Wine brewing tank real-time monitoring method and system based on artificial intelligence

Abstract

The invention discloses a grape wine brewing tank real-time monitoring method and system based on artificial intelligence, and relates to the technical field of grape wine brewing equipment. The method comprises the following steps: constructing a database storing the total sugar content of a wine sample at each time stage in the wine brewing process and near infrared spectrum data of the sample, extracting characteristic parameters of the near infrared spectrum data of the wine sample, and performing grey correlation analysis on the total sugar content and the characteristic parameter data of the wine sample; obtaining a grey correlation degree between each characteristic parameter and the total sugar content, setting a grey correlation degree threshold, screening out a master control characteristic parameter, and establishing a quantitative relation between the master control characteristic parameter of the wine sample and the total sugar content by using a BP neural network; and inputting the main control characteristic parameters of the near infrared spectrum data of the wine sample in the brewing tank to be monitored, which are acquired in real time, into the established quantitative relation to obtain the total sugar content predicted value of the wine sample in the brewing tank to be monitored.

Description

Technical Field

[0001] The present invention relates to the technical field of wine brewing equipment, and in particular to a real-time monitoring method and system for wine brewing tanks based on artificial intelligence. Background Art

[0002] Fermentation is a series of biochemical reactions caused by yeast and fungi, and is also the process of converting grape juice into wine. The fermentation process is mainly divided into alcoholic fermentation and malic acid-lactic acid fermentation. Among them, alcoholic fermentation is the process of sugar fermented into alcohol by yeast, which is the core step in the winemaking process. The sugar contained in grape juice is the basic raw material required for fermentation. At the right temperature, all sugars will be converted into alcohol. If the alcohol content of the fermentation process is too high, the action of yeast will be inhibited, and the fermentation process will be terminated, resulting in some unfermented residual sugars. Therefore, the level of sugar content determines the level of alcohol content and taste type of wine. The level of alcohol content also affects the fermentation process. Together, they affect the balance of wine flavor. In the process of winemaking, it is necessary to monitor the total sugar content and alcohol content changes in the fermentation process in real time and accurately to obtain wine with better flavor and taste.

[0003] Since the fermentation process is carried out in a closed wine brewing tank, if you want to obtain the total sugar content data during the fermentation process, you first need to use the sampling method to sample the raw materials, and then use high-performance liquid chromatography, direct titration, potassium permanganate oxidation or chemical methods in the laboratory to measure the total sugar content of the sampled samples. The detection process is cumbersome, and the traditional detection methods mentioned above have the disadvantages of complex sample processing, long analysis time, and delayed detection results. They cannot meet the current production requirements of wine companies for detection timeliness. To this end, we propose a real-time monitoring method and system for wine brewing tanks based on artificial intelligence. Summary of the invention

[0004] The main purpose of the present invention is to provide a real-time monitoring method and system for wine brewing tanks based on artificial intelligence, which can effectively solve the problems in the background technology.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] The real-time monitoring method of wine brewing tanks based on artificial intelligence includes:

[0007] Constructing a database, wherein the database stores the total sugar content of wine samples measured at each time stage during the winemaking process and the near-infrared spectrum data of the samples, wherein the characteristic parameters of the near-infrared spectrum data include statistical characteristics, frequency domain characteristics, time domain characteristics, and wavelet transform characteristics, wherein the statistical characteristics include the mean, variance, maximum, and minimum of the spectrum data; the frequency domain characteristics include the frequency domain energy and frequency domain mean of the spectrum data; the time domain characteristics include the waveform peak value and waveform width of the spectrum data; and the wavelet transform characteristics are wavelet coefficients extracted after the spectrum data is decomposed into signals of multiple scales and frequencies through wavelet transform;

[0008] Extract characteristic parameters of the near infrared spectrum data of the wine sample in the database, perform grey correlation analysis on the total sugar content and characteristic parameter data of the wine sample, and obtain the grey correlation degree between each characteristic parameter and the total sugar content. The acquisition process includes the following steps:

[0009] Construct the characteristic parameter data sequence E and total sugar content data sequence F of the wine sample, where E = {E1, E2, ..., E i}, F = {F1}, E i It is represented as the i-th characteristic parameter data set; F1 is represented as the total sugar content data set;

[0010] The kth data value e of the i-th feature parameter in the data sequence E ik and the kth data value f of the total sugar content in the data sequence F 1k Perform standardization and obtain standardized data e' ik and f' jk , the data standardization formula is:

[0011] Calculate the correlation coefficient between all data points and characteristic parameters in the data sequence F. The calculation formula is: In the formula, r 1k It is expressed as the correlation coefficient between the kth data value of total sugar content and the characteristic parameter; ρ is the constant coefficient;

[0012] Δ min and Δ max They are the minimum difference between the two levels and the maximum difference between the two levels of the standardized data of the characteristic parameter data and the total sugar content data, and the calculation formulas are:

[0013] Δ min =min i (min k (e' ik -f' 1k ));

[0014] Δ max =maxi (max k (e' ik -f' 1k ));

[0015] In the formula, min i (min k (e' ik -f' 1k )) means to take e' first ik -f' jk |The minimum value of all calculated results, and then take the minimum value of all minimum values; max i (max k (e' ik -f' jk )) means to take |e' first ik -f' jk |, and then take the maximum value of all the maximum values, where k = 1, 2, ..., Q, where Q is the total amount of data;

[0016] S24: Based on the obtained correlation coefficient, calculate the grey correlation degree R between the i-th characteristic parameter and the total sugar content i1 , the calculation formula is:

[0017] The main control feature parameters are screened out by setting the gray correlation threshold, wherein the interval of the gray correlation threshold is Among them, R max , R min are the maximum and minimum values ​​of the grey correlation between the characteristic parameter and the total sugar content;

[0018] The BP neural network is used to establish the quantitative relationship between the main control characteristic parameters and the total sugar content of the wine sample, wherein the input layer of the BP neural network is the main control characteristic parameters, the input layer is the total sugar content, and the number of neurons in the middle layer is based on the formula: Determine, where q represents the type of main control feature parameter; p is the number of neurons in the output layer, in this neural network, p = 1; a is a positive integer from 1 to 9;

[0019] Real-time acquisition of the main control characteristic parameters of the near infrared spectrum data of the wine sample in the brewing tank to be monitored, and obtaining the predicted value of the total sugar content of the wine sample in the brewing tank to be monitored by inputting the acquired main control characteristic parameter sampling values ​​into the established quantitative relationship;

[0020] The method further comprises:

[0021] Update the sampled data in the database in real time, and adjust the number of neurons in the middle layer of the BP neural network according to the data update results until its prediction accuracy meets the expected value; the calculation formula of the expected value is:

[0022]

[0023] Where E(Y) is the expected value of prediction accuracy; N is the number of neural network input samples; f(X t ) is the output function of the neural network; X t is the tth output sample of the neural network.

[0024] A real-time monitoring system for wine brewing tanks based on artificial intelligence, including a sample database, a sample data analysis module, a neural network module, a sample collection module, a near-infrared light emission module, a spectral data acquisition module, and a data processing module;

[0025] The sample database is used to store the total sugar content of wine samples at each time stage measured during the winemaking process and the near infrared spectrum data of the samples;

[0026] The sample data analysis module is used to extract characteristic parameters of the near-infrared spectrum data of the wine samples in the database, perform grey correlation analysis on the total sugar content and characteristic parameter data of the wine samples, obtain the grey correlation degree between each characteristic parameter and the total sugar content, and select the main control characteristic parameters by setting the grey correlation degree threshold;

[0027] The neural network module is used to construct a BP neural network with the main control characteristic parameters as input and the total sugar content data as output, and use the BP neural network to establish a quantitative relationship between the main control characteristic parameters and the total sugar content of the wine sample, wherein the input layer of the BP neural network is each of the main control characteristic parameters, the input layer is the total sugar content, and the number of neurons in the middle layer is according to the formula: Determine, where q represents the type of main control feature parameter; p is the number of neurons in the output layer, in this neural network, p = 1; a is a positive integer from 1 to 9;

