Intelligent protein quality detection method and system based on spectral analysis

Abstract

The present application relates to the field of spectral analysis technology, and specifically to a protein quality intelligent detection method and system based on spectral analysis, the method comprising: collecting infrared spectrum data of a preset number of protein samples; determining the local fluctuation coefficient of transmittance at each wavelength; determining the fluctuation coefficient of infrared spectrum data of each protein sample; screening the infrared spectrum data of all protein samples based on the fluctuation coefficient to obtain each suspected impurity fluctuation data; dividing each suspected impurity fluctuation data into each segment; determining the abnormal index of the suspected impurity fluctuation data in each segment; modifying the spatial domain weight and value range weight of each filter window in the bilateral filtering algorithm based on the abnormal index, and denoising the infrared spectrum data of the protein sample; determining the difference coefficient of the infrared spectrum data of each protein sample, and detecting the protein quality. Thus, the accuracy of protein quality detection is improved.

Description

Technical Field

[0001] The present application relates to the technical field of spectral analysis, and specifically to an intelligent protein quality detection method and system based on spectral analysis. Background Art

[0002] As one of the key nutrients in food, the quality of protein directly affects the nutritional value of food and human health. Traditional protein quality detection methods often rely on chemical reagents and complex laboratory equipment. In recent years, spectral analysis technology has been widely used in the food and biomedical fields due to its rapid, non-destructive and high-precision characteristics. In particular, infrared spectroscopy technology, as a non-invasive detection method, can obtain the spectral information of samples in a short time, providing new possibilities for intelligent detection of protein quality.

[0003] However, the existing spectral detection methods will cause fluctuations in the infrared spectrum detection results and generate noise due to differences in detection instruments and detection environments. In addition, there may be impurities that do not belong to proteins inside the test object, which will also interfere with the collected infrared spectrum data. The existing bilateral filtering algorithms often use fixed filtering parameters, which leads to the filtering out of noises generated by factors such as the detection instrument, the detection environment, and impurities inside the test object. However, the impurities inside the test object have an important impact on the detection of protein quality. If they are filtered out, they will interfere with the protein quality detection results and cause errors in the protein quality detection results. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of this application is to provide a protein quality intelligent detection method and system based on spectral analysis. The technical solutions adopted are as follows:

[0005] In a first aspect, the present invention provides a method for intelligent protein quality detection based on spectral analysis, the method comprising the following steps:

[0006] Collect infrared spectral data of a preset number of protein samples;

[0007] Analyze the difference between the transmittance of each wavelength in the infrared spectrum data of each protein sample and the transmittance of the wavelength in its neighborhood range to determine the local fluctuation coefficient of the transmittance of each wavelength; analyze the difference between the infrared spectrum data of each protein sample and the remaining protein samples, and determine the fluctuation coefficient of the infrared spectrum data of each protein sample in combination with the local fluctuation coefficient;

[0008] Screening the infrared spectrum data of all protein samples based on the fluctuation coefficient to obtain each suspected impurity fluctuation data; dividing each suspected impurity fluctuation data into each segment; and determining the abnormal index of the suspected impurity fluctuation data in each segment based on the distribution trend of the suspected impurity fluctuation data in each segment;

[0009] Based on the abnormal index, the spatial domain weight and the value range weight of each filtering window in the bilateral filtering algorithm are corrected, the corrected bilateral filtering algorithm is used to denoise the suspected impurity fluctuation data, and the bilateral filtering algorithm without the correction is used to denoise the infrared spectrum data except the suspected impurity fluctuation data; the protein quality is detected based on the denoising result.

[0010] In one embodiment, the local fluctuation coefficient is the average of the differences between the transmittance of each wavelength and the transmittance of all wavelengths in its neighborhood.

[0011] In one embodiment, determining the fluctuation coefficient includes:

[0012] Calculate the integrated difference of infrared spectrum data between any protein sample and the remaining protein samples;

[0013] Determine the absolute difference between any one protein sample and the remaining protein samples based on the average level of the integrated difference and the difference between the local fluctuation coefficients of any one protein sample and the remaining protein samples;

[0014] The coefficient of fluctuation of the infrared spectrum data of any one of the protein samples is the mean of all the absolute differences of any one of the protein samples.

