Rice oil detection method based on Raman spectrum

Abstract

The invention relates to the technical field of spectrum detection, in particular to a Raman spectrum-based rice oil detection method, which comprises the following steps: extracting main and auxiliary peaks to construct a peak group structure, performing sliding comparison to output a conformity region, identifying jump to remove an abnormal signal, calibrating a peak position to generate a correction structure, and performing homing numbering to complete a detection scheme. According to the method, the feature matching rule is constructed by combining the peak position spacing, the intensity ratio and the wavenumber index, the recognition precision of the multi-peak structure in the spectrum is enhanced, non-continuous response signals are recognized and eliminated through derivative change trend and proportion deviation constraint, and the accuracy of stable response extraction is improved. Dynamic calibration and position adjustment of peak positions are realized by utilizing an offset trend sequence, the consistency of a peak group structure in an overall direction and the continuity of a local structure are ensured, a numbering system and structure attribution division is completed after homing sorting, and the integrity of feature expression in a complex sample and the orderliness and reliability of spectral data processing are improved.

Description

Technical Field

[0001] This invention relates to the field of spectroscopic detection technology, and in particular to a method for detecting rice bran oil based on Raman spectroscopy. Background Technology

[0002] The field of spectroscopic detection technology involves methods that utilize the interaction of electromagnetic waves in matter to identify and quantify the composition and structure of substances by analyzing the spectral characteristics of absorption, scattering, or emission. It is an important branch of modern analytical testing. This technology encompasses various detection methods, including atomic spectroscopy, molecular spectroscopy, Raman spectroscopy, infrared spectroscopy, and ultraviolet-visible spectroscopy, and is widely used in food safety, biomedicine, environmental monitoring, and chemical analysis. Raman spectroscopy, in particular, has attracted significant attention due to its minimal sample damage, lack of complex pretreatment, and ability to achieve rapid and non-destructive detection, making it especially suitable for the qualitative and quantitative analysis of complex organic systems. Traditional methods for detecting rice bran oil refer to the detection operations performed on technical matters such as component content, safety indicators, or adulteration in rice bran oil. Commonly used methods include liquid chromatography, gas chromatography, mass spectrometry, and infrared spectroscopy. These methods generally require multiple steps, such as sample extraction, chemical reagent treatment, separation, and purification, to complete the analysis, making the process relatively cumbersome and involving some losses. The Raman spectroscopy-based method for detecting rice bran oil collects the spectral characteristics of rice bran oil in Raman scattering, and combines the information on the peak positions or intensity distribution of spectral characteristics to identify and discriminate components, thereby enabling the identification and monitoring of the properties or quality of rice bran oil.

[0003] Existing technologies generally rely on static peak position characteristics for spectral detection and lack effective analysis mechanisms for the continuity of band response morphology. When faced with sample types with complex information distribution between peak positions or frequent spectral signal perturbations, it is often impossible to distinguish between the target band and the interference response. Especially when there is a large shift in the signal or a sudden change in intensity gradient, the original methods are difficult to adapt to dynamic changes in feature recognition, leading to judgment distortion. For example, abnormal response interference is prone to occur in rice bran oil samples containing dopants or oxide residues, affecting the accuracy of component identification and the stability of data processing. Summary of the Invention

[0004] To address the technical problems existing in the prior art, this invention provides a method for detecting rice bran oil based on Raman spectroscopy, comprising the following steps: To achieve the above objectives, the present invention adopts the following technical solution: a method for detecting rice bran oil based on Raman spectroscopy, comprising the following steps: S1: Obtain the Raman spectrum of rice bran oil, extract the main peak and adjacent secondary peaks, record the peak position, full width at half maximum (FWHM), peak height ratio, and distance between adjacent peaks, establish a peak position sequence index according to wavenumber, and merge and construct a standard peak group structure sequence based on structural information; S2: Based on the standard peak group structure sequence, slide the extraction window along the spectrum to be measured, analyze the peak position interval, intensity ratio, and sorting sequence within the window, compare it with the corresponding position of the reference structure, set and determine whether the difference is within the preset range, and output the matching area. S3: For the main peak of the matching region, collect the signal derivative sequence, mark the slope jump point and calculate the difference, compare it with the difference of the adjacent band, and if the deviation exceeds the proportional deviation threshold, it is removed to obtain a set of stable response segments. S4: Generate an offset sequence based on the change in the distance between the peak position and the reference peak position of the set of stable response segments, determine whether the directions are consistent, if consistent, adjust the whole, if alternating, adjust according to the adjacent difference, and obtain the corrected peak position structure. S5: Call the corrected peak structure, classify it into the original spectral segment in ascending order of wavenumber, establish the correspondence between number and peak group structure, classify and merge the numbered blocks to form a rice bran oil detection scheme.

[0005] As a further aspect of the present invention, the standard peak group structure sequence includes the position of the main peak, the position of the secondary peak, the distance difference between peaks, the full width at half maximum (FWHM) value, and the peak height ratio. The matching region includes peak position interval matching information, peak intensity ratio matching information, band sorting sequence, and reference structure comparison results. The stable response segment set includes the main peak derivative change characteristics, slope jump point distribution, jump amplitude and natural difference ratio deviation, and effective bands after screening. The corrected peak position structure includes a unified offset direction, adjusted reference benchmark, offset trend sequence, and corrected peak position set. The rice bran oil detection scheme includes a peak position repositioning sequence, spectral response segment numbering information, peak group structure attribution relationship, and merging processing record.

[0006] As a further aspect of the present invention, the proportional deviation threshold refers to the allowable error range used to determine the difference in peak derivative jump points.

[0007] As a further aspect of the present invention, setting and judging whether the difference is within a preset range means judging whether it falls within a preset error range by comparing the differences between the peak group to be tested and the reference structure in terms of peak position, interval, and intensity.

