A method for decolorizing and preserving the fragrance of sweet orange flower essential oil

Abstract

The present invention relates to the technical field of essential oil production and manufacturing, and in particular to a method for decolorizing and preserving fragrance of sweet orange flower essential oil, the method comprising: using a gas chromatography-mass spectrometry method to detect the content of the main components of the essential oil in the decolorizing and preserving fragrance stage, integrating the content data of each main component at different times to obtain the content sequence of each component; analyzing the component content sequence, obtaining the raw material pollution index of each data point caused by the uneven mixing of essential oils, performing correlation analysis and piecewise linear fitting between different component content sequences, and combining the raw material pollution index and LOF algorithm to monitor the decolorizing and preserving fragrance process in real time. Thus, the decolorizing and preserving fragrance of sweet orange flower essential oil is achieved, the comprehensive influence of normal mutations caused by reagent addition and changes in all sequences is evaluated, the recognition accuracy of the mutation of essential oil content components caused by abnormal reasons is improved, and the stability of the essential oil during the decolorizing and preserving fragrance process is ensured.

Description

Technical Field

[0001] The present application relates to the technical field of essential oil production and manufacturing, and in particular to a method for decolorizing and preserving the fragrance of sweet orange flower essential oil. Background Art

[0002] Orange blossom essential oil is a natural plant essential oil extracted from orange blossoms. It is widely used in aromatherapy to enhance mood, reduce stress and anxiety, and its fragrant aroma helps balance and relax the body and mind. In addition, orange blossom essential oil can also be used in skin care to relieve dry, sensitive or inflamed skin, and is also considered to have a sleep-inducing effect; its anti-depressant, anti-anxiety, anti-inflammatory and antioxidant properties make it highly valued in natural therapy. Orange blossom essential oil is usually extracted by water distillation, placing the petals of orange blossoms in a distiller and using steam to extract the essential oil from the plant material. In addition, solvent extraction is sometimes used to extract the essential oil using solvents such as ethanol, or the outer skin of orange blossoms is mechanically pressed by cold pressing to obtain the essential oil. In the extraction process, the key is to ensure the use of high-quality plant raw materials, appropriate extraction methods, and quality control to ensure the purity and efficacy of orange blossom essential oil.

[0003] The decolorization and aroma preservation process of sweet orange flower essential oil aims to improve its appearance, stability and aroma characteristics. Decolorization helps to remove pigments and impurities, improve the transparency and purity of essential oils, and improve their stability and extend the shelf life. Aroma preservation ensures that the original aromatic characteristics of essential oils are not lost during the processing process, prevents oxidation, and retains its authentic and lasting aroma. However, the changes in the dosage of decolorizers and aroma preservatives during the decolorization process will cause sudden changes in the content of certain components. In addition, since different batches of essential oils are mixed during decolorization and aroma preservation, different essential oils have different chemical compositions, volatility, concentration and other characteristics because the raw materials of different batches of essential oils are different, which will lead to unevenness in the mixing process. It will also cause sudden changes in the content of certain components, and this kind of mutation is a mutation caused by normal changes. Traditional anomaly detection algorithms, due to their lack of understanding of the scene, will confuse this normal mutation with abnormal mutations such as equipment abnormalities, causing a series of adverse consequences such as quality fluctuations, resource waste, system instability and quality damage. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a method for decolorizing and preserving the fragrance of sweet orange flower essential oil to solve the existing problems.

[0005] A method for decolorizing and preserving the fragrance of sweet orange flower essential oil of the present invention adopts the following technical scheme:

[0006] One embodiment of the present invention provides a method for decolorizing and preserving the fragrance of sweet orange flower essential oil, the method comprising the following steps:

[0007] Collect the content data of various main components in the sweet orange flower essential oil at each time during the decolorization and fragrance preservation process, and for each main component, use the sequence composed of the normalized values ​​of the content of the main component at all times as the component content sequence of the main component;

[0008] According to the changes of outliers in the component content sequence, each subsequence in the component content sequence is obtained; according to the changes of data in the frequency domain signal of the subsequence, the spectral density sequence of each subsequence is obtained; according to the changes of the neighborhood of each outlier in the component content sequence, the raw material contamination discrimination factor of each outlier in the component content sequence is obtained; according to the raw material contamination discrimination factor and the spectral density sequence of the subsequence, the raw material contamination index of each data point in the component content sequence is obtained; according to the correlation between each component content sequence, a correlation matrix is ​​constructed; the slope of the fitting straight line of all data points in each subsequence is used as the slope of the fitting straight line of each subsequence; according to the slope of the fitting straight line of each subsequence and the coefficient of variation of each row of data in the correlation matrix, the reagent interference influence characteristic value of the main component is obtained; according to the raw material contamination index and the reagent interference influence characteristic value, the abnormal discrimination confidence of each data point in the component content sequence is obtained; according to the abnormal discrimination confidence combined with the LOF algorithm, the local abnormal factor improvement value of each data point in the component content sequence is obtained;

[0009] Decolorization and fragrance preservation operation abnormality judgment is made based on the improved value of the local abnormal factor.