[0015] In one embodiment, the absolute difference is calculated as follows:

[0016] ;

[0017] In the formula, is the absolute difference between the rth protein sample and the tth protein sample, is the integral difference between the infrared spectrum data of the rth protein sample and the tth protein sample, is the horizontal coordinate length of the spectral curve corresponding to the infrared spectrum data of the rth protein sample, is the difference in the local fluctuation coefficient of the i-th wavelength between the r-th protein sample and the t-th protein sample, and I is the number of wavelengths in the infrared spectrum data of the r-th protein sample.

[0018] In one embodiment, the determination of the suspected impurity fluctuation data includes:

[0019] A segmentation threshold is determined based on the fluctuation coefficients of all protein samples, and infrared spectrum data of protein samples whose fluctuation coefficients are greater than the segmentation threshold are used as suspected impurity fluctuation data.

[0020] In one embodiment, determining the abnormality index includes:

[0021] Obtain all trough points of the suspected impurity fluctuation data in each segment, calculate the discrete degree of transmittance of all the trough points in each segment, calculate the curvature of each trough point, calculate the difference between the wavelength corresponding to the minimum value of the curvature in all the trough points and the wavelength corresponding to the minimum value of the transmittance, and record it as the wavelength difference; determine the abnormality index based on the discrete degree, the minimum value of the curvature and the wavelength difference;

[0022] The abnormal index is positively correlated with the discrete degree, and negatively correlated with the minimum value of the curvature and the wavelength difference.

[0023] In one embodiment, the modifying of the spatial domain weight and the range weight of each filter window in the bilateral filtering algorithm based on the abnormality index includes:

[0024] For any suspected impurity fluctuation data, if the filtering window of the bilateral filtering algorithm spans the nth segment and the n+1th segment of the suspected impurity fluctuation data, the area ratio of the filtering window in the nth segment and the n+1th segment is calculated respectively, recorded as the first ratio and the second ratio, and the fusion result of the abnormal index of the nth segment and the first ratio, and the abnormal index of the n+1th segment and the second ratio is calculated;

[0025] Taking the difference between the minimum value of the abnormal index of all segments of any suspected impurity fluctuation data and the fusion result as the proportional coefficient of the filter window;

[0026] The products of the proportional coefficient and the original spatial domain weight and range weight of the filter window are respectively used as the modified spatial domain weight and range weight.

[0027] In one embodiment, the detecting of protein quality based on the denoising result comprises:

[0028] The difference between the denoised infrared spectral data of each protein sample and the standard protein spectral data is analyzed to determine the difference coefficient of the infrared spectral data of each protein sample. If the difference coefficient of the protein sample is greater than a preset threshold, the protein sample is judged to be of unqualified quality; otherwise, the protein sample is judged to be of qualified quality.

[0029] In one embodiment, the coefficient of difference is calculated as follows:

[0030] ;

[0031] In the formula, is the coefficient of difference of the infrared spectrum data of the rth protein sample, is the transmittance of the ith wavelength of the infrared spectrum data of the rth protein sample, is the transmittance of the i-th wavelength of the standard protein spectrum data, and I is the number of wavelengths in the infrared spectrum data of the r-th protein sample.

[0032] In a second aspect, an embodiment of the present application also provides an intelligent protein quality detection system based on spectral analysis, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of any one of the above methods when executing the computer program.

[0033] This application has at least the following beneficial effects:

[0034] The present application collects infrared spectral data of a preset number of protein samples; analyzes the difference between the transmittance of each wavelength in the infrared spectral data of each protein sample and the transmittance of the wavelength in its neighborhood range, and determines the local fluctuation coefficient of the transmittance of each wavelength; analyzes the difference between the infrared spectral data of each protein sample and the remaining protein samples, and determines the fluctuation coefficient of the infrared spectral data of each protein sample in combination with the local fluctuation coefficient; the fluctuation coefficient measures whether the abnormal fluctuation of the infrared spectral data is a local abnormality, thereby improving the accuracy of subsequent judgment of the local fluctuation area in the infrared spectral data; based on the fluctuation coefficient, the infrared spectral data of all protein samples are screened to obtain each suspected impurity fluctuation data; the suspected impurity fluctuation data corresponds to the infrared spectral data of the protein sample that is most likely to be interfered by impurities; each suspected impurity fluctuation data is equally divided into each segment; based on the distribution trend of the suspected impurity fluctuation data in each segment, the abnormal index of the suspected impurity fluctuation data in each segment is determined; the abnormal index reflects the degree of influence of impurity interference on the suspected impurity fluctuation data, and the abnormal index reflects the degree of influence of impurity interference on the suspected impurity fluctuation data. The constant index helps to enhance the accuracy of weight adjustment in subsequent bilateral filtering; further, based on the abnormal index, the spatial domain weight and the value domain weight of each filtering window in the bilateral filtering algorithm are corrected, the corrected bilateral filtering algorithm is used to denoise the suspected impurity fluctuation data, and the bilateral filtering algorithm without the correction is used to denoise the infrared spectrum data other than the suspected impurity fluctuation data; the denoising effect of the infrared spectrum data of each protein sample is improved; the difference between the denoised infrared spectrum data of each protein sample and the standard protein spectrum data is analyzed, the difference coefficient of the infrared spectrum data of each protein sample is determined, and the protein quality is tested; according to the different characteristics of the noise caused by impurities in the test sample and the noise caused by the test instrument or the test environment, when bilateral filtering the infrared spectrum data, the weight of the bilateral filtering is adaptively adjusted according to the characteristics of the infrared spectrum data, so that both the filtering effect of the global noise is guaranteed and the local details of the absorption peak abnormality caused by impurities are paid attention to, thereby improving the accuracy of the final protein quality detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application 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 application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0036] Figure 1 A flowchart of the steps of a protein quality intelligent detection method based on spectral analysis provided in one embodiment of the present application;

[0037] Figure 2 Construct a flow chart for the coefficient of variation of protein samples. DETAILED DESCRIPTION

[0038] In order to further explain the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the following is a detailed description of the protein quality intelligent detection method and system based on spectral analysis proposed in the present application, its specific implementation method, structure, characteristics 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.

[0039] 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 application belongs.

[0040] The specific scheme of the protein quality intelligent detection method and system based on spectral analysis provided by the present application is described in detail below with reference to the accompanying drawings.

[0041] See also Figure 1 , which shows a flowchart of a method for intelligent protein quality detection based on spectral analysis provided by an embodiment of the present application, the method comprising the following steps:

[0042] S1, collecting infrared spectral data of a preset number of protein samples.

[0043] In this embodiment, 100 egg samples of the same variety are selected as each protein sample, and the infrared spectrum curve of each protein sample is measured by a spectrometer, wherein the acquisition parameters of the spectrometer in this embodiment are set as follows: spectral band 4000~7000nm, resolution 1nm, accumulation times 3 times. Then, with 1nm as the spectral interval, discrete infrared spectrum data of the infrared spectrum curve of each protein sample are obtained respectively, recorded as infrared spectrum data, and the infrared spectrum data represents the transmittance of each wavelength. Rubberband correction is applied to the infrared spectrum data, and infrared spectrum data after baseline correction is output. Rubberband correction is an existing well-known technology, and the specific process is not repeated here.

[0044] It should be noted that the number of protein samples, the acquisition parameters of the spectrometer, and the spectral interval can be set by the implementer according to the actual situation, and this embodiment does not limit this.

[0045] S2, analyzing the difference between the transmittance of each wavelength in the infrared spectrum data of each protein sample and the transmittance of the wavelength in its neighborhood range, and determining the local fluctuation coefficient of the transmittance of each wavelength; analyzing the difference between the infrared spectrum data of each protein sample and the remaining protein samples, and determining the fluctuation coefficient of the infrared spectrum data of each protein sample in combination with the local fluctuation coefficient.

[0046] The infrared spectrum of ovalbumin contains amide I band, amide II band, , -folding absorption peaks, these absorption peaks mainly reflect the secondary structure of the protein. The secondary structure of the protein is an important aspect of the protein structure, which has a direct impact on the functional performance of the protein. However, due to the detection instrument, detection environment and impurities in the protein sample, noise is introduced into the spectral curve, which has a negative impact on the detection of protein quality. The noise can be filtered out by bilateral filtering. However, the noise caused by the detection instrument and the detection environment has nothing to do with the protein quality and can be directly filtered out; however, the error caused by the impurities in the protein sample is related to the protein quality, and filtering will affect the accuracy of protein quality detection.