[0028] The sample collection module is used to perform real-time quantitative sampling of wine samples in the monitored brewing tank, and transport the acquired wine samples to the sample chamber;

[0029] The near-infrared light emitting module is used to emit near-infrared light with an incident angle of θ to the sample chamber;

[0030] The spectrum data acquisition module is used to collect diffuse reflection light generated when the near-infrared light passes through the sample chamber, and acquire near-infrared spectrum data of the wine sample by analyzing the diffuse reflection light;

[0031] The data processing module is used to obtain the main control characteristic parameters of the near infrared spectrum data of the wine sample in the brewing tank to be monitored, and input the sampling value of the main control characteristic parameter into the established quantitative relationship to obtain the predicted value of the total sugar content of the wine sample in the brewing tank to be monitored;

[0032] The system also includes a data updating module, which is used to update the sampled data in the database in real time and adjust the number of neurons in the middle layer of the BP neural network according to the data updating result until its prediction accuracy meets the expected value;

[0033] The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor.

[0034] The present invention has the following beneficial effects:

[0035] Compared with the prior art, a database is constructed, wherein the database stores the total sugar content of wine samples and near-infrared spectrum data of the samples measured at each time stage during the winemaking process, characteristic parameters of the near-infrared spectrum data of the wine samples in the database are extracted, grey correlation analysis is performed on the total sugar content and characteristic parameter data of the wine samples, the grey correlation degree between each characteristic parameter and the total sugar content is obtained, the main control characteristic parameters are screened out by setting a grey correlation degree threshold, the quantitative relationship between the main control characteristic parameters and the total sugar content of the wine samples is established by using a BP neural network, the main control characteristic parameters of the near-infrared spectrum data of the wine samples in the monitored brewing tanks are collected in real time, and the predicted value of the total sugar content of the wine samples in the monitored brewing tanks is obtained by inputting the acquired sampling values ​​of the main control characteristic parameters into the established quantitative relationship, thereby providing a convenient and fast real-time detection method for the total sugar content of wine, which can shorten the detection time of the total sugar content of wine in the brewing tanks during the fermentation process, simplify the detection process, and meet the production requirements of wine companies for detection timeliness. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a flow chart of a method for real-time monitoring of wine brewing tanks based on artificial intelligence according to the present invention;

[0037] Figure 2 This is a structural block diagram of the real-time monitoring system for wine brewing tanks based on artificial intelligence of the present invention;

[0038] Figure 3 It is a schematic structural diagram of a wine sample collection device proposed in an embodiment of the present invention;

[0039] Figure 4 This is a schematic diagram of the principle of measuring near infrared spectroscopy data of wine samples in the scheme of the present invention;

[0040] Figure 5 This is an example of the spectrum obtained;

[0041] Figure 6 It is a schematic diagram of the structure of the BP neural network constructed in the solution of the present invention.

[0042] In the figure, 1. brewing tank; 2. sample room; 3. quantitative circulation pump; 4. connecting pipe between quantitative circulation pump and brewing tank; 5. connecting pipe between sample room and brewing tank. DETAILED DESCRIPTION

[0043] The present invention will be further described below in conjunction with specific implementation methods, wherein the accompanying drawings are only used for exemplary descriptions and represent only schematic diagrams rather than actual drawings, and should not be understood as limiting the present invention. In order to better illustrate the specific implementation methods of the present invention, some parts of the accompanying drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product.

[0044] The specific implementation process of the technical solution of the present invention includes the following steps:

[0045] Step 1: construct a database, wherein the database stores the total sugar content of wine samples measured at each time stage during the winemaking process and the near infrared spectrum data of the samples;

[0046] The total sugar content of wine samples can be determined by direct titration, and the near infrared spectroscopy data of wine samples can be obtained by the following method: Figure 4 The measurement principle shown is that near-infrared light is emitted to the wine sample in the sample chamber. When the near-infrared light passes through the sample, diffuse reflection occurs. The diffuse reflection light is collected and the spectrum is stored in the form of absorbance log. The average of multiple measurements is taken as the final spectrum data of the sample. The obtained spectrum diagram is as follows: Figure 5 shown.

