A process for the manufacture of a chicory preparation

Abstract

The present application relates to the technical field of data processing, and especially relates to a process detection method for chicory preparation manufacturing process, which obtains concentrated liquid concentration data sequence and its corresponding chicory acid yield data sequence in the manufacturing process of chicory preparation under each ethanol concentration; according to the data change characteristics in each concentrated liquid concentration data sequence, each concentrated liquid concentration data sequence is divided into at least two stage sequences; each concentrated liquid concentration data sequence is divided into at least two subsequences respectively, according to the data difference of the subsequence contained in each stage sequence in each concentrated liquid concentration data sequence, the fluctuation characteristics of the concentrated liquid concentration data in each stage sequence, and the similarity degree between each concentrated liquid concentration data sequence and its corresponding chicory acid yield data sequence, the stage influence weight of each stage sequence in each concentrated liquid concentration data sequence is obtained, a regression analysis model is constructed, and the accuracy of the regression analysis result is improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a process detection method for the manufacturing process of chicory preparations. Background Technology

[0002] Modern pharmacological studies have shown that chicory is rich in active ingredients such as phenolic acids (e.g., chicoric acid), coumarins (e.g., aesculin), sesquiterpenes (e.g., lactucin), polysaccharides, and alkaloids. It possesses multiple pharmacological effects, including lowering blood sugar, lowering blood lipids, lowering uric acid, protecting the liver, and regulating immunity. Based on its dual nature as both food and medicine, chicory has been developed into various formulations. However, with increasing market demand, the quality control of chicory formulations has become increasingly prominent. Existing detection methods face technical bottlenecks in areas such as quantification of active ingredients, assessment of process stability, and impurity control, necessitating the development of systematic process detection methods to ensure product quality.

[0003] In the manufacturing process of chicory preparations, the extraction process directly affects the manufacturing results. The main extraction process is ethanol reflux extraction, in which parameters such as ethanol concentration, extraction temperature, time, and solid-liquid ratio affect the extraction rate of active ingredients such as chicoric acid and chlorogenic acid. Therefore, regression analysis can be used to establish a mathematical model between process parameters and product quality indicators to optimize extraction conditions (parameter settings for ethanol reflux extraction). However, regression analysis is suitable for steady-state processes. In ethanol reflux extraction, there are dynamic changes in the concentration of the concentrate over time due to solvent evaporation and solute enrichment, dynamic adjustments in component dissolution equilibrium, and the degradation of heat-sensitive components. This results in poor adaptability of regression analysis to dynamic changes in continuous production, which may lead to deviations in the analytical results and inaccurate conclusions.

[0004] Therefore, improving the adaptability of regression analysis to dynamic changes in continuous production and enhancing the accuracy of analysis results have become urgent problems to be solved. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a process detection method for chicory formulation manufacturing process to address the problem of how to improve the adaptability of regression analysis to dynamic changes in continuous production and improve the accuracy of analysis results.

[0006] This invention provides a process detection method for chicory formulation manufacturing, the method comprising the following steps:

[0007] In the manufacturing process of chicory formulations at at least two ethanol concentrations, the concentration of concentrate and the yield of chicoric acid are detected in real time according to a preset detection frequency, so as to obtain at least two concentrate concentration data sequences and their corresponding chicoric acid yield data sequences.

[0008] Based on the data change characteristics in each concentrate concentration data sequence, obtain the stage coefficient of each concentrate concentration data in each concentrate concentration data sequence, and divide each concentrate concentration data sequence into at least two stage sequences based on the stage coefficient of each concentrate concentration data in each concentrate concentration data sequence.

[0009] Each concentrated solution concentration data sequence is divided into at least two subsequences. Based on the data differences of the subsequences contained in each stage sequence of each concentrated solution concentration data sequence, the fluctuation characteristics of the concentrated solution concentration data in each stage sequence, and the similarity between each concentrated solution concentration data sequence and its corresponding chicoric acid yield data sequence, the stage influence weight of each stage sequence in each concentrated solution concentration data sequence is obtained.

[0010] Based on the stage influence weight of each stage sequence in each concentrate concentration data sequence, a regression analysis model for chicory preparations in the manufacturing process is constructed to detect and adjust the process parameters of chicory preparations in the manufacturing process.

[0011] Preferably, the step of obtaining the stage coefficient of each concentrate concentration data in each concentrate concentration data sequence based on the data change characteristics in each concentrate concentration data sequence includes:

[0012] The formula for calculating the stage coefficient of the b-th concentrate concentration data in any concentrate concentration data sequence is as follows:

[0013] ;

[0014] in, This is the stage coefficient for the b-th concentrate concentration data; For the first The stage coefficient of the concentrate concentration data; This represents the number of concentrate concentration data points preceding the b-th concentrate concentration data point. This is the first preset quantity; This represents the difference between the concentration data of the b-th concentrate and its left adjacent concentrate concentration data. The concentration difference between the b-th concentrate data and its right adjacent concentrate data; N is the second preset quantity; Let be the variance of the data sequence consisting of the b-th concentrate concentration data and the previous n concentrate concentration data; Let be the variance of the data sequence consisting of the b-th concentrate concentration data and the preceding n-1 concentrate concentration data. This represents the number of concentrate concentration data points following the b-th concentrate concentration data point. This is the third preset quantity; The stage coefficient for the left adjacent concentrate concentration data of the last concentrate concentration data; It is the absolute value symbol; This is the normalization function.

