A kind of bean product production process intelligent management method and system

Abstract

The invention discloses an intelligent management method and system for a bean product production process, belonging to the technical field of production management. The invention identifies abnormal soybeans through an abnormal soybean identification module, retains qualified soybeans, avoids the subsequent prediction data and product quality being affected by the abnormal soybeans, and ensures the accuracy of the prediction data; predicts the optimal soaking water temperature and the optimal soaking time for soaking qualified soybeans through a soaking prediction module, provides the optimal soaking water temperature and time for the soaking process, and improves production efficiency and product quality; through the interaction of a water quality change estimation module and a water change judgment module, determines whether water needs to be changed according to the current accumulated impurities in the water, avoids manually determining the number of water changes and the time for water changes, and further improves production efficiency and product quality.

Description

Technical Field

[0001] The present invention relates to the technical field of production management, and in particular to an intelligent management method and system for a bean product production process. Background Art

[0002] The production process of soy products generally includes soybean quality inspection, soybean soaking, pulping, pulp cooking and product forming. Among them, the soybean soaking stage is extremely easy to be overlooked, but it is an extremely important link, which not only affects the production efficiency of subsequent links, but also affects the quality of the product.

[0003] In the soybean soaking process, if the soaking water temperature is too low, the soybean water absorption and expansion speed will slow down, prolonging the soaking time and reducing production efficiency; if the soaking water temperature is too high, the soybean surface will expand rapidly while the inside has not completely absorbed water, resulting in uneven texture, thereby reducing product quality; if the soaking time is too short, the soybean may not be able to fully absorb water and expand, affecting subsequent processing steps, thereby leading to low production efficiency and product quality; if the soaking time is too long, it may cause the loss of nutrients in the soybean and affect the soybean texture, thereby leading to low product quality. Therefore, choosing the right water temperature and soaking time has a great impact on the production efficiency and product quality of soybean products. In the existing soybean product production management technology, most of them only set the soaking time and temperature based on manual experience;

[0004] At the same time, during the soybean soaking process, long-term soaking may lead to an increase in the number of microorganisms in the water, or an increase in impurities released by soybeans into the water, which reduces the soaking effect and thus affects the product quality. Therefore, the soaking water needs to be replaced. However, in the existing soybean product production management technology, the number and frequency of water changes are only manually set, which leads to reduced production efficiency and product quality.

[0005] Therefore, in order to solve the technical problems existing in the prior art, the present invention provides a method and system for intelligent management of a soy product production process. Summary of the invention

[0006] In order to solve the technical problems existing in the prior art, the present invention provides an intelligent management system for the production process of soy products, comprising:

[0007] The abnormal soybean identification module is used to identify and remove abnormal soybeans from soybeans, retain qualified soybeans and calculate the total weight G of qualified soybeans;

[0008] A soaking prediction module, used to predict the optimal soaking water temperature T and the optimal soaking time J for soaking the qualified soybeans;

[0009] The water quality change estimation module is used to calculate the current accumulated impurities in the water, specifically:

[0010]

[0011]

[0012] Where Q(t) represents the accumulated impurity after time t, Indicates the time when the water was last changed. If the qualified soybeans were soaked before the first water change, , R(t) represents the impurity release rate after time t, K represents the preset water quality deterioration coefficient, and λ represents the preset attenuation rate;

[0013] The water change judgment module is used to judge whether the water needs to be changed according to the current accumulated impurities in the water. Specifically:

[0014] like = ZQ, and the corresponding t satisfies J-t≥yt, then it is determined that water needs to be changed, and the current corresponding t is assigned to ; Where ZQ represents the preset impurity tolerance, and yt represents the preset remaining time.

[0015] Furthermore, the abnormal soybeans include shriveled soybeans;

[0016] The abnormal soybean identification module comprises:

[0017] An image acquisition module, used for acquiring soybean images;

[0018] The shriveled identification module is used to identify shriveled soybeans through the soybean image.

