阅读说明：本技术 一种地贞颗粒醇提过程在线质量监控方法 (Online quality monitoring method for ligustrum japonicum granules alcohol extraction process ) 是由 刘爽 谈宗华 吴统选 朱丹 王晓 昝琼 刘月 于 2021-07-20 设计创作，主要内容包括：本发明公开了一种地贞颗粒醇提过程在线质量监控方法,其监控方法步骤如下：步骤一：提取液样品收集；步骤二：取样试验设计；步骤三：近红外光谱采集；步骤四：醇提样品含量测定；步骤五：建立定量模型；步骤六：建立多变量统计过程控制(MSPC)质量监控模型；采用PCA得分图、Hotelling T2和DModX控制图绘制过程轨迹图,步骤七：验证多变量统计过程控制(MSPC)质量监控模型,选择A4,A5两个正常批次对所建立的统计监控模型进行验证,本发明提出了一种地贞颗粒醇提过程在线质量监控方法,实现地贞颗粒醇提过程有效成分含量的在线监测；进一步实现提取终点的快速判断和实时放行；进一步提高了地贞颗粒乙醇提取液质量的批次间稳定性,提高了产品质量。(The invention discloses an online quality monitoring method for an alcohol extraction process of Ligustrum japonicum Kishinouye granules, which comprises the following steps: the method comprises the following steps: collecting an extract sample; step two: designing a sampling test; step three: collecting near infrared spectrum; step four: measuring the content of the alcohol extraction sample; step five: establishing a quantitative model; step six: establishing a Multivariate Statistical Process Control (MSPC) quality monitoring model; drawing a process track diagram by using a PCA score diagram, Hotelling T2 and a DModX control diagram, and performing the following steps: verifying a Multivariate Statistical Process Control (MSPC) quality monitoring model, and selecting two normal batches of A4 and A5 to verify the established statistical monitoring model, the invention provides an online quality monitoring method for the alcohol extraction process of Ligustrum japonicum Kishinouye granules, which realizes the online monitoring of the content of active ingredients in the alcohol extraction process of Ligustrum japonicum Kishinouye granules; further realizing the rapid judgment and real-time release of the extraction end point; further improves the batch stability of the quality of the Ligustrum lucidum particle ethanol extract and improves the product quality.)

1. An online quality monitoring method for an alcohol extraction process of Ligustrum japonicum Thunb granules is characterized in that: the monitoring method comprises the following steps:

the method comprises the following steps: collecting an extract sample;

weighing 60g of each of a schisandra chinensis medicinal material and a glossy privet fruit medicinal material, grinding into coarse powder, sieving by a second sieve, placing into a 1000mL three-neck flask, placing the three-neck flask into a water bath kettle at 85 ℃ for preheating, installing a reflux device, adding 95% ethanol with the temperature of 75 ℃ and the volume of 840mL, timing from the time when a solvent is completely added, sampling 3mL of ethanol extract every 5min, wherein the whole extraction time is 90min, and obtaining 18 ethanol extract samples in total;

step two: designing a sampling test;

designing 12 batches of tests, including A1-A8 and B1-B4, wherein A1-A8 is a normal batch, and B1-B2 is a process abnormal batch, wherein B1 removes a heating source immediately after extraction starts, B2 keeps the extraction temperature at 40 ℃, abnormal conditions such as equipment damage or power failure in the production process of a factory are simulated, B3-B4 simulates abnormal feeding conditions, B3 reduces the feeding amount of raw medicinal materials, and B4 increases the feeding amount of the raw medicinal materials;

step three: collecting near infrared spectrum;

immediately scanning near infrared transmission spectrum of the alcohol extraction sample after the alcohol extraction sample is taken out, and setting the resolution ratio to be 8cm by taking air as background^-1Optical path of 2mm, scanning times of 32 times, collection of 10000-4000cm^-1Scanning each sample three times and obtaining an average spectrum;

step four: measuring the content of the alcohol extraction sample;

step five: establishing a quantitative model;

the establishing method comprises the following steps:

s1, abnormal point elimination and sample set division;

the Monte Carlo cross validation algorithm is realized through MATLAB R2018b software to remove abnormal samples, and the sample set is divided into a correction set and a validation set according to the proportion of 3:1 by adopting the SPXY algorithm;

s2, characteristic wavelength screening;

the data of effective components of specnuezhenide, schizandrin A, deoxyschizandrin and schisandrin B are screened by using the characteristic wavelengths of siPLS, CARS, RF and MC-UVE. The screening combination interval of the specnuezhenide is 5403.5-5600.2, 6005.2-6201.9, 8010.8-8207.5 and 8211.4-8408.1 cm < -1 >; the screening combination range of the schizandrol A is 5457.5-5696.7, 6417.9-6653.2, 7135.3-7370.6 and 8809.2-9044.5 cm < -1 >; the screening combination intervals of the schizandrin A are 6005.2-6201.9, 7609.7-7806.4, 7810.3-8007.0 and 8612.5-8809.2 cm < -1 >; the screening combination interval of the schisandrin B is 5403.5-5600.2, 5604.1-5800.8, 6005.2-6201.9 and 6406.3-6603.0 cm < -1 >;

s3 quantitative model establishment and evaluation

After the elimination of abnormal samples and the characteristic wavelength screening are completed, a PLSR quantitative analysis model of each index component is respectively established;

step six: establishing a Multivariate Statistical Process Control (MSPC) quality monitoring model;

the process trajectory plot was plotted using the PCA score plot, Hotelling T2, and the DModX control chart.

Step seven: validating a Multivariate Statistical Process Control (MSPC) quality monitoring model;

two normal batches, A4 and A5, were selected to validate the established statistical monitoring model.

2. The online quality monitoring method for the alcohol extraction process of the Ligustrum lucidum granules according to claim 1, characterized in that: the fourth step comprises the following specific steps:

s1: the chromatographic conditions were as follows:

the column was Xselect HST 3 (4.6X 250mm,5 μm); the detection wavelength is 230nm, and the flow rate is 1 mL/min^-1The sample amount was 10 μ L, and 0.1% phosphoric acid and acetonitrile were used as mobile phases a and B, and the elution was carried out in the following procedure: 0-10 min: 18-20% of B, 10-20 min: 20-25% of B, 20-35 min: 25-70% of B, 35-50 min: 70-75% of B, 50-60 min: 75% of B;

s2: preparing a reference substance solution;

precisely weighing appropriate amount of schisandrin A, deoxyschizandrin, schisandrin B and specnuezhenide reference substances, and adding appropriate amount of 95% ethanol to prepare a mixed reference substance solution containing 1099 μ g of specnuezhenide, 504.0 μ g of schisandrin A, 113 μ g of deoxyschizandrin A and 302 μ g of schisandrin B per 1 mL;

s3: preparing a test solution.

Taking a proper amount of alcohol extract samples, filtering, and taking subsequent filtrate to obtain the test solution.

3. The online quality monitoring method for the alcohol extraction process of the Ligustrum lucidum granules according to claim 1, characterized in that: the model evaluation indexes in the fifth step comprise a correction set error Root Mean Square (RMSECV), a correction set correlation coefficient (Rc), a correction set error Root Mean Square (RMSEC), a correction set relative deviation (RSEC), a verification set correlation coefficient (Rp), a verification set error Root Mean Square (RMSEP) and a verification set relative deviation (RSEP).

4. The online quality monitoring method for the alcohol extraction process of the Ligustrum lucidum granules according to claim 1, characterized in that: the PCA score chart in the sixth step shows the variation trend of the PC1 score of a certain batch of samples in the alcohol extraction process, and the variation trend of the PC1 basically represents the variation trend of the PC1 accounting for 94.46 percentThe overall trend of the sample is shown. Hotelling T²The control chart displays the distance from the sample to the principal component model, and abnormality is prompted according to the deviation degree of the control chart, the DModX control chart is an external residual error matrix of the principal component model and reflects external data change of the model, and the two control charts are usually combined and mutually supplemented, so that the process judgment is more accurate; PC1 score, Hotelling T, of training set data calculated by SIMCA 14.1 data analysis software²And (3) calculating the value and the value DModX, simultaneously calculating the average value (Avg) and the standard deviation (sigma) of the PC1 scores of the training data at each time point of alcohol extraction, setting the upper limit and the lower limit as the average value +/-3 sigma, setting 95 percent of Hotelling T2 as the upper control limit, and setting the average value +3 sigma as the upper control limit for DModX.

Technical Field

The invention relates to the field of traditional Chinese medicine pharmacy, in particular to an online quality monitoring method for an alcohol extraction process of ligustrum japonicum granules.

Background

The Dizhen granule is prepared from eight medicines of cortex lycii radicis, glossy privet fruit, yerbadetajo herb, Chinese magnoliavine fruit, flatstem milkvetch seed, silktree albizzia bark, liquoric root and turmeric root-tuber, and has the main effects of clearing deficiency heat, nourishing liver and kidney and calming heart-mind. Can be used for treating female climacteric syndrome with syndrome of yin deficiency and internal heat, manifested by symptoms of fever, perspiration, vexation, irascibility, feverish sensation in palms and soles, insomnia, dreaminess, soreness of waist and knees, dry mouth, constipation, etc.

The productive process flow of the glossy privet fruit granules comprises the steps of alcohol extraction, water extraction, concentration, drying, granulation and the like, wherein the alcohol extraction process comprises the steps of crushing the Chinese magnoliavine fruit and the glossy privet fruit into coarse powder and extracting the coarse powder by using an ethanol solution. The extraction process of the traditional Chinese medicine is a process of continuously dissolving out the effective components in the traditional Chinese medicine, the extraction process link is crucial to the production and the quality of the traditional Chinese medicine, and the quality of the traditional Chinese medicine extracting solution influences the quality of the final traditional Chinese medicine product. If an effective quality detection means is lacked, the content change of the effective components in the extraction process cannot be monitored in real time, and the uniformity and the stability of the product are influenced. At present, the quality detection of the alcohol extraction process of the glossy privet fruit granules adopts a method of manual sampling-laboratory chromatography detection, the detection time is long, and online and real-time monitoring and release control cannot be realized.

Therefore, the online quality monitoring method for the alcohol extraction process of the ligustrum lucidum granules is researched, so that online real-time monitoring of active ingredients in the alcohol extraction process of the ligustrum lucidum granules is realized, the quality stability of an extracting solution is guaranteed, and the uniformity and the stability of the quality of a final product are guaranteed.

Disclosure of Invention

The invention aims to provide an online quality monitoring method for an alcohol extraction process of Ligustrum japonicum Thunb granules, so as to solve the problems in the background art.

In order to achieve the purpose, the invention provides the following technical scheme:

an online quality monitoring method for an alcohol extraction process of Ligustrum japonicum Kishinouye granules comprises the following steps:

the method comprises the following steps: collecting an extract sample;

weighing 60g of each of a schisandra chinensis medicinal material and a glossy privet fruit medicinal material, grinding into coarse powder, sieving by a second sieve, placing into a 1000mL three-neck flask, placing the three-neck flask into a water bath kettle at 85 ℃ for preheating, installing a reflux device, adding 95% ethanol with the temperature of 75 ℃ and the volume of 840mL, timing from the time when a solvent is completely added, sampling 3mL of ethanol extract every 5min, wherein the whole extraction time is 90min, and obtaining 18 ethanol extract samples in total;

step two: designing a sampling test;

designing 12 batches of tests, including A1-A8 and B1-B4, wherein A1-A8 is a normal batch, and B1-B2 is a process abnormal batch, wherein B1 removes a heating source immediately after extraction starts, B2 keeps the extraction temperature at 40 ℃, abnormal conditions such as equipment damage or power failure in the production process of a factory are simulated, B3-B4 simulates abnormal feeding conditions, B3 reduces the feeding amount of raw medicinal materials, and B4 increases the feeding amount of the raw medicinal materials;