[0015] Preferably, dividing each concentrate concentration data sequence into at least two stage sequences based on the stage coefficient of each concentrate concentration data in each concentrate concentration data sequence includes:

[0016] A set of stage coefficients is formed by combining the stage coefficients of each concentrate concentration data in all concentrate data sequences. The peak values ​​in the set of stage coefficients are obtained. The peak values ​​are removed from the set of stage coefficients to obtain a first set of stage coefficients. The average value of the stage coefficients in the first set of stage coefficients is obtained. Stage coefficients that are greater than the average value of the stage coefficients are removed from the first set of stage coefficients to obtain a second set of stage coefficients. The peak values ​​in the second set of stage coefficients are obtained, and the corresponding average value of the peak values ​​is obtained.

[0017] For any concentrate concentration data sequence, the peak average value is used to divide the concentrate concentration data sequence into at least two stage sequences.

[0018] Preferably, the step of obtaining the stage influence weight of each stage sequence in each concentrated solution concentration data sequence based on the data differences of the subsequences contained in each stage sequence in each concentrated solution concentration data sequence, the fluctuation characteristics of the concentrated solution concentration data in each stage sequence, and the similarity between each concentrated solution concentration data sequence and its corresponding chicoric acid yield data sequence includes:

[0019] For any concentrate concentration data sequence, the dynamic hysteresis coefficient of each subsequence is obtained based on the data differences of the concentrate concentration data in each subsequence of the any concentrate concentration data sequence.

[0020] Based on the dynamic lag influence coefficient of the subsequences contained in each stage sequence of any concentrate concentration data sequence, and the fluctuation characteristics of the concentrate concentration data in each stage sequence, the importance of the influence coefficient of each stage sequence is obtained.

[0021] The importance of the influence coefficients of each stage sequence in any concentrate concentration data sequence is accumulated to obtain the accumulated value of the importance of the influence coefficients. The ratio of the importance of the influence coefficients of each stage sequence in any concentrate concentration data sequence to the accumulated value of the importance of the influence coefficients is obtained to obtain the stage weight of each stage sequence.

[0022] Based on the stage weight of each stage sequence in any given concentrate concentration data sequence, and the similarity between any given concentrate concentration data sequence and its corresponding chicoric acid yield data sequence, the stage influence weight of each stage sequence is obtained.

[0023] Preferably, the step of obtaining the dynamic hysteresis coefficient of each subsequence based on the data differences of the concentrate concentration data in each subsequence of any concentrate concentration data sequence includes:

[0024] For any subsequence, based on the concentrate concentration data excluding the last concentrate concentration data in the subsequence, the ARIMA algorithm is used to obtain the predicted value of the last concentrate concentration data in the subsequence. The absolute value of the difference between the last concentrate concentration data and the predicted value in the subsequence is obtained to obtain the predicted difference value. The predicted difference value is normalized to obtain the dynamic lag influence coefficient of the subsequence.

[0025] Preferably, the step of obtaining the importance of the influence coefficient of each stage sequence based on the dynamic lag influence coefficient of the subsequences contained in each stage sequence of any concentrate concentration data sequence, and the fluctuation characteristics of the concentrate concentration data in each stage sequence, includes:

[0026] For any stage sequence in any concentrated solution concentration data sequence, the average value of the dynamic lag influence coefficient of the subsequences contained in any stage sequence is obtained and denoted as the mean dynamic lag influence coefficient.

[0027] The stability is obtained by taking the reciprocal of the sum of the variance of the sequence at any stage and the constant 1, and the fluctuation is obtained by taking the difference between the constant 1 and the stability.

[0028] The importance of the influence coefficient of any stage sequence is obtained by multiplying the mean of the dynamic lag influence coefficient by the degree of fluctuation.

[0029] Preferably, the step of obtaining the stage influence weight of each stage sequence based on the stage weight of each stage sequence in any concentrate concentration data sequence and the similarity between any concentrate concentration data sequence and its corresponding chicoric acid yield data sequence includes:

[0030] For any stage sequence in any concentrate concentration data sequence, obtain the first-order difference data sequence of the concentrate concentration data sequence to obtain the concentrate concentration difference sequence. Obtain the first-order difference data sequence of the chicoric acid yield data sequence corresponding to the concentrate concentration data sequence to obtain the chicoric acid yield difference sequence. Calculate the similarity between the concentrate concentration difference sequence and the chicoric acid yield difference sequence using the DTW algorithm. Obtain the product of the stage weight of any stage sequence and the similarity to obtain the stage influence weight of any stage sequence.