[0047] Under normal circumstances, the fluctuations in the spectral curve of a protein are small. However, when impurities exist in the protein, the originally smooth absorption peak will become irregular or a new absorption peak will be introduced, resulting in relatively large local fluctuations in the spectral curve.

[0048] Based on the above analysis, the fluctuation coefficient of the infrared spectral data of each protein sample can be calculated according to the difference between the transmittance of each wavelength in the infrared spectral data of the same protein sample and the transmittance of wavelengths in its neighborhood range, as well as the fluctuation difference of the transmittance of the same wavelength in the infrared spectral data of different protein samples. This is used to measure the fluctuation change of the infrared spectral data of each protein sample and the range of fluctuation.

[0049] In this embodiment, for each wavelength of the infrared spectrum data of each protein sample, the J wavelengths closest to it are selected as the neighborhood range of each wavelength. In this embodiment, J=6, and the implementer can set it according to the actual situation. This embodiment does not limit it here.

[0050] The calculation method of the local fluctuation coefficient of the transmittance at each wavelength in the infrared spectrum data of each protein sample is:

[0051] ;

[0052] In the formula, It is Infrared spectrum data of protein samples The local fluctuation coefficient of transmittance at each wavelength, It is Infrared spectrum data of protein samples The transmittance at each wavelength is is the number of wavelengths in the neighborhood of the i-th wavelength in the infrared spectrum data of the r-th protein sample, is the transmittance of the jth wavelength in the neighborhood of the ith wavelength in the infrared spectrum data of the rth protein sample.

[0053] The local fluctuation coefficient measures the average variation trend of the transmittance of each wavelength in the infrared spectrum data, avoiding the errors caused by the influence of local abnormal noise.

[0054] Secondly, calculate the integral difference of the infrared spectrum data of any protein sample and the rest of the protein samples. The specific calculation method is:

[0055] ;

[0056] In the formula, represents the integral difference between the infrared spectrum data of the rth protein sample and the infrared spectrum data of the tth protein sample, Represents the fitting function of the spectral curve corresponding to the infrared spectrum data of the rth protein sample, Indicates the minimum value of the horizontal coordinate of the spectral curve corresponding to the infrared spectrum data of the rth protein sample, Indicates the maximum value of the abscissa of the spectral curve corresponding to the infrared spectrum data of the rth protein sample, Represents the fitting function of the spectral curve corresponding to the infrared spectrum data of the t-th protein sample, Indicates the minimum value of the horizontal coordinate of the spectral curve corresponding to the infrared spectrum data of the t-th protein sample, Represents the maximum value of the abscissa of the spectral curve corresponding to the infrared spectrum data of the t-th protein sample.

[0057] Based on the integral difference and the local fluctuation coefficient, the absolute difference between any protein sample and the remaining protein samples is determined, and the specific calculation method is:

[0058] ;

[0059] In the formula, is the absolute difference between the rth protein sample and the tth protein sample, is the integral difference between the infrared spectrum data of the rth protein sample and the tth protein sample, is the horizontal coordinate length of the spectral curve corresponding to the infrared spectrum data of the rth protein sample, is the difference in the local fluctuation coefficient of the i-th wavelength between the r-th protein sample and the t-th protein sample, and I is the number of wavelengths in the infrared spectrum data of the r-th protein sample.

[0060] In this embodiment, the difference between the local fluctuation coefficients of the i-th wavelength in the r-th protein sample and the t-th protein sample is calculated by the absolute value of the difference, that is, is the absolute value of the difference between the local fluctuation coefficients at the i-th wavelength in the r-th protein sample and the t-th protein sample. The implementer may also choose other calculation methods that reflect the difference, such as ratio, square of the difference, etc.

[0061] Further, based on the absolute difference, the fluctuation coefficient of the infrared spectrum data of each protein sample is calculated, and the specific calculation method is:

[0062] ;

[0063] In the formula, It is The fluctuation coefficient of the infrared spectrum data of protein samples, is the number of protein samples. .

[0064] It can be understood that if there is a difference between the two spectral curves, it will affect the transmittance of each wavelength in the infrared spectrum data, so that the transmittance of each wavelength will increase or decrease to a certain extent, and the difference in the integrated area between the two spectral curves can be used as the overall difference between the two spectral curves. It can represent the average difference of transmittance at each wavelength. The larger the average difference, and The larger the value is, the greater the difference between the infrared spectral data of the two corresponding protein samples is, the greater the fluctuation coefficient of the infrared spectral data of the protein samples is, and the greater the possibility of protein impurities affecting the data.

[0065] S3, based on the fluctuation coefficient, screening the infrared spectrum data of all protein samples to obtain each suspected impurity fluctuation data; dividing each suspected impurity fluctuation data into each segment; based on the distribution trend of the suspected impurity fluctuation data in each segment, determining the abnormal index of the suspected impurity fluctuation data in each segment.

[0066] When using the bilateral filtering algorithm to denoise infrared spectral data, if fixed spatial domain weights and range weights are used, the detection of protein quality will be affected, because if larger spatial domain weights and range weights are used, the denoising effect will be enhanced, so that the noise caused by the detection instrument or the detection environment will be better filtered out, but at the same time, the details will be blurred, and the local fluctuations caused by impurities in the protein sample are displayed through some details, such as the curvature change of the absorption peak; if smaller spatial domain weights and range weights are used, although the details can be better preserved, the noise filtering effect caused by the detection instrument or the detection environment is poor, which will have a negative impact on the subsequent protein quality detection. In order to resolve this conflict, it is necessary to distinguish the infrared spectral data.

[0067] When using infrared spectroscopy to detect protein quality, the presence of impurities may introduce new absorption peaks. The reason for these absorption peaks is that the main components of the impurities are different from the main components of the protein. Secondly, impurities may also make the originally smooth absorption peaks irregular or produce new shoulder peaks. The reason for this is that impurities affect the absorption intensity of the main components of the protein at a specific wavelength. The irregularity of this absorption peak will cause changes in the local details of the infrared spectrum data.

[0068] In this embodiment, the upper four digits of the fluctuation coefficients of the infrared spectrum data of all protein samples are obtained as the segmentation threshold, and the infrared spectrum data of the protein samples with fluctuation coefficients greater than the segmentation threshold are taken as suspected impurity fluctuation data.

[0069] It should be noted that the implementer may select other existing feasible methods for obtaining the segmentation threshold according to actual conditions, such as the Otsu threshold method, etc., and this embodiment does not limit this.

[0070] The protein samples with suspected impurity fluctuation data are recorded as suspected impurity protein samples. The suspected impurity fluctuation data of each suspected impurity protein sample is divided into N segments. In this embodiment, N=60. The implementer can set it according to the actual situation, and this embodiment does not limit it here.

[0071] For any suspected impurity fluctuation data, taking any fragment therein as an example, the transmittance of all wavelengths in the fragment is organized into a data sequence of the fragment in ascending order of wavelength. The data sequence is taken as input, and a comparison and discrimination method is adopted to output each trough point in the data sequence. The comparison and discrimination method is an existing well-known technology and will not be described in detail.

[0072] Based on the above analysis, the abnormal index of each fragment in each suspected impurity fluctuation data is calculated to determine whether there is a detailed change in the absorption peak caused by impurities in each fragment. The specific calculation method is:

[0073] ;

[0074] In the formula, The suspected impurity fluctuation data The abnormal index of the fragments, The suspected impurity fluctuation data The data set consisting of all trough points in the segment, To calculate the discrete degree of transmittance of all trough points in the data set, The suspected impurity fluctuation data The minimum curvature of all trough points in the segment, The suspected impurity fluctuation data The wavelength corresponding to the trough point with the smallest curvature in the segment, The suspected impurity fluctuation data The wavelength corresponding to the trough point with the minimum transmittance in the segment. Recorded as wavelength difference.

[0075] It should be noted that the degree of discreteness can be calculated by variance, standard deviation, coefficient of variation, etc., and the present embodiment adopts variance as the calculation method of the degree of discreteness; the calculation of curvature is an existing well-known technology, and the specific process will not be described in detail; each trough point in the data sequence represents each transmittance, that is, all the data in the data set are specific values ​​of the transmittance.

[0076] It can be understood that when one section of the infrared spectrum data is greatly affected by impurities and there is a change in the absorption peak caused by impurities, it will cause the absorption peak to be irregular or generate a new shoulder peak. Under normal circumstances, the contour of the absorption peak is sharp or arc-shaped, so the curvature at the corresponding trough is relatively large. The irregular absorption peak caused by impurities will distort the contour of the absorption peak, thereby reducing the curvature corresponding to the trough, that is, is smaller; secondly, when a new shoulder peak is generated, the shoulder peak is also a trough and the corresponding transmittance is relatively small, which makes Increase; Since the transmittance of the absorption peak of the main component in the protein is relatively small, the trough point with the smallest transmittance is selected as the main component in the protein sample. If there are impurities in the protein sample, the curvature corresponding to the trough point will be reduced. Therefore, in a segment, if the trough point with the smallest curvature is closer to the trough point with the smallest transmittance, that is, The smaller the value, the more serious the impact of impurities on the absorption peak of the main components of the protein, and the more likely it is that the absorption peak is abnormal due to impurities; therefore, when one of the fragments of the infrared spectrum data has an absorption peak change caused by impurities, the abnormal index of the fragment On the contrary, when a fragment is less affected by impurities, the corresponding anomaly index Also smaller.

[0077] S4, based on the abnormality index, the spatial domain weight and the value range weight of each filter window in the bilateral filtering algorithm are corrected, the corrected bilateral filtering algorithm is used to denoise the suspected impurity fluctuation data, and the bilateral filtering algorithm without the correction is used to denoise the infrared spectrum data except the suspected impurity fluctuation data.

[0078] The infrared spectrum data of each protein sample is denoised using a bilateral filtering algorithm. In this embodiment, the size of the filter window is 1×5. The implementer can set the size of the filter window on his own, provided that the size of the filter window spans at most two segments of the infrared spectrum data. For the infrared spectrum data of protein samples that are not suspected impurity protein samples, the standard deviation of the spatial domain Gaussian kernel function of the bilateral filtering algorithm is and the standard deviation of the range Gaussian kernel function Set to: , the infrared spectrum data of the protein samples that are not suspected impurity protein samples are denoised using a bilateral filtering algorithm, wherein the bilateral filtering algorithm is a known technology and will not be described in detail in this embodiment. The standard deviation of the spatial domain Gaussian kernel function is recorded as the spatial domain weight, and the standard deviation of the range Gaussian kernel function is recorded as the range weight.

[0079] For the infrared spectral data of protein samples that are not suspected impurity protein samples, they are affected by the global errors caused by the detection instrument or the detection environment. Therefore, larger spatial domain weights and value range weights can be used for filtering to achieve better denoising effects and eliminate the global errors caused by the detection instrument or the detection environment. Therefore, a fixed and Perform filtering.

[0080] For each suspected impurity fluctuation data, it may cause detailed changes in the absorption peak of the main component of the protein. If a larger spatial domain weight and value range weight are used for filtering, these detailed changes will be ignored, resulting in errors in the detection of protein quality. Therefore, the weight can be adaptively adjusted according to the abnormal index of each fragment in the suspected impurity fluctuation data. Taking the kth filter window as an example, assuming that the kth filter window spans the nth fragment and the n+1th fragment in the suspected impurity fluctuation data, the proportional coefficient of the kth filter window is calculated as follows:

[0081] ;

[0082] In the formula, is the proportional coefficient of the kth filter window, is the abnormal index of the nth segment in the suspected impurity fluctuation data, is the abnormal index of the n+1th segment in the suspected impurity fluctuation data, is the abnormal index of the e-th fragment in the suspected impurity fluctuation data, N is the number of fragments in the suspected impurity fluctuation data, min{} is the minimum function, Norm[] is the normalization function, is the area ratio of the kth filter window in the nth segment of the suspected impurity fluctuation data, recorded as the first ratio, is the area ratio of the kth filter window in the n+1th segment of the suspected impurity fluctuation data, recorded as the second ratio. Recorded as a fusion result, it should be understood that fusion means combining multiple variables, which can be calculated by addition, multiplication, or a combination of addition and multiplication.

