An intelligent detection method and workstation for mycotoxins

Abstract

The present invention relates to the technical field of determining material properties for analysis, and specifically to an intelligent detection method and workstation for mycotoxins. The present invention analyzes the spectrum curves of the original chromatographic signal and the IMF component signal to obtain the noise possibility and abnormal interference factor of each frequency; according to the abnormal interference factor of each frequency, all frequencies are clustered to obtain frequency clustering clusters; according to the distribution characteristics of frequencies in different frequency clustering clusters, high noise component signals are screened out; according to the noise possibility distribution of different frequencies in the spectrum curve, the interference influence degree of the high noise component signal is obtained; combined with the degree of confusion of the difference between adjacent extreme values ​​in the high noise component signal, the frequency abnormality factor of the high noise component signal is obtained, and the abnormal cutoff frequency is obtained; the original chromatographic signal is filtered to obtain a filtered chromatographic signal. The present invention improves the accuracy of mycotoxin detection by obtaining accurate signal performance during mycotoxin detection.

Description

Technical Field

[0001] The present invention relates to the technical field of analyzing the properties of materials, and in particular to an intelligent detection method and workstation for mycotoxins. Background Art

[0002] Intelligent detection of mycotoxins refers to the use of advanced technologies and equipment, combined with intelligent analytical methods, to quickly, accurately and sensitively detect and monitor mycotoxins in food, feed or environmental samples. It has significant technical significance in improving detection efficiency, sensitivity and accuracy, real-time monitoring and early warning, cost-effectiveness, and scientific and technological progress.

[0003] In the prior art, high performance liquid chromatography is used to obtain a chromatogram through a designated instrument, and the presence of fungal toxins is detected by analyzing the chromatogram; however, due to the presence of noise and abnormal interference from other microorganisms, the accuracy of the signal data of the chromatogram obtained by liquid chromatography is not high, resulting in poor accuracy in the detection of fungal toxins. Summary of the invention

[0004] In order to solve the technical problem that the detection accuracy of mycotoxins is poor due to the presence of noise and abnormal interference from other microorganisms, the purpose of the present invention is to provide an intelligent detection method and workstation for mycotoxins. The technical solutions adopted are as follows:

[0005] The present invention proposes an intelligent detection method for mycotoxins, which comprises:

[0006] Acquire the original chromatographic signal and multiple IMF component signals of the fungal toxin, and obtain a corresponding spectrum curve, wherein the spectrum curve includes signal amplitudes at different frequencies;

[0007] For the original chromatographic signal, the noise possibility of each frequency is obtained according to the fluctuation trend of the signal amplitude corresponding to different frequencies in the spectrum curve;

[0008] For any IMF component signal, the abnormal interference factor of each frequency is obtained according to the signal amplitude, noise possibility and corresponding frequency value of each frequency in the spectrum curve; according to the abnormal interference factor of each frequency, all frequencies are clustered to obtain frequency clustering clusters, and the frequency clustering clusters include interference frequency clustering clusters and normal frequency clustering clusters; according to the distribution characteristics of frequencies in different frequency clustering clusters, high noise component signals are screened out;

[0009] For any high noise component signal, the interference influence degree of the high noise component signal is obtained according to the noise possibility distribution of different frequencies in the spectrum curve; the frequency anomaly factor of the high noise component signal is obtained according to the confusion degree of the difference between adjacent extreme values ​​in the high noise component signal and the interference influence degree; the abnormal cutoff frequency is obtained according to the frequency anomaly factors corresponding to all high noise component signals at different frequencies;

[0010] The original chromatographic signal is filtered based on the abnormal cutoff frequency to obtain a filtered chromatographic signal.

[0011] Furthermore, the method for obtaining the noise possibility includes:

[0012] For the original chromatographic signal, the second-order derivative of the spectrum curve at each frequency is obtained, and the frequency corresponding to the second-order derivative being 0 is used as the segmentation frequency to obtain the curve segments between adjacent segmentation frequencies;

[0013] The average value of the slope between each frequency on the spectrum curve and other frequencies on the corresponding curve segment is obtained as the local amplitude fluctuation coefficient of each frequency;

[0014] A negative correlation mapping is performed on the local amplitude fluctuation coefficient of each frequency, and the product of the negative correlation mapping result and the number of frequencies in the corresponding curve segment is calculated as the noise possibility of each frequency.