[0031] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows:

[0032] In this invention, stage coefficients are obtained to divide each concentrate concentration data sequence into at least two stage sequences. That is, the ethanol reflux extraction process is divided into at least two stages according to the change in concentrate concentration, which can better capture the stage-based dynamic change characteristics of the concentrate concentration data. Then, the stage influence weight of each stage sequence in each concentrate concentration data sequence is obtained. This allows for a better analysis of the impact of concentrate concentration changes on chicoric acid yield in different stages of the ethanol reflux extraction process, by combining the dynamic lag and stage change characteristics of the concentrate concentration data. Finally, a regression analysis model for chicory preparations in the manufacturing process is constructed to improve the adaptability of regression analysis to the dynamic changes in concentrate concentration in continuous production and improve the accuracy of the analysis results. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the 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.

[0034] Figure 1 This is a flowchart of a process detection method for a chicory preparation manufacturing process provided in Embodiment 1 of the present invention. Detailed Implementation

[0035] Embodiments of this disclosure are described in detail below, with examples of these embodiments illustrated in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this disclosure, and should not be construed as limiting it.

[0036] It should be noted that the terms "first," "second," etc., used in this disclosure and the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure.

[0037] To illustrate the technical solution of the present invention, specific embodiments are described below.

[0038] See Figure 1 This is a flowchart of a process detection method for a chicory preparation manufacturing process provided in Embodiment 1 of the present invention, as shown below. Figure 1 As shown, the method may include:

[0039] Step S101: During the manufacturing process of chicory preparations at at least two ethanol concentrations, the concentration of the concentrate and the yield of chicoric acid are detected in real time according to a preset detection frequency to obtain at least two concentrate concentration data sequences and their corresponding chicoric acid yield data sequences.

[0040] In the manufacturing process of chicory preparations, the extraction process directly affects the manufacturing results. The main extraction process is ethanol reflux extraction. The ethanol reflux extraction process involves pulverizing and drying chicory roots or leaves, adding chicory powder and ethanol solution according to a set material-to-liquid ratio, and heating and reflux extraction under controlled temperature. After extraction, filtration and separation are performed, and the filtrate is concentrated to dryness under reduced pressure. The ethanol is recovered (and can be recycled), yielding an extract. The target components are then further separated using methods such as macroporous resin adsorption and column chromatography, followed by vacuum drying or spray drying to obtain a powdered extract. Parameters such as ethanol concentration, extraction temperature, time, and material-to-liquid ratio affect the extraction rates of active ingredients such as chicoric acid and chlorogenic acid. Therefore, regression analysis can be used to establish a mathematical model between process parameters and product quality indicators to optimize the extraction conditions (parameter settings during ethanol reflux extraction).

[0041] However, regression analysis is applicable to steady-state processes. In ethanol reflux extraction, there are dynamic changes in the concentration of the concentrate over time, such as solvent evaporation and solute enrichment, dynamic adjustment of component dissolution balance, and degradation of heat-sensitive components. This makes regression analysis less adaptable to dynamic changes in continuous production, which may lead to deviations in the analysis results and inaccurate results.

[0042] Since ethanol concentration is one of the key factors affecting the concentration of the concentrate, and the concentration of the concentrate is one of the key factors affecting the purity of the purified product, this embodiment takes ethanol concentration as an example for analysis. Different concentrations of ethanol are set and chicory powder and ethanol solution are added according to the common material-to-liquid ratio (1:10-1:20). The temperature is controlled near the boiling point of ethanol (78-80℃), and the reflux time is usually 1-3 hours. The concentration of the concentrate (i.e., the concentrated extract after the extract has been evaporated by rotary evaporation) and the yield of chicoric acid are detected in real time according to the preset detection frequency. The concentrate concentration data sequence and its corresponding chicoric acid yield data sequence are formed. That is, a concentrate concentration data sequence and its corresponding chicoric acid yield data sequence are obtained in the ethanol reflux extraction process at each ethanol concentration, which is used to analyze the dynamic changes of the concentrate concentration in the ethanol reflux extraction process. In this embodiment, the preset detection frequency is set to once every 2 minutes. This is not limited and can be set according to the specific implementation scenario. The concentration of the concentrate is obtained by direct weighing or drying the residue. The chicoric acid yield is obtained by HPLC. Then, the yield of chicoric acid is obtained by dividing the mass of the extract by the mass of the raw material.

[0043] Step S102: Based on the data change characteristics in each concentrate concentration data sequence, obtain the stage coefficient of each concentrate concentration data in each concentrate concentration data sequence, and divide each concentrate concentration data sequence into at least two stage sequences based on the stage coefficient of each concentrate concentration data in each concentrate concentration data sequence.

[0044] During ethanol reflux extraction, the concentration of the concentrate changes dynamically over time. This change is influenced by multiple factors, including solvent evaporation, component dissolution equilibrium, and degradation of heat-sensitive components. For example, during ethanol reflux, ethanol continuously vaporizes and condenses back into the solvent. Some ethanol is lost due to evaporation or leakage, leading to a reduction in the total solvent volume. As the solvent decreases, the concentration of solutes (such as chicoric acid and polysaccharides) in the remaining solvent gradually increases, creating a concentration effect. At different stages, such as the initial stage where the solvent is plentiful and the solute dissolves rapidly, resulting in a rapid increase in concentration; the middle stage where dissolution approaches equilibrium and the concentration increase slows down; and the later stage where the solvent is nearly exhausted and the concentration reaches saturation or supersaturation, the change in the concentrate concentration itself exhibits a phased nature.

[0045] Therefore, based on the data change characteristics in each concentrate concentration data sequence, the stage coefficient of each concentrate concentration data in each concentrate concentration data sequence can be obtained, and then based on the stage coefficient of each concentrate concentration data in each concentrate concentration data sequence, each concentrate concentration data sequence can be divided into at least two stage sequences.

[0046] The method for obtaining the stage coefficient of each concentrated concentration data in each concentrated concentration data sequence based on the data change characteristics in each concentrated concentration data sequence is as follows:

[0047] The formula for calculating the stage coefficient of the b-th concentrate concentration data in any concentrate concentration data sequence is as follows:

[0048]

[0049] in, This is the stage coefficient for the b-th concentrate concentration data; For the first The stage coefficient of the concentrate concentration data; This represents the number of concentrate concentration data points preceding the b-th concentrate concentration data point. This is the first preset quantity; This represents the difference between the concentration data of the b-th concentrate and its left adjacent concentrate concentration data. The concentration difference between the b-th concentrate data and its right adjacent concentrate data; N is the second preset quantity; Let be the variance of the data sequence consisting of the b-th concentrate concentration data and the previous n concentrate concentration data; Let be the variance of the data sequence consisting of the b-th concentrate concentration data and the preceding n-1 concentrate concentration data. This represents the number of concentrate concentration data points following the b-th concentrate concentration data point. This is the third preset quantity; The stage coefficient for the left adjacent concentrate concentration data of the last concentrate concentration data; It is the absolute value symbol; This is the normalization function.

[0050] It should be noted that in this embodiment... Set to 3; there are no restrictions here, and the setting can be adjusted according to the specific implementation scenario. When the concentration data of the concentrate before the b-th concentrate concentration data is insufficient to analyze the characteristics of data change, and there will be no stage change at the beginning, the stage coefficient of the first 3 concentrate concentration data is set to be the same as the stage coefficient of the fourth concentrate concentration data.

[0051] when and hour, The larger the value, the greater the difference in concentration before and after the b-th concentrate concentration data point, indicating a higher probability that the b-th concentrate concentration data point represents a stage point. The larger it is; This represents the difference in average variance between the b-th concentrate concentration data and the data sequence consisting of the previous n-1 concentrate concentration data. The larger the value, the greater the likelihood that the b-th concentrate concentration data point exists within the range of the previous n-1 concentrate concentration data points. The larger the value, the better. In order to reduce the influence of historical concentrate concentration data and ensure that there is enough data for analysis, N is set to 10 in this embodiment. There is no limit here, and it can be set according to the specific implementation scenario. If the number of concentrate concentration data before the b-th concentrate concentration data is less than 10, then N is processed according to the number of the b-th concentrate concentration data.

[0052] In this embodiment Set to 1; there are no restrictions here, and the setting can be adjusted according to the specific implementation scenario. When the b-th concentrate concentration data is the last concentrate concentration data, it is impossible to analyze the concentration change between the b-th concentrate concentration data and the concentrate concentration data before and after it. Therefore, the stage coefficient of the last concentrate concentration data is set to be the same as the stage coefficient of its left adjacent concentrate concentration data.

[0053] Following the method for obtaining the stage coefficient of the b-th concentrated solution concentration data in any concentrated solution concentration data sequence, the stage coefficient of each concentrated solution concentration data in any concentrated solution concentration data sequence is obtained. Similarly, the stage coefficient of each concentrated solution concentration data in each concentrated solution concentration data sequence is obtained.

[0054] Furthermore, based on the stage coefficient of each concentrate concentration data in each concentrate concentration data sequence, each concentrate concentration data sequence is divided into at least two stage sequences, as follows:

[0055] A set of stage coefficients is formed by combining the stage coefficients of each concentrate concentration data in all concentrate data sequences. A statistical histogram of the stage coefficient set is obtained, with the horizontal axis representing the stage coefficient and the vertical axis representing the frequency corresponding to the stage coefficient. The peak values ​​in the stage coefficient set are obtained from the statistical histogram. Since the stage coefficients of other data points are relatively large except for the stage points, and their numbers are much larger than the frequency distribution of the stage points, the peak data are likely to be stage coefficients of non-stage points. Therefore, the peak values ​​are removed from the stage coefficient set to obtain the first stage coefficient set. Since the stage coefficients of the stage points are relatively large, the average value of the stage coefficients of the first stage coefficient set is obtained. Stage coefficients greater than the average value of the stage coefficients are removed from the first stage coefficient set to obtain the second stage coefficient set. At the same time, since the stage-based changes in concentrate concentration data are regular and exist at different ethanol concentrations, they have a certain degree of repetition. Therefore, the peak values ​​of the second stage coefficient set are obtained, and the average value of the corresponding peak values ​​is used as a reference threshold for the stage points.

[0056] For any concentrate concentration data sequence, the concentrate concentration data with a stage coefficient greater than a reference threshold are taken as stage points of the concentrate concentration data sequence, and the concentrate concentration data sequence is divided into at least two stage sequences. For example, for the concentrate concentration data sequence (1, 2, 3, 4, 5, 6, 7, 8, 9, 10), assuming that concentrate concentration data 3 and 7 are stage points, the concentrate concentration data sequence is divided into three stage sequences: (1, 2), (3, 4, 5, 6), and (7, 8, 9, 10).

[0057] Similarly, each concentrate concentration data sequence is divided into at least two phase sequences.

[0058] Step S103: Divide each concentrate concentration data sequence into at least two subsequences. Based on the data differences of the subsequences contained in each stage sequence of each concentrate concentration data sequence, the fluctuation characteristics of the concentrate concentration data in each stage sequence, and the similarity between each concentrate concentration data sequence and its corresponding chicoric acid yield data sequence, obtain the stage influence weight of each stage sequence in each concentrate concentration data sequence.

[0059] Since the dynamic nature of concentrate concentration data is mainly reflected in the different changes at different stages, the stage influence weight of each stage sequence in each concentrate concentration data sequence can be obtained to build a regression model and enhance the dynamic adaptability of the regression model.

[0060] Since the concentrate concentration data has time-varying characteristics, and traditional regression models have limitations such as static assumptions and neglect of lag effects, a sliding window with a length of 10 and a step size of 1 is used. This is not limited here and can be set according to the specific implementation scenario. Starting from the first concentrate concentration data in each concentrate concentration data sequence, each concentrate concentration data sequence is divided into at least two subsequences. Based on the data differences of the subsequences contained in each stage sequence of each concentrate concentration data sequence, the dynamic lag effect coefficient of each subsequence in each concentrate concentration data sequence is obtained.

[0061] The method for obtaining the dynamic hysteresis coefficient of each subsequence in each concentrated solution concentration data sequence, based on the data differences of the subsequences contained in each stage sequence, is as follows:

[0062] For any subsequence in any concentrate concentration data sequence, based on the concentrate concentration data in any subsequence excluding the last concentrate concentration data, the ARIMA algorithm is used to obtain the predicted value of the last concentrate concentration data in any subsequence. The ARIMA algorithm is an existing technology and will not be described in detail here. The absolute value of the difference between the last concentrate concentration data in any subsequence and its predicted value is obtained to obtain the predicted difference value. The predicted difference value is normalized to obtain the dynamic lag influence coefficient of any subsequence.

[0063] In one embodiment, taking the a-th subsequence in any concentrate concentration data sequence as an example, the formula for calculating the dynamic hysteresis effect coefficient of the a-th subsequence is:

[0064]

[0065] in, Let be the dynamic lag effect coefficient of the a-th subsequence; This refers to the concentration data of the last concentrate in the a-th subsequence; This is the predicted value of the concentration data of the last concentrate in the a-th subsequence; It is the absolute value symbol; This is the normalization function.

[0066] It should be noted that, The larger the value, the more dynamic the data in the a-th subsequence is, indicating that the data is not fully captured. This means that the concentration data of the last concentrate in the a-th subsequence may be influenced by more historical concentrate concentration data, resulting in a longer period of influence. The larger it is.

[0067] Furthermore, since the changes in concentrate concentration data are phased, with each phase corresponding to an extraction result (chicoric acid yield) of a concentrate concentration, if the dynamic lag influence coefficient of the subsequences contained in each phase changes significantly, the regression model may underestimate the true impact of process parameters. When the changes in concentrate concentration data are large, the impact on the extraction result is also large. Therefore, the phase influence weight of each phase sequence in each concentrate concentration data sequence can be obtained based on the dynamic lag influence coefficient of the subsequences contained in each phase sequence in each concentrate concentration data sequence, the fluctuation characteristics of the concentrate concentration data in each phase sequence, and the similarity between each concentrate concentration data sequence and its corresponding chicoric acid yield data sequence.

[0068] The method for assigning stage influence weights to each stage of a given concentrate concentration data sequence is as follows:

[0069] (1) Based on the dynamic lag influence coefficient of the subsequence contained in each stage sequence of any concentrate concentration data sequence, and the fluctuation characteristics of the concentrate concentration data in each stage sequence, obtain the importance of the influence coefficient of each stage sequence.

[0070] Specifically, for any stage sequence in any concentrate concentration data sequence, the average value of the dynamic lag influence coefficient of the subsequences contained in any stage sequence is obtained and denoted as the mean dynamic lag influence coefficient.