[0047] Step 2: Extract characteristic parameters of near infrared spectral data of wine samples in the database;

[0048] Among them, the characteristic parameters of near-infrared spectral data include statistical characteristics, frequency domain characteristics, time domain characteristics, and wavelet transform characteristics; statistical characteristics include the mean, variance, maximum value, and minimum value of spectral data; frequency domain characteristics include the frequency domain energy and frequency domain mean of spectral data; time domain characteristics include the waveform peak value and waveform width of spectral data; wavelet transform characteristics are the wavelet coefficients extracted after decomposing spectral data into signals of multiple scales and frequencies through wavelet transform;

[0049] Step 3: Perform grey correlation analysis on the total sugar content and characteristic parameter data of the wine sample to obtain the grey correlation between each characteristic parameter and the total sugar content. The acquisition process includes the following steps:

[0050] Step 31: Construct the characteristic parameter data sequence E and the total sugar content data sequence F of the wine sample, where E = {E1, E2, ..., E i}, F = {F1}, E i It is represented as the i-th characteristic parameter data set; F1 is represented as the total sugar content data set;

[0051] Step 32: The kth data value e of the i-th feature parameter in the data sequence E ik and the kth data value f of the total sugar content in the data sequence F 1k Perform standardization and obtain standardized data e' ik and f' jk , the data standardization formula is:

[0052] Step 33: Calculate the correlation coefficient between all data points and characteristic parameters in the data sequence F. The calculation formula is: In the formula, r 1k It is expressed as the correlation coefficient between the kth data value of total sugar content and the characteristic parameter; ρ is a constant coefficient, generally ρ = 0.5; where Δ min and Δ max They are the minimum difference between the two levels and the maximum difference between the two levels of the standardized data of the characteristic parameter data and the total sugar content data, and the calculation formulas are:

[0053] Δ min =min i (min k (e' ik -f' 1k ));

[0054] Δ max =max i (max k (e' ik -f' 1k ));

[0055] In the formula, mini(mink(e'ik-f'1k)) means taking the minimum value of all calculated results in e'ik-f'jk| first, and then taking the minimum value of all minimum values; max i (max k (e' ik -f' jk )) means to take |e' first ik -f' jk |, and then take the maximum value of all the maximum values, where k = 1, 2, ..., Q, where Q is the total amount of data;

[0056] Step 34: Based on the obtained correlation coefficient, calculate the grey correlation degree R between the i-th characteristic parameter and the total sugar content i1 , the calculation formula is:

[0057] Step 4: Filter out the main control feature parameters by setting the grey correlation threshold, where the interval of the grey correlation threshold is Among them, R max , R min are the maximum and minimum values ​​of the grey correlation between the characteristic parameter and the total sugar content;

[0058] Step 5: Use BP neural network to establish the quantitative relationship between the main control characteristic parameters of wine samples and the total sugar content;

[0059] Among them, the input layer of the BP neural network is the main control feature parameters, the input layer is the total sugar content, and the number of neurons in the middle layer is based on the formula: To determine, where q represents the type of main control feature parameter; p is the number of neurons in the output layer. In this neural network, p = 1; a is a positive integer from 1 to 9. The constructed BP neural network structure is as follows Figure 6 As shown;

[0060] Step 7: Real-time acquisition of the main control characteristic parameters of the near infrared spectrum data of the wine sample in the brewing tank to be monitored;

[0061] In this embodiment, a schematic diagram of the structure of a wine sample collection device is provided. Figure 3 As shown, the quantitative circulation pump 3 pumps the wine sample in the brewing tank 1 through the pipeline 4 to the sample chamber 5, and adopts Figure 4 The principle of the determination method is to obtain the near-infrared spectrum data of the wine sample in the sample chamber 5, and obtain the main control characteristic parameters of the spectrum data. After the data determination is completed, the collected sample is pumped back to the brewing tank 1 through the quantitative circulation pump 3. In this way, not only the near-infrared spectrum data of the sample can be collected in real time, but also the wine sample to be measured does not need to be sampled outside the brewing tank 1, thereby reducing the waste of wine samples.