[0008] As a further aspect of the present invention, the specific steps of S1 are as follows: S101: Obtain the response spectrum of the rice bran oil Raman detection sample, scan the continuously changing waveform signal on the frequency axis, extract the complete main peak and adjacent sub-peaks, mark their positions on the wavenumber axis, and generate a peak wavenumber position set; S102: Based on the peak wavenumber position set, extract the spectral intensity values ​​of the main peak and the secondary peak, extract the frequency range at half height, calculate the half height width, and perform a ratio calculation on the intensity values ​​of the main peak and the secondary peak to obtain the peak shape structure parameter set. S103: Call the peak wavenumber, half width at half maximum, peak height ratio and adjacent peak spacing in the peak structure parameter group, establish an index in wavenumber order, merge peak groups with similar structures according to the index, and construct a standard peak group structure sequence.

[0009] As a further aspect of the present invention, the specific steps of S2 are as follows: S201: Based on the standard peak group structure sequence, slide along the spectrum to be measured to extract equal-width windows in the wavenumber direction, and deconstruct the waveform signal in each window to extract the peak spacing, intensity ratio and peak order of multiple peaks in the window, and obtain the window waveform feature matrix; S202: Call the peak spacing, intensity ratio and sorting order in the window waveform feature matrix, compare them position by position with the same features in the standard peak group structure sequence, calculate the difference amplitude of the features, and compare the amplitude with the corresponding preset feature difference threshold item by item to obtain the feature difference judgment mark group; S203: Based on the judgment result in the mark group according to the feature difference, filter all window position indices that meet the judgment result range, record the corresponding wavenumber range information, and obtain the matching area.

[0010] As a further aspect of the present invention, the specific steps of S3 are as follows: S301: For the main peak position in the matching region, collect the numerical sequence of the corresponding signal intensity changing with the wavenumber, calculate the first derivative between adjacent wavenumber points, construct a continuous derivative sequence, and extract the jump point index according to the amplitude gradient of the derivative value change to obtain the slope jump position sequence. S302: Call the slope jump position sequence, calculate the signal strength difference between adjacent wavenumber positions before and after the jump point, extract the difference corresponding to the jump point, and perform a ratio operation with the strength difference in the natural change section of the adjacent waveband to obtain the jump difference ratio group; S303: Based on the proportional values ​​in the jump difference proportional group, compare them sequentially with the set proportional deviation threshold, mark the bands whose proportional values ​​exceed the proportional deviation threshold and remove them from the original band set to obtain a stable response segment set.

[0011] As a further aspect of the present invention, the specific steps of S4 are as follows: S401: Based on the wavenumber position of the main peak in the set of stable response segments and the reference peak position in the standard peak group structure sequence, calculate the wavenumber interval difference of each pair of corresponding peak positions one by one, and arrange the difference set in order to generate a peak position shift trend sequence. S402: Call the peak position offset trend sequence, statistically analyze the positive and negative distribution of all differences in the sequence, determine whether the change in a single direction is maintained in the whole segment, if all are differences in the same direction, record them as a unified offset mark, if there are alternating directions, compare the difference directions of adjacent items, and obtain the offset adjustment control signal set. S403: Based on the offset adjustment control signal set, the wavenumber coordinates of all peak positions in the stable response segment are corrected. If it is a uniform offset, a fixed value is added or subtracted uniformly. If it is an alternating offset, the peak position values ​​are adjusted according to the adjacent difference direction to generate the corrected peak position structure.

[0012] As a further aspect of the present invention, the specific steps of S5 are as follows: S501: Call up all peak information in the corrected peak structure, sort them in ascending order according to the wavenumber values ​​of the peaks, and sequentially assign multiple peaks to the corresponding original spectral response segments, and label the spectral segments to which the assigned peaks belong, and obtain the peak assignment mapping table. S502: Based on the response segment number information of each peak position in the peak position attribution mapping table, number and group them according to the peak group distribution structure, and mark the peak group sequence in the same group to construct the attribution record with the number as the index and generate the peak group number structure index table. S503: Based on the set of peak groups in the same numbered block in the peak group numbering structure index table, perform aggregation and merging operations on the peak position sequences in the numbered block, and combine the merged structure output results to form a rice bran oil detection scheme.

[0013] As a further aspect of the present invention, the spectral segment refers to a continuous interval in the original spectrum that carries a specific peak response, divided according to the wavenumber range, and identifies the position of the peak in the spectrum.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, feature matching rules are constructed by combining peak spacing, intensity ratio, and wavenumber index to enhance the recognition accuracy of multi-peak structures in the spectrum. Discontinuous response signals are identified and eliminated by the derivative change trend and proportional deviation constraint, thereby improving the accuracy of stable response extraction. The offset trend sequence is used to realize the dynamic calibration and position adjustment of peak positions, ensuring the consistency of peak group structure in the overall direction and the continuity of local structure. After sorting and repositioning, the numbering system and structural classification are completed, improving the integrity of feature expression in complex samples and the orderliness and reliability of spectral data processing. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the steps of the present invention; Figure 2 This is a detailed schematic diagram of S1 of the present invention; Figure 3 This is a detailed schematic diagram of S2 of the present invention; Figure 4 This is a detailed schematic diagram of S3 of the present invention; Figure 5 This is a detailed schematic diagram of S4 of the present invention; Figure 6 This is a detailed schematic diagram of S5 of the present invention; Figure 7 This is a schematic diagram of the data graph of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0019] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, their intended meanings are consistent. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, their intended meanings are consistent.