[0019] Furthermore, the step of identifying shriveled soybeans through the soybean image is specifically as follows:

[0020] Identifying the actual contour of each soybean in the soybean image by a contour detection algorithm;

[0021] According to the actual contour of soybean, the contour length C and contour range area S are analyzed, and the circularity is calculated. ;

[0022] The convex hull contour corresponding to the actual contour of each soybean is analyzed by the convex hull algorithm, and then the number of convex hull defects and the corresponding convex hull defect depth of each soybean are analyzed according to the convex hull contour;

[0023] The shrinkage value GBZ is calculated by the roundness of soybeans, the number of convex hull defects n and the corresponding convex hull defect depth L. ,in, represents the first preset weight, represents a second preset weight;

[0024] Soybeans with a shriveled value greater than or equal to a preset shriveled threshold value are judged as shriveled soybeans.

[0025] Furthermore, the analysis of the convex hull contour corresponding to the actual contour of each soybean is specifically analyzed using the convexHull function of OpenCV; the analysis of the number of convex hull defects and the corresponding convex hull defect depth of each soybean is specifically analyzed using the convexityDefects function of OpenCV.

[0026] Furthermore, the soaking prediction module comprises:

[0027] A data acquisition module, used for acquiring the target water absorption multiple mb and the type of the qualified soybeans;

[0028] A water temperature prediction module, used for predicting the optimal soaking water temperature according to the type of the qualified soybeans using a trained first deep learning model;

[0029] The duration prediction module is used to predict the optimal soaking duration according to the target water absorption multiple and the optimal soaking water temperature.

[0030] Furthermore, the predicted optimal soaking time is specifically:

[0031] The second deep learning model that has been trained obtains an average water absorption rate v according to the optimal soaking water temperature and the type of the qualified soybeans;

[0032] The optimal soaking time is calculated according to the target water absorption multiple and the average water absorption rate, J=(mb-1) / v.

[0033] Furthermore, the abnormal soybeans also include moldy soybeans; the abnormal soybean identification module also includes a moldy identification module, which is used to identify moldy soybeans through the soybean image.

[0034] Furthermore, the identifying moldy soybeans through the soybean image is specifically:

[0035] The soybean image is segmented into a plurality of single soybean images, and the single soybean images are used to identify whether the corresponding soybeans are moldy soybeans through the trained third deep learning model.

[0036] Furthermore, the third deep learning model consists of a deep convolutional generative adversarial network and a two-dimensional convolutional neural network.

[0037] The present invention also provides a method for intelligent management of a bean product production process, comprising the following steps:

[0038] Identify abnormal soybeans from the soybeans and remove them, retain qualified soybeans and calculate the total weight of qualified soybeans;

[0039] Predicting the optimal soaking water temperature T and optimal soaking time J for soaking the qualified soybeans;

[0040] Calculate the current accumulated impurities in the water, specifically:

[0041]

[0042]

[0043] Where Q(t) represents the accumulated impurity after time t, Indicates the time when the water was last changed. If it is before the first water change, , R(t) represents the impurity release rate after time t, K represents the preset water quality deterioration coefficient, and λ represents the preset attenuation rate;

[0044] Determine whether the water needs to be changed based on the current amount of accumulated impurities in the water:

[0045] like = ZQ, and the corresponding t satisfies J-t≥yt, then it is determined that water needs to be changed, and the current corresponding t is assigned to ; Where ZQ represents the preset impurity tolerance, and yt represents the preset remaining time.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] The present invention identifies abnormal soybeans through an abnormal soybean identification module, retains qualified soybeans, avoids the subsequent prediction data and product quality being affected by the abnormal soybeans, and ensures the accuracy of the prediction data; predicts the optimal soaking water temperature and optimal soaking time for soaking qualified soybeans through a soaking prediction module, provides the optimal soaking water temperature and time for the soaking process, and improves production efficiency and product quality; through the interaction of a water quality change estimation module and a water change judgment module, determines whether water needs to be changed according to the current accumulated impurities in the water, avoids manually determining the number of water changes and the time for water changes, and further improves production efficiency and product quality;

[0048] The present invention calculates the roundness of soybeans, the number of convex hull defects and the corresponding convex hull defect depths, and comprehensively analyzes the shriveling value, thereby improving the accuracy of identifying shriveled soybeans.