[0071] The stability is obtained by taking the reciprocal of the sum of the variance of the sequence at any stage and the constant 1, and the fluctuation is obtained by taking the difference between the constant 1 and the stability.

[0072] The importance of the influence coefficient of any stage sequence is obtained by multiplying the mean of the dynamic lag influence coefficient by the degree of fluctuation.

[0073] In one embodiment, taking the v-th stage sequence in any concentrate concentration data sequence as an example, the formula for calculating the importance of the influence coefficient of the v-th stage sequence is as follows:

[0074]

[0075] in, represents the importance of the influence coefficient of the v-th stage sequence; M is the number of subsequences contained in the v-th stage sequence; The dynamic lag effect coefficient of the m-th subsequence contained in the v-th stage sequence; Let be the variance of the concentrate concentration data in the v-th stage sequence.

[0076] It should be noted that, This represents the mean of the dynamic lag effect coefficient. The larger the value, the greater the influence of the dynamic coefficient of the subsequences contained in the v-th stage sequence. The larger it is; For the degree of fluctuation, The larger the value, the greater the data volatility of the sequence in the v-th stage. The larger the value, the greater the impact on the extraction results (chicoric acid yield). The larger it is.

[0077] Similarly, the importance of the influence coefficient of each stage sequence is obtained.

[0078] (2) Obtain the stage weight of each stage sequence.

[0079] Specifically, the importance of the influence coefficients of each stage sequence in any concentrate concentration data sequence is accumulated to obtain the accumulated value of the influence coefficient importance. The ratio of the importance of the influence coefficients of each stage sequence in any concentrate concentration data sequence to the accumulated value of the influence coefficient importance is obtained to obtain the stage weight of each stage sequence.

[0080] In one embodiment, taking the v-th stage sequence in any concentrate concentration data sequence as an example, the formula for calculating the stage weight of the v-th stage sequence is:

[0081]

[0082] in, Let be the stage weight of the v-th stage sequence; denoted by , where is the influence coefficient importance of the v-th stage sequence; V is the number of stage sequences in any concentrate concentration data sequence.

[0083] It should be noted that the influence coefficient of the v-th stage sequence is of varying importance. The larger the value, the more significant the impact of changes in the concentrate concentration data in the v-th stage sequence on the extraction results. The larger it is.

[0084] Similarly, obtain the stage weight of each stage sequence.

[0085] (3) Based on the stage weight of each stage sequence in any concentrate concentration data sequence and the similarity between any concentrate concentration data sequence and its corresponding chicoric acid yield data sequence, obtain the stage influence weight of each stage sequence.

[0086] Specifically, for any stage sequence in any concentrate concentration data sequence, the first-order difference data sequence of the concentrate concentration data sequence is obtained to obtain the concentrate concentration difference sequence. The first-order difference data sequence of the chicoric acid yield data sequence corresponding to the concentrate concentration data sequence is obtained to obtain the chicoric acid yield difference sequence. The similarity between the concentrate concentration difference sequence and the chicoric acid yield difference sequence is calculated using the DTW algorithm. The stage weight of any stage sequence is obtained by multiplying the similarity by the stage weight to obtain the stage influence weight of any stage sequence. The DTW algorithm for obtaining similarity is existing technology and will not be elaborated here.

[0087] In one embodiment, taking the v-th stage sequence in any concentrate concentration data sequence as an example, the formula for calculating the stage influence weight of the v-th stage sequence is as follows:

[0088]

[0089] in, Let V be the stage influence weight of the v-th stage sequence; Let be the stage weight of the v-th stage sequence; The similarity is between the concentration difference sequence of the concentrate and the yield difference sequence of chicoric acid.

[0090] in, The larger the value, the more significant the impact of changes in the concentrate concentration data in the v-th stage sequence on the extraction results. The larger it is; The larger the value, the more similar the concentration difference sequence of the concentrate to the chicoric acid yield difference sequence. In other words, the more similar the data changes of any concentrate concentration data sequence are to their corresponding chicoric acid yield data sequence, the greater the impact of changes in concentrate concentration on changes in chicoric acid yield. The larger it is.

[0091] Similarly, the stage influence weight of each stage sequence of any concentrate concentration data sequence is obtained.

[0092] The stage influence weight of each stage sequence in each concentrated solution concentration data sequence is obtained by following the method for obtaining the stage influence weight of each stage sequence in any concentrated solution concentration data sequence.

[0093] Step S104: Based on the stage influence weight of each stage sequence in each concentrate concentration data sequence, construct a regression analysis model for chicory preparations during the manufacturing process, which is used to detect and adjust the process parameters of chicory preparations during the manufacturing process.