[0020] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0021] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0022] Please see Figure 1 This invention provides a method for detecting rice bran oil based on Raman spectroscopy, comprising the following steps: S1: Obtain the response spectrum of the rice bran oil Raman spectroscopy sample, extract the main peak and adjacent sub-peaks with complete waveforms and continuous morphology, record the peak position, full width at half maximum (FWHM), peak height ratio, and distance difference between adjacent peaks for each group of peaks, establish a peak position sequence index according to wavenumber order, call the structural information under the index to perform a merge judgment, and construct a standard peak group structure sequence; for details of the response spectrum, please refer to [link to relevant documentation]. Figure 7 ; S2: Based on the standard peak group structure sequence, the waveform within the window is extracted by sliding along the wavenumber direction of the spectrum to be measured. The peak position interval, intensity ratio and band sorting sequence between multiple peaks in the current window are analyzed. The data of the sequence and the corresponding position in the reference structure are compared. The difference is judged in turn to determine whether it meets the preset condition range and the matching area is output. S3: For the main peak in the matching region, collect the derivative sequence of signal intensity as a function of wavenumber, mark the slope jump point and calculate the signal difference before and after, analyze the proportional deviation between the jump difference value and the natural difference value of the adjacent band, if the deviation exceeds the set proportional deviation threshold, it is judged as an abnormal response and removed, and the filtered result is defined as the set of stable response segments. S4: Based on the change in the distance between the peak position and the reference structure peak position in the set of stable response segments, generate an offset trend sequence, determine whether the overall direction change of the sequence is consistent, if it is consistent, adjust the peak position uniformly according to the overall offset direction, if the offset direction alternates, adjust according to the adjacent difference direction, and generate the corrected peak position structure. S5: Call up all peak information in the corrected peak structure, and classify them into the original spectral response segments in order of increasing wave value. After corresponding classification, establish numbered records according to the peak group distribution structure, and classify and merge them according to the numbered blocks to form a rice bran oil detection scheme.

[0023] The standard peak group structure sequence includes the position of the main peak, the position of the secondary peak, the distance difference between peaks, the half-width at half-maximum (FWHM) value, and the peak height ratio. The matching region includes peak position interval matching information, peak intensity ratio matching information, band sorting sequence, and reference structure comparison results. The stable response segment set includes the main peak derivative change characteristics, slope jump point distribution, jump amplitude and natural difference ratio deviation, and the effective bands after screening. The corrected peak position structure includes unified offset direction, adjusted reference benchmark, offset trend sequence, and corrected peak position set. The rice bran oil detection scheme includes peak position repositioning sequence, spectral response segment numbering information, peak group structure affiliation relationship, and merging processing record.

[0024] Please see Figure 2 The specific steps of S1 are as follows: S101: Obtain the response spectrum of the rice bran oil Raman detection sample, scan the continuously changing waveform signal on the frequency axis, extract the complete main peak and adjacent sub-peaks, mark their positions on the wavenumber axis, and generate a peak wavenumber position set; When performing Raman spectroscopy on the sample, first prepare the rice bran oil sample and evenly drop it onto the surface of a clean and dry quartz glass slide, ensuring that the sample is spread evenly without air bubbles or thickness differences. Then, fix the glass slide on the sample stage of the Raman detector, set the excitation wavelength to 785nm, control the laser power to within 100mW to prevent sample ablation, initially set the integration time to 2 seconds to obtain a stable signal, and set the scanning range to 200 to 1800 cm⁻¹. -1 Step length set to 1cm -1 The light source is moved at a fixed frequency across the entire wavenumber range, and signals are collected sequentially. In the acquired two-dimensional spectrum, each wavenumber point corresponds to a light intensity value, forming a continuous waveform. The spectrum is then processed, and peak identification is performed. A sliding window with 21 points is used for scanning, and the maximum intensity value in each window is extracted as a peak candidate point. The criteria for determining whether a point is a main peak are: its intensity value must be greater than the light intensity of the five adjacent points, and this value must be greater than five times the average baseline intensity of the entire spectrum. The baseline intensity is determined by selecting a wavenumber of 200-400 cm⁻¹. -1 Intensity mean estimation within the featureless absorption region: if the mean is 40 counts, then the main peak identification threshold is 200 counts, and the secondary peak identification criterion is that the point falls within ±25 cm of the main peak. -1 Within the specified range, the intensity must be more than three times the average baseline intensity, i.e., 120 counts. The identified primary and secondary peaks are marked as valid peak points based on their specific positions on the wavenumber axis, ultimately generating a complete set of peak wavenumber positions. For example, if the primary peak is identified as being located at 1655 cm⁻¹... -1 The secondary peak is located at 1632cm. -1 and 1678cm -1 Then the wavenumber location set is {1632, 1655, 1678}.

[0025] S102: Based on the peak wavenumber position set, extract the spectral intensity values ​​of the main peak and the secondary peak, truncate the frequency range at half height, calculate the half height width, and perform a ratio calculation on the intensity values ​​of the main peak and the secondary peak to obtain the peak shape structure parameter set; Extract the spectral intensity information corresponding to each peak one by one. The extraction method is to locate the wavenumber position point and retrieve the light intensity value of the corresponding point in the original spectral data, such as the main peak at 1655 cm⁻¹. -1 The intensity at the peak is 1120 counts, and the secondary peak is 1632 cm⁻¹. -1 With 1678cm -1If the wavenumbers are 460 and 580 respectively, then the set of peak intensities is {1120, 460, 580}. Then, a half-width at half-maximum (WHM) operation is performed at each peak point. The WHM is defined as half the peak intensity. For example, if the main peak is 1120 counts, then the WHM is 560 counts. The first point with an intensity lower than 560 counts is found to the left and right of this wavenumber position, and its wavenumber value is recorded. The difference between the two wavenumber values ​​is the WHM of the main peak, for example, 1646 cm on the left. -1 The right side is 1663cm -1 The half-height and width are 17cm. -1 The same operation is performed on the secondary peaks to extract their full width at half maximum (FWHM). Then, the intensity ratio between the primary and secondary peaks is calculated by dividing the secondary peak intensity by the primary peak intensity, resulting in relative intensity ratios of 460 / 1120 = 0.41 and 580 / 1120 = 0.52, forming a set of peak structure parameters. If multiple secondary peaks are detected, they are sorted and filtered according to their distance from the primary peak and their intensity ratio. The effective secondary peak filtering threshold can be set to 0.25 times the primary peak intensity. This value is determined by reference... In actual testing, the background noise intensity determines the peak intensity. If the background noise is below 80 counts, the main peak intensity is usually greater than 1000 counts. Therefore, setting the threshold to 250 counts is more reasonable to filter out low-intensity false peaks. For example, if the secondary peak is 200 counts, its ratio is 200 / 1120=0.18. If it is below 0.25, it will not be included in the structural parameter group. Finally, the wavenumber, intensity, full width at half maximum, intensity ratio and other information of each main and secondary peak combination are extracted to form the peak structure parameter group.

[0026] S103: Call the peak wavenumber, half width at half maximum, peak height ratio and adjacent peak spacing in the peak structure parameter group, build an index in wavenumber order, merge peak groups with similar structures according to the index, and construct a standard peak group structure sequence. The structural parameters of each main peak and its sub-peaks are processed to extract their respective peak wavenumbers, full width at half maximum (FWHM), main-sub-peak intensity ratios, and inter-peak spacings. First, all peak wavenumbers are sorted in ascending order to create a numbered index table, for example: 1632cm. -1 Number 1, 1655cm -1 Number 2, 1678cm -1 Numbered 3, indexed sequentially, each main peak and its neighboring secondary peaks are merged. The criteria for determining whether a peak group with similar structure is based on three dimensions: (1) the wavenumber interval must be less than or equal to 25cm. -1 This value is based on the standard wavenumber drift tolerance of Raman equipment, which is ±5cm. -1 Set it to 5 times; (2) The difference between half height and width shall not exceed 3cm. -1The value is obtained by averaging the peak width differences of ten groups of samples; (3) The difference between the intensity ratios of the main peak and the secondary peak shall not be greater than 0.1. This threshold is set based on experience to ensure the stability of the correlation between the main peak and the secondary peak. For example, if the intensity ratios of the main peak and the secondary peak are 0.41 and 0.49 respectively, the difference is 0.08, which meets the merging condition. If a secondary peak meets the above three judgment criteria at the same time, it will be classified under the main peak and form a standard peak group structure sequence unit. When processing multiple main peaks and their secondary peak combinations, the parameter groups are traversed in turn. The above conditions are used to determine whether they should be merged into the same peak group. If they are not satisfied, another peak group sequence is started. Finally, each peak in each peak group is arranged in the order of wavenumber index to form a structure sequence. For example, if the peak group {1632, 1655, 1678} is identified, the standard peak group structure sequence is [{main 1655, secondary 1632, secondary 1678}]. All structure sequences maintain the original index correlation of the data for subsequent comparison or identification.

[0027] Please see Figure 3 The specific steps of S2 are as follows: S201: Based on the standard peak group structure sequence, slide along the spectrum to be measured to extract equal-width windows in the wavenumber direction, and deconstruct the waveform signal in each window to extract the peak spacing, intensity ratio and peak order of multiple peaks in the window, and obtain the window waveform feature matrix; First, the window width needs to be set according to the maximum wavenumber span in the standard peak group structure sequence. For example, the farthest distance between peaks in the standard peak group is 48cm. -1 Set the window width to 50cm -1 To fully cover the entire characteristic peak group structure, the sliding window starts at the position of the minimum wavenumber of the spectral data, for example, at 200 cm⁻¹. -1 Each time 5cm -1 To ensure the window continuously covers the entire spectral band, the step size slides to the right. All data points within the window are extracted, including all wavenumbers and their corresponding light intensity values ​​within the band. Within each window, the light intensity values ​​of adjacent wavenumber points are compared point-by-point to identify points with local maxima as candidate peaks. The peak identification criterion is that the intensity of this point is greater than the intensities of the two points before and after it, and the intensity value is not less than 1.5 times the average light intensity of the window region. For example, if the average light intensity of a window region is 200 counts, then the identification threshold is 300 counts. Valid peaks are extracted based on this condition, and their corresponding wavenumber positions are recorded. For example, if wavenumbers 265, 278, and 293 cm⁻¹ are identified within the window... -1 The three peaks were identified, and then the peak spacing was calculated by subtracting the previous peak from the next peak, resulting in a spacing value of 13cm. -1 With 15cm -1Simultaneously, the light intensity of each peak is recorded, such as 420, 500, and 545 counts. The intensity ratios between adjacent peaks are calculated as 500 / 420 and 545 / 500, which are 1.19 and 1.09 respectively. This order is recorded from left to right and labeled as sequence [1, 2, 3]. The three indicators extracted above—peak spacing, intensity ratio, and sorting order—are used as the waveform feature parameters of this window. The window waveform feature matrix is ​​constructed in sequence. The window is then slid to the next segment, and the above extraction, calculation, and recording steps are repeated until the entire spectrum is covered, finally obtaining the complete window waveform feature matrix.

[0028] S202: Call the peak spacing, intensity ratio and sorting order in the window waveform feature matrix, compare them position by position with the same features in the standard peak group structure sequence, calculate the difference amplitude of the features, and compare the amplitude with the corresponding preset feature difference threshold item by item to obtain the feature difference judgment mark group; First, each indicator in the standard peak group structure sequence is labeled with its serial number, and a one-to-one correspondence is established with the same type of indicator in the window feature matrix. Then, the peak spacing data extracted from each window is processed by difference. The difference is calculated by subtracting the actual spacing in the window from the corresponding spacing value in the standard peak group and taking the absolute value. For example, if the spacing between two peaks in the window is 13cm... -1 The corresponding spacing in the standard structure is 15cm. -1 The difference is 2cm -1 This difference is compared with a preset peak spacing difference threshold. If it is within the allowable error range, it is considered a match. The peak spacing difference threshold is set to ≤5cm. -1 This value is set with reference to the frequency drift range of the Raman instrument under normal conditions, and confirmed by statistical analysis of the maximum deviation after repeated sample testing. Then, the intensity ratio is compared item by item, and the difference between the window ratio and the standard ratio is calculated. For example, if the window ratio is 1.19 and the standard ratio is 1.15, the difference is 0.04. The intensity ratio difference threshold is set to ≤0.20, which is set based on the maximum light intensity fluctuation ratio in the experimental group samples and does not expand with sample changes, excluding abnormal fluctuation points. Finally, the sorting order is compared by directly comparing whether the numbering order in the two sequences is completely consistent. For example, if the standard is [1, 2, 3] and the window is [1, 2, 3], it is considered a match; otherwise, it is a mismatch. Each of the three comparisons is recorded as "1" if it meets the matching condition, and as "0" otherwise, forming a complete judgment mark group. For example, a window comparison result is: a spacing difference of 2cm. -1 → "1", intensity ratio difference 0.04 → "1", sorting order inconsistent → "0", then the final judgment label group of this window is [1, 1, 0]. Process all window feature matrices in this way to obtain the corresponding feature difference judgment label group.