[0049] The present invention identifies moldy soybeans through a third deep learning model composed of a deep convolutional generative adversarial network and a two-dimensional convolutional neural network, so that the moldy soybean images of the expanded data are closer to the real moldy soybean images, ensuring the accuracy of the moldy features, and thereby improving the recognition accuracy of the third deep learning model. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0052] Figure 1 It is a structural diagram of the intelligent management system for the bean product production process of the present invention;

[0053] Figure 2 It is a structural block diagram of an abnormal soybean identification module of the soybean product production process intelligent management system of the present invention;

[0054] Figure 3 It is a structural block diagram of the soaking prediction module of the intelligent management system for the bean product production process of the present invention;

[0055] Figure 4 It is a flow chart of the intelligent management method of the bean product production process of the present invention;

[0056] Figure 5 It is a flow chart of step S12 of the intelligent management method for the bean product production process of the present invention. DETAILED DESCRIPTION

[0057] 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.

[0058] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0059] In addition, the descriptions of "first", "second", etc. in the present invention are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0060] Embodiment 1

[0061] See also Figure 1 As shown, the present invention provides an intelligent management system for the bean product production process, comprising:

[0062] An abnormal soybean identification module, a soaking prediction module, a water quality change estimation module and a water change judgment module, and a central processing unit that communicates with the above modules.

[0063] The abnormal soybean identification module is used to identify abnormal soybeans from soybeans and remove them, retain qualified soybeans and calculate the total weight G of qualified soybeans.

[0064] The abnormal soybeans include shriveled soybeans.

[0065] See also Figure 2 As shown, the abnormal soybean identification module includes: an image acquisition module and a shriveled identification module.

[0066] The image acquisition module is used to acquire soybean images.

[0067] The soybean image is an image that has undergone image preprocessing.

[0068] The shriveled identification module is used to identify shriveled soybeans through the soybean image.

[0069] Shriveled soybeans usually become shriveled due to lack of water and nutrition due to various reasons during growth or storage. The internal tissue of shriveled soybeans is poorly developed and may contain more cellulose and less soluble protein. This will reduce the quality of the final bean product, and its water absorption capacity is worse than that of normal soybeans, which means that during the soaking process, they may not fully expand, thereby affecting the efficiency of subsequent processing. Therefore, the present invention identifies shriveled soybeans through a shriveled identification module to improve the production efficiency and production quality of bean product production.

[0070] The identifying of shriveled soybeans through the soybean image is specifically as follows:

[0071] Identifying the actual contour of each soybean in the soybean image by a contour detection algorithm;

[0072] Analyze the contour length and contour range area according to the actual contour of soybeans, and calculate the circularity;

[0073] The convex hull contour corresponding to the actual contour of each soybean is analyzed by the convex hull algorithm, and then the number of convex hull defects and the corresponding convex hull defect depth of each soybean are analyzed according to the convex hull contour;

[0074] The shrinkage value is calculated by the roundness of soybeans, the number of convex hull defects and the corresponding convex hull defect depth;

[0075] Soybeans with a shriveled value greater than or equal to a preset shriveled threshold value are judged as shriveled soybeans and discarded, while qualified soybeans are retained and their total weight is calculated.

[0076] The calculation formula of the circularity is specifically:

[0077]

[0078] Among them, Circularity represents the circularity of the corresponding soybean, S represents the contour range area of ​​the corresponding soybean, and C represents the contour length of the corresponding soybean.

[0079] The roundness value ranges from 0 to 1, where 1 represents a perfect circle and the closer it is to 0, the further the shape deviates from a circle. Normally shaped soybeans are usually oval or round, and their roundness can be maintained at a high level, while shriveled soybeans may lose water, and the surface may become irregular or even concave, causing their roundness to decrease and become closer to 0.