[0015] Furthermore, the method for obtaining the abnormal interference factor includes:

[0016] For any IMF component signal, the signal amplitude of each frequency in the spectrum curve is negatively correlated with the abnormal interference factor, and the noise possibility and the corresponding frequency value are positively correlated with the abnormal interference factor.

[0017] Furthermore, the method for obtaining the frequency clustering clusters includes:

[0018] According to the abnormal interference factor of each frequency, K-means clustering is performed on all frequencies to obtain frequency clustering clusters.

[0019] Furthermore, the method for acquiring the high noise component signal includes:

[0020] For any IMF component signal, the noise performance of each IMF component signal is obtained according to the distribution characteristics of frequencies in different frequency clusters;

[0021] If the noise level of the IMF component signal is greater than a preset level threshold, the corresponding IMF component signal is regarded as a high noise component signal.

[0022] Furthermore, the method for obtaining the noise performance degree includes:

[0023] For any IMF component signal, the noise performance degree of the corresponding IMF component signal is obtained according to the fluctuation degree of all frequencies between different frequency clusters and the difference in average frequency. The fluctuation degree of all frequencies and the difference in average frequency are positively correlated with the noise performance degree.

[0024] Furthermore, the method for obtaining the interference impact degree includes:

[0025] For any high noise component signal, obtain the fluctuation degree of all signal peak widths;

[0026] The mean of the noise possibilities corresponding to all frequencies in the frequency curve is obtained as the overall noise possibility; the possibility difference between the noise possibility of each frequency on the frequency curve and the overall noise possibility is obtained;

[0027] The fluctuation degree and possibility difference of signal peak width are positively correlated with the degree of interference impact.

[0028] Furthermore, the method for obtaining the frequency anomaly factor includes:

[0029] The information entropy of the difference between all adjacent extreme value points in each high noise component signal is obtained as the degree of chaos;

[0030] The degree of confusion and interference influence of each high noise component signal are fused and normalized to obtain the frequency anomaly factor of the corresponding high noise component signal.

[0031] Furthermore, the method for obtaining the abnormal cutoff frequency includes:

[0032] For the original chromatographic signal, the mean value of the frequency anomaly factors of all high noise components corresponding to each frequency in the spectrum curve is obtained as the overall anomaly coefficient of each frequency;

[0033] According to the overall abnormal coefficient of each frequency, K-means clustering is performed on all frequencies in the spectrum curve to obtain frequency categories;

[0034] The mean frequency in each frequency category is obtained as the overall frequency level; the frequency category with the largest overall frequency level value is selected, and the corresponding frequency category is used as the abnormal frequency category; the frequency category with the smallest frequency value is selected in the abnormal frequency category, and the corresponding frequency is used as the abnormal cutoff frequency.

[0035] The present invention also proposes an intelligent detection workstation for fungal toxins, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of any one of the intelligent detection methods for fungal toxins are implemented.

[0036] The present invention has the following beneficial effects:

[0037] The present invention analyzes the spectrum curves of the original chromatographic signal and the IMF component signal to obtain the noise possibility and abnormal interference factor of each frequency; according to the abnormal interference factor of each frequency, all frequencies are clustered to obtain frequency clustering clusters, and the frequency clustering clusters include interference frequency clustering clusters and normal frequency clustering clusters. The frequencies are divided into two categories: interference frequency and normal frequency, which helps to more clearly understand which frequencies in the signal are subject to abnormal interference; according to the distribution characteristics of frequencies in different frequency clustering clusters, high-noise component signals are screened out, which helps to focus on the frequency components that have the greatest impact on signal quality; for any high-noise component signal, according to the noise possibility distribution of different frequencies in the spectrum curve, the frequency components are obtained. The interference influence degree of the high noise component signal can more accurately understand the influence degree of the high noise component signal on the overall signal quality; according to the degree of confusion of the difference between adjacent extreme values ​​in the high noise component signal and the interference influence degree, the frequency anomaly factor of the high noise component signal is obtained to more accurately determine which frequency components have a higher abnormal level; according to the frequency anomaly factors on all high noise component signals corresponding to different frequencies, the abnormal cutoff frequency is obtained, and by setting the abnormal cutoff frequency, a clear parameter basis can be provided for subsequent filtering processing; the original chromatographic signal is filtered to obtain a filtered chromatographic signal, which can effectively remove the noise and abnormal interference components in the original chromatographic signal and improve the quality and reliability of the signal. The present invention improves the accuracy of mycotoxin detection by obtaining accurate signal performance during mycotoxin detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0039] Figure 1 A flow chart of an intelligent detection method for fungal toxins provided by one embodiment of the present invention;

[0040] Figure 2 A flow chart of a method for obtaining noise probability provided by an embodiment of the present invention;

[0041] Figure 3 A flow chart of a method for obtaining an abnormal cutoff frequency provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0042] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the intelligent detection method and workstation of mycotoxins proposed by the present invention, its specific implementation, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0043] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0044] The specific scheme of the intelligent detection method of mycotoxins and the workstation provided by the present invention is described in detail below with reference to the accompanying drawings.

[0045] See also Figure 1 , which shows a method flow chart of an intelligent detection method of fungal toxins provided by an embodiment of the present invention, specifically comprising:

[0046] Step S1: obtaining the original chromatographic signal of the fungal toxin and a plurality of IMF component signals, and obtaining a corresponding spectrum curve, wherein the spectrum curve includes signal amplitudes at different frequencies.

[0047] In an embodiment of the present invention, since noise and interference from other microorganisms may exist during the detection of fungal toxins, resulting in low detection accuracy, it is necessary to process the signal data of the fungal toxins to eliminate the influence of other noises and other microorganisms; since different substances exhibit different frequency distributions, the fungal toxins are detected by high performance liquid chromatography, and a detector is used to obtain a liquid chromatogram for initial analysis to obtain the original chromatographic signal of the fungal toxin and multiple IMF component signals.

[0048] It should be noted that, considering that the original chromatographic signal often contains multiple frequency components and fluctuation patterns, in order to better understand and analyze the different components in the signal, in an embodiment of the present invention, the original chromatographic signal is subjected to EMD decomposition to obtain multiple IMF component signals; the specific EMD decomposition is a technical means well known to those skilled in the art and will not be elaborated here.

[0049] Since the molecular structure of fungal toxins determines their vibration mode at specific frequencies, the signals of fungal toxins in the frequency domain are often concentrated near one or several specific frequencies. By converting the signal from the time domain to the frequency domain, the frequency components and amplitude characteristics of the signal can be analyzed more intuitively. In one embodiment of the present invention, a method for obtaining a spectrum curve includes: performing Fourier transform on the acquired original chromatographic signal and multiple IMF component signals to obtain a corresponding spectrum graph; performing curve fitting on the signal amplitudes corresponding to all frequencies in the spectrum graph to obtain a corresponding spectrum curve, and the spectrum curve includes the signal amplitudes at different frequencies. The specific Fourier transform algorithm is a technical means well known to those skilled in the art and is not limited or elaborated herein.

[0050] It should be noted that, in some embodiments of the present invention, curve fitting can be performed using fitting methods such as the least squares method and polynomial fitting. The specific means are technical means well known to those skilled in the art and will not be described in detail here.

[0051] Step S2: For the original chromatographic signal, the noise possibility of each frequency is obtained according to the fluctuation trend of the signal amplitude corresponding to different frequencies in the spectrum curve.

[0052] Since the frequency range corresponding to noise in the spectrum is a long and relatively flat curve, while the frequency range corresponding to fungal toxins is a short and sharp curve, analyzing the fluctuation trend of the signal amplitude corresponding to different frequencies helps to understand the characteristics and behavior of the signal, identify the main frequency components in the signal, and quantify the possibility of the frequency being noise; for the original chromatographic signal, the noise possibility of each frequency is obtained based on the fluctuation trend of the signal amplitude corresponding to different frequencies in the spectrum curve.

[0053] Preferably, in one embodiment of the present invention, the method for obtaining the noise possibility can be found in Figure 2 , which shows a flow chart of a method for obtaining noise possibility, including:

[0054] Step S201: for the original chromatographic signal, obtain the second-order derivative of the spectrum curve at each frequency, take the frequency corresponding to the second-order derivative being 0 as the segmentation frequency, and obtain the curve segments between adjacent segmentation frequencies.

[0055] The second-order derivative can more sensitively capture changes in the signal peak. The position where the second-order derivative is 0 is at the inflection point of the peak. Segmentation can make the signal components in each segment more uniform, which helps to identify the interactions and interferences between different frequency components.

[0056] Step S202: obtaining the average of the slopes between each frequency on the spectrum curve and other frequencies in the corresponding curve segment as the local amplitude fluctuation coefficient of each frequency.