[0029] S203: Based on the feature difference, determine the judgment result in the label group, filter all window position indices that meet the judgment result range, record the corresponding wavenumber range information, and obtain the matching area; First, the basic requirement for a valid window determination is that at least two of the three comparison dimensions must be "1". This determination threshold is set based on the matching intensity distribution analysis of structural peak groups in the sample. Since the peak spacing and intensity ratio are more affected by physical structure, and the sorting order is more affected by noise, any two of the three dimensions matching can be considered as a potential matching window. The specific operation is to read the judgment mark group of each window, count the number of "1"s, and if the result is 2 or 3, the window position is recorded as a matching window. For example, if the judgment mark group of window W8 is [1, 1, 0], it meets the standard. Record the window index value and its corresponding starting and ending wavenumber ranges. For example, if the window covers a band of 370cm... -1 Up to 420cm -1 The result is recorded as W8: 370–420cm -1 The same filtering and recording operations are performed on all windows in the entire spectrum. Finally, all window indices that meet the judgment criteria are sorted and collected to construct a set consisting of window indices and their corresponding wavenumber intervals. All window indices must maintain the same sequential numbering and record their respective wavenumber interval information to generate the final matching region.

[0030] Please see Figure 4 The specific steps of S3 are as follows: S301: For the main peak position in the matching region, collect the numerical sequence of the corresponding signal intensity as a function of wavenumber, calculate the first derivative between adjacent wavenumber points, construct a continuous derivative sequence, and extract the jump point index based on the amplitude gradient of the derivative value change to obtain the slope jump position sequence. First, locate the main peak wavenumber point within each matched band, and then extend a fixed number of wavenumber points forward and backward from this main peak to form a signal segment within a symmetrical range. A typical value is ±10 wavenumber points to ensure the complete waveform structure is included. For example, when the main peak wavenumber is 1655 cm⁻¹... -1 At that time, the cropping range was set to 1645cm. -1 Up to 1665cm -1 The corresponding wavenumber interval is 1cm. -1A total of 21 data points were collected, each corresponding to a Raman intensity value. For example, the constructed light intensity sequence is [400, 430, 470, 510, 550, 600, 660, 730, 800, 870, 940, 910, 880, 850, 810, 770, 730, 690, 650, 620, 590]. Then, the light intensities of any two adjacent points in this sequence are subtracted term by term to generate a first derivative sequence containing 20 elements. Its physical meaning is the rate of increase or decrease of light intensity per unit wavenumber. For example, 430 minus 400 equals 30, then 470 minus 430 equals 40, and so on. This derivative sequence records the slope trend change at each wavenumber point, in order to identify abrupt changes. The trend is observed by monitoring the magnitude of changes between adjacent derivative values ​​in the derivative sequence to determine the location of abrupt change points. For example, if two derivative values ​​change from +80 to -60, the gradient change is 140 counts, which serves as a reference value for gradient abrupt changes. A slope change threshold of 40 counts is set for identifying abrupt change points. This value is calculated based on the average of the maximum fluctuation range of the derivative sequence in real experimental samples, then weighted with a 20% offset. This serves as a critical standard for distinguishing between continuous and abrupt changes. Any two adjacent derivatives in the sequence whose difference exceeds this threshold are considered to have a slope abrupt change, and their position index in the original wavenumber sequence is recorded. For example, if the abrupt change occurs at the 11th point, the wavenumber is 1655 cm⁻¹. -1 After processing all the main peak bands in this way, a sequence of jump points indexed by wavenumber position is formed, and the slope jump position sequence is obtained.

[0031] S302: Call the slope jump position sequence, calculate the signal strength difference between adjacent wavenumber positions before and after the jump point, extract the difference corresponding to the jump point, and perform a ratio operation with the strength difference in the natural change section of the adjacent waveband to obtain the jump difference ratio group; Each transition point is traversed sequentially, and the wavenumber positions before and after the transition point are used as comparison boundaries. The corresponding light intensity values ​​are collected, and the light intensity difference is calculated as the numerical representation of the abrupt change signal. For example, if the transition point is located at 1655 cm⁻¹... -1 The previous point is 1654cm -1 The light intensity value is 940 counts, and the next point is 1656 cm. -1 The light intensity value is 910 counts, and the difference between the two points is 30 counts. This is the abrupt change in intensity at the jump point. To establish a reference standard for this difference, a naturally varying segment with relatively uniform light intensity fluctuations is selected within the continuous stable interval on both sides of this jump point. The segment length is within 5 wavenumber points, and the difference should not exceed 10 counts. (If at 1650cm...) -1 Up to 1654cm -1Within the range, the light intensity ranges from 870 to 940, per 1cm -1 The increment is approximately 14 counts. This segment is designated as the natural variation segment. The difference between two adjacent points is extracted as a representative value. For example, the first difference is 880 minus 870, which equals 10 counts. This 10 counts is then used as the natural variation value of the jump point. The judgment benchmark needs to consider the local average change amplitude and fluctuation range. The area within 2 times the standard deviation of the change fluctuation is set as the natural background change. This natural variation value will be compared with the abrupt change value to form the jump difference ratio of the jump point. Other jump points are processed, and the ratio of their intensity difference to their respective background change values ​​is recorded. Finally, a jump difference ratio group is formed, in which each item contains the jump wavenumber index and its corresponding difference ratio.