[0080] The convex hull contour corresponding to the actual contour of each soybean is analyzed by using a convex hull algorithm. Specifically, the convexHull function of OpenCV can be used to analyze the convex hull contour corresponding to the actual contour of each soybean.

[0081] The convex hull contour is the smallest convex polygon that can enclose all points of the corresponding actual contour and is connected by the points corresponding to the actual contour.

[0082] The number of convex hull defects and the corresponding convex hull defect depth of each soybean are analyzed according to the convex hull contour. Specifically, the convexityDefects function of OpenCV can be used to analyze the number of convex hull defects and the corresponding convex hull defect depth of each soybean.

[0083] The convex hull defect refers to the concave area on the surface of soybeans. For soybeans of normal shape, their actual contour is relatively smooth, so the corresponding convex hull contour is more consistent with their actual contour, and thus the number of convex hull defects is relatively small. For shriveled soybeans, there may be more or deeper depressions on their surface, so there are more or larger gaps between their convex hull contour and the actual contour. In this gap, the longest distance from the actual contour to the convex hull contour is the convex hull defect depth of the corresponding convex hull defect.

[0084] The calculation formula of the shrinkage value is specifically:

[0085]

[0086] Among them, GBZ represents the shrinkage value of the corresponding soybean, n represents the number of convex hull defects of the corresponding soybean, L represents the convex hull defect depth corresponding to the convex hull defect, represents the first preset weight, Indicates the second preset weight.

[0087] The soaking prediction module is used to predict the optimal soaking water temperature T and the optimal soaking time J for soaking the qualified soybeans.

[0088] If the soaking water temperature is too low, the speed at which soybeans absorb water and swell will slow down, thus prolonging the soaking time and reducing production efficiency. If the soaking water temperature is too high, the surface of the soybeans will swell quickly while the interior has not completely absorbed water, resulting in uneven texture and lowering product quality. If the soaking time is too short, the soybeans may not be able to fully absorb water and swell, affecting subsequent processing steps and leading to low production efficiency and product quality. If the soaking time is too long, the nutrients in the soybeans may be lost and the texture of the soybeans may be affected, leading to low product quality. Therefore, the present invention predicts the optimal soaking water temperature and optimal soaking time through a soaking prediction module to improve production efficiency and product quality.

[0089] See also Figure 3 As shown, the immersion prediction module includes: a data acquisition module, a water temperature prediction module and a duration prediction module.

[0090] The data acquisition module is used to obtain the target water absorption multiple mb and the type of the qualified soybeans.

[0091] The target water absorption multiple refers to the ratio of the target weight of soybeans after soaking to the weight of soybeans before soaking, and is set by the management personnel according to the type of soy products to be produced.

[0092] The water temperature prediction module is used to predict the optimal soaking water temperature according to the type of the qualified soybeans using a trained first deep learning model.

[0093] The first deep learning model is trained using different soybean varieties, different water temperatures, and corresponding times to reach optimal soaking quality as sample data.

[0094] The duration prediction module is used to predict the optimal soaking duration according to the target water absorption multiple and the optimal soaking water temperature.

[0095] The predicted optimal soaking time is specifically:

[0096] The second deep learning model that has been trained obtains an average water absorption rate v according to the optimal soaking water temperature and the type of the qualified soybeans;

[0097] The optimal soaking time is calculated according to the target water absorption multiple and the average water absorption rate, J=(mb-1) / v.

[0098] The water quality change estimation module is used to calculate the current accumulated impurities in the water, specifically:

[0099]

[0100]

[0101] Where Q(t) represents the accumulated impurity after time t, Indicates the time when the water was last changed. If the qualified soybeans were soaked before the first water change, , R(t) represents the impurity release rate after time t, K represents the preset water quality deterioration coefficient, and λ represents the preset attenuation rate.