[0094] After obtaining the stage influence weights of each stage sequence in the concentration data series of each concentrate, the yield changes of chicoric acid under different ethanol concentrations and concentrate concentrations were analyzed according to the single-factor influence mechanism. For example, the solubility of chicoric acid in the ethanol-water system showed an inverted U-shaped change with increasing ethanol concentration: in the low concentration area (<40%), chicoric acid exists in the form of salts, and the solubility increases with increasing ethanol concentration (due to increased solvent polarity); in the high concentration area (>60%), the solvent polarity decreases, the solubility of chicoric acid salts decreases, and at the same time, the precipitation of co-soluble impurities (such as polysaccharides and proteins) may encapsulate chicoric acid, further inhibiting the extraction efficiency. Then the analysis results showed that increasing the concentrate concentration significantly increases the solution viscosity, hindering the diffusion of chicoric acid from the raw material cells to the solvent. In the low concentration area (1.0-1.2 g / mL), the viscosity is low, the diffusion resistance is small, and the yield of chicoric acid increases with increasing concentration (due to the increase in total solute); in the high concentration area (>1.3 g / mL), the viscosity increases sharply, the diffusion rate decreases, and the rate of increase in chicoric acid yield slows down or even decreases. Therefore, based on ethanol concentration data at different concentrations, concentrated solution concentration data sequences at different ethanol concentrations and their corresponding chicoric acid yield data sequences, and the stage influence weights of each stage sequence in each concentrated solution concentration data sequence, an autoregressive distributed lag model is established. The establishment of the autoregressive distributed lag model is based on existing technology and will not be elaborated here.

[0095] Based on an autoregressive lag model, the lag effect of adjusting ethanol and concentrate concentrations under different conditions was obtained. Adjustments were then made based on the predicted chicoric acid yield. For example, the autoregressive lag model predicted that when the ethanol concentration was 60%, the concentrate concentration was 1.20 g / mL, and the extraction time was 90 min, the chicoric acid yield could reach 1.03% (peak). According to the model, if the ethanol concentration decreased from 60% to 55% after 60 min of extraction, the chicoric acid yield would decrease by 5% after 30 min. In this case, the predicted yield could be maintained by replenishing the ethanol concentration to 60%. The method of obtaining the lag effect of adjusting ethanol and concentrate concentrations under different conditions using an autoregressive lag model is existing technology and will not be elaborated upon here.

[0096] In summary, the embodiments of this invention obtain stage coefficients to divide each concentrate concentration data sequence into at least two stage sequences, that is, to divide the ethanol reflux extraction process into at least two stages according to the change in concentrate concentration, which can better capture the stage-based dynamic change characteristics of concentrate concentration data; then, to obtain the stage influence weight of each stage sequence in each concentrate concentration data sequence, which can combine the dynamic lag and stage change characteristics of concentrate concentration data to better analyze the impact of concentrate concentration changes on chicoric acid yield in different stages of the ethanol reflux extraction process; finally, to construct a regression analysis model for chicory preparations in the manufacturing process, improve the adaptability of regression analysis to the dynamic changes in concentrate concentration in continuous production, and improve the accuracy of analysis results.

[0097] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A process detection method for the manufacturing process of chicory preparations, characterized in that, The process detection method for the manufacturing process of chicory preparations includes: In the manufacturing process of chicory formulations at at least two ethanol concentrations, the concentration of concentrate and the yield of chicoric acid are detected in real time according to a preset detection frequency, so as to obtain at least two concentrate concentration data sequences and their corresponding chicoric acid yield data sequences. Based on the data change characteristics in each concentrate concentration data sequence, obtain the stage coefficient of each concentrate concentration data in each concentrate concentration data sequence, and divide each concentrate concentration data sequence into at least two stage sequences based on the stage coefficient of each concentrate concentration data in each concentrate concentration data sequence. Each concentrated solution concentration data sequence is divided into at least two subsequences. Based on the data differences of the subsequences contained in each stage sequence of each concentrated solution concentration data sequence, the fluctuation characteristics of the concentrated solution concentration data in each stage sequence, and the similarity between each concentrated solution concentration data sequence and its corresponding chicoric acid yield data sequence, the stage influence weight of each stage sequence in each concentrated solution concentration data sequence is obtained. Based on the stage influence weight of each stage sequence in each concentrate concentration data sequence, a regression analysis model for chicory preparations in the manufacturing process is constructed to detect and adjust the process parameters of chicory preparations in the manufacturing process. The step of obtaining the stage coefficient of each concentrate concentration data in each concentrate concentration data sequence based on the data change characteristics in each concentrate concentration data sequence includes: The formula for calculating the stage coefficient of the b-th concentrate concentration data in any concentrate concentration data sequence is as follows: ； in, The stage coefficient for the b-th concentrate concentration data; For the first The stage coefficient of the concentrate concentration data; This represents the number of concentrate concentration data points preceding the b-th concentrate concentration data point. This is the first preset quantity; This represents the difference between the concentration data of the b-th concentrate and its left adjacent concentrate concentration data. The concentration difference between the b-th concentrate data and its right adjacent concentrate data; N is the second preset quantity; Let be the variance of the data sequence consisting of the b-th concentrate concentration data and the previous n concentrate concentration data; Let be the variance of the data sequence consisting of the b-th concentrate concentration data and the preceding n-1 concentrate concentration data. This represents the number of concentrate concentration data points following the b-th concentrate concentration data point. This is the third preset quantity; The stage coefficient for the left adjacent concentrate concentration data of the last concentrate concentration data; It is the absolute value symbol; This is the normalization function.