[0032] S303: Based on the proportional values ​​in the jump difference proportional group, compare them sequentially with the set proportional deviation threshold, mark the bands whose proportional values ​​exceed the proportional deviation threshold and remove them from the original band set to obtain a stable response segment set; Each proportional value was compared sequentially with a preset deviation threshold, which was set at 2.0. The criterion was as follows: in 30 randomly selected rice bran oil samples, the intensity difference and natural fluctuation ratio of all non-mutation regions were statistically analyzed. The results showed that the proportion of stable regions in more than 95% of cases was less than 1.8. Therefore, 2.0 was selected as the deviation threshold. All abrupt changes with proportions higher than this value were identified as unstable signals. The wavenumber position corresponding to the abrupt change was recorded and extended ±2cm before and after it. -1 The specified bands are used as markers for unstable bands. For example, a certain transition point ratio of 2.8 corresponds to a wavenumber of 1655cm. -1 The labeled band is 1653–1657 cm. -1 The band is removed from the original band index set. The proportion values ​​of the remaining transition points are checked in turn. The judgment, marking and removal process is repeated. After all the data is processed, the unmarked band segments are reorganized and merged according to their original index order, and finally a set of multiple continuous stable bands is formed to obtain the set of stable response segments.

[0033] Please see Figure 5 The specific steps of S4 are as follows: S401: Based on the wavenumber position of the main peak in the set of stable response segments and the position of the reference peak in the standard peak group structure sequence, calculate the wavenumber interval difference of each pair of corresponding peaks one by one, arrange the difference set in order, and generate the peak position shift trend sequence. First, arrange the two sets of peak position data in ascending order and establish a one-to-one correspondence. Let the reference peak position sequence of the standard peak group be [1632, 1655, 1678] cm. -1The main peak positions identified in the stable response range are [1633.7, 1656.2, 1679.1] cm. -1 After matching each pair, the wavenumber interval difference is calculated for each pair of matches. The calculation method is to subtract the reference peak position from the actual detected peak position, i.e., 1633.7–1632=+1.7, 1656.2–1655=+1.2, 1679.1–1678=+1.1. All differences retain the positive or negative sign and are not taken as absolute values. They are used to determine the overall offset direction. The resulting difference sequence is [+1.7, +1.2, +1.1]. This operation requires ensuring that each detected peak in the sample is within ±5cm of the reference peak. -1 Within a reasonable pairing tolerance, otherwise the point will not be included in the comparison; the reference range is ±5cm. -1 The settings are derived from the maximum Raman drift error range of the device combined with empirical values. All peak position differences that meet the range and are successfully paired are recorded and output strictly in the order of the reference peaks to form a set of differences describing the shift of the peak position of the detected sample relative to the standard structure. Finally, the set of differences is arranged in order to generate a peak position shift trend sequence.

[0034] S402: Call the peak position offset trend sequence, statistically analyze the positive and negative distribution of all differences in the sequence, determine whether the change in the whole segment is maintained in a single direction, if all are differences in the same direction, record them as a unified offset mark, if there are alternating directions, compare the difference directions of adjacent items, and obtain the offset adjustment control signal set. First, iterate through the difference sequence [+1.7, +1.2, +1.1]. All terms are positive and are recorded as "positive shift". If a similar sequence [+1.7, -0.8, +1.2] appears, indicating alternating positive and negative terms, then proceed with the alternation analysis process. Compare adjacent terms to see if their difference direction changes. For example, if +1.7 and -0.8 have different directions, record the direction change at the second term. Then compare -0.8 and +1.2; if the direction changes again, record the change at the third term. If two or more direction change points appear in the entire sequence, the peak shift trend is determined to be a non-uniform pattern, requiring local adjustment. Conversely, if there is no direction change or the change does not exceed one time, it is considered a uniform shift state. The standard for determining a uniform shift is: all differences are positive or all are negative, and the difference between the maximum and minimum differences does not exceed 1.0 cm. -1 This threshold is derived from the systematic error fluctuation range after the Raman spectrometer is calibrated, typically within ±0.5 cm⁻¹. -1 If the condition is met, the control signal is marked as "uniform offset"; otherwise, it is marked as "alternating offset". If it is alternating offset, all position indices where direction switching occurs are marked, and the corresponding peak wave number is recorded so that local adjustment operations can be performed later to finally obtain the offset adjustment control signal set.

[0035] S403: Based on the offset adjustment control signal set, the wavenumber coordinates of all peak positions in the stable response section are corrected. If the offset is uniform, a fixed value is added or subtracted uniformly. If the offset is alternating, the peak position values ​​are adjusted according to the direction of adjacent difference to generate the corrected peak position structure. Wavenumber coordinate correction is performed on all peak positions in the stable response range. If the previous step determined a uniform shift, the average value of the overall shift is first calculated as the correction reference value. For example, if the difference sequence is [+1.7, +1.2, +1.1], its average shift is (1.7+1.2+1.1) / 3=1.33cm. -1 Take an approximate value of +1.3cm. -1 As a uniform correction, all detection peaks in the stable response section will be uniformly reduced by 1.3cm. -1 The process involves adjusting the peak values ​​from 1633.7 to 1632.4, 1656.2 to 1654.9, and 1679.1 to 1677.8. All corrected peak values ​​are then rearranged according to their original indices to form the corrected peak structure. If the previous step determined an alternating shift, for example, a difference sequence of [+1.6, -1.2, +1.8], then no uniform processing is performed. Instead, each peak is corrected sequentially based on the direction of change between adjacent differences. The specific rule is: if the current difference is in the same direction as the previous difference, the current peak correction value remains consistent; if the direction is opposite, the current peak correction value is halved according to the direction of the previous correction value. For example, the first value, +1.6, is set as a correction of +1.6 cm. -1 The second term is -1.2, which is in the opposite direction, so the correction is -0.8cm. -1 If the third item changes direction again, the correction amount will be adjusted to +0.9cm. -1 The maximum correction range is set to ±2.0cm. -1 Based on the maximum average change range of frequency drift data statistically analyzed by the equipment within 30 minutes, an independent correction operation is performed on all peak positions, and the position changes before and after the correction are recorded in the structure table, finally generating the corrected peak position structure.