[0057] By calculating the slope, the changing trend of the signal amplitude between each frequency and its surrounding frequencies can be quantified, thereby more accurately describing the local characteristics of the spectrum. The larger the slope, the greater the local amplitude fluctuation coefficient.

[0058] It should be noted that the first-order derivative of the spectrum curve at each frequency is obtained to obtain the slope of the corresponding frequency.

[0059] Step S203: negative correlation mapping is performed on the local amplitude fluctuation coefficient of each frequency, and the product of the negative correlation mapping result and the number of frequencies in the corresponding curve segment is calculated as the noise possibility of each frequency.

[0060] In one embodiment of the present invention, the formula for noise probability is expressed as:

[0061] ;

[0062] in, Indicates The noise probability of each frequency; Indicates The frequency corresponds to the number of frequencies on the curve segment; Indicates the frequency spectrum curve The frequency and the corresponding curve segment The slope between the other frequencies; Indicates taking the absolute value; Represents an exponential function with a natural constant as its base.

[0063] In the formula for noise probability, Perform negative correlation mapping; It means calculating the slope mean of the data points between each frequency on the spectrum curve and other frequencies on the corresponding curve segment, that is, the local amplitude fluctuation coefficient. The larger the local amplitude fluctuation coefficient, the larger the slope mean, indicating that the inclination between the data points is greater, the more frequencies there are on the curve segment, the higher the peak-shaped curve, the more likely it is the frequency corresponding to the fungal toxin, and the smaller the possibility of noise; the smaller the local amplitude fluctuation coefficient, the smaller the slope mean, indicating that the inclination between the data points is smaller, the flatter the curve, the greater the possibility of noise; The more frequencies there are on the frequency corresponding curve segment, the longer the curve is, the more likely it is a noise curve, and the greater the possibility of noise; the fewer frequencies there are, the shorter the curve is, the more likely it is a fungal toxin curve, and the smaller the possibility of noise.

[0064] Step S3: For any IMF component signal, obtain the abnormal interference factor of each frequency according to the signal amplitude, noise possibility and corresponding frequency value of each frequency in the spectrum curve; cluster all frequencies according to the abnormal interference factor of each frequency to obtain frequency clustering clusters, and the frequency clustering clusters include interference frequency clustering clusters and normal frequency clustering clusters; according to the distribution characteristics of frequencies in different frequency clustering clusters, screen out high noise component signals.

[0065] Each IMF component signal represents an inherent mode in the original chromatographic signal and has specific frequency and amplitude characteristics, so for any IMF component signal, analysis is performed;

[0066] The noise possibility can evaluate the quality of the signal and understand the distribution of noise at different frequencies. The greater the noise possibility, the greater the possibility that the corresponding frequency will be interfered with. Since other microorganisms are produced through fungal metabolism and their amount is relatively small, the amplitude corresponding to other microorganisms is smaller than that corresponding to mycotoxins, and the amplitude of irregular noise is also relatively small. The frequency of other microorganisms and noise is relatively large relative to that of mycotoxins. The signal amplitude and frequency are combined to further analyze the interference of noise and other microorganisms. According to the signal amplitude, noise possibility and corresponding frequency value of each frequency in the spectrum curve, the abnormal interference factor of each frequency is obtained.

[0067] Preferably, in one embodiment of the present invention, the method for obtaining the abnormal interference factor includes:

[0068] For any IMF component signal, the signal amplitude of each frequency in the spectrum curve is negatively correlated with the abnormal interference factor, and the noise possibility and the corresponding frequency value are positively correlated with the abnormal interference factor.

[0069] Among them, the larger the signal amplitude, the more likely it is to be fungal toxins, the smaller the abnormal interference factor, and the negative correlation is; the greater the possibility of noise, the larger the corresponding frequency value, the more likely it is noise and other microbial interference, the larger the abnormal interference factor, and the positive correlation is.

[0070] In one embodiment of the present invention, for any IMF component signal, the formula of the abnormal interference factor of each frequency is expressed as:

[0071] ;

[0072] in, Indicates Abnormal interference factors of frequencies; Indicates the frequency spectrum curve The signal amplitude of the frequency; Indicates The noise probability of each frequency; Indicates The frequency value of the frequency; Represents the normalization function.

[0073] It should be noted that, in one embodiment of the present invention, the normalization function is maximum and minimum value normalization, and the specific means are technical means well known to those skilled in the art and will not be elaborated here.

[0074] The abnormal interference factor reflects the degree to which each frequency is affected by abnormal interference. By clustering the frequencies of similar abnormal interference factors into one category, it helps to more clearly understand the distribution and change trend of abnormal interference frequencies; according to the abnormal interference factor of each frequency, all frequencies are clustered to obtain frequency clustering clusters.

[0075] Preferably, in one embodiment of the present invention, the method for acquiring frequency clustering clusters includes:

[0076] According to the abnormal interference factor of each frequency, K-means clustering is performed on all frequencies to obtain frequency clustering clusters.

[0077] The K-means clustering algorithm clusters multiple cluster objects into specified K clusters based on the similarity between the cluster objects. Each cluster object belongs to and only belongs to one cluster with the smallest distance to the center of the frequency cluster.

[0078] It should be noted that, considering the need to distinguish the frequencies of interference from mycotoxins, noise and other microbial factors, in one embodiment of the present invention, the number of clusters K is set to 2, that is, interference frequency clustering clusters and normal frequency clustering clusters are obtained. The specific K-means clustering algorithm is a technical means well known to those skilled in the art and is not limited or elaborated here.

[0079] Taking into account the fact that all frequencies are distributed relatively concentratedly before being interfered by noise, and are distributed relatively dispersedly after being interfered by noise due to the randomness of noise, the degree to which the frequencies are interfered by noise is quantified; based on the distribution characteristics of frequencies in different frequency clusters, high noise component signals are screened out.

[0080] Preferably, in one embodiment of the present invention, the method for acquiring a high noise component signal includes:

[0081] For any IMF component signal, the noise performance of each IMF component signal is obtained according to the distribution characteristics of frequencies in different frequency clusters;

[0082] If the noise level of the IMF component signal is greater than a preset level threshold, the corresponding IMF component signal is regarded as a high noise component signal.

[0083] It should be noted that, in one embodiment of the present invention, the size of the preset degree threshold is 0.5; in other embodiments of the present invention, the size of the preset degree threshold can be set according to specific circumstances, and is not limited or elaborated here.

[0084] Preferably, in one embodiment of the present invention, the method for obtaining the noise performance degree includes:

[0085] For any IMF component signal, the noise performance degree of the corresponding IMF component signal is obtained according to the fluctuation degree of all frequencies between different frequency clusters and the difference in average frequency. The fluctuation degree of all frequencies and the difference in average frequency are positively correlated with the noise performance degree.

[0086] Among them, the greater the fluctuation degree of all frequencies, the greater the difference in average frequency, which means that the frequency distribution is more discrete, the greater the possibility of being interfered by noise, and the greater the degree of noise performance, which is a positive correlation.

[0087] In one embodiment of the present invention, the degree of fluctuation is expressed by calculating the standard deviation. The larger the standard deviation, the greater the degree of fluctuation, and the smaller the standard deviation, the smaller the degree of fluctuation. In other embodiments of the present invention, the variance may also be used to express the degree of fluctuation. The specific means are technical means well known to those skilled in the art and will not be elaborated here.

[0088] In one embodiment of the present invention, for any IMF component signal, the formula for the noise performance degree is expressed as:

[0089] ;

[0090] in, Indicates the degree of noise performance of the IMF component signal; Indicates The standard deviation of all frequencies in a frequency cluster; Indicates The standard deviation of all frequencies in a frequency cluster; Indicates The mean of all frequencies in the frequency cluster, i.e. The average frequency of the frequency clusters; Indicates The mean of all frequencies in the frequency cluster, i.e. The average frequency of the frequency clusters; represents the linear normalization function; Indicates taking the absolute value.

[0091] Step S4: For any high noise component signal, the interference influence degree of the high noise component signal is obtained according to the noise possibility distribution of different frequencies in the spectrum curve; the frequency anomaly factor of the high noise component signal is obtained according to the confusion degree of the difference between adjacent extreme values ​​in the high noise component signal and the interference influence degree; the abnormal cutoff frequency is obtained according to the frequency anomaly factors of different frequencies on all high noise component signals.

[0092] The distribution of noise possibilities at different frequencies in the spectrum curve reflects the noise characteristics of the frequency and quantifies the impact of noise on the signal. The more uneven the distribution of noise possibilities at the frequency, the greater the degree of interference impact. For any high-noise component signal, the degree of interference impact of the high-noise component signal is obtained based on the distribution of noise possibilities at different frequencies in the spectrum curve.

[0093] Preferably, considering that mycotoxins can be quickly separated from a mixture under specific chromatographic conditions, resulting in a clear and concentrated peak shape, the peak width is usually small; metabolites of other microorganisms have complex molecular structures, diverse types, overlapping or incomplete separation of multiple components, and their peak width is wider than that of mycotoxins; if there is interference from other microorganisms, the peak width will continue to change. In one embodiment of the present invention, the method for obtaining the degree of interference includes:

[0094] For any high noise component signal, obtain the fluctuation degree of all signal peak widths;

[0095] The mean of the noise possibilities corresponding to all frequencies in the frequency curve is obtained as the overall noise possibility; the possibility difference between the noise possibility of each frequency on the frequency curve and the overall noise possibility is obtained;

[0096] The fluctuation degree and possibility difference of signal peak width are positively correlated with the degree of interference impact.

[0097] Among them, the greater the fluctuation of the signal peak width, the more uneven the distribution of the peak width is, and the more interference from other microorganisms exists; the greater the possibility difference, the greater the possibility of frequency noise and the greater the interference impact.

[0098] In one embodiment of the present invention, the formula for the interference impact degree is expressed as:

[0099] ;

[0100] in, Indicates The interference influence degree of a high noise component signal; Indicates The spectrum curve of the high noise component signal The noise probability of each frequency; Indicates The overall noise probability corresponding to the high noise component signal; Indicates The number of frequencies in the spectrum curve of the high noise component signal; Indicates The degree of fluctuation of the signal peak width in a high noise component signal.

[0101] In the formula for the degree of interference impact, It indicates the mean of the difference between the noise possibility of all frequencies on the calculated frequency curve and the overall noise possibility. The larger the mean, the greater the difference in possibility, the more inconsistent the noise possibility change of the frequency, and the greater the interference impact. The greater the fluctuation of the signal peak width in a high noise component signal, the more uneven the peak width distribution, the greater the signal change, and the greater the interference impact.

[0102] It should be noted that, in one embodiment of the present invention, the method for obtaining the signal peak width is: using professional analysis software to measure the chromatogram obtained during liquid chromatography detection to obtain the signal peak width.

[0103] Since the original chromatographic signal of mycotoxins is greatly affected by interference, the more abnormal the component signal fluctuation is, the less affected it is by interference, and the more normal and regular the component signal fluctuation is, the volatility of the signal is reflected by analyzing the degree of confusion of the difference between adjacent extreme values. The greater the degree of confusion of the difference, the more unstable the difference change, and the greater the frequency anomaly factor; the degree of interference influence evaluates the influence of interference on the frequency. The greater the degree of interference influence, the greater the possibility of anomaly; the frequency anomaly factor of the high noise component signal is obtained according to the degree of confusion of the difference between adjacent extreme values ​​in the high noise component signal and the degree of interference influence.

[0104] Preferably, in one embodiment of the present invention, the method for obtaining the frequency anomaly factor includes:

[0105] The information entropy of the difference between all adjacent extreme value points in each high noise component signal is obtained as the degree of chaos;

[0106] The degree of confusion and interference influence of each high noise component signal are fused and normalized to obtain the frequency anomaly factor of the corresponding high noise component signal.

[0107] Information entropy can reflect the randomness and complexity of difference distribution. The larger the entropy value, the more diverse the differences between adjacent extreme points and the larger the frequency anomaly factor.

[0108] It should be noted that, in some embodiments of the present invention, the derivative method and Newton's method can be used to obtain the extreme points on the signal, and the addition or multiplication method can be used for fusion. The specific means are technical means well known to those skilled in the art and will not be elaborated here.

[0109] In one embodiment of the present invention, the formula of the frequency anomaly factor is expressed as:

[0110] ;

[0111] in, Indicates The frequency anomaly factor of the high noise component signal; Indicates The information entropy of the difference between all adjacent extreme points in a high noise component signal, that is, the degree of chaos; Indicates The interference influence degree of a high noise component signal; Represents the normalization function.

[0112] In the formula of the frequency anomaly factor, the larger the information entropy, the more inconsistent the size of the difference, the greater the fluctuation, the greater the degree of interference influence, and the larger the frequency anomaly factor of the high noise component signal.