[0010] Preferably, the obtaining of each subsequence in the component content sequence according to the change of the outlier in the component content sequence is specifically: using each outlier in the component content sequence as a segmentation point to segment the component content sequence to obtain each subsequence.

[0011] Preferably, the obtaining of the frequency spectrum density sequence of each subsequence according to the data change in the frequency domain signal of the subsequence is specifically:

[0012] The frequency domain signal of each subsequence is obtained by the frequency domain analysis algorithm, and the spectrum density at each frequency in the frequency domain signal of each subsequence is calculated by the second-order difference discriminant method; the sequence composed of the spectrum density at all frequencies in the frequency domain signal of each subsequence is taken as the spectrum density sequence of each subsequence.

[0013] Preferably, the raw material contamination discrimination factor of each outlier in the component content sequence is obtained according to the neighborhood change of each outlier in the component content sequence, and the expression is:

[0014]

[0015] In the formula, Cf u is the raw material contamination discriminant factor of the u-th outlier in the Q-th component content sequence, e * is an exponential function with e as the base, N is the length of the neighborhood window of the outlier point, x u 、xu +p and x u -p are the times corresponding to the u-th outlier point, the p-th data point before the u-th outlier point, and the p-th data point after the u-th outlier point in the Q-th component content sequence, respectively. and are the x in the Qth component content sequence u Time, x u +p moment and x u -The element value corresponding to the time p, It is the mean of all data points from the center data point to the last data point in the neighborhood window of the u-th outlier point.

[0016] Preferably, the raw material pollution index of each data point in the component content sequence is obtained according to the raw material pollution discrimination factor and the spectral density sequence of the subsequence, specifically including:

[0017] In the Qth component content sequence, the troughs in each subsequence frequency domain signal and the peaks in each subsequence spectrum density sequence are obtained by the second-order difference discrimination method; the average value of the absolute value of the difference between the corresponding frequency values ​​of all two adjacent troughs in the uth subsequence frequency domain signal is calculated. Get the number of peaks B in the u-th subsequence spectral density sequence u ; Calculate the raw material pollution index Ri of the cth data point in the Qth component content sequence Q,c The expression is:

[0018]

[0019] In the formula, Cf u-1 , Cf u are the raw material contamination discriminant factors of the u-1th and uth outliers in the Qth component content sequence, respectively, e * is an exponential function with base e.

[0020] Preferably, the correlation matrix is ​​constructed according to the correlation between the content sequences of each component, specifically:

[0021] The Pearson correlation coefficient between the i-th and j-th component content sequences is taken as the element of the i-th row and j-th column; the elements in other positions are obtained by obtaining the elements of the i-th row and j-th column; and the matrix composed of all elements is taken as the correlation matrix.

[0022] Preferably, the step of obtaining the reagent interference effect characteristic value of the main component according to the slope of the fitted straight line of each subsequence and the coefficient of variation of each row of data in the correlation matrix specifically includes:

[0023] Get the data point at the same time as the u′th outlier in the i-th component content sequence in the j-th component content sequence; the expression for calculating the characteristic value Ht of the reagent interference effect is:

[0024]

[0025] In the formula, m is the number of main components, e * is an exponential function with base e, r i,j is the Pearson correlation coefficient between the i-th and j-th component content sequences, X i,j is the non-reagent synergy weight between the i-th and j-th component content sequences, Co i is the coefficient of variation of the data in the i-th row of the correlation matrix, L i is the number of outliers in the i-th component content sequence, A i,u′+1 is the slope of the fitted straight line of the u′+1th subsequence of the i-th component content sequence, A j,u″ is the slope of the fitted straight line of the subsequence of the data points at the same moment.

[0026] Preferably, the abnormality discrimination confidence is: the product of the reagent interference impact characteristic value and the raw material contamination index of the cth data point in the Qth component content sequence is used as the abnormality discrimination confidence of the cth data point in the Qth component content sequence.

[0027] Preferably, the method of obtaining the local anomaly factor improvement value of each data point in the component content sequence according to the anomaly discrimination confidence combined with the LOF algorithm specifically includes:

[0028] The local anomaly factor of each data point in the component content sequence is calculated by the LOF algorithm; the calculation result of an exponential function with a natural constant as the base and the inverse of the anomaly discrimination confidence of each data point as the exponent is calculated; and the product of the calculation result and the local anomaly factor is used as the improved value of the local anomaly factor of each data point.

