β-glucan fermentation strain screening method, device and β-glucan

Abstract

The present invention belongs to the technical field of fermentation engineering, and discloses a method and device for screening fermentation strains of beta-glucan, and beta-glucan; the method comprises: step 100, collecting fermentation control data of fermentation strains in a fermentation device within a unit time, and collecting growth image information of the fermentation strains at the end of the unit time; step 101, extracting features from the fermentation control data to obtain fermentation feature data; step 102, inputting pre-collected strain type information, initial inoculation data, fermentation feature data and growth image information into a pre-trained yield prediction model, and outputting a yield prediction value of beta-glucan after a future unit time; step 103, judging whether to generate a correction instruction for the fermentation control data based on a comparison result between the yield prediction value and a preset yield threshold, and improving the efficiency of the overall screening process of the fermentation strains.

Description

Technical Field

[0001] The present invention relates to the technical field of fermentation engineering, and more specifically, to a method and device for screening fermentation strains of beta-glucan and beta-glucan. Background Art

[0002] β-Glucan is an important functional polysaccharide widely used in food, medicine, beverage and cosmetics industries.

[0003] The existing patent application with the publication number CN114134073A discloses a method for screening high-yield strains of biological dust suppressants. The method uses the hemolytic characteristics of the strain in the process of producing dust suppressants and the ability of the strain to synthesize dust suppressants for screening. It mainly includes two parts: screening of high-yield strains and verification of screened strains; by fully diluting the original strain after expansion culture, inoculating it on a blood agar plate for culture, quantitatively screening the strains according to the size of the hemolytic ring formed by each strain on the blood agar plate, and fermenting and culturing the strains with larger hemolytic rings, extracting the biological dust suppressants after fermentation and culture by an acidification precipitation method, and measuring the final dust suppressant yield synthesized by different strains to determine the best high-yield strain. This technical solution carries out screening based on the characteristics of the strain itself, selects green and environmentally friendly culture, and uses the final yield as the evaluation standard. It has the advantages of simple screening method, green and environmentally friendly process, and high screening efficiency, and provides a new method for screening strains in the preparation process of biological dust suppressants.

[0004] The existing technology still has the following problems:

[0005] Traditional fermentation strain screening methods usually rely on the analysis and determination of the final target product; this method requires a complete fermentation cycle, and it is difficult to detect problems in advance during the culture process, which will also affect the yield of the final target product of the fermentation strain. It not only consumes a lot of time, but also increases the culture cost; especially when screening multiple groups of fermentation strains at the same time, the consumption of resources and time delays are more significant; there will be problems such as potential excellent strains being ignored or the screening process being too lengthy, resulting in low screening efficiency.

[0006] In view of this, the present invention provides a method and device for screening fermentation strains of β-glucan and β-glucan to solve the above problems. Summary of the invention

[0007] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solution: a method for screening a fermentation strain of β-glucan, applied in a fermentation device, comprising:

[0008] Step 100, collecting fermentation control data of the fermentation strains in the fermentation equipment per unit time, and collecting growth image information of the fermentation strains at the end of the unit time;

[0009] Step 101, extracting features from fermentation control data to obtain fermentation feature data;

[0010] Step 102: input the pre-collected strain type information, initial inoculation data, fermentation characteristic data and growth image information into the pre-trained yield prediction model, and output the predicted value of β-glucan yield after a unit time in the future;

[0011] Step 103: based on the comparison result between the output prediction value and the preset output threshold, determine whether to generate a correction instruction for the fermentation control data;

[0012] Step 104: if no correction instruction is generated, the fermentation strain in the corresponding fermentation equipment is used as the fermentation strain of β-glucan; if a correction instruction is generated, the strain type information, initial inoculation data, fermentation characteristic data and yield prediction value are input into the pre-trained feedback control model, and the correction data of the fermentation control data is output to obtain new fermentation control data, and replace the fermentation control data in step 100;

[0013] Step 105, execute steps 100 to 104, and end when the execution times reaches a preset number.

[0014] Furthermore, the training method of the yield prediction model includes:

[0015] Collecting a yield data set, the yield data set includes i groups of yield data, the yield data includes strain type information, initial inoculation data, fermentation characteristic data, growth image information and corresponding yield prediction value, i is an integer greater than 1;

[0016] The yield data set is divided into a training set and a test set, and a classifier is constructed. The strain type information, initial inoculation data, fermentation characteristic data and growth image information in the training set are used as input data, and the yield prediction value in the training set is used as output data. The classifier that meets the preset accuracy is output as a yield prediction model; the yield prediction model is a convolutional neural network CNN model.

[0017] Furthermore, the correction data includes an ambient temperature correction value, a ventilation volume correction value, a pH value correction value and a stirring speed correction value;

[0018] The training method of the feedback control model includes:

[0019] A feedback data set is collected, the feedback data set includes s groups of feedback data, the feedback data includes strain type information, initial inoculation data, fermentation characteristic data, yield prediction value and corresponding correction data, i is an integer greater than 1; the yield data set is divided into a training set and a test set, and a classifier is constructed, the strain type information, initial inoculation data, fermentation characteristic data and yield prediction value in the training set are used as input data, and the correction data in the training set is used as output data, and a classifier that meets a preset accuracy is output as a yield prediction model; the yield prediction model is a convolutional neural network RNN ​​model.