[0113] The frequency anomaly factor reflects the overall anomaly of the frequency in the high noise component signal. By integrating the frequency anomaly factor in the high noise component signal corresponding to each frequency, the higher the frequency anomaly factor, the higher the degree of confusion and interference of the noise on the frequency, which can reflect the abnormal level of each frequency and help determine the frequency distinction boundary;

[0114] The abnormal cutoff frequency is obtained according to the frequency abnormal factors on all high noise component signals corresponding to different frequencies.

[0115] Preferably, in one embodiment of the present invention, the abnormal cutoff frequency is Figure 3 , which shows a flow chart of a method for obtaining an abnormal cutoff frequency, including:

[0116] Step S301: For the original chromatographic signal, obtain the mean value of the frequency anomaly factors of all high noise components corresponding to each frequency in the spectrum curve as the overall anomaly coefficient of each frequency.

[0117] The overall frequency anomaly of each frequency is quantified by taking the average value to reflect the overall anomaly level of each frequency.

[0118] It should be noted that, for frequencies not included in the high noise component signal, the corresponding frequency anomaly factor is 0.

[0119] Step S302: Perform K-means clustering on all frequencies in the frequency spectrum curve according to the overall abnormal coefficient of each frequency to obtain frequency categories.

[0120] By clustering the frequencies of similar abnormal features into one category, the similarities and differences of different frequencies in the spectrum curve are reflected, which helps to intuitively understand the distribution and changes of spectrum characteristics.

[0121] It should be noted that the K value is determined by the elbow rule; the specific means are well known to those skilled in the art and will not be described in detail here.

[0122] Step S303: Obtain the mean value of the frequency in each frequency category as the overall frequency level; select the frequency category with the largest overall frequency level value and use the corresponding frequency category as the abnormal frequency category; select the frequency category with the smallest frequency value and use the corresponding frequency as the abnormal cutoff frequency.

[0123] The overall frequency level can reflect the frequency average characteristics of the frequency category. Since the frequencies of other microorganisms and noise are relatively large compared to mycotoxins, the higher the overall frequency level, the more likely it is to contain more high-frequency noise and interference signals. Selecting the category corresponding to the largest overall frequency level as the abnormal frequency category helps to focus on frequency components with greater influence. Selecting the smallest frequency component can effectively suppress or process frequency components with larger frequencies and greater influences.

[0124] Step S5: filtering the original chromatogram signal based on the abnormal cutoff frequency to obtain a filtered chromatogram signal.

[0125] The original chromatographic signal is filtered by a low-pass filter based on the abnormal cutoff frequency, the frequencies higher than the abnormal cutoff frequency are removed, and the frequencies less than or equal to the abnormal cutoff frequency are retained to obtain a filtered chromatographic signal, which helps to improve the accuracy of subsequent fungal toxin detection.

[0126] In summary, the present invention analyzes the spectrum curves of the original chromatographic signal and the IMF component signal to obtain the noise possibility and abnormal interference factor of each frequency; according to the abnormal interference factor of each frequency, all frequencies are clustered to obtain frequency clustering clusters; according to the distribution characteristics of frequencies in different frequency clustering clusters, high noise component signals are screened out; according to the noise possibility distribution of different frequencies in the spectrum curve, the interference influence degree of the high noise component signal is obtained; combined with the degree of confusion of the difference between adjacent extreme values ​​in the high noise component signal, the frequency abnormality factor of the high noise component signal is obtained, and the abnormal cutoff frequency is obtained; the original chromatographic signal is filtered to obtain a filtered chromatographic signal. The present invention improves the accuracy of mycotoxin detection by obtaining accurate signal performance during mycotoxin detection.

[0127] The present invention also proposes an intelligent detection workstation for mycotoxins, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any step of an intelligent detection method for mycotoxins.

[0128] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0129] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. An intelligent detection method for mycotoxins, characterized in that: The method comprises: Acquire the original chromatographic signal and multiple IMF component signals of the fungal toxin, and obtain a corresponding spectrum curve, wherein the spectrum curve includes signal amplitudes at different frequencies; For the original chromatographic signal, the noise possibility of each frequency is obtained according to the fluctuation trend of the signal amplitude corresponding to different frequencies in the spectrum curve; For any IMF component signal, the abnormal interference factor of each frequency is obtained according to the signal amplitude, noise possibility and corresponding frequency value of each frequency in the spectrum curve; according to the abnormal interference factor of each frequency, all frequencies are clustered to obtain frequency clustering clusters, and the frequency clustering clusters include interference frequency clustering clusters and normal frequency clustering clusters; according to the distribution characteristics of frequencies in different frequency clustering clusters, high noise component signals are screened out; For any high noise component signal, the interference influence degree of the high noise component signal is obtained according to the noise possibility distribution of different frequencies in the spectrum curve; the frequency anomaly factor of the high noise component signal is obtained according to the confusion degree of the difference between adjacent extreme values ​​in the high noise component signal and the interference influence degree; the abnormal cutoff frequency is obtained according to the frequency anomaly factors on all high noise component signals corresponding to different frequencies; The original chromatographic signal is filtered based on the abnormal cutoff frequency to obtain a filtered chromatographic signal.