[0029] Preferably, the abnormality judgment of the decolorization and fragrance preservation operation based on the improved value of the local abnormality factor is specifically as follows:

[0030] When the local anomaly factor improvement value of the data point of any component content sequence at the current moment is greater than or equal to the preset anomaly threshold, the bleaching and fragrance preservation operation at the current moment is abnormal; otherwise, the bleaching and fragrance preservation operation at the current moment is normal.

[0031] The present invention has at least the following beneficial effects:

[0032] The invention proposes a decolorization and fragrance preservation method for sweet orange flower essential oil. Aiming at the problem of normal mutation of component content caused by different raw materials, the invention constructs a raw material pollution index at each data point based on fast Fourier transform and neighborhood window analysis; aiming at the problem that normal mutation of the content of some components of sweet orange flower essential oil caused by the addition of a fragrance preservative and a decolorizing agent is considered as an outlier, the data change in each component content sequence is analyzed by density clustering and linear fitting, the reagent interference influence characteristic value in all sequences is constructed, and the comprehensive influence degree of normal mutation caused by reagent addition and change in all sequences is evaluated; and the local abnormal factor of each data point is improved in combination with the raw material pollution index, so as to ensure that the normal mutation caused by the reagent and the raw material is considered in the calculation process of the local abnormal factor, thereby enhancing the robustness and accuracy of the algorithm, so that the algorithm can more accurately identify the essential oil content component mutation caused by abnormal reasons in the decolorization and fragrance preservation process of the sweet orange flower essential oil, and ensure the stability of the essential oil in the decolorization and fragrance preservation process. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0034] Figure 1 A flow chart of a method for decolorizing and preserving the fragrance of sweet orange flower essential oil provided by the present invention;

[0035] Figure 2 The present invention is a schematic diagram of the steps of a method for decolorizing and preserving the fragrance of sweet orange flower essential oil. DETAILED DESCRIPTION

[0036] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the method for decolorizing and preserving the fragrance of sweet orange flower essential oil proposed by the present invention, 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.

[0037] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0038] The specific scheme of the method for decolorizing and preserving the fragrance of sweet orange flower essential oil provided by the present invention is described in detail below with reference to the accompanying drawings.

[0039] An embodiment of the present invention provides a method for decolorizing and preserving the fragrance of sweet orange flower essential oil.

[0040] Specifically, a method for decolorizing and preserving the fragrance of sweet orange flower essential oil is provided as follows. Figure 1 , the method comprises the following steps:

[0041] Step S001, using gas chromatography-mass spectrometry to detect the content of the main components of the essential oil in the bleaching and fragrance preservation stage, and integrating the content data of each main component at different times after pretreatment to obtain the content sequence of each component.

[0042] During the decolorization and fragrance preservation process of sweet orange flower essential oil, the main chemical components of the essential oil are analyzed by gas chromatography-mass spectrometry, wherein the main chemical components of the essential oil include: main volatile components (such as limonene, etc.), main ketone components (such as α-ionone, etc.), main aldehydes and alcohol components (such as linalool, etc.) and main aromatic ester components (such as neryl acetate, etc.). The main volatile components are one of the key factors affecting the aroma. By analyzing the relative content of the main volatile components, it is possible to evaluate whether the loss or change of volatile components has occurred during the decolorization and fragrance preservation process; the ketone components play a role in aroma regulation. Understanding the presence and content of these components can help determine whether the decolorization and fragrance preservation process affects the harmony and depth of the aroma; the aldehyde and alcohol components are of great significance for judging the mildness and fragrance properties of essential oils. By analyzing these components, it is possible to understand whether the essential oils have maintained their softness and fragrance properties during the decolorization and fragrance preservation process; the aromatic ester components play a key role in aroma contribution. Detecting the presence and relative content of these components is helpful in evaluating the overall quality and complexity of the aroma. It should be noted that the content of the components is the concentration of the components in sweet orange flower essential oil.

[0043] First, various main components in sweet orange flower essential oil are obtained by gas chromatography-mass spectrometer, assuming that there are m main components in total; then, at the beginning of the decolorization and fragrance-preserving operation of sweet orange flower essential oil, the processed essential oil is discharged by flow injection technology at a time interval of t, and the sample discharged each time is separated by a gas chromatographic column, and the content data of m main components in each sample are detected and obtained, so as to realize the real-time collection of sweet orange flower essential oil component content data. The content of the m main components collected in each sample is normalized by the Z-Score algorithm to eliminate the influence of dimension, and then the content of the Qth main component in all moments of the normalized sample is arranged in ascending order of time, and the obtained sequence is recorded as the component content sequence of the Qth main component, and the component content sequence of various main components is obtained by the acquisition mode of the component content sequence of the Qth main component, and the component content sequence of the Qth main component is recorded as the Qth component content sequence. Wherein, the gas chromatographic column and the Z-Score algorithm are known technologies, and the specific process is not repeated. It should be noted that the value of t can be set by the implementer. In this embodiment, the value of t is set to 1 second.