[0036] Please see Figure 6 The specific steps of S5 are as follows: S501: Call up all peak information in the corrected peak structure, sort them in ascending order according to the wavenumber values ​​of the peaks, and assign multiple peaks to the corresponding original spectral response segments in sequence. Then, label the spectral segments to which the assigned peaks belong and obtain the peak assignment mapping table. First, extract the wavenumber values ​​for each peak in the peak structure to form an initial wavenumber set. Sort this set in ascending order to ensure the sorting conforms to the Raman spectrum frequency axis arrangement from low to high wavenumbers. For example, the corrected peak positions are [1678.2, 1655.0, 1632.3] cm⁻¹. -1The sorted values ​​are [1632.3, 1655.0, 1678.2] cm. -1 After sorting, each peak is segmented. The original spectral response segments are divided into fixed-width, equally spaced segments, with each segment width set to 50cm. -1 The dividing point is set at 200cm. -1 The finish line is set at 1800cm. -1 The project is planned to be divided into 32 sections. The width of each section is derived from the previous summary and statistics of the distribution characteristics of standard peak groups, considering that the span of typical peak groups is concentrated in the range of 30-45cm. -1 To avoid truncating the peak group, a 50cm interval was selected. -1 To establish a reasonable upper limit, each sorted peak is then assigned to a specific segment based on its wavenumber. This is done by comparing the start and end boundary values ​​of each segment. If a peak's wavenumber falls within a segment (greater than or equal to the start value, less than the end value), it is assigned to that segment. For example, peak 1655.0 cm⁻¹ -1 It falls at (1650–1700) cm -1 Within a given segment, the peak position is assigned to spectral response segment number 30. The peak wavenumber is paired with the corresponding segment number, and the pairing relationship is recorded. After completing the segment assignment operation for all peak positions in sequence, a mapping data table of "peak wavenumber → response segment number" is constructed to record the assignment correspondence between the peak position and its original spectral segment, and finally, the peak position assignment mapping table is obtained.

[0037] S502: Based on the response segment number information of each peak in the peak attribution mapping table, number and group them according to the peak group distribution structure, and mark the peak group sequence in the same group to construct the attribution record with the number as the index and generate the peak group number structure index table. All peak positions are sorted in ascending order of their numerical values ​​and grouped to create a multi-level structure with the segment number as the primary index. Each number contains several assigned peak position data. Furthermore, within each group, the peak position sequences are again sorted in ascending order of their wave values ​​and numbered starting from 1. For example, segment number 29 contains peak positions 1632.3 cm and 1635.4 cm. -1 After sorting, they are marked as serial numbers 1 and 2 respectively. For the 30th segment, 1655.0 and 1659.1cm... -1Number them sequentially as 1 and 2, repeating this operation to ensure that the wavenumber sequence within each group is bound to its local position identifier. During this process, it should be noted that if there is only a single peak in a certain segment, the peak should still be assigned the number "1" for compatibility maintenance. After completing the numbering work within all groups, a structured record table is generated. Each record contains three types of information: "response segment number", "peak wavenumber contained in the segment", and "peak ranking number within the group". This structure facilitates the rapid location of peak group members and their relative positions in subsequent aggregation and merging operations. It can also serve as a unique index reference to realize the number-driven rice oil spectrum identification path record, and finally generate a peak group numbering structure index table.

[0038] S503: Based on the set of peak groups in the same numbered block in the peak group numbering structure index table, the peak position sequence in the numbered block is aggregated and merged, and combined with the merged structure output results, a rice bran oil detection scheme is formed. Each numbered block is read separately, and the set of peaks under that number is taken as the candidate aggregation object for the same segment. In each group, adjacent peaks are compared pairwise to determine if they meet the aggregation merging conditions. The aggregation judgment is based on the wavenumber spacing between peaks, with an upper limit of 25cm. -1 This threshold is derived from the statistical results of the average wavenumber difference between the main and secondary peaks in a large number of rice bran oil sample test data, combined with a reasonable safety upper limit considering the measurement accuracy deviation of the equipment. Specifically, if the wavenumber difference between two adjacent peaks is less than or equal to 25cm... -1 If so, they are grouped into a peak group, such as 1632.3 and 1635.4 cm in segment 29. -1 The spacing is 3.1cm. -1 If they are all peaks, they are grouped into the same peak group and assigned the number 29-1; if a third peak exists within the segment, such as 1649.8cm... -1 The distance between it and the previous peak is 14.4 cm. -1 It is also less than 25cm -1 If the distance between a peak and the previous peak is greater than 25cm, then the three peaks are grouped together; -1 If the peaks are not grouped together, they are grouped separately and numbered 29-2. Following this logic, the peak sequence within each segment is fully judged and grouped. All merged peak groups are combined according to their segment and group sequence order to form a complete peak group structure definition record. The record indicates the segment number, peak group number, and all wavenumber information contained in the peak group. After aggregation, all merged results are integrated and output as the final basis for constructing the characteristic peak group system of rice bran oil Raman spectrum, thus forming a rice bran oil detection scheme.

[0039] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for detecting rice bran oil based on Raman spectroscopy, characterized in that, Includes the following steps: S1: Obtain the Raman spectrum of rice bran oil, extract the main peak and adjacent secondary peaks, record the peak position, full width at half maximum (FWHM), peak height ratio, and distance between adjacent peaks, establish a peak position sequence index according to wavenumber, and merge and construct a standard peak group structure sequence based on structural information; S2: Based on the standard peak group structure sequence, slide the extraction window along the spectrum to be measured, analyze the peak position interval, intensity ratio, and sorting sequence within the window, compare it with the corresponding position of the standard peak group structure, set and determine whether the difference is within the preset range, and output the matching area. S3: For the main peak of the matching region, collect the signal derivative sequence, mark the slope jump point and calculate the difference, compare it with the difference of the adjacent band, and if the deviation exceeds the proportional deviation threshold, it is removed to obtain a set of stable response segments. S4: Generate an offset sequence based on the change in the distance between the peak position and the reference peak position of the set of stable response segments, determine whether the directions are consistent, if consistent, adjust the whole, if alternating, adjust according to the adjacent difference, and obtain the corrected peak position structure. S5: Call the corrected peak structure, classify it into the original spectral segment in ascending order of wavenumber, establish a numbering system corresponding to the standard peak group structure sequence, assign and merge the numbered blocks to form a rice bran oil detection scheme.

