Gradation automatic regulation and control system and method based on aggregate image recognition

Abstract

The invention discloses an automatic gradation regulation and control system and method based on aggregate image recognition, relates to the technical field of automation of asphalt mixture production equipment, and solves the technical problem of deviation in gradation parameter setting caused by the fact that actual transportation aggregate is not recognized during gradation regulation during asphalt mixture production in the prior art. