold is calculated, and a proportional factor is multiplied by the difference to obtain the increased number. The sum of the increased number and the lower limit is used as the number of EMD decompositions. If the calculated number of EMD decompositions is greater than the upper limit, the upper limit is taken as the number of EMD decompositions.

[0037] Preferably, the variance of adding white noise when determining each EMD decomposition based on the average value is specifically: Determine a plurality of coefficients according to the number of EMD decompositions performed, each EMD decomposition corresponding to a coefficient; The square of the product of the coefficient and the mean value is taken as the variance of white noise added during EMD decomposition.

[0038] Preferably, the first modal component mean is obtained by removing outliers from the first modal components obtained by decomposing all EMDs, specifically: Obtain a feature matrix according to the first modal components obtained by all EMD classifications, wherein each row of the feature matrix is ​​a first modal component obtained by EMD decomposition; The feature matrix is ​​smoothed by using Gaussian filtering or mean filtering; A column average operation is performed on the smoothed feature matrix to obtain a mean value of the first modal component after removing outliers.

[0039] In a third embodiment of the present invention, the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method described in the first embodiment is implemented.

[0040] In a fourth embodiment of the present invention, the present invention further provides a computer device, which includes at least a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the method described in the first embodiment is implemented.

[0041] Through the description of the above implementation methods, technicians in this field can clearly understand that each implementation method can be implemented by adding a necessary general hardware platform, and of course can also be implemented by combining hardware and software. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a computer product, and the present invention can be in the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0042] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it, and other embodiments may also be used. Although the present invention has been described in detail with reference to the aforementioned embodiments, ordinary technicians in this field should understand that they can still modify the technical solutions recorded in the aforementioned embodiments, or replace some of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting mycotoxins in feed, characterized in that: The method comprises the following steps: Acquiring a data sequence containing the content of mycotoxins in feed measured by a detector of an HPLC device; Obtaining the time range in which each fungal toxin appears when subjected to HPLC, determining the average value of data sequence fragments that are not within the time range, and determining the permutation entropy of the data sequence fragments, determining the number of EMD decompositions according to the permutation entropy, and determining the variance of white noise added during each EMD decomposition based on the average value, performing EMD decomposition after adding white noise to the data sequence, and obtaining the first modal component mean after removing outliers from all first modal components obtained by EMD decomposition; The noise is obtained according to the mean value of the first modal component, and the noise is subtracted from the data sequence to obtain a denoised data sequence. The permutation entropy of the denoised data sequence fragments is calculated. If the decrease in the permutation entropy is less than a preset value, the above process is repeated for the denoised data sequence. Otherwise, the content of fungal toxins is obtained based on the denoised data sequence.

2. The method according to claim 1, characterized in that The determination of the permutation entropy of the data sequence fragment is specifically: Calculate the upper envelope and the lower envelope after smoothing the data sequence segments not within the time range; The average of the permutation entropy of the upper envelope and the permutation entropy of the lower envelope is taken as the permutation entropy of the data sequence segment.

3. The method according to claim 1, characterized in that The number of EMD decompositions to be determined according to the permutation entropy is specifically: Get the permutation entropy threshold and the upper and lower limits of the number of decompositions; If the permutation entropy is not greater than the permutation entropy threshold, the number of EMD decompositions is the lower limit; If the permutation entropy is greater than the permutation entropy threshold, the difference between the permutation entropy and the permutation entropy threshold is calculated, and a proportional factor is multiplied by the difference to obtain the increased number. The sum of the increased number and the lower limit is used as the number of EMD decompositions. If the calculated number of EMD decompositions is greater than the upper limit, the upper limit is taken as the number of EMD decompositions.

4. The method according to claim 1, characterized in that The variance of adding white noise when each EMD decomposition is determined based on the average value is specifically: Determine a plurality of coefficients according to the number of EMD decompositions performed, each EMD decomposition corresponding to a coefficient; The square of the product of the coefficient and the mean value is taken as the variance of white noise added during EMD decomposition.

5. The method according to claim 1, characterized in that The first modal component mean is obtained by removing outliers from the first modal components obtained by decomposing all EMDs, specifically: Obtain a feature matrix according to the first modal components obtained by all EMD classifications, wherein each row of the feature matrix is ​​a first modal component obtained by EMD decomposition; The feature matrix is ​​smoothed by using Gaussian filtering or mean filtering; A column average operation is performed on the smoothed feature matrix to obtain a mean value of the first modal component after removing outliers.

6. A system for detecting mycotoxins in feed, characterized in that: The system includes the following modules: A data acquisition module for acquiring a data sequence containing the mycotoxin content in the feed measured by a detector of the HPLC device; A decomposition module, used for obtaining the time range in which each fungal toxin appears when performing HPLC, determining the average value of data sequence fragments not within the time range, and determining the permutation entropy of the data sequence fragments, determining the number of EMD decompositions according to the permutation entropy, and determining the variance of white noise added during each EMD decomposition based on the average value, performing EMD decomposition after adding white noise to the data sequence, and obtaining the first modal component mean after removing outliers from all first modal components obtained by EMD decomposition; The content determination module is used to obtain noise according to the mean value of the first modal component, subtract the noise from the data sequence to obtain a denoised data sequence, calculate the permutation entropy of the denoised data sequence fragments, and repeat the above process for the denoised data sequence if the decrease in the permutation entropy is less than a preset value, otherwise obtain the content of fungal toxins based on the denoised data sequence.

7. The system according to claim 6, characterized in that The determination of the permutation entropy of the data sequence fragment is specifically: Calculate the upper envelope and the lower envelope after smoothing the data sequence segments not within the time range; The average of the permutation entropy of the upper envelope and the permutation entropy of the lower envelope is taken as the permutation entropy of the data sequence segment.

8. The system according to claim 6, characterized in that The number of EMD decompositions to be determined according to the permutation entropy is specifically: Get the permutation entropy threshold and the upper and lower limits of the number of decompositions; If the permutation entropy is not greater than the permutation entropy threshold, the number of EMD decompositions is the lower limit; If the permutation entropy is greater than the permutation entropy threshold, the difference between the permutation entropy and the permutation entropy threshold is calculated, and a proportional factor is multiplied by the difference to obtain the increased number. The sum of the increased number and the lower limit is used as the number of EMD decompositions. If the calculated number of EMD decompositions is greater than the upper limit, the upper limit is taken as the number of EMD decompositions.

9. The system according to claim 6, characterized in that The variance of adding white noise when each EMD decomposition is determined based on the average value is specifically: Determine a plurality of coefficients according to the number of EMD decompositions performed, each EMD decomposition corresponding to a coefficient; The square of the product of the coefficient and the mean value is taken as the variance of white noise added during EMD decomposition.

10. A computer storage device, wherein a computer program is stored on the storage device, characterized in that: When the computer program is executed by a processor, the computer program implements the method according to any one of claims 1 to 5.

Images