[0044] Step S002, analyzing the component content sequence, obtaining the raw material pollution index of each data point caused by the uneven mixing of essential oils, performing correlation analysis and piecewise linear fitting between different component content sequences, and improving the LOF algorithm in combination with the raw material pollution index.

[0045] Before the bleaching and fragrance preservation of the sweet orange flower essential oil, the preparation of the sweet orange flower essential oil is not prepared from the sweet orange flowers collected at the same time. Since the sweet orange flowers in different seasons may contain different chemical components, the sweet orange flowers in different seasons may contain different proportions of volatile components, ketone components, aldehydes and alcohol components and aromatic ester components, etc. When these seasonal components are mixed together, the relative content of each component in the mixed essential oil fluctuates at different times, and the sweet orange flowers collected in different seasons may have differences in growth environment, climate, etc., resulting in heterogeneity in the mixed essential oil. The heterogeneity will make the chemical composition of the mixed essential oil unevenly distributed, and some regional outliers will appear.

[0046] Due to the unevenness of mixed essential oils, some components may undergo large concentration changes in a short period of time, and abnormalities due to other reasons are mostly chaotic and disordered fluctuations after mutations. The abnormalities caused by the unevenness of mixed essential oils are long-term stable trend fluctuations after mutations caused by volatilization or reaction with reagents at the uneven mixing point. It should be noted that the abnormalities of mixed essential oils may mainly affect certain specific components, and have no or very weak effects on other components.

[0047] Taking the Qth component content sequence as an example, the component content sequence is input into the CURE density clustering algorithm, the data points in the sequence are clustered, and the outliers in the component content sequence are output. Each outlier in the component content sequence is used as a segmentation point to segment the component content sequence, that is, the sequence between the first data point and the first outlier in the component content sequence is used as the first subsequence of the component content sequence, and the sequence between the first and second outliers is used as the second subsequence of the component content sequence. And so on, the sequence between the last outlier and the last data point of the component content sequence is used as the last subsequence to obtain each subsequence in the component content sequence. Taking the uth subsequence as input, the fast Fourier transform (FFT) algorithm is used to convert the uth subsequence from a time series signal to a frequency domain signal; taking the frequency domain signal of the uth subsequence as input, the second-order difference discrimination method is used to perform trough judgment, and the trough in the frequency domain signal is obtained, and the frequency value corresponding to each trough is recorded, and the average value of the absolute value of the difference between all two adjacent trough frequency values ​​is calculated. In the frequency domain signal of the u-th subsequence, the spectrum density at each frequency is calculated, and the spectrum densities at all frequencies are arranged in ascending order according to the corresponding frequency. The composed sequence is recorded as the spectrum density sequence of the u-th subsequence. The spectrum density sequence is used as the input of the second-order difference discrimination method for peak judgment, and the number of peaks in the spectrum density sequence is obtained. u Among them, the fast Fourier transform algorithm, the second-order difference discrimination method and the spectrum density calculation are all well-known technologies, and the specific process will not be repeated here.

[0048] Take the uth outlier in the component content sequence as the central data point, construct a neighborhood window of size 1×N, and calculate the mean of all data points from the central data point to the last data point in the window It should be noted that the value of N can be set by the implementer. In this embodiment, the value of N is set to 5. The CURE density clustering algorithm is a well-known technology, and the specific process is not repeated. Taking the cth data point in the Qth component content sequence as an example, assuming that the cth data point is in the uth subsequence of the Qth component content sequence, the expression for calculating the raw material pollution index of each data point in each component content sequence is:

[0049]

[0050] In the formula, Ri Q,c is the raw material contamination index of the cth data point in the Qth component content sequence, Cf u-1 , Cf u are the raw material contamination discriminant factors of the u-1th and uth outliers in the Qth component content sequence, respectively, e * is an exponential function with base e, is the average value of the absolute value of the difference between all two adjacent trough frequency values ​​in the frequency domain signal of the u-th subsequence, B u is the number of peaks in the frequency spectrum density sequence of the u-th subsequence, N is the length of the neighborhood window of each outlier, x u 、x u +p and x u -p are the times corresponding to the u-th outlier point, the p-th data point before the u-th outlier point, and the p-th data point after the u-th outlier point in the Q-th component content sequence, respectively. and are the x in the Qth component content sequence u Time, x u +p moment and x u -The element value corresponding to the time p, It is the mean of all data points from the center data point to the last data point in the neighborhood window of the u-th outlier point.