[0020] Furthermore, the method for determining whether to generate a correction instruction for fermentation control data includes:

[0021] When the output prediction value is less than the preset output threshold, a correction instruction is generated; when the output prediction value is greater than or equal to the preset output threshold, no correction instruction is generated.

[0022] Furthermore, the strain type information includes yeast strains, fungal strains and bacterial strains; the initial inoculation data includes inoculation volume ratio, culture medium type and sugar concentration; the inoculation volume ratio is the initial inoculation amount of the fermentation strain in the culture medium; the sugar concentration includes glucose concentration, fructose concentration and sucrose concentration; the fermentation control data includes ambient temperature, pH value, stirring speed, ventilation volume and carbon-nitrogen ratio.

[0023] Further, the fermentation characteristic data include the average value of ambient temperature per unit time, the standard deviation of ambient temperature, the average value of pH value, the standard deviation of pH value, the average value of stirring speed, the standard deviation of stirring speed, the average value of ventilation, the standard deviation of ventilation, the average value of carbon-nitrogen ratio and the standard deviation of carbon-nitrogen ratio;

[0024] The method for obtaining the temperature average value and the temperature standard deviation is as follows:

[0025] Establish a temperature analysis set based on the ambient temperature in unit time, and calculate the temperature average and temperature standard deviation in the temperature analysis set;

[0026] The pH values ​​in a unit time are used to establish a pH value analysis set, and the average pH value and the pH value standard deviation in the pH value analysis set are calculated;

[0027] Establish a stirring speed analysis set based on the stirring speed in unit time, and calculate the stirring speed average value and stirring speed standard deviation in the stirring speed analysis set;

[0028] Establish a ventilation analysis set based on the ventilation in unit time, and calculate the ventilation mean value and ventilation standard deviation in the ventilation analysis set;

[0029] A carbon-nitrogen ratio analysis set is established based on the carbon-nitrogen ratio within a unit time, and the average carbon-nitrogen ratio and the standard deviation of the carbon-nitrogen ratio in the carbon-nitrogen ratio analysis set are calculated.

[0030] Furthermore, the growth image information includes fermentation pictures, fermentation time, ambient brightness and shooting distance;

[0031] The fermentation picture is captured by a microscopic camera; the shooting distance is the distance between the microscopic camera and the surface of the culture medium of the fermentation strain to be detected when the fermentation picture is captured by the microscopic camera, and the distance acquisition steps are as follows:

[0032] When the microscopic camera takes pictures of fermentation, it sends short pulses of light to the surface of the culture medium, and the light pulse emission time T1 is recorded by the microscopic camera;

[0033] The light signal reflected from the surface of the culture medium is received by the microscopic camera, and the moment T2 when the light signal is received is recorded by the microscopic camera;

[0034] By calculating the time difference between the emission and reception of the short pulse of light, the time it takes for the light signal to reach the surface of the culture medium from the microscope camera and then return is determined and recorded as the flight time FXT, FXT = |T2-T1|;

[0035] By formula Calculate the distance between the culture medium surface and the microscope camera;

[0036] Wherein, DB is the distance between the culture medium surface and the microscope camera; c is the speed of light.

[0037] Furthermore, between step 101 and step 102, the method further includes: inputting the pre-collected strain type information, initial inoculation data and growth image information into the pre-trained state evaluation model, and outputting the evaluation result of the fermentation strain; the evaluation result includes abnormal and normal;

[0038] In step 102, the input data of the yield prediction model also includes evaluation results, and the yield data also includes evaluation results;

[0039] In step 104, the input data of the feedback control model further includes an evaluation result, and the feedback data further includes an evaluation result;

[0040] The training methods of the state assessment model include:

[0041] Collecting a state assessment data set, wherein the state assessment data set includes r groups of state assessment data, the state assessment data includes strain type information, initial inoculation data, growth image information and corresponding assessment results, where r is an integer greater than 1;

[0042] The state evaluation data set is divided into a training set and a test set, and a classifier is constructed. The strain type information, initial inoculation data, and growth image information in the training set are used as input data, and the evaluation results in the training set are used as output data. The classifier is trained to obtain an initial classifier, and the initial classifier is tested using the test set. A classifier that meets the preset accuracy is output as a state evaluation model; the state evaluation model is a ResNet50 model.

[0043] The β-glucan fermentation strain screening device is used in a fermentation device to implement the β-glucan fermentation strain screening method, comprising:

[0044] A collection module is used to collect fermentation control data of the fermentation strains in the fermentation equipment per unit time, and to collect growth image information of the fermentation strains at the end of the unit time;

[0045] A feature extraction module is used to extract features from fermentation control data to obtain fermentation feature data;

[0046] The yield prediction module inputs the pre-collected strain type information, initial inoculation data, fermentation characteristic data and growth image information into the pre-trained yield prediction model, and outputs the predicted value of β-glucan yield after a unit time in the future;

[0047] A judgment module, based on a comparison result between the output prediction value and a preset output threshold value, determines whether to generate a correction instruction for the fermentation control data;

[0048] The instruction execution module, if no correction instruction is generated, uses the fermentation strain in the corresponding fermentation equipment as the fermentation strain of β-glucan; if the correction instruction is generated, inputs the strain type information, initial inoculation data, fermentation characteristic data and yield prediction value into the pre-trained feedback control model, outputs the correction data of the fermentation control data, obtains the new fermentation control data, and replaces the fermentation control data in the acquisition module;

[0049] The loop control module executes the corresponding steps of the acquisition module, feature extraction module, yield prediction module, judgment module, and instruction execution module in sequence, and ends when the number of executions reaches a preset number.