[0062] Step 8: Obtain a predicted value of the total sugar content of the wine sample in the brewing tank to be monitored by inputting the obtained sampling value of the main control characteristic parameter into the established quantitative relationship;

[0063] Step 9: Update the sampled data in the database in real time, and adjust the number of neurons in the middle layer of the BP neural network according to the data update result until its prediction accuracy meets the expected value; wherein the calculation formula of the expected value is:

[0064]

[0065] Where E(Y) is the expected value of prediction accuracy; N is the number of neural network input samples; f(X t ) is the output function of the neural network; X t is the tth output sample of the neural network.

[0066] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A real-time monitoring method for wine brewing tanks based on artificial intelligence, characterized in that: include: Constructing a database, wherein the database stores total sugar content of wine samples at various time stages measured during the winemaking process and near infrared spectrum data of the samples; Extracting characteristic parameters of the near-infrared spectrum data of the wine samples in the database, performing grey correlation analysis on the total sugar content and characteristic parameter data of the wine samples, obtaining the grey correlation degree between each characteristic parameter and the total sugar content, and screening out the main control characteristic parameters by setting a grey correlation degree threshold; The BP neural network is used to establish the quantitative relationship between the main control characteristic parameters and the total sugar content of the wine sample, wherein the input layer of the BP neural network is the main control characteristic parameters, the input layer is the total sugar content, and the number of neurons in the middle layer is based on the formula: Determine, where q represents the type of main control feature parameter; p is the number of neurons in the output layer, in this neural network, p = 1; a is a positive integer from 1 to 9; The main control characteristic parameters of the near-infrared spectral data of the wine sample in the monitored brewing tank are collected in real time, and the predicted value of the total sugar content of the wine sample in the monitored brewing tank is obtained by inputting the obtained main control characteristic parameter sampling values ​​into the established quantitative relationship.

2. The method for real-time monitoring of wine brewing tanks based on artificial intelligence according to claim 1, characterized in that: The method further comprises: Update the sampled data in the database in real time, and adjust the number of neurons in the middle layer of the BP neural network according to the data update results until its prediction accuracy meets the expected value; the calculation formula of the expected value is: Where E(Y) is the expected value of prediction accuracy; N is the number of neural network input samples; f(X t ) is the output function of the neural network; X t is the tth output sample of the neural network.

3. The method for real-time monitoring of wine brewing tanks based on artificial intelligence according to claim 1, characterized in that: The characteristic parameters of near-infrared spectral data include statistical characteristics, frequency domain characteristics, time domain characteristics, and wavelet transform characteristics, wherein the statistical characteristics include the mean, variance, maximum value, and minimum value of the spectral data; the frequency domain characteristics include the frequency domain energy and frequency domain mean of the spectral data; the time domain characteristics include the waveform peak value and waveform width of the spectral data; and the wavelet transform characteristics are the wavelet coefficients extracted after decomposing the spectral data into signals of multiple scales and frequencies through wavelet transform.