[0051] From the above analysis, we can see that Cf u By taking the endpoint of the u-th subsequence as the window center, The difference between the corresponding positions on both sides of the window is judged to determine the probability that this type of mutation is caused by material contamination. The larger the value, the greater the difference between the two sides centered on the endpoint, that is, the mutation generated is a situation where the value after the mutation changes greatly from the value before the mutation, rather than the situation where the value after the mutation only returns to the size range before the mutation at the center of the window. In this case, the greater the probability of the mutation being caused by material contamination, Cf u The bigger, Ri Q,c It increases accordingly; It reflects the difference between the value after the mutation and the value at the mutation point, that is, the difference between the stable state of the component content after the mutation and the mutation point. The larger the value, the smaller the connection between the stable trend of the component content and the mutation point. At this time, the center of the window is less likely to be a mutation point caused by material contamination. u The smaller the Ri Q,c decreases accordingly; Represents the average width of all peaks in the subsequence frequency domain signal. In the frequency domain signal, a stable signal has a narrower width, while an unstable or fluctuating signal has a wider width; B u represents the number of peaks in the spectral density sequence. The more unstable the signal, the more dispersed the energy is or the more frequencies are distributed. In this case, the spectral density sequence has more peaks. When the value is larger, it is less consistent with the long-term stable trend fluctuation after the mutation caused by volatilization or reaction with reagents in the uneven mixing analyzed above. At this time, the smaller the raw material pollution index, Ri, is given to the data point Q,cThe smaller.

[0052] During the decolorization and fragrance preservation process of sweet orange blossom essential oil, if the processing conditions change, such as improper addition of decolorizing agents or fragrance preservatives, it will lead to a sudden change in the amount of normal ingredients, which will be mistakenly identified as an outlier by the algorithm. For example, the selected reagents may cause changes in the extraction efficiency of volatile components, ketone components, aldehydes and alcohol components, aromatic ester components, etc. If the amount of reagents used is changed during the bleaching and fragrance preservation process, the number of normal components will mutate suddenly and be misjudged as abnormal by the algorithm. In addition, changes in temperature during the processing process will affect the volatility and relative content of various components in the sweet orange flower essential oil. If the temperature changes during processing, the concentration of normal components will volatilize abnormally, resulting in abnormal values ​​in the detected data. The abnormal changes in the content of the detected components due to changes in the amount of reagent added or the temperature are all normal changes caused by changes in processing conditions, rather than problems caused by the quality of the sweet orange flower essential oil itself or the raw materials. However, due to the lack of understanding of normal changes or insensitivity to changes in operating conditions by the abnormal component content detection algorithm, the normal changes caused by changes in processing conditions will be misjudged as abnormal changes, resulting in a series of adverse consequences such as decreased production efficiency, product quality fluctuations, waste of resources, system instability and quality damage.

[0053] Regarding the abnormal changes in the content of ingredients caused by changes in the amount of reagents such as flavoring agents and decolorizing agents required for processing, different reagents have different extraction efficiencies, that is, reagents can more effectively extract specific ingredients from raw materials. Therefore, changes in the amount of reagents added will directly affect the content of different ingredients in the extraction process. If the amount of reagents added changes, it will directly affect the extraction efficiency of the main ingredients, resulting in an increase or decrease in their content, that is, the changes between different main ingredients due to changes in the amount of reagents added are usually presented as real-time correlations. For example, when ethanol is used as a decolorizing agent to remove color components from plant extracts, ethanol will increase the solubility of volatile oils, prompting more volatile compounds to be released from plant materials into the extract, resulting in an increase in the content of volatile oils, such as terpenoids, alcohols, aldehydes, etc., while the increase in ethanol leads to a decrease in the relative concentration of water-soluble compounds, resulting in a relative decrease in the content of some water-soluble components, such as phenols and phenolic alcohols in the extract.

[0054] The component content sequences of the i-th and j-th main components of sweet orange flower essential oil were analyzed, and the Pearson correlation coefficient r between the two sequences was calculated. i,j , where the Pearson correlation coefficient is a well-known technique, and the specific process is not repeated here. The Pearson correlation coefficient r between the i-th and j-th component content sequences is i,jAs the element value of the i-th row and j-th column in the correlation matrix, the element values ​​of other positions in the correlation matrix are obtained by obtaining the element value, thereby constructing a correlation matrix of size m×m. Calculate the coefficient of variation Co of the i-th row data in the correlation matrix i , where the coefficient of variation is a well-known technique, and the specific process will not be repeated here.

[0055] Get the slope of the fitted line for each subsequence. Specifically, input all data points in each subsequence into the straight line fitting algorithm for fitting, and use the slope of the fitted line obtained as the slope of the fitted line for each subsequence. Get the u′th outlier point D in the i-th component content sequence i,u′ The data points at the same time in the other component content series, among which the outlier point D i,u′ The data point at the same time in the jth component content sequence is denoted as D j,u″ . Get the slope of the fitted line A of the u′+1th subsequence in the i-th component content sequence i,u′+1 ; Get the data point D in the jth component content sequence j,u″ The slope of the fitted line of the subsequence A j,u″ Among them, if the data point D j,u″ is an outlier in the jth component content sequence. Assuming that the outlier is the μth outlier in the jth component content sequence, the slope of the fitting line of the μ+1th subsequence in the jth component content sequence is recorded as A j,u″ .