[0050] The β-glucan is prepared by the fermentation strain obtained by the β-glucan fermentation strain screening method.

[0051] The technical effects and advantages of the β-glucan fermentation strain screening method and device and the β-glucan of the present invention are as follows:

[0052] When screening multiple groups of β-glucan fermentation strains, the β-glucan yield of the fermentation strains is predicted in advance by building a yield prediction model, and the fermentation control data is adaptively corrected according to the prediction results, so that the fermentation strains can produce β-glucan under the most appropriate conditions. This not only avoids missing potential excellent fermentation strains, but also reduces the cultivation cost and time of low-yield β-glucan fermentation strains, thereby improving the efficiency of screening β-glucan fermentation strains.

[0053] Specifically, by constructing a yield prediction model based on a convolutional neural network, and inputting multi-dimensional data, including strain type information, initial inoculation data, fermentation characteristic data and growth image information into the yield prediction model, the accuracy of the β-glucan yield prediction value is effectively improved; then, by inputting the strain type information, initial inoculation data, fermentation characteristic data and yield prediction value into a pre-trained feedback control model, correction data is obtained, so that the ambient temperature, ventilation volume, pH value and stirring speed can be adaptively adjusted to avoid missing the fermentation strains of β-glucan, thereby improving the effectiveness of screening the fermentation strains; finally, according to the comparison result between the yield prediction value and the preset yield threshold, the cultivated fermentation strains are screened, and the β-glucan fermentation strains and the corresponding strain type information, initial inoculation data and fermentation control data are recorded, thereby improving the efficiency of the overall screening process. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a schematic diagram of a β-glucan fermentation strain screening device according to Example 1 of the present invention;

[0055] Figure 2 This is a flow chart of the method for screening β-glucan fermentation strains according to Example 2 of the present invention;

[0056] Figure 3 This is a schematic diagram of an electronic device according to Embodiment 3 of the present invention;

[0057] Figure 4 This is a schematic diagram of a storage medium according to Embodiment 4 of the present invention. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] Example 1

[0060] See also Figure 1As shown, the β-glucan fermentation strain screening device described in this embodiment is applied to fermentation equipment, including a collection module, a feature extraction module, a state evaluation module, a yield prediction module, a judgment module, an instruction execution module and a cycle control module, and each module is connected by wire and / or wirelessly.

[0061] The acquisition module is used to collect fermentation control data of the fermentation strains in the fermentation equipment per unit time, and to collect growth image information of the fermentation strains at the end of the unit time; the fermentation control data includes ambient temperature, pH value, stirring speed, ventilation volume and carbon-nitrogen ratio; the growth image information includes fermentation pictures, fermentation time, ambient brightness and shooting distance.

[0062] Ambient temperature is an important data in the fermentation process of fermentation strains. Different fermentation strains have different optimal growth temperature ranges. Exceeding the optimal growth temperature range will inhibit the metabolic activity of the fermentation strain, thereby affecting the production of β-glucan. By obtaining the ambient temperature, it is helpful to improve the accuracy of the subsequent yield prediction model. At the same time, by adjusting the ambient temperature, the fermentation strain culture process can be ensured to be stable, avoiding temperature fluctuations that lead to reduced efficiency of the fermentation strain in producing β-glucan.

[0063] During the fermentation process of the fermentation strain, the fermentation strain will produce acidic or alkaline metabolites, resulting in changes in the pH value of the culture medium. The pH value affects the enzyme activity of the fermentation strain, thereby affecting the normal metabolism of the fermentation strain and further affecting the yield of β-glucan. By obtaining the pH value, it is helpful to improve the accuracy of the subsequent yield prediction model. At the same time, by adjusting the pH value, the negative impact of pH fluctuations on the yield of β-glucan can be avoided.

[0064] Dissolved oxygen is a key factor for aerobic fermentation strains to carry out metabolic activities; the higher the stirring speed, the higher the dissolved oxygen concentration, which helps to improve the metabolic efficiency of aerobic fermentation strains, thereby increasing the yield of β-glucan; at the same time, too high a stirring speed will cause mechanical damage to the fermentation strain, affecting the metabolic activity of the fermentation strain, thereby reducing the yield of β-glucan; by obtaining the stirring speed, it is helpful to improve the accuracy of the subsequent yield prediction model, and at the same time, the stirring speed can be adaptively adjusted to optimize the fermentation conditions of the fermentation strain and increase the yield of β-glucan.

[0065] The ventilation volume directly affects the dissolved oxygen concentration in the culture medium for cultivating fermentation strains, which in turn affects the metabolic efficiency of the fermentation strains and the yield of β-glucan. At the same time, the ventilation volume also affects the discharge of carbon dioxide produced during the fermentation process. Excessive carbon dioxide will inhibit the growth of fermentation strains and reduce fermentation efficiency. By obtaining the ventilation volume, the subsequent yield prediction model can output a more accurate output value of β-glucan production. The ventilation volume can be adaptively adjusted to ensure a stable supply of dissolved oxygen in the culture medium, which helps to increase the yield of β-glucan.

[0066] When the carbon source is sufficient but the nitrogen source is relatively insufficient, the fermentation strain tends to synthesize carbon storage metabolites, such as lipids and sugars, which are conducive to the synthesis and production of β-glucan; when the nitrogen source is sufficient but the carbon source is relatively insufficient, the fermentation strain will give priority to the synthesis of proteins and nucleic acids to support cell growth and reproduction, but will reduce the synthesis efficiency of target products, such as β-glucan; therefore, by obtaining the carbon-nitrogen ratio, the accuracy of the subsequent yield prediction model can be effectively improved.

[0067] By regularly collecting growth image information and obtaining the dynamic growth process of the fermentation strain during the fermentation process, it is helpful to improve the accuracy of subsequent evaluation of the health status of the fermentation strain, and thus help improve the accuracy of the subsequent yield prediction model.

[0068] The fermentation picture is the picture of the fermentation strain to be analyzed, which serves as the input of the subsequent status evaluation model; by recording the fermentation time, it is helpful to identify the growth stage of the strain, so as to judge the current environmental brightness, which affects the contrast and clarity of the fermentation picture, affects the observation of the color of the fermentation strain, and thus affects the accuracy of the subsequent evaluation of the health status of the fermentation flora. The shooting distance affects the size of the fermentation strain in the fermentation picture and the cell size of the fermentation strain, thereby affecting the accuracy of the subsequent evaluation of the health status of the fermentation flora.

[0069] The ambient temperature is collected by a temperature sensor; the pH value is collected by a PH sensor; the stirring speed is collected by a speed sensor installed on the agitator shaft; the ventilation volume is collected by a gas flow meter installed in the ventilation pipe; and the carbon-nitrogen ratio is collected by an online analyzer.

[0070] The fermentation picture is captured by a microscope camera; the fermentation time is captured by a system clock; and the ambient brightness is captured by an ambient light sensor.

[0071] The shooting distance is the distance between the microscopic camera and the surface of the culture medium of the fermentation strain to be detected when the fermentation picture is taken by the microscopic camera. The distance acquisition steps are as follows:

[0072] When the microscope camera takes fermentation pictures, it sends short pulses of light to the surface of the culture medium, and records the light pulse emission time T1 through the microscope camera.

[0073] The light signal reflected from the surface of the culture medium is received by the microscope camera, and the moment T2 when the light signal is received is recorded by the microscope camera.

[0074] By calculating the time difference between the emission and reception of the short pulse of light, the time it takes for the light signal to reach the culture medium surface from the microscope camera and then return is determined and recorded as the flight time FXT, FXT = |T2-T1|.

[0075] By formula Calculate the distance between the culture medium surface and the microscope camera.

[0076] Wherein, DB is the distance between the culture medium surface and the microscope camera; c is the speed of light.

[0077] The state evaluation module inputs the trained state evaluation model based on the strain type information, initial inoculation data and growth image information to obtain the current evaluation result of the fermentation strain, which includes abnormal and normal.

[0078] The feature extraction module is used to extract features from the fermentation control data to obtain fermentation feature data; the fermentation feature data includes the average ambient temperature per unit time, the standard deviation of the ambient temperature, the average pH value, the standard deviation of the pH value, the average stirring speed, the standard deviation of the stirring speed, the average ventilation volume, the standard deviation of the ventilation volume, the average carbon-nitrogen ratio, and the standard deviation of the carbon-nitrogen ratio.

[0079] Specifically, the method for obtaining the temperature mean and temperature standard deviation is as follows:

[0080] The ambient temperature in unit time is used to establish a temperature analysis set, and the temperature average and temperature standard deviation in the temperature analysis set are calculated.

[0081]

[0082] In the formula, WD is the standard deviation of ambient temperature per unit time; W h is the hth temperature in the temperature analysis set; is the average temperature in the temperature analysis set; H is the number of temperatures in the temperature analysis set, and H is a positive real number greater than 1.

[0083] The same method for obtaining the temperature average and temperature standard deviation was used to obtain the pH average, pH standard deviation, stirring speed average, stirring speed standard deviation, aeration average, aeration standard deviation, carbon-nitrogen ratio average and carbon-nitrogen ratio standard deviation.

[0084] The state evaluation module inputs the pre-collected strain type information, initial inoculation data and growth image information into the pre-trained state evaluation model, and outputs the evaluation result of the fermentation strain; the evaluation result includes abnormal and normal; the strain type information and the initial inoculation data are pre-collected by the staff before the fermentation of the fermentation strain, and input into the fermentation strain screening device, the strain type information includes yeast strains, fungal strains and bacterial strains; the initial inoculation data includes the inoculation volume ratio, the culture medium type and the sugar concentration; the inoculation volume ratio is the initial inoculation amount of the fermentation strain in the culture medium; the culture medium types include YDP culture medium, PDA culture medium and LB culture medium; the sugar concentration includes glucose concentration, fructose concentration and sucrose concentration.

[0085] Due to different strain type information, different requirements are placed on the culture environment when cultivating fermentation strains to produce β-glucan.

[0086] When cultivating yeast strains to produce β-glucan, YPD medium is selected, the inoculation volume ratio is controlled between 1% and 10%, the ambient temperature range is controlled between 28 and 30 degrees Celsius, the pH range is controlled between 4.5 and 6.0, the stirring speed is controlled between 150 and 200 rpm, the ventilation volume is controlled between 0.5 and 1 vvm, and the carbon-nitrogen ratio is controlled between 20 and 40.

[0087] When cultivating fungal strains to produce β-glucan, PDA medium is selected, the inoculation volume ratio is controlled between 5% and 10%, the ambient temperature range is controlled between 25 and 28 degrees Celsius, the pH range is controlled between 4.0 and 6.0, the stirring speed is controlled between 100 and 150 rpm, the ventilation volume is controlled between 0.5 and 1 vvm, and the carbon-nitrogen ratio is controlled between 20 and 30.

[0088] When cultivating bacterial strains to produce β-glucan, LB medium is selected, the inoculation volume ratio is controlled between 1% and 5%, the ambient temperature range is controlled between 30 and 37 degrees Celsius, the pH range is controlled between 6.0 and 6.5, the stirring speed is controlled between 200 and 250 rpm, the ventilation volume is controlled between 1 and 2 vvm, and the carbon-nitrogen ratio is controlled between 10 and 20.

[0089] Obtaining the strain type information of the fermentation strain can help in the subsequent selection of the culture medium type, and also help in the subsequent determination of the range of fermentation control data that is most suitable for the fermentation strain to produce β-glucan.

[0090] The inoculation volume ratio of the fermentation strain affects the fermentation efficiency and production output of the fermentation strain in producing β-glucan. Selecting an appropriate inoculation volume ratio can enable the fermentation strain to quickly enter the logarithmic growth phase and avoid a long lag phase, thereby improving the fermentation efficiency and production output. Therefore, by obtaining the inoculation volume ratio, the accuracy of the subsequent yield prediction model in predicting the β-glucan yield after unit time is improved.

[0091] Different types of fermentation strains have different requirements for culture media. Different types of culture media are selected based on the strain type information to provide a suitable nutritional environment for the fermentation strain so that the fermentation strain can carry out normal metabolic activities. At the same time, the composition of the culture medium will affect the type, quality and quantity of metabolites during the cultivation of the fermentation strain. By selecting the appropriate type of culture medium, the yield of metabolites, such as β-glucan, can be increased.

[0092] Sugar is the main carbon source for fermentation strains during the fermentation process. Different sugar concentrations will affect the growth rate and metabolic pathways of the fermentation strains, thereby affecting the target product, such as the yield of β-glucan. Glucose, fructose and sucrose are commonly used carbon sources, and each has different effects on the metabolism of the strains. For example, glucose is the main carbon source for yeast. By obtaining the sugar concentration, the actual yield of β-glucan can be accurately determined, thereby improving the accuracy of the subsequent yield prediction model.

[0093] The strain type information is obtained through the label corresponding to the strain preservation tube; the inoculation volume ratio is pre-set and obtained by professionals in this field according to the strain type information of the fermentation strain; the culture medium type is matched and obtained according to the strain type information of the fermentation strain; and the sugar concentration is obtained according to the culture medium type.

[0094] For example, if the strain type information of the fermentation strain is a yeast strain, the corresponding inoculation volume ratio is 5%, the culture medium type is YPD culture medium, the glucose concentration in the sugar concentration is 20 g / L, the fructose concentration is 15 g / L, and the sucrose concentration is 10 g / L.

[0095] A microscope camera is used to capture the growth image of the fermentation strain. Combined with the ambient brightness, shooting distance and fermentation time, the strain type information and the initial inoculation data, the health status of the fermentation strain is judged in real time through the pre-trained state assessment model. This can effectively distinguish between normal fermentation strains and abnormal fermentation strains, ensuring the accuracy of screening the fermentation strains.

[0096] The training methods of the state assessment model include:

[0097] A state assessment data set is collected, wherein the state assessment data set includes r groups of state assessment data, the state assessment data includes r groups of strain type information, initial inoculation data, growth image information and corresponding assessment results, and r is an integer greater than 1.

[0098] It should be noted that the evaluation results are calibrated by professionals in this field based on the strain type information, initial inoculation data, and characteristic information in the fermentation pictures in the growth image information. The characteristic information in the fermentation pictures includes the bacterial color of the fermentation strain, the size of the fermentation strain, and the cell size of the fermentation strain.

[0099] The state evaluation data set is divided into a training set and a test set, and a classifier is constructed. The strain type information, initial inoculation data, and growth image information in the training set are used as input data, and the evaluation results in the training set are used as output data. The classifier is trained to obtain an initial classifier, and the initial classifier is tested using the test set. A classifier that meets the preset accuracy is output as a state evaluation model; the state evaluation model is a ResNet50 model or other suitable models.

[0100] For example, the strain type information is a yeast strain, in the initial inoculation data, the inoculation volume accounts for 5%, the culture medium type is YPD culture medium, the sugar concentration is 20 g / L, and in the growth picture information, the fermentation time is 4 hours, the ambient brightness is 500 LUX, and the shooting distance is 10 cm. In the feature information of the fermentation picture, the bacterial body color is white, the size is 10 microns, and the cell size is 3 microns, then the evaluation result is normal.

[0101] The strain type information is yeast, the initial inoculation data shows that the inoculation volume accounts for 5%, the culture medium type is YPD culture medium, the sugar concentration is 30 g / L, the growth picture information shows that the fermentation time is 4 hours, the ambient brightness is 500 LUX, the shooting distance is 10 cm, and the characteristic information in the fermentation picture shows that the bacterial color is white, the size is 13 microns, and the cell size is 4 microns. The evaluation result is abnormal;

[0102] The yield prediction module inputs the pre-collected strain type information, initial inoculation data, fermentation characteristic data and growth image information into the pre-trained yield prediction model, and outputs the predicted value of β-glucan yield after a unit time in the future.

[0103] The training method of the yield prediction model includes:

[0104] A yield data set is collected, where the yield data set includes i groups of yield data, including i groups of strain type information, initial inoculation data, fermentation characteristic data, growth image information, evaluation results, and corresponding β-glucan yield prediction values, where i is an integer greater than 1.

[0105] It should be noted that the yield data set is obtained through historical experimental data or laboratory experiments, and the β-glucan yield prediction value is calibrated to the actual value of β-glucan yield after a unit time in the future.

[0106] The yield data set is divided into a training set and a test set, and a classifier is constructed. The strain type information, initial inoculation data, fermentation characteristic data, growth image information and evaluation results in the training set are used as input data, and the β-glucan yield prediction value in the training set is used as output data. A classifier that meets the preset accuracy is output as a yield prediction model; the yield prediction model is a convolutional neural network CNN or other suitable models.

[0107] Exemplarily, the strain type information is a yeast strain, in the initial inoculation data, the inoculation volume accounts for 5%, the culture medium type is YPD culture medium, the sugar concentration is 20 g / L, in the growth picture information, the fermentation time is 4 hours, the ambient brightness is 500 LUX, and the shooting distance is 10 cm. In the characteristic information in the fermentation picture, the bacterial body color is white, the size is 10 microns and the cell size is 3 microns. In the fermentation characteristic data, the ambient temperature is 29 degrees Celsius, the pH value is 5.5, the stirring speed is 180 rpm, the ventilation volume is 0.8 vvm, the carbon-nitrogen ratio is 30, the evaluation result is normal, and the predicted value of β-glucan production is 150 g / L.

[0108] The judgment module determines whether to generate a correction instruction for the fermentation control data based on a comparison result between the output prediction value and a preset output threshold.

[0109] The determination method for generating a correction instruction for fermentation control data includes:

[0110] When the output prediction value is less than the preset output threshold, a correction instruction is generated; when the output prediction value is greater than or equal to the preset output threshold, no correction instruction is generated.

[0111] The instruction execution module, if no correction instruction is generated, uses the fermentation strain in the corresponding fermentation equipment as the fermentation strain of β-glucan; if the correction instruction is generated, inputs the strain type information, initial inoculation data, fermentation characteristic data and yield prediction value into the pre-trained feedback control model, outputs the correction data of the fermentation control data, obtains the new fermentation control data, and replaces the fermentation control data in the acquisition module; the correction data includes the correction value of the ambient temperature, the correction value of the ventilation volume, the correction value of the pH value and the correction value of the stirring speed;

[0112] The training method of the feedback control model includes:

[0113] A feedback data set is collected, wherein the feedback data set includes s groups of strain type information, initial inoculation data, fermentation characteristic data, evaluation results, β-glucan yield prediction values ​​and corresponding correction data, where i is an integer greater than 1; the yield data set is divided into a training set and a test set, and a classifier is constructed, wherein the strain type information, initial inoculation data, fermentation characteristic data, evaluation results and β-glucan yield prediction values ​​in the training set are used as input data, and the correction data in the training set are used as output data, and a classifier that meets a preset accuracy is output as a yield prediction model; the yield prediction model is a convolutional neural network RNN ​​or other suitable models.

[0114] It should be noted that the feedback data set was obtained through fermentation experiments.

[0115] For example, the strain type information is a fungal strain, in the initial inoculation data, the inoculation volume accounts for 6%, the culture medium type is PDA culture medium, the sugar concentration is 25 g / L, and in the fermentation characteristic data, the ambient temperature is 27 degrees Celsius, the pH value is 5.0, the stirring speed is 130 rpm, the ventilation volume is 0.6 vvm, the carbon-nitrogen ratio is 25, the evaluation result is abnormal, and the β-glucan yield prediction value is 110 g / L. In the output correction data, the ambient temperature correction value is +2 degrees Celsius, the ventilation correction value is +0.3 vvm, the pH correction value is -0.5, and the stirring speed correction value is +10 rpm.

[0116] The loop control module executes the corresponding steps of the acquisition module, feature extraction module, state evaluation module, output prediction module, judgment module, and instruction execution module in sequence, and ends when the number of executions reaches a preset number.