2. The process detection method for chicory preparation manufacturing according to claim 1, characterized in that, The step of dividing each concentrate concentration data sequence into at least two stage sequences based on the stage coefficient of each concentrate concentration data in each concentrate concentration data sequence includes: A set of stage coefficients is formed by combining the stage coefficients of each concentrate concentration data in all concentrate data sequences. The peak values ​​in the set of stage coefficients are obtained. The peak values ​​are removed from the set of stage coefficients to obtain a first set of stage coefficients. The average value of the stage coefficients in the first set of stage coefficients is obtained. Stage coefficients that are greater than the average value of the stage coefficients are removed from the first set of stage coefficients to obtain a second set of stage coefficients. The peak values ​​in the second set of stage coefficients are obtained, and the corresponding average value of the peak values ​​is obtained. For any concentrate concentration data sequence, the peak average value is used to divide the concentrate concentration data sequence into at least two stage sequences.

3. The process detection method for chicory preparation manufacturing according to claim 1, characterized in that, The step-by-step influence weight of each stage sequence in each concentrated solution concentration data sequence is obtained based on the data differences of subsequences contained in each stage sequence in each concentrated solution concentration data sequence, the fluctuation characteristics of concentrated solution concentration data in each stage sequence, and the similarity between each concentrated solution concentration data sequence and its corresponding chicoric acid yield data sequence, including: For any concentrate concentration data sequence, the dynamic hysteresis coefficient of each subsequence is obtained based on the data differences of the concentrate concentration data in each subsequence of the any concentrate concentration data sequence. Based on the dynamic lag influence coefficient of the subsequences contained in each stage sequence of any concentrate concentration data sequence, and the fluctuation characteristics of the concentrate concentration data in each stage sequence, the importance of the influence coefficient of each stage sequence is obtained. The importance of the influence coefficients of each stage sequence in any concentrate concentration data sequence is accumulated to obtain the accumulated value of the importance of the influence coefficients. The ratio of the importance of the influence coefficients of each stage sequence in any concentrate concentration data sequence to the accumulated value of the importance of the influence coefficients is obtained to obtain the stage weight of each stage sequence. Based on the stage weight of each stage sequence in any given concentrate concentration data sequence, and the similarity between any given concentrate concentration data sequence and its corresponding chicoric acid yield data sequence, the stage influence weight of each stage sequence is obtained.

4. The process detection method for chicory preparation manufacturing according to claim 3, characterized in that, The step of obtaining the dynamic hysteresis coefficient of each subsequence based on the data differences of the concentrate concentration data in each subsequence of any concentrate concentration data sequence includes: For any subsequence, based on the concentrate concentration data excluding the last concentrate concentration data in the subsequence, the ARIMA algorithm is used to obtain the predicted value of the last concentrate concentration data in the subsequence. The absolute value of the difference between the last concentrate concentration data and the predicted value in the subsequence is obtained to obtain the predicted difference value. The predicted difference value is normalized to obtain the dynamic lag influence coefficient of the subsequence.

5. The process detection method for chicory preparation manufacturing according to claim 3, characterized in that, The step of obtaining the importance of the influence coefficient of each stage sequence based on the dynamic lag influence coefficient of the subsequences contained in each stage sequence of any concentrate concentration data sequence, and the fluctuation characteristics of the concentrate concentration data in each stage sequence, includes: For any stage sequence in any concentrated solution concentration data sequence, the average value of the dynamic lag influence coefficient of the subsequences contained in any stage sequence is obtained and denoted as the mean dynamic lag influence coefficient. The stability is obtained by taking the reciprocal of the sum of the variance of the sequence at any stage and the constant 1, and the fluctuation is obtained by taking the difference between the constant 1 and the stability. The importance of the influence coefficient of any stage sequence is obtained by multiplying the mean of the dynamic lag influence coefficient by the degree of fluctuation.

6. The process detection method for chicory preparation manufacturing according to claim 3, characterized in that, The step of obtaining the stage influence weight of each stage sequence based on the stage weight of each stage sequence in any concentrate concentration data sequence and the similarity between any concentrate concentration data sequence and its corresponding chicoric acid yield data sequence includes: For any stage sequence in any concentrate concentration data sequence, obtain the first-order difference data sequence of the concentrate concentration data sequence to obtain the concentrate concentration difference sequence. Obtain the first-order difference data sequence of the chicoric acid yield data sequence corresponding to the concentrate concentration data sequence to obtain the chicoric acid yield difference sequence. Calculate the similarity between the concentrate concentration difference sequence and the chicoric acid yield difference sequence using the DTW algorithm. Obtain the product of the stage weight of any stage sequence and the similarity to obtain the stage influence weight of any stage sequence.

Images