4. The method for real-time monitoring of wine brewing tanks based on artificial intelligence according to claim 1, characterized in that: The process of obtaining the grey relational degree includes the following steps: Construct the characteristic parameter data sequence E and total sugar content data sequence F of the wine sample, where E = {E1, E2, ..., E i }, F = {F1}, E i It is represented as the i-th characteristic parameter data set; F1 is represented as the total sugar content data set; The kth data value e of the i-th feature parameter in the data sequence E ik and the kth data value f of the total sugar content in the data sequence F 1k Perform standardization and obtain standardized data e' ik and f' jk , the data standardization formula is: Calculate the correlation coefficient between all data points and characteristic parameters in the data sequence F. The calculation formula is: In the formula, r 1k It is expressed as the correlation coefficient between the kth data value of total sugar content and the characteristic parameter; ρ is the constant coefficient; Δ min and Δ max They are the minimum difference between the two levels and the maximum difference between the two levels of the standardized data of the characteristic parameter data and the total sugar content data, and the calculation formulas are: D min =min i (min) k (|e' ik -f' 1k |)); D max =max i (max. k (|e' ik -f' 1k |)); In the formula, min i (min k (|e' ik -f' 1k |)) means to take |e' first ik -f' jk |The minimum value of all calculated results, and then take the minimum value of all minimum values; max i (max k (|e' ik -f' jk |)) means to take |e' first ik -f' jk |, and then take the maximum value of all the maximum values, where k = 1, 2, ..., Q, where Q is the total amount of data; S24: Based on the obtained correlation coefficient, calculate the grey correlation degree R between the i-th characteristic parameter and the total sugar content i1 , the calculation formula is:

5. The method for real-time monitoring of wine brewing tanks based on artificial intelligence according to claim 1, characterized in that: The interval of the gray relational threshold is Among them, R max , R min are the maximum and minimum values ​​of the grey correlation between the characteristic parameter and the total sugar content.

6. Real-time monitoring system for wine brewing tanks based on artificial intelligence, characterized in that: It includes a sample database, a sample data analysis module, a neural network module, a sample collection module, a near-infrared light emission module, a spectral data acquisition module, and a data processing module; The sample database is used to store the total sugar content of wine samples at each time stage measured during the winemaking process and the near infrared spectrum data of the samples; The sample data analysis module is used to extract the characteristic parameters of the near-infrared spectral data of the wine samples in the database, perform grey correlation analysis on the total sugar content and characteristic parameter data of the wine samples, obtain the grey correlation between each characteristic parameter and the total sugar content, and select the main control characteristic parameter by setting a grey correlation threshold, wherein, when the grey correlation between the characteristic parameter and the total sugar content is within the set grey correlation threshold, it is determined to be the main control characteristic parameter; the interval of the grey correlation threshold is Among them, R max , R min are the maximum and minimum values ​​of the grey correlation between the characteristic parameter and the total sugar content; The neural network module is used to construct a BP neural network with the main control characteristic parameters as input and the total sugar content data as output, and use the BP neural network to establish a quantitative relationship between the main control characteristic parameters and the total sugar content of the wine sample, wherein the input layer of the BP neural network is each of the main control characteristic parameters, the input layer is the total sugar content, and the number of neurons in the middle layer is according to the formula: Determine, where q represents the type of main control feature parameter; p is the number of neurons in the output layer, in this neural network, p = 1; a is a positive integer from 1 to 9; The sample collection module is used to perform real-time quantitative sampling of wine samples in the monitored brewing tank, and transport the acquired wine samples to the sample chamber; The near-infrared light emitting module is used to emit near-infrared light with an incident angle of θ to the sample chamber; The spectrum data acquisition module is used to collect diffuse reflection light generated when the near-infrared light passes through the sample chamber, and acquire near-infrared spectrum data of the wine sample by analyzing the diffuse reflection light; The data processing module is used to obtain the main control characteristic parameters of the near-infrared spectrum data of the wine sample in the brewing tank to be monitored, and input the sampling value of the main control characteristic parameter into the established quantitative relationship to obtain the predicted value of the total sugar content of the wine sample in the brewing tank to be monitored.

7. The artificial intelligence-based real-time monitoring system for wine brewing tanks according to claim 6 is characterized in that: The system also includes a data updating module, which is used to update the sampled data in the database in real time and adjust the number of neurons in the middle layer of the BP neural network according to the data updating result until its prediction accuracy meets the expected value; wherein the calculation formula of the expected value is: Where E(Y) is the expected value of prediction accuracy; N is the number of neural network input samples; f(X t ) is the output function of the neural network; X t is the tth output sample of the neural network.

8. The artificial intelligence-based real-time monitoring system for wine brewing tanks according to claim 6 is characterized in that: The system comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the steps of the method according to any one of claims 1 to 5 are implemented when the processor executes the program.

Images