[0056] Taking the cth data point in the Qth component content sequence as an example, calculate the abnormality discrimination confidence of each data point in each component content sequence:

[0057] Rm Q,c =Ht×Ri Q,c

[0058]

[0059] In the formula, Rm Q,c is the abnormality discrimination confidence of the cth data point in the Qth component content sequence, Ht is the characteristic value of reagent interference, Ri Q,c is the raw material pollution index of the cth data point in the Qth component content sequence, m is the number of main component types, e * is an exponential function with base e, r i,j is the Pearson correlation coefficient between the i-th and j-th component content sequences, X i,j is the non-reagent synergy weight between the i-th and j-th component content sequences, Co i is the coefficient of variation of the data in the i-th row of the correlation matrix, L iis the number of outliers in the i-th component content sequence, A i,u′+1 is the slope of the fitted straight line of the u′+1th subsequence of the i-th component content sequence, A j,u″ is the u′th outlier point D in the i-th component content sequence i,u′ The slope of the fitted straight line corresponding to the subsequence of the data points at the same time in the j-th component content sequence.

[0060] X i,j The outliers of the i-th component content sequence are used to judge the component content mutations caused by various reasons, and the simultaneous data points in each component content sequence are determined based on the mutation time of the i-th component content sequence. The slope of the fitted straight line of the subsequence between two adjacent outliers is analyzed. The larger the difference in the slope of the fitted straight line of the subsequences corresponding to the i-th and j-th component content sequences, the greater the difference in the slope of the fitted straight line of the subsequences corresponding to the i-th and j-th component content sequences. The larger the value, the greater the difference in the degree of change and the change law of the two components under the influence of the same mutation, that is, the smaller the possibility that the component change in this subsequence is caused by the influence of the reagent, and the non-reagent synergy weight X i,j The larger the value, the smaller the weight given to the calculation of the interference degree of subsequent reagents. The smaller the Ht, the greater the Rm Q,c The correlation r i,j represents the degree of synergy between the i-th and j-th component content sequences. The closer its value is to 1, the stronger the real-time correlation between the data changes in the two component content sequences caused by the reagent changes. i,j -1|The smaller, Rm Q,c As it increases, Co i It represents the overall correlation change between the content of the ith component and the contents of all other components. The larger its value is, the more unstable the correlation distribution between the content of the ith component and the contents of all other components is, that is, the possibility that the change in the content of the ith component and the change in the content of all other components are caused by the change of the same reagent is lower. At this time, the interference of reagent changes on abnormal detection is smaller, Ht is smaller, and Rm Q,c Then it decreases; finally combined with Ri Q,c The influence of raw materials on the data points is reflected in Ht as the reagent influence weight. After multiplying the two, it can be shown that the cth data point in the Qth component content sequence is affected by raw materials and reagents. The larger the value, the greater the degree of normal mutation of the data point, that is, the greater the confidence that it will be misjudged as an abnormal mutation.

[0061] Afterwards, each component content sequence is used as input, and the LOF local outlier factor algorithm is used to detect anomalies in the component content sequence to obtain the local anomaly factor of each data point in each component content sequence, wherein the LOF local outlier factor algorithm is a well-known technology, and the specific process is not repeated. However, due to changes in the amount of reagents added and material contamination, the component content that belongs to the normal range will cause mutations, but because the LOF algorithm cannot correctly determine whether the mutation is a normal range mutation, it will cause these normal range mutations to be misjudged as abnormal mutations. Abnormal mutations include component content mutations caused by machine failures or abnormal reagent additions. To address this problem, taking the cth data point in the Qth component content sequence as an example, the local anomaly factor of each data point in each component content sequence calculated by the LOF algorithm is improved:

[0062]

[0063] Where LOF′ Q,c is the improved value of the local anomaly factor of the cth data point in the Qth component content sequence, e * is an exponential function with base e, Rm Q,c is the abnormality discrimination confidence of the cth data point in the Qth component content sequence, LOF Q,c It is the local anomaly factor of the cth data point in the Qth component content sequence calculated by the original algorithm.

[0064] The improved LOF algorithm is used to detect anomalies in all component content sequences, so that the algorithm can introduce reagent addition and material anomalies to redefine and calculate the local anomaly factor of each data point. When the raw material pollution index of the data point is larger and the reagent interference influence characteristic value is larger, the data point is given a lower local anomaly factor to reduce the misjudgment of normal mutations by the LOF algorithm. Finally, the improved LOF algorithm is applied to the decolorization and fragrance preservation process of sweet orange flower essential oil, and a real-time detection system is used to detect anomalies in the content of each component in the decolorization and fragrance preservation process to ensure that the product quality and aroma characteristics meet the standards and ensure that the operation can proceed smoothly.