[0083] The specific calculation method of the first proportion is: take the ratio of the amount of data contained in the nth segment of the suspected impurity fluctuation data of the kth filtering window to the total amount of data in the nth segment as the first proportion, and for the n+1th segment, use the same calculation method as the first proportion to obtain the second proportion.

[0084] It should be noted that, assuming that the k-th filter window is only located in the n-th segment of the suspected impurity fluctuation data, the proportional coefficient of the k-th filter window is calculated as follows: .

[0085] For the kth filtering window, the spatial domain weight and the range weight of the bilateral filtering algorithm are modified based on the proportional coefficient, specifically in the following manner:

[0086] ;

[0087] In the formula, For the The standard deviation of the spatial domain Gaussian kernel function after the filter window correction is For the The proportionality factor of the filter window, is the standard deviation of the original spatial domain Gaussian kernel function.

[0088] Similarly, the standard deviation of the Gaussian kernel function in the corrected value range is: ; In the formula, For the The standard deviation of the Gaussian kernel function of the value range after the filter window correction, is the standard deviation of the original range Gaussian kernel function.

[0089] Based on the modified standard deviation of the spatial domain Gaussian kernel function and the range Gaussian kernel function, the bilateral filtering algorithm is used to denoise the suspected impurity fluctuation data.

[0090] It can be understood that when the area where the filter window is located is greatly affected by impurities, the corresponding abnormal index is also larger, so that the proportional coefficient of the filter window The smaller the scale factor, the smaller the weight will be, so that more details can be retained during filtering. On the contrary, when the area where the filter window is located is less affected by impurities, the scale factor of the filter window is The larger the value, the greater the weight, which can improve the filtering effect on noise.

[0091] S5, analyzing the difference between the infrared spectrum data of each protein sample after denoising and the standard protein spectrum data, determining the difference coefficient of the infrared spectrum data of each protein sample, and detecting the protein quality.

[0092] Obtain the standard infrared spectrum data for protein detection from historical data, recorded as standard protein spectrum data, and judge the protein quality by comparing the difference between the infrared spectrum data of each protein sample after denoising and the standard protein spectrum data. Specifically, calculate the difference coefficient of the infrared spectrum data of each protein sample. The specific calculation method is:

[0093] ;

[0094] In the formula, is the coefficient of difference of the infrared spectrum data of the rth protein sample, is the transmittance of the ith wavelength of the infrared spectrum data of the rth protein sample, is the transmittance of the i-th wavelength of the standard protein spectrum data, and I is the number of wavelengths in the infrared spectrum data of the r-th protein sample. The flowchart for constructing the protein sample difference coefficient is shown in Figure 2 shown.

[0095] Preset Threshold , in this embodiment , the implementer can set it according to the actual situation, and this embodiment does not limit it. When The quality of protein samples was unqualified; When If the quality of each protein sample is qualified, the same method is used to test each protein sample to obtain the test results of each protein sample.

[0096] Based on the same inventive concept as the above method, an embodiment of the present application also provides a protein quality intelligent detection system based on spectral analysis, including a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, the steps of any one of the above-mentioned protein quality intelligent detection methods based on spectral analysis are implemented.

[0097] It should be noted that the above sequence of the embodiments of the present application is for description only and does not represent the advantages and disadvantages of the embodiments. The above is a description of a specific embodiment of this specification. In addition, 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.

[0098] 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.

[0099] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present application should be included in the protection scope of the present application.

Claims

1. A protein quality intelligent detection method based on spectral analysis, characterized in that: The method comprises the following steps: Collect infrared spectral data of a preset number of protein samples; Determine the local fluctuation coefficient of the transmittance of each wavelength as the average value of the difference between the transmittance of each wavelength in the infrared spectrum data of each protein sample and the transmittance of all wavelengths in its neighborhood; calculate the integral difference between the infrared spectrum data of any protein sample and the remaining protein samples; determine the absolute difference between any protein sample and the remaining protein samples based on the average level of the integral difference and the difference in the local fluctuation coefficient between any protein sample and the remaining protein samples; determine the fluctuation coefficient of the infrared spectrum data of each protein sample as the average value of all the absolute differences; Based on the fluctuation coefficient, the infrared spectrum data of all protein samples are screened to obtain each suspected impurity fluctuation data; each suspected impurity fluctuation data is equally divided into each segment; all trough points of the suspected impurity fluctuation data in each segment are obtained, and based on the discrete degree of transmittance of the trough point and the curvature of the trough point, and the difference between the wavelength corresponding to the minimum value of the curvature in the trough point and the wavelength corresponding to the minimum value of the transmittance, the abnormal index of the suspected impurity fluctuation data in each segment is determined; Based on the abnormal index, the spatial domain weight and the value range weight of each filter window in the bilateral filtering algorithm are corrected, the suspected impurity fluctuation data are denoised using the corrected bilateral filtering algorithm, and the infrared spectrum data other than the suspected impurity fluctuation data are denoised using the bilateral filtering algorithm without the correction; the protein quality is detected based on the denoising results of each protein sample; The calculation formula of the absolute difference is: ； In the formula, is the absolute difference between the rth protein sample and the tth protein sample, is the integral difference between the infrared spectrum data of the rth protein sample and the tth protein sample, is the horizontal coordinate length of the spectral curve corresponding to the infrared spectrum data of the rth protein sample, is the difference in the local fluctuation coefficient of the i-th wavelength between the r-th protein sample and the t-th protein sample, and I is the number of wavelengths in the infrared spectrum data of the r-th protein sample.

2. The protein quality intelligent detection method based on spectral analysis according to claim 1, characterized in that: The determination of the suspected impurity fluctuation data includes: A segmentation threshold is determined based on the fluctuation coefficients of all protein samples, and infrared spectrum data of protein samples whose fluctuation coefficients are greater than the segmentation threshold are used as suspected impurity fluctuation data.

3. The protein quality intelligent detection method based on spectral analysis according to claim 1, characterized in that: The determination of the abnormality index includes: Calculating the degree of dispersion of the transmittance of all the trough points in each segment, calculating the curvature of each trough point, and recording the difference between the wavelength corresponding to the minimum value of the curvature in all the trough points and the wavelength corresponding to the minimum value of the transmittance as the wavelength difference; determining the abnormality index based on the degree of dispersion, the minimum value of the curvature and the wavelength difference; The abnormal index is positively correlated with the discrete degree, and negatively correlated with the minimum value of the curvature and the wavelength difference.

4. The protein quality intelligent detection method based on spectral analysis according to claim 1, characterized in that: The modifying of the spatial domain weight and the range weight of each filter window in the bilateral filtering algorithm based on the abnormal index includes: For any suspected impurity fluctuation data, if the filtering window of the bilateral filtering algorithm spans the nth segment and the n+1th segment of the suspected impurity fluctuation data, the area ratio of the filtering window in the nth segment and the n+1th segment is calculated respectively, recorded as the first ratio and the second ratio, and the fusion result of the abnormal index of the nth segment and the first ratio, and the abnormal index of the n+1th segment and the second ratio is calculated; Taking the difference between the minimum value of the abnormal index of all segments of any suspected impurity fluctuation data and the fusion result as the proportional coefficient of the filter window; The products of the proportional coefficient and the original spatial domain weight and range weight of the filter window are respectively used as the modified spatial domain weight and range weight.

5. The protein quality intelligent detection method based on spectral analysis according to claim 1, characterized in that: The protein quality is detected based on the denoising result, including: The difference between the denoised infrared spectral data of each protein sample and the standard protein spectral data is analyzed to determine the difference coefficient of the infrared spectral data of each protein sample. If the difference coefficient of the protein sample is greater than a preset threshold, the protein sample is judged to be of unqualified quality; otherwise, the protein sample is judged to be of qualified quality.

6. The protein quality intelligent detection method based on spectral analysis according to claim 5, characterized in that: The coefficient of variation is calculated as follows: ； In the formula, is the coefficient of difference of the infrared spectrum data of the rth protein sample, is the transmittance of the ith wavelength of the infrared spectrum data of the rth protein sample, is the transmittance of the i-th wavelength of the standard protein spectrum data, and I is the number of wavelengths in the infrared spectrum data of the r-th protein sample.

7. A protein quality intelligent detection system based on spectral analysis, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

Images