step three: collecting near infrared spectrum;

immediately scanning the near-infrared transmission spectrum of the alcohol extraction sample after the alcohol extraction sample is taken out, setting the resolution as 8cm < -1 >, the optical path as 2mm and the scanning frequency as 32 times by taking air as a background, collecting 10000-plus 4000cm < -1 > spectrum information, scanning each sample for three times, and obtaining an average spectrum;

step four: measuring the content of the alcohol extraction sample;

step five: establishing a quantitative model;

the establishing method comprises the following steps:

s1, abnormal point elimination and sample set division;

the Monte Carlo cross validation algorithm is realized through MATLAB R2018b software to remove abnormal samples, and the sample set is divided into a correction set and a validation set according to the proportion of 3:1 by adopting the SPXY algorithm;

s2, characteristic wavelength screening;

the data of effective components of specnuezhenide, schizandrin A, deoxyschizandrin and schisandrin B are screened by using the characteristic wavelengths of siPLS, CARS, RF and MC-UVE. The screening combination interval of the specnuezhenide is 5403.5-5600.2, 6005.2-6201.9, 8010.8-8207.5 and 8211.4-8408.1 cm < -1 >; the screening combination range of the schizandrol A is 5457.5-5696.7, 6417.9-6653.2, 7135.3-7370.6 and 8809.2-9044.5 cm < -1 >; the screening combination intervals of the schizandrin A are 6005.2-6201.9, 7609.7-7806.4, 7810.3-8007.0 and 8612.5-8809.2 cm < -1 >; the screening combination interval of the schisandrin B is 5403.5-5600.2, 5604.1-5800.8, 6005.2-6201.9 and 6406.3-6603.0 cm < -1 >;

s3 quantitative model establishment and evaluation

After the elimination of abnormal samples and the characteristic wavelength screening are completed, a PLSR quantitative analysis model of each index component is respectively established;

step six: establishing a Multivariate Statistical Process Control (MSPC) quality monitoring model;

the process trajectory plot was plotted using the PCA score plot, Hotelling T2, and the DModX control chart.

Step seven: validating a Multivariate Statistical Process Control (MSPC) quality monitoring model;

two normal batches, A4 and A5, were selected to validate the established statistical monitoring model.

As a further scheme of the invention: the fourth step comprises the following specific steps:

s1: the chromatographic conditions were as follows:

the column was Xselect HST 3 (4.6X 250mm,5 μm); the detection wavelength is 230nm, the flow rate is 1mL min < -1 >, the sample injection amount is 10 mu L, 0.1 percent phosphoric acid is taken as a mobile phase A, acetonitrile is taken as a mobile phase B, and the gradient elution is carried out according to the following procedures: 0-10 min: 18-20% of B, 10-20 min: 20-25% of B, 20-35 min: 25-70% of B, 35-50 min: 70-75% of B, 50-60 min: 75% of B;

s2: preparing a reference substance solution;

precisely weighing appropriate amount of schisandrin A, deoxyschizandrin, schisandrin B and specnuezhenide reference substances, and adding appropriate amount of 95% ethanol to prepare a mixed reference substance solution containing 1099 μ g of specnuezhenide, 504.0 μ g of schisandrin A, 113 μ g of deoxyschizandrin A and 302 μ g of schisandrin B per 1 mL;

s3: preparing a test solution.

Taking a proper amount of alcohol extract samples, filtering, and taking subsequent filtrate to obtain the test solution.

As a still further scheme of the invention: the model evaluation indexes in the fifth step comprise a correction set error Root Mean Square (RMSECV), a correction set correlation coefficient (Rc), a correction set error Root Mean Square (RMSEC), a correction set relative deviation (RSEC), a verification set correlation coefficient (Rp), a verification set error Root Mean Square (RMSEP) and a verification set relative deviation (RSEP).