[0065] Step S003, real-time monitoring of the decolorization and fragrance preservation process based on the improved LOF algorithm.

[0066] An abnormal threshold is set. When the local abnormal factor improvement value of the data point of any component content sequence at the current moment is greater than or equal to the preset threshold, it indicates that the bleaching and fragrance preservation operation at the current moment is abnormal. At this time, an alarm is issued to notify the relevant processing personnel to adjust the bleaching and fragrance preservation operation; when the local abnormal factor improvement value of the data point of all component content sequences at the current moment is less than the preset threshold, it indicates that the bleaching and fragrance preservation operation at the current moment is normal, and the subsequent process continues. It should be noted that the implementer of the abnormal threshold can set it by himself. In this embodiment, the abnormal threshold is set to 0.5.

[0067] After the preparation and extraction of the sweet orange flower essential oil, the sweet orange flower essential oil is subjected to a decolorization and fragrance preservation step. The decolorization and fragrance preservation process uses an integrated device. The original flower essential oil that has been prepared and extracted from the sweet orange flower is placed in the equipment, and a decolorizer and a fragrance preservative are added to mix evenly. The mixture is stirred at an appropriate temperature to promote the combination of the decolorizer with the impurities in the flower essential oil, ensure the volatility of the fragrance, and detect the content of the main components of the essential oil. During the processing, the improved LOF algorithm is used to perform abnormal detection on the content sequence of each component in the sweet orange flower essential oil, so as to control the accuracy of the decolorization and fragrance preservation operation in real time and achieve the desired decolorization effect, so that the equipment can issue an alarm in time according to the abnormal situation of the fragrance content, and make adjustments and responses. After the decolorization process, the decolorizer and residue are separated by filtration and other methods, and a washing step is performed to eliminate the residue. The schematic diagram of the steps of the above method is shown as follows. Figure 2 shown.

[0068] In summary, the embodiments of the present invention aim at the problem of normal mutation of component content caused by different raw materials, and construct a raw material pollution index at each data point based on fast Fourier transform and neighborhood window analysis; aim at the problem that normal mutation of the content of some components of sweet orange flower essential oil caused by the addition of flavor preservatives and decolorizing agents is considered as outliers, and the data changes in each component content sequence are analyzed by density clustering and straight line fitting, and the reagent interference influence characteristic values ​​in all sequences are constructed, and the comprehensive influence degree of normal mutation caused by reagent addition and change in all sequences is evaluated; and the local abnormal factor of each data point is improved in combination with the raw material pollution index, so as to ensure that the normal mutation caused by reagent reasons and raw material reasons is taken into account in the calculation process of the local abnormal factor, thereby enhancing the robustness and accuracy of the algorithm, so that the algorithm can more accurately identify the essential oil content component mutation caused by abnormal reasons in the decolorization and fragrance preservation process of sweet orange flower essential oil, and ensure the stability of the essential oil in the decolorization and fragrance preservation process.

[0069] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The 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.

[0070] 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.

[0071] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Modifications to the technical solutions recorded in the aforementioned embodiments, or equivalent replacement of some of the technical features therein, do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for decolorizing and preserving the fragrance of sweet orange flower essential oil, characterized in that: The method comprises the following steps: Collect the content data of various main components in the sweet orange flower essential oil at each time during the decolorization and fragrance preservation process, and for each main component, use the sequence composed of the normalized values ​​of the content of the main component at all times as the component content sequence of the main component; According to the changes of outliers in the component content sequence, each subsequence in the component content sequence is obtained; according to the changes of data in the frequency domain signal of the subsequence, the spectral density sequence of each subsequence is obtained; according to the changes of the neighborhood of each outlier in the component content sequence, the raw material contamination discrimination factor of each outlier in the component content sequence is obtained; according to the raw material contamination discrimination factor and the spectral density sequence of the subsequence, the raw material contamination index of each data point in the component content sequence is obtained; according to the correlation between each component content sequence, a correlation matrix is ​​constructed; the slope of the fitting straight line of all data points in each subsequence is used as the slope of the fitting straight line of each subsequence; according to the slope of the fitting straight line of each subsequence and the coefficient of variation of each row of data in the correlation matrix, the reagent interference influence characteristic value of the main component is obtained; according to the raw material contamination index and the reagent interference influence characteristic value, the abnormal discrimination confidence of each data point in the component content sequence is obtained; according to the abnormal discrimination confidence combined with the LOF algorithm, the local abnormal factor improvement value of each data point in the component content sequence is obtained; Decolorization and fragrance preservation operation abnormality judgment is performed based on the improved value of the local abnormal factor; The step of obtaining each subsequence in the component content sequence according to the change of the outlier point in the component content sequence is specifically: using each outlier point in the component content sequence as a segmentation point to segment the component content sequence to obtain each subsequence; The frequency spectrum density sequence of each subsequence is obtained according to the data change in the frequency domain signal of the subsequence, specifically: The frequency domain signal of each subsequence is obtained by a frequency domain analysis algorithm, and the spectrum density at each frequency in the frequency domain signal of each subsequence is calculated by a second-order difference discriminant method; the sequence composed of the spectrum densities at all frequencies in the frequency domain signal of each subsequence is taken as the spectrum density sequence of each subsequence; The raw material contamination discrimination factor of each outlier in the component content sequence is obtained according to the neighborhood change of each outlier in the component content sequence, and the expression is: In the formula, Cf u is the raw material contamination discriminant factor of the u-th outlier in the Q-th component content sequence, e * is an exponential function with e as the base, N is the length of the neighborhood window of the outlier point, x u 、x u +p and x u -p are the times corresponding to the u-th outlier point, the p-th data point before the u-th outlier point, and the p-th data point after the u-th outlier point in the Q-th component content sequence, respectively. and are the x in the Qth component content sequence u Time, x u +p moment and x u -The element value corresponding to the time p, is the mean of all data points from the center data point to the last data point in the neighborhood window of the u-th outlier point; The raw material pollution index of each data point in the component content sequence is obtained according to the raw material pollution discrimination factor and the spectrum density sequence of the subsequence, specifically including: In the Qth component content sequence, the troughs in each subsequence frequency domain signal and the peaks in each subsequence spectrum density sequence are obtained by the second-order difference discrimination method; the average value of the absolute value of the difference between the corresponding frequency values ​​of all two adjacent troughs in the uth subsequence frequency domain signal is calculated. Get the number of peaks B in the u-th subsequence spectral density sequence u ; Calculate the raw material pollution index Ri of the cth data point in the Qth component content sequence Q,c The expression is: In the formula, Cf u-1 , Cf u are the raw material contamination discriminant factors of the u-1th and uth outliers in the Qth component content sequence, respectively, e * is an exponential function with base e; The correlation matrix is ​​constructed according to the correlation between the content sequences of each component, specifically: The Pearson correlation coefficient between the i-th and j-th component content sequences is used as the element of the i-th row and j-th column; the elements in other positions are obtained by obtaining the elements of the i-th row and j-th column; the matrix composed of all elements is used as the correlation matrix; The slope of the fitted straight line is: input all data points in each subsequence into a straight line fitting algorithm for fitting, and use the slope of the obtained fitted straight line as the slope of the fitted straight line of each subsequence; The step of obtaining the reagent interference effect characteristic value of the main component according to the slope of the fitted straight line of each subsequence and the coefficient of variation of each row of data in the correlation matrix specifically includes: Get the data point at the same time as the u′th outlier in the i-th component content sequence in the j-th component content sequence; the expression for calculating the characteristic value Ht of the reagent interference effect is: In the formula, m is the number of main components, e * is an exponential function with base e, r i,j is the Pearson correlation coefficient between the i-th and j-th component content sequences, X i,j is the non-reagent synergy weight between the i-th and j-th component content sequences, Co i is the coefficient of variation of the data in the i-th row of the correlation matrix, L i is the number of outliers in the i-th component content sequence, A i,u′+1 is the uth sequence of the i-th component content ′ +1 slope of the fitted line of the subsequence, A j,u″ is the slope of the fitted straight line of the subsequence of the data points at the same moment; The abnormality discrimination confidence is: the product of the reagent interference influence characteristic value and the raw material contamination index of the cth data point in the Qth component content sequence is used as the abnormality discrimination confidence of the cth data point in the Qth component content sequence; The method of obtaining the local anomaly factor improvement value of each data point in the component content sequence according to the anomaly discrimination confidence combined with the LOF algorithm specifically includes: The local anomaly factor of each data point in the component content sequence is calculated by the LOF algorithm; the calculation result of an exponential function with a natural constant as the base and the inverse of the anomaly discrimination confidence of each data point as the exponent is calculated; and the product of the calculation result and the local anomaly factor is used as the improved value of the local anomaly factor of each data point.

2. A method for decolorizing and preserving the fragrance of sweet orange flower essential oil as claimed in claim 1, characterized in that: The abnormal judgment of the decolorization and fragrance preservation operation based on the improved value of the local abnormal factor is specifically as follows: When the local anomaly factor improvement value of the data point of any component content sequence at the current moment is greater than or equal to the preset anomaly threshold, the bleaching and fragrance preservation operation at the current moment is abnormal; otherwise, the bleaching and fragrance preservation operation at the current moment is normal.

Images