2. The intelligent detection method of mycotoxins according to claim 1, characterized in that: The method for obtaining the noise possibility includes: For the original chromatographic signal, the second-order derivative of the spectrum curve at each frequency is obtained, and the frequency corresponding to the second-order derivative being 0 is used as the segmentation frequency to obtain the curve segments between adjacent segmentation frequencies; The average value of the slope between each frequency on the spectrum curve and other frequencies on the corresponding curve segment is obtained as the local amplitude fluctuation coefficient of each frequency; A negative correlation mapping is performed on the local amplitude fluctuation coefficient of each frequency, and the product of the negative correlation mapping result and the number of frequencies in the corresponding curve segment is calculated as the noise possibility of each frequency.

3. The intelligent detection method of mycotoxins according to claim 1, characterized in that: The method for obtaining the abnormal interference factor includes: For any IMF component signal, the signal amplitude of each frequency in the spectrum curve is negatively correlated with the abnormal interference factor, and the noise possibility and the corresponding frequency value are positively correlated with the abnormal interference factor.

4. The intelligent detection method of mycotoxins according to claim 1, characterized in that: The method for obtaining the frequency clustering clusters includes: According to the abnormal interference factor of each frequency, K-means clustering is performed on all frequencies to obtain frequency clustering clusters.

5. The intelligent detection method of mycotoxins according to claim 1, characterized in that: The method for acquiring the high noise component signal comprises: For any IMF component signal, the noise performance of each IMF component signal is obtained according to the distribution characteristics of frequencies in different frequency clusters; If the noise level of the IMF component signal is greater than a preset level threshold, the corresponding IMF component signal is regarded as a high noise component signal.

6. The intelligent detection method of mycotoxins according to claim 5, characterized in that: The method for obtaining the noise performance degree includes: For any IMF component signal, the noise performance degree of the corresponding IMF component signal is obtained according to the fluctuation degree of all frequencies between different frequency clusters and the difference in average frequency. The fluctuation degree of all frequencies and the difference in average frequency are positively correlated with the noise performance degree.

7. The intelligent detection method of mycotoxins according to claim 1, characterized in that: The method for obtaining the interference impact degree includes: For any high noise component signal, obtain the fluctuation degree of all signal peak widths; The mean of the noise possibilities corresponding to all frequencies in the frequency curve is obtained as the overall noise possibility; the possibility difference between the noise possibility of each frequency on the frequency curve and the overall noise possibility is obtained; The fluctuation degree and possibility difference of signal peak width are positively correlated with the degree of interference impact.

8. The intelligent detection method of mycotoxins according to claim 1, characterized in that: The method for obtaining the frequency anomaly factor includes: The information entropy of the difference between all adjacent extreme value points in each high noise component signal is obtained as the degree of chaos; The degree of confusion and interference influence of each high noise component signal are fused and normalized to obtain the frequency anomaly factor of the corresponding high noise component signal.

9. The intelligent detection method of mycotoxins according to claim 1, characterized in that: The method for obtaining the abnormal cutoff frequency includes: For the original chromatographic signal, the mean value of the frequency anomaly factors of all high noise components corresponding to each frequency in the spectrum curve is obtained as the overall anomaly coefficient of each frequency; According to the overall abnormal coefficient of each frequency, K-means clustering is performed on all frequencies in the spectrum curve to obtain frequency categories; The mean frequency in each frequency category is obtained as the overall frequency level; the frequency category with the largest overall frequency level value is selected, and the corresponding frequency category is used as the abnormal frequency category; the frequency category with the smallest frequency value is selected in the abnormal frequency category, and the corresponding frequency is used as the abnormal cutoff frequency.

10. An intelligent detection workstation for mycotoxins, the workstation comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the intelligent detection method of a fungal toxin as described in any one of claims 1 to 9 are implemented.

Images