As a still further scheme of the invention: in the sixth step, the PCA score chart shows the variation trend of the PC1 score of a certain batch of samples in the alcohol extraction process, and since the PC1 explains 94.46%, the variation trend of the PC1 basically represents the overall variation trend of the samples. Hotelling T²The control chart displays the distance from the sample to the principal component model, and abnormality is prompted according to the deviation degree of the control chart, the DModX control chart is an external residual error matrix of the principal component model and reflects external data change of the model, and the two control charts are usually combined and mutually supplemented, so that the process judgment is more accurate; PC1 score, Hotelling T, of training set data calculated by SIMCA 14.1 data analysis software²And (3) calculating the value and the value DModX, simultaneously calculating the average value (Avg) and the standard deviation (sigma) of the PC1 scores of the training data at each time point of alcohol extraction, setting the upper limit and the lower limit as the average value +/-3 sigma, setting 95 percent of Hotelling T2 as the upper control limit, and setting the average value +3 sigma as the upper control limit for DModX.

Compared with the prior art, the invention has the beneficial effects that:

1. the invention provides an online quality monitoring method for an alcohol extraction process of Ligustrum japonicum Kishinouye granules, which realizes online monitoring of the content of active ingredients in the alcohol extraction process of Ligustrum japonicum Kishinouye granules;

2. further realizing the rapid judgment and real-time release of the extraction end point;

3. further improves the batch stability of the quality of the Ligustrum lucidum particle ethanol extract and improves the product quality.

Drawings

Fig. 1 is a correlation diagram of predicted values and measured values of each index component NIRs in the correction set and the verification set (a. specnuezhenide, b. schizandrin, c. schizandrin a, d. schizandrin b).

Fig. 2 is a model diagram of PCA monitoring during alcohol extraction.

FIG. 3 is a schematic diagram of a Hotelling T2 monitoring model in an alcohol extraction process.

Fig. 4 is a model diagram of DModX monitoring during alcohol extraction.

FIG. 5 is a model A4 and A5 for monitoring alcohol extraction process.

FIG. 6 is a model B1 and B2 for alcohol extraction process monitoring.

FIG. 7 is a model B3 and B4 for alcohol extraction process monitoring.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Referring to fig. 1 to 7, in an embodiment of the present invention, an online quality monitoring method for an alcohol extraction process of ligustrum lucidum ait particles includes the following steps:

the method comprises the following steps: collecting an extract sample;

weighing 60g of each of a schisandra chinensis medicinal material and a glossy privet fruit medicinal material, grinding into coarse powder, sieving by a second sieve, placing into a 1000mL three-neck flask, placing the three-neck flask into a water bath kettle at 85 ℃ for preheating, installing a reflux device, adding 95% ethanol with the temperature of 75 ℃ and the volume of 840mL, timing from the time when a solvent is completely added, sampling 3mL of ethanol extract every 5min, wherein the whole extraction time is 90min, and obtaining 18 ethanol extract samples in total;

step two: designing a sampling test;

12 batches of tests are designed, the test conditions are shown in Table 1, and the tests comprise A1-A8 and B1-B4, A1-A8 is a normal batch, B1-B2 is a process abnormal batch, wherein B1 removes a heating source immediately after extraction starts, B2 keeps the extraction temperature at 40 ℃, abnormal conditions such as equipment damage or power failure in the production process of a factory are simulated, B3-B4 simulates abnormal feeding conditions, B3 reduces the feeding amount of raw medicinal materials, and B4 increases the feeding amount of the raw medicinal materials;

TABLE 1 Experimental design for alcohol extraction Process

Step three: collecting near infrared spectrum;

immediately scanning a near-infrared transmission spectrum of an alcohol extraction sample after the alcohol extraction sample is taken out, setting the resolution as 8cm & lt-1 & gt, the optical path as 2mm and the scanning times as 32 times by taking air as a background, collecting 10000-4000cm & lt-1 & gt spectrum information, scanning each sample for three times, and obtaining an average spectrum;

step four: measuring the content of the alcohol extraction sample;

the determination method comprises the following specific steps:

s1: the chromatographic conditions were as follows:

the column was Xselect HST 3 (4.6X 250mm,5 μm); the detection wavelength is 230nm, the flow rate is 1mL min < -1 >, the sample injection amount is 10 mu L, 0.1 percent phosphoric acid is taken as a mobile phase A, acetonitrile is taken as a mobile phase B, and the gradient elution is carried out according to the following procedures: 0-10 min: 18-20% of B, 10-20 min: 20-25% of B, 20-35 min: 25-70% of B, 35-50 min: 70-75% of B, 50-60 min: 75% of B;

s2: preparing a reference substance solution;

precisely weighing appropriate amount of schisandrin A, deoxyschizandrin, schisandrin B and specnuezhenide reference substances, and adding appropriate amount of 95% ethanol to prepare a mixed reference substance solution containing 1099 μ g of specnuezhenide, 504.0 μ g of schisandrin A, 113 μ g of deoxyschizandrin A and 302 μ g of schisandrin B per 1 mL;

s3: preparation of test solution

Taking a proper amount of alcohol extract samples, filtering, and taking subsequent filtrate to obtain the test solution.

Step five: establishing a quantitative model;

the model evaluation indexes comprise a correction set error Root Mean Square (RMSECV), a correction set correlation coefficient (Rc), a correction set error Root Mean Square (RMSEC), a correction set relative deviation (RSEC), a verification set correlation coefficient (Rp), a verification set error Root Mean Square (RMSEP) and a verification set relative deviation (RSEP); generally, the closer Rc and Rp are to 1, the stronger the correlation between the fitting value of the correction set, the predicted value of the verification set and the measured value of the sample is, and when the three indexes of RMSECV, RMSEC and RMSEP are smaller and closer to each other, RSEC and RSEP are smaller, the higher the accuracy and stability of the established quantitative analysis model is, and the better the performance is;

the establishing method comprises the following steps:

s1, abnormal point elimination and sample set division;

the Monte Carlo cross validation algorithm is realized through MATLAB R2018b software to remove abnormal samples, and the sample set is divided into a correction set and a validation set according to the proportion of 3:1 by adopting the SPXY algorithm;

s2, characteristic wavelength screening;

the data of effective components of specnuezhenide, schizandrin A, deoxyschizandrin and schisandrin B are screened by using the characteristic wavelengths of siPLS, CARS, RF and MC-UVE. The screening combination interval of the specnuezhenide is 5403.5-5600.2, 6005.2-6201.9, 8010.8-8207.5 and 8211.4-8408.1 cm < -1 >; the screening combination range of the schizandrol A is 5457.5-5696.7, 6417.9-6653.2, 7135.3-7370.6 and 8809.2-9044.5 cm < -1 >; the screening combination intervals of the schizandrin A are 6005.2-6201.9, 7609.7-7806.4, 7810.3-8007.0 and 8612.5-8809.2 cm < -1 >; the screening combination interval of the schisandrin B is 5403.5-5600.2, 5604.1-5800.8, 6005.2-6201.9 and 6406.3-6603.0 cm < -1 >;

s3 quantitative model establishment and evaluation

After the elimination of abnormal samples and the characteristic wavelength screening are completed, PLSR quantitative analysis models of all the index components are respectively established, parameters of the PLSR models established by different wavelength screening methods of all the index components are summarized in a table 2, and the table 2 shows that compared with PLSR quantitative models established by full spectrum, key variables obtained by four wavelength screening methods are improved to a certain extent in model performance, and the number of the variables required by modeling is greatly reduced. In the four wavelength screening methods, the RF shows excellent characteristic extraction capability in 4 index models, and the method can remove most of irrelevant variables in the near infrared spectrum while ensuring the excellent model precision and stability. In the PLSR quantitative model with the best performance, RMSEC values of specnuezhenide, schizandrin A and schizandrin B are 0.0570, 0.0060, 0.0018 and 0.0031 respectively, RMSEP values are 0.0613, 0.0054, 0.0034 and 0.0037 respectively, the RMSEC and RMSEP values are closer and smaller, Rc and Rp of the four index models are closer or higher than 0.9, RSEP in a verification set is less than 5%, RPD values are more than 2, and the requirements of process analysis are basically met. The correlation graph of the verification set and the correction concentrated measured value and the predicted value of the specnuezhenide, the schizandrin, the deoxyschizandrin and the schisandrin B is shown in figure 1, the correction concentrated measured value and the predicted value point are uniformly distributed on two sides of a straight line y ═ x, the measured value and the predicted value of the verification set are closer, and the requirement of content prediction of unknown samples can be met;