2. The method for detecting rice bran oil based on Raman spectroscopy according to claim 1, characterized in that, The standard peak group structure sequence includes the position of the main peak, the position of the secondary peak, the distance difference between peaks, the half-width at half-maximum (FWHM) value, and the peak height ratio. The matching region includes peak position interval matching information, peak intensity ratio matching information, band sorting sequence, and reference structure comparison results. The stable response segment set includes the main peak derivative change characteristics, slope jump point distribution, jump amplitude and natural difference ratio deviation, and effective bands after screening. The corrected peak position structure includes unified offset direction, adjusted reference benchmark, offset trend sequence, and corrected peak position set. The rice bran oil detection scheme includes peak position repositioning sequence, spectral response segment numbering information, peak group structure affiliation relationship, and merging processing record.

3. The method for detecting rice bran oil based on Raman spectroscopy according to claim 1, characterized in that, The proportional deviation threshold refers to the allowable error range used to determine the difference in peak derivative jump points.

4. The method for detecting rice bran oil based on Raman spectroscopy according to claim 1, characterized in that, The setting and judging whether the difference is within the preset range refers to judging whether it falls within the preset error range by comparing the differences between the peak group to be tested and the reference structure in terms of peak position, interval, and intensity.

5. The method for detecting rice bran oil based on Raman spectroscopy according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Obtain the response spectrum of the rice bran oil Raman detection sample, scan the continuously changing waveform signal on the frequency axis, extract the complete main peak and adjacent sub-peaks, mark their positions on the wavenumber axis, and generate a peak wavenumber position set; S102: Based on the peak wavenumber position set, extract the spectral intensity values ​​of the main peak and the secondary peak, extract the frequency range at half height, calculate the half height width, and perform a ratio calculation on the intensity values ​​of the main peak and the secondary peak to obtain the peak shape structure parameter set. S103: Call the peak wavenumber, half width at half maximum, peak height ratio and adjacent peak spacing in the peak structure parameter group, establish an index in wavenumber order, merge peak groups with similar structures according to the index, and construct a standard peak group structure sequence.

6. The method for detecting rice bran oil based on Raman spectroscopy according to claim 1, characterized in that, The specific steps of S2 are as follows: S201: Based on the standard peak group structure sequence, slide along the spectrum to be measured to extract equal-width windows in the wavenumber direction, and deconstruct the waveform signal in each window to extract the peak spacing, intensity ratio and peak order of multiple peaks in the window, and obtain the window waveform feature matrix; S202: Call the peak spacing, intensity ratio and sorting order in the window waveform feature matrix, compare them position by position with the same features in the standard peak group structure sequence, calculate the difference amplitude of the features, and compare the amplitude with the corresponding preset feature difference threshold item by item to obtain the feature difference judgment mark group; S203: Based on the judgment result in the mark group according to the feature difference, filter all window position indices that meet the judgment result range, record the corresponding wavenumber range information, and obtain the matching area.

7. The method for detecting rice bran oil based on Raman spectroscopy according to claim 1, characterized in that, The specific steps for S3 are as follows: S301: For the main peak position in the matching region, collect the numerical sequence of the corresponding signal intensity changing with the wavenumber, calculate the first derivative between adjacent wavenumber points, construct a continuous derivative sequence, and extract the jump point index according to the amplitude gradient of the derivative value change to obtain the slope jump position sequence. S302: Call the slope jump position sequence, calculate the signal strength difference between adjacent wavenumber positions before and after the jump point, extract the difference corresponding to the jump point, and perform a ratio operation with the strength difference in the natural change section of the adjacent waveband to obtain the jump difference ratio group; S303: Based on the proportional values ​​in the jump difference proportional group, compare them sequentially with the set proportional deviation threshold, mark the bands whose proportional values ​​exceed the proportional deviation threshold and remove them from the original band set to obtain a stable response segment set.

8. The method for detecting rice bran oil based on Raman spectroscopy according to claim 1, characterized in that, The specific steps of S4 are as follows: S401: Based on the wavenumber position of the main peak in the set of stable response segments and the reference peak position in the standard peak group structure sequence, calculate the wavenumber interval difference of each pair of corresponding peak positions one by one, and arrange the difference set in order to generate a peak position shift trend sequence. S402: Call the peak position offset trend sequence, statistically analyze the positive and negative distribution of all differences in the sequence, determine whether the change in a single direction is maintained in the whole segment, if all are differences in the same direction, record them as a unified offset mark, if there are alternating directions, compare the difference directions of adjacent items, and obtain the offset adjustment control signal set. S403: Based on the offset adjustment control signal set, the wavenumber coordinates of all peak positions in the stable response segment are corrected. If it is a uniform offset, a fixed value is added or subtracted uniformly. If it is an alternating offset, the peak position values ​​are adjusted according to the adjacent difference direction to generate the corrected peak position structure.

9. The method for detecting rice bran oil based on Raman spectroscopy according to claim 1, characterized in that, The specific steps of S5 are as follows: S501: Call up all peak information in the corrected peak structure, sort them in ascending order according to the wavenumber values ​​of the peaks, and sequentially assign multiple peaks to the corresponding original spectral response segments, and label the spectral segments to which the assigned peaks belong, and obtain the peak assignment mapping table. S502: Based on the response segment number information of each peak position in the peak position attribution mapping table, number and group them according to the peak group distribution structure, and mark the peak group sequence in the same group to construct the attribution record with the number as the index and generate the peak group number structure index table. S503: Based on the set of peak groups in the same numbered block in the peak group numbering structure index table, perform aggregation and merging operations on the peak position sequences in the numbered block, and combine the merged structure output results to form a rice bran oil detection scheme.

10. The method for detecting rice bran oil based on Raman spectroscopy according to claim 1, characterized in that, The spectral segment refers to the continuous interval in the original spectrum that carries the peak response, divided according to the wavenumber range, and identifies the position of the peak in the spectrum.

Images