[0102] For example, if you have not changed the water, , if water needs to be changed after a period of time, then in the calculation of Q after the water change, a.

[0103] When soaking soybeans, the more soybeans there are, the more impurities and dissolved components they release, which may cause the soaking water quality to deteriorate faster. At the same time, water temperature affects the growth rate of microorganisms, which also causes the soaking water quality to deteriorate. Therefore, the present invention estimates the current accumulated impurities in the water through a water quality change estimation module, providing a basis for subsequent water change judgments.

[0104] The preset water quality deterioration coefficient can be obtained by conducting water quality change experimental tests through different soybean types and water temperatures, and a water quality deterioration model is established and trained based on the test data. Then, the preset water quality deterioration coefficient is obtained through the water quality deterioration model based on the type of qualified soybeans and the optimal soaking water temperature; it can also be obtained by relevant staff through empirical calculations.

[0105] As time goes by, the impurity content in the water will gradually increase, but this increase will gradually slow down over time because as the soaking time increases, the impurities that soybeans can release into the water will gradually decrease over time. Therefore, a preset attenuation rate is added to improve the accuracy of the accumulated impurity amount.

[0106] The water change judgment module is used to judge whether the water needs to be changed according to the current accumulated impurities in the water. Specifically:

[0107] like = ZQ, and the corresponding t satisfies J-t≥yt, then it is determined that water needs to be changed, and the current corresponding t is assigned to ; Where ZQ represents the preset impurity tolerance, and yt represents the preset remaining time.

[0108] In the production of soy products, long-term soaking of soybeans may lead to an increase in the number of microorganisms in the water, or an increase in impurities released by the soybeans into the water, which will reduce the soaking effect and further affect the product quality. Therefore, the soaking water needs to be replaced. The present invention uses a water change judgment module to judge whether the water needs to be changed in real time according to the current accumulated impurities in the water, thereby avoiding manual setting of the number and frequency of water changes, thereby improving production efficiency while ensuring product quality.

[0109] In some embodiments, the abnormal soybeans further include moldy soybeans.

[0110] The abnormal soybean identification module also includes a mold identification module.

[0111] The moldy identification module is used to identify moldy soybeans through the soybean image.

[0112] Moldy soybeans may contain various molds, which pose a threat to health and greatly affect the quality of soy products. Therefore, the present invention also identifies moldy soybeans through a mold identification module to ensure the safety of soy product production and improve production quality.

[0113] The method of identifying moldy soybeans through the soybean image is specifically as follows:

[0114] The soybean image is segmented into a plurality of single soybean images, and the single soybean images are used to identify whether the corresponding soybeans are moldy soybeans through the trained third deep learning model.

[0115] The third deep learning model consists of a deep convolutional generative adversarial network and a two-dimensional convolutional neural network.

[0116] The training of the third deep learning model is specifically as follows:

[0117] Several moldy soybean images acquired in advance are used as the initial sample data set. After the deep convolutional generative adversarial network is trained using the initial sample data set, the initial sample data set is expanded to obtain an expanded data set, and then the initial sample data set and the expanded data set are input into a two-dimensional convolutional neural network for training.

[0118] The training of the deep convolutional adversarial generation network is specifically as follows:

[0119] Random noise is input into the generator with a transposed convolutional neural network to generate new image data, and then the initial sample data set and the new image data are input into the discriminator with a convolutional neural network to output the identification result. The generator calculates the generator loss value through the generator loss function according to the identification result, and updates the generator parameters according to the generator loss value. The discriminator calculates the discriminator loss value through the discriminator loss function according to the identification result, and then updates the discriminator parameters according to the discriminator loss value. The above operations are repeated until the discriminator can accurately identify the real data or the preset number of training rounds is reached.

[0120] The deep convolutional adversarial generative network uses a transposed convolutional neural network as a generator, which can convert low-resolution images back to high-resolution images, because the transposed convolution layer is equivalent to the inverse operation of the convolution layer, which increases the size of the image through upsampling and expansion operations; and the transposed convolution allows the generator to maintain more details and structures when generating images, because the transposed convolution can better retain local information while upsampling;

[0121] In summary, the deep convolutional generative adversarial network of this scheme can improve the moldy soybean images of the expanded data to be closer to the real moldy soybean images, ensure the accuracy of the moldy features, and thus improve the recognition accuracy of the third deep learning model.

[0122] Embodiment 2

[0123] See also Figure 4 As shown, the present invention also provides a method for intelligent management of a bean product production process, comprising the following steps:

[0124] S1. Identify and remove abnormal soybeans from the soybeans, retain qualified soybeans and calculate the total weight of qualified soybeans;

[0125] S2, predicting the optimal soaking water temperature T and optimal soaking time J for soaking the qualified soybeans;

[0126] S3, calculating the current accumulated impurities in the water;

[0127] S4. Determine whether the water needs to be changed based on the current amount of accumulated impurities in the water.

[0128] Regarding step S1, the step of identifying abnormal soybeans from soybeans specifically includes the following steps:

[0129] S11, obtaining a soybean image;

[0130] S12, identifying shriveled soybeans through the soybean image.

[0131] See also Figure 5 As shown, for step S12, the step of identifying shriveled soybeans through the soybean image specifically includes the following steps:

[0132] T1, identifying the actual contour of each soybean in the soybean image by a contour detection algorithm;

[0133] T2. Analyze the contour length and contour range area according to the actual contour of soybeans, and calculate the circularity;

[0134] T3, analyzing the convex hull contour corresponding to the actual contour of each soybean by a convex hull algorithm, and then analyzing the number of convex hull defects and the corresponding convex hull defect depth of each soybean according to the convex hull contour;

[0135] T4. Calculate the shrinkage value by the roundness of soybeans, the number of convex hull defects and the corresponding convex hull defect depth;

[0136] T5. The soybeans whose shriveling values ​​are greater than or equal to the preset shriveling threshold value are judged as shriveled soybeans.

[0137] For step T2, the calculation formula of the circularity is specifically:

[0138]

[0139] Among them, Circularity represents the circularity of the corresponding soybean, S represents the contour range area of ​​the corresponding soybean, and C represents the contour length of the corresponding soybean.

[0140] For step T3, the convex hull contour corresponding to the actual contour of each soybean is analyzed by a convex hull algorithm. Specifically, the convexHull function of OpenCV can be used to analyze the convex hull contour corresponding to the actual contour of each soybean;

[0141] The number of convex hull defects and the corresponding convex hull defect depth of each soybean are analyzed according to the convex hull contour. Specifically, the convexityDefects function of OpenCV can be used to analyze the number of convex hull defects and the corresponding convex hull defect depth of each soybean.

[0142] For step T4, the calculation formula of the shrinkage value is specifically:

[0143]

[0144] Among them, GBZ represents the shrinkage value of the corresponding soybean, n represents the number of convex hull defects of the corresponding soybean, L represents the convex hull defect depth corresponding to the convex hull defect, represents the first preset weight, Indicates the second preset weight.

[0145] For step S2, the prediction of the optimal soaking water temperature T and the optimal soaking time J for soaking the qualified soybeans specifically includes the following steps:

[0146] S21, obtaining the target water absorption multiple and the type of the qualified soybeans;

[0147] S22, predicting the optimal soaking water temperature using the trained first deep learning model according to the type of the qualified soybeans;

[0148] S23, predicting an optimal soaking time according to the target water absorption multiple and the optimal soaking water temperature.

[0149] For step S23, predicting the optimal soaking time according to the target water absorption multiple and the optimal soaking water temperature specifically includes the following steps:

[0150] U1, using the trained second deep learning model to obtain the average water absorption rate v according to the optimal soaking water temperature and the type of qualified soybeans;

[0151] U2. Calculate the optimal soaking time according to the target water absorption multiple and the average water absorption rate, J = (mb-1) / v.

[0152] For step S3, the current accumulated impurity amount in the water is calculated, specifically:

[0153]

[0154]

[0155] Where Q(t) represents the accumulated impurity after time t, Indicates the time when the water was last changed. If it is before the first water change, , R(t) represents the impurity release rate after time t, K represents the preset water quality deterioration coefficient, and λ represents the preset attenuation rate.

[0156] For step S4, judging whether water needs to be changed according to the current accumulated impurities in the water is as follows:

[0157] like = ZQ, and the corresponding t satisfies J-t≥yt, then it is determined that water needs to be changed, and the current corresponding t is assigned to ; Where ZQ represents the preset impurity tolerance, and yt represents the preset remaining time.

[0158] In some embodiments, in step S12, the moldy soybeans are further identified through the soybean image, specifically:

[0159] The soybean image is segmented into a plurality of single soybean images, and the single soybean images are identified using a trained second deep learning model to identify whether the corresponding soybeans are moldy soybeans.

[0160] The third deep learning model consists of a deep convolutional generative adversarial network and a two-dimensional convolutional neural network.

[0161] Embodiment 3

[0162] The present invention also provides an electronic device, comprising: a processor, a sending device, an input device, an output device and a memory, the memory is used to store computer program code, the computer program code includes computer instructions, when the processor executes the computer instructions, the electronic device executes a method as any one of the possible implementation methods described above.

[0163] Embodiment 4

[0164] The present invention also provides a computer-readable storage medium, in which a computer program is stored. The computer program includes program instructions. When the program instructions are executed by a processor of an electronic device, the processor executes a method as described in any possible implementation manner.

[0165] The beneficial effects of the present invention are:

[0166] The present invention identifies abnormal soybeans through an abnormal soybean identification module, retains qualified soybeans, avoids the subsequent prediction data and product quality being affected by the abnormal soybeans, and ensures the accuracy of the prediction data; predicts the optimal soaking water temperature and optimal soaking time for soaking qualified soybeans through a soaking prediction module, provides the optimal soaking water temperature and time for the soaking process, and improves production efficiency and product quality; through the interaction of a water quality change estimation module and a water change judgment module, determines whether water needs to be changed according to the current accumulated impurities in the water, avoids manually determining the number of water changes and the time for water changes, and further improves production efficiency and product quality;

[0167] The present invention calculates the roundness of soybeans, the number of convex hull defects and the corresponding convex hull defect depths, and comprehensively analyzes the shriveling value, thereby improving the accuracy of identifying shriveled soybeans.

[0168] The present invention identifies moldy soybeans through a third deep learning model composed of a deep convolutional generative adversarial network and a two-dimensional convolutional neural network, so that the moldy soybean images of the expanded data are closer to the real moldy soybean images, ensuring the accuracy of the moldy features, and thereby improving the recognition accuracy of the third deep learning model.

[0169] In the description of the specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0170] The foregoing is merely a specific embodiment of the present invention, which enables those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features claimed herein.

Claims

1. An intelligent management system for the production process of soy products, characterized in that: include: The abnormal soybean identification module is used to identify and remove abnormal soybeans from soybeans, retain qualified soybeans and calculate the total weight G of qualified soybeans; A soaking prediction module, used to predict the optimal soaking water temperature T and the optimal soaking time J for soaking the qualified soybeans; The water quality change estimation module is used to calculate the current accumulated impurities in the water, specifically: ， ， Where Q(t) represents the accumulated impurity after time t, Indicates the time when the water was last changed. If the qualified soybeans were soaked before the first water change, , R(t) represents the impurity release rate after time t, K represents the preset water quality deterioration coefficient, and λ represents the preset attenuation rate; The water change judgment module is used to judge whether the water needs to be changed according to the current accumulated impurities in the water. Specifically: like = ZQ, and the corresponding t satisfies J-t≥yt, then it is determined that water needs to be changed, and the current corresponding t is assigned to ; Where ZQ represents the preset impurity tolerance, and yt represents the preset remaining time.

2. The intelligent management system for bean product production process according to claim 1, characterized in that: The abnormal soybeans include shriveled soybeans; The abnormal soybean identification module comprises: An image acquisition module, used for acquiring soybean images; The shriveled identification module is used to identify shriveled soybeans through the soybean image.

3. The intelligent management system for bean product production process according to claim 2, characterized in that: The identifying of shriveled soybeans through the soybean image is specifically as follows: Identifying the actual contour of each soybean in the soybean image by a contour detection algorithm; According to the actual contour of soybean, the contour length C and contour range area S are analyzed, and the circularity is calculated. ; The convex hull contour corresponding to the actual contour of each soybean is analyzed by the convex hull algorithm, and then the number of convex hull defects and the corresponding convex hull defect depth of each soybean are analyzed according to the convex hull contour; The shrinkage value GBZ is calculated by the roundness of soybeans, the number of convex hull defects n and the corresponding convex hull defect depth L. ,in, represents the first preset weight, represents a second preset weight; Soybeans with a shriveled value greater than or equal to a preset shriveled threshold value are judged as shriveled soybeans.

4. The intelligent management system for bean product production process according to claim 3 is characterized in that: The analysis of the convex hull contour corresponding to the actual contour of each soybean is specifically analyzed using the convexHull function of OpenCV; the analysis of the number of convex hull defects and the corresponding convex hull defect depth of each soybean is specifically analyzed using the convexityDefects function of OpenCV.

5. The intelligent management system for bean product production process according to claim 1, characterized in that: The soaking prediction module comprises: A data acquisition module, used for acquiring the target water absorption multiple mb and the type of the qualified soybeans; A water temperature prediction module, used for predicting the optimal soaking water temperature according to the type of the qualified soybeans using a trained first deep learning model; The duration prediction module is used to predict the optimal soaking duration according to the target water absorption multiple and the optimal soaking water temperature.

6. The intelligent management system for bean product production process according to claim 5, characterized in that: The predicted optimal soaking time is specifically: The second deep learning model that has been trained obtains an average water absorption rate v according to the optimal soaking water temperature and the type of the qualified soybeans; The optimal soaking time is calculated according to the target water absorption multiple and the average water absorption rate, J=(mb-1) / v.

7. The intelligent management system for bean product production process according to claim 2, characterized in that: The abnormal soybeans also include moldy soybeans; the abnormal soybean identification module also includes a moldy identification module, which is used to identify moldy soybeans through the soybean image.

8. The intelligent management system for bean product production process according to claim 7, characterized in that: The method of identifying moldy soybeans through the soybean image is specifically as follows: The soybean image is segmented into a plurality of single soybean images, and the single soybean images are used through the trained third deep learning model to identify whether the corresponding soybeans are moldy soybeans.

9. The intelligent management system for bean product production process according to claim 8, characterized in that: The third deep learning model consists of a deep convolutional generative adversarial network and a two-dimensional convolutional neural network.

10. A method for intelligent management of a bean product production process, applied to the intelligent management system for a bean product production process as claimed in any one of claims 1 to 9, comprising the following steps: Identify abnormal soybeans from the soybeans and remove them, retain qualified soybeans and calculate the total weight of qualified soybeans; Predicting the optimal soaking water temperature T and optimal soaking time J for soaking the qualified soybeans; Calculate the current accumulated impurities in the water, specifically: ， ， Where Q(t) represents the accumulated impurity after time t, Indicates the time when the water was last changed. If it is before the first water change, , R(t) represents the impurity release rate after time t, K represents the preset water quality deterioration coefficient, and λ represents the preset attenuation rate; Determine whether the water needs to be changed based on the current amount of accumulated impurities in the water: like = ZQ, and the corresponding t satisfies J-t≥yt, then it is determined that water needs to be changed, and the current corresponding t is assigned to ; Where ZQ represents the preset impurity tolerance, and yt represents the preset remaining time.

Images