TABLE 2 alcohol extraction sample index componentsNIRsModel parameters

Step six: establishing a Multivariate Statistical Process Control (MSPC) quality monitoring model;

and drawing a process track graph by adopting a PCA score graph, Hotelling T2 and a DModX control graph, wherein the PCA score graph shows the change trend of the PC1 score of a certain batch of samples in the alcohol extraction process, and the change trend of the PC1 basically represents the overall change trend of the samples because the PC1 explains 94.46%. Hotelling T²The control chart shows the distance from the sample to the principal component model, and abnormality is indicated according to the degree of deviation thereof, and the DModX control chart isThe external residual matrix of the principal component model reflects the external data change of the model, and the two control charts are commonly combined and mutually supplemented, so that the process judgment is more accurate.

PC1 score, Hotelling T, of training set data calculated by SIMCA 14.1 data analysis software²And (3) calculating the value and the value DModX, simultaneously calculating the average value (Avg) and the standard deviation (sigma) of the PC1 scores of the training data at each time point of alcohol extraction, setting the upper limit and the lower limit as the average value +/-3 sigma, setting 95 percent of Hotelling T2 as the upper control limit, and setting the average value +3 sigma as the upper control limit for DModX.

Step seven: validating a Multivariate Statistical Process Control (MSPC) quality monitoring model;

two normal batches of A4 and A5 are selected to verify the established statistical monitoring model, the alcohol extraction process trend locus diagram is shown in figure 5, the batches of A4 and A5 do not exceed the set control limit in the three monitoring models, the trend is basically consistent with that of the training samples of 6 modeled batches, and the situation that the established model has no fault and false alarm is shown.

Note that:

1. equipment faults such as process parameter setting errors, power failure and the like are frequently encountered in the production process, if the faults can be found and the waste of raw materials can be greatly reduced in time, the batches B1 and B2 respectively simulate process abnormal conditions, B1 stops heating immediately after extraction starts, B2 is lower than the extraction temperature, and the extraction process is completed at 40 ℃. Fig. 6 shows the condition of batches B1 and B2 in the MSPC model, and in the PCA score chart, the principal components of B1 and B2 are both below the lower score limit during the whole alcohol extraction process. Hotelling T²Control chart does not exceed statistic T²95% control Limit, but far above the T of the normal batch samples²Statistics, in the DModX control map, both B1 and B2 are above the DModX control upper limit;

2. in addition to the process abnormality, the abnormal feeding solid-liquid ratio can also cause larger quality fluctuation in the production process, B3 and B4 simulate the situation of too little or too much feeding of medicinal materials respectively, as shown in FIG. 7, a PCA score chart shows that the PC1 score of B3 runs around the lower control limit of the PC1 score of the B4 score of the B3 is lower than the normal situation, and the B4 score of the B4 is higher than the normal situationThe medicine material feeding is higher than normal, the PC1 score is in a controlled range in the initial stage, and the PC1 score gradually rises to exceed the upper control limit after 20min, which indicates that the intrinsic ingredients are higher than normal. Hotelling T²The control charts show that B3 and B4 run near the 95% control limit and eventually B4 is abnormal beyond the control limit. In the DModX control chart, both B3 and B4 are higher than the upper control limit, and the two abnormal conditions prove that the established monitoring model can be used as a tool for judging process abnormality and can play a role in timely reporting errors in the production process.

Although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that various changes in the embodiments and/or modifications of the invention can be made, and equivalents and modifications of some features of the invention can be made without departing from the spirit and scope of the invention.