[0117] Example 2

[0118] See also Figure 2 As shown, this embodiment provides a method for screening β-glucan fermentation strains, which is applied to a fermentation device, comprising:

[0119] Step 100, collecting fermentation control data of the fermentation strains in the fermentation equipment per unit time, and collecting growth image information of the fermentation strains at the end of the unit time;

[0120] Step 101, extracting features from fermentation control data to obtain fermentation feature data;

[0121] Step 102: input the pre-collected strain type information, initial inoculation data, fermentation characteristic data and growth image information into the pre-trained yield prediction model, and output the predicted value of β-glucan yield after a unit time in the future;

[0122] Step 103: based on the comparison result between the output prediction value and the preset output threshold, determine whether to generate a correction instruction for the fermentation control data;

[0123] Step 104: if no correction instruction is generated, the fermentation strain in the corresponding fermentation equipment is used as the fermentation strain of β-glucan; if a correction instruction is generated, the strain type information, initial inoculation data, fermentation characteristic data and yield prediction value are input into the pre-trained feedback control model, and the correction data of the fermentation control data is output to obtain new fermentation control data, and replace the fermentation control data in step 100;

[0124] Step 105, execute steps 100 to 104, and end when the execution times reaches a preset number.

[0125] Example 3

[0126] See also Figure 3 As shown, the present application also provides an electronic device 500. The electronic device 500 may include one or more processors and one or more memories. The memories store computer readable codes, and when the computer readable codes are run by one or more processors, the above-mentioned method for screening fermentation strains of β-glucan can be executed.

[0127] The method or system according to the embodiment of the present application can also be used Figure 3 The electronic device architecture shown in FIG. Figure 3 As shown, the electronic device 500 may include a bus 501, one or more CPUs 502, ROM 503, RAM 504, a communication port 505 connected to a network, an input / output 506, a hard disk 507, etc. The storage device in the electronic device 500, such as ROM 503 or hard disk 507, may store the fermentation strain screening method for β-glucan provided in the present application. Further, the electronic device 500 may also include a user interface 508. Of course, Figure 3 The architecture shown is only exemplary and can be omitted according to actual needs when implementing different devices. Figure 3 One or more components of an electronic device are shown.

[0128] Example 4

[0129] See also Figure 4 As shown, the present application also provides a computer-readable storage medium 600. Computer-readable instructions are stored on the computer-readable storage medium 600. When the computer-readable instructions are executed by the processor, the fermentation strain screening method of β-glucan according to the embodiment of the present application described with reference to the above figures can be executed. The storage medium 600 includes, but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0130] In addition, according to the embodiments of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the present application provides a non-transitory machine-readable storage medium, which stores machine-readable instructions, and the machine-readable instructions can be executed by a processor to execute instructions corresponding to the method steps provided in the present application, for example: a method for screening fermentation strains of β-glucan. When the computer program is executed by a central processing unit (CPU), the above functions defined in the method of the present application are executed.

[0131] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0132] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for screening β-glucan fermentation strains, applied to a fermentation device, characterized in that: include: Step 100, collecting fermentation control data of the fermentation strains in the fermentation equipment per unit time, and collecting growth image information of the fermentation strains at the end of the unit time; Step 101, extracting features from fermentation control data to obtain fermentation feature data; Step 102: input the pre-collected strain type information, initial inoculation data, fermentation characteristic data and growth image information into the pre-trained yield prediction model, and output the predicted value of β-glucan yield after a unit time in the future; Step 103: based on the comparison result between the output prediction value and the preset output threshold, determine whether to generate a correction instruction for the fermentation control data; Step 104: if no correction instruction is generated, the fermentation strain in the corresponding fermentation equipment is used as the fermentation strain of β-glucan; if a correction instruction is generated, the strain type information, initial inoculation data, fermentation characteristic data and yield prediction value are input into the pre-trained feedback control model, and the correction data of the fermentation control data is output to obtain new fermentation control data, and replace the fermentation control data in step 100; Step 105, execute steps 100 to 104, and end when the execution times reaches a preset number.

2. The method for screening β-glucan fermentation strains according to claim 1, characterized in that: The training method of the yield prediction model includes: Collecting a yield data set, the yield data set includes i groups of yield data, the yield data includes strain type information, initial inoculation data, fermentation characteristic data, growth image information and corresponding yield prediction value, i is an integer greater than 1; The yield data set is divided into a training set and a test set, and a classifier is constructed. The strain type information, initial inoculation data, fermentation characteristic data and growth image information in the training set are used as input data, and the yield prediction value in the training set is used as output data. The classifier that meets the preset accuracy is output as a yield prediction model; the yield prediction model is a convolutional neural network CNN model.

3. The method for screening β-glucan fermentation strains according to claim 1, characterized in that: The correction data include an ambient temperature correction value, a ventilation volume correction value, a pH value correction value and a stirring speed correction value; The training method of the feedback control model includes: A feedback data set is collected, the feedback data set includes s groups of feedback data, the feedback data includes strain type information, initial inoculation data, fermentation characteristic data, yield prediction value and corresponding correction data, i is an integer greater than 1; the yield data set is divided into a training set and a test set, and a classifier is constructed, the strain type information, initial inoculation data, fermentation characteristic data and yield prediction value in the training set are used as input data, and the correction data in the training set is used as output data, and a classifier that meets a preset accuracy is output as a yield prediction model; the yield prediction model is a convolutional neural network RNN ​​model.

4. The method for screening β-glucan fermentation strains according to claim 1, characterized in that: The determination method for generating a correction instruction for fermentation control data includes: When the output prediction value is less than the preset output threshold, a correction instruction is generated; when the output prediction value is greater than or equal to the preset output threshold, no correction instruction is generated.

5. The method for screening β-glucan fermentation strains according to claim 3, characterized in that: The strain type information includes yeast strains, fungal strains and bacterial strains; the initial inoculation data includes inoculation volume ratio, culture medium type and sugar concentration; the inoculation volume ratio is the initial inoculation amount of the fermentation strain in the culture medium; the sugar concentration includes glucose concentration, fructose concentration and sucrose concentration; the fermentation control data includes ambient temperature, pH value, stirring speed, ventilation volume and carbon-nitrogen ratio.

6. The method for screening β-glucan fermentation strains according to claim 5, characterized in that: The fermentation characteristic data include the average value of ambient temperature per unit time, the standard deviation of ambient temperature, the average value of pH value, the standard deviation of pH value, the average value of stirring speed, the standard deviation of stirring speed, the average value of ventilation, the standard deviation of ventilation, the average value of carbon-nitrogen ratio and the standard deviation of carbon-nitrogen ratio; The method for obtaining the temperature average value and the temperature standard deviation is as follows: Establish a temperature analysis set based on the ambient temperature in unit time, and calculate the temperature average and temperature standard deviation in the temperature analysis set; The pH values ​​in a unit time are used to establish a pH value analysis set, and the average pH value and the pH value standard deviation in the pH value analysis set are calculated; Establish a stirring speed analysis set based on the stirring speed in unit time, and calculate the stirring speed average value and stirring speed standard deviation in the stirring speed analysis set; Establish a ventilation analysis set based on the ventilation in unit time, and calculate the ventilation mean value and ventilation standard deviation in the ventilation analysis set; A carbon-nitrogen ratio analysis set is established based on the carbon-nitrogen ratio within a unit time, and the average carbon-nitrogen ratio and the standard deviation of the carbon-nitrogen ratio in the carbon-nitrogen ratio analysis set are calculated.

7. The method for screening β-glucan fermentation strains according to claim 4, characterized in that: The growth image information includes fermentation pictures, fermentation time, ambient brightness and shooting distance; The fermentation picture is captured by a microscopic camera; the shooting distance is the distance between the microscopic camera and the surface of the culture medium of the fermentation strain to be detected when the fermentation picture is captured by the microscopic camera, and the distance acquisition steps are as follows: When the microscopic camera takes pictures of fermentation, it sends short pulses of light to the surface of the culture medium, and the light pulse emission time T1 is recorded by the microscopic camera; The light signal reflected from the surface of the culture medium is received by the microscopic camera, and the moment T2 when the light signal is received is recorded by the microscopic camera; By calculating the time difference between the emission and reception of the short pulse of light, the time it takes for the light signal to reach the surface of the culture medium from the microscope camera and then return is determined and recorded as the flight time FXT, FXT = |T2-T1|; By formula Calculate the distance between the culture medium surface and the microscope camera; Wherein, DB is the distance between the culture medium surface and the microscope camera; c is the speed of light.

8. The method for screening β-glucan fermentation strains according to claim 4, characterized in that: Between step 101 and step 102, the method further includes: inputting the pre-collected strain type information, initial inoculation data and growth image information into the pre-trained state evaluation model, and outputting the evaluation result of the fermentation strain; the evaluation result includes abnormal and normal; In step 102, the input data of the yield prediction model also includes evaluation results, and the yield data also includes evaluation results; In step 104, the input data of the feedback control model further includes an evaluation result, and the feedback data further includes an evaluation result; The training methods of the state assessment model include: Collecting a state assessment data set, wherein the state assessment data set includes r groups of state assessment data, the state assessment data includes strain type information, initial inoculation data, growth image information and corresponding assessment results, where r is an integer greater than 1; The state evaluation data set is divided into a training set and a test set, and a classifier is constructed. The strain type information, initial inoculation data, and growth image information in the training set are used as input data, and the evaluation results in the training set are used as output data. The classifier is trained to obtain an initial classifier, and the initial classifier is tested using the test set. A classifier that meets the preset accuracy is output as a state evaluation model; the state evaluation model is a ResNet50 model.

9. A device for screening fermentation strains of β-glucan, used in fermentation equipment, characterized in that: The method for screening a β-glucan fermentation strain according to any one of claims 1 to 8 comprises: A collection module is used to collect fermentation control data of the fermentation strains in the fermentation equipment per unit time, and to collect growth image information of the fermentation strains at the end of the unit time; A feature extraction module is used to extract features from fermentation control data to obtain fermentation feature data; The yield prediction module inputs the pre-collected strain type information, initial inoculation data, fermentation characteristic data and growth image information into the pre-trained yield prediction model, and outputs the predicted value of β-glucan yield after a unit time in the future; A judgment module, based on a comparison result between the output prediction value and a preset output threshold value, determines whether to generate a correction instruction for the fermentation control data; The instruction execution module, if no correction instruction is generated, uses the fermentation strain in the corresponding fermentation equipment as the fermentation strain of β-glucan; if the correction instruction is generated, inputs the strain type information, initial inoculation data, fermentation characteristic data and yield prediction value into the pre-trained feedback control model, outputs the correction data of the fermentation control data, obtains the new fermentation control data, and replaces the fermentation control data in the acquisition module; The loop control module executes the corresponding steps from the acquisition module to the instruction execution module in sequence, and ends when the number of executions reaches a preset number.

10. β-glucan, characterized in that The method is prepared by using a fermentation strain obtained by the method for screening a fermentation strain of β-glucan according to any one of claims 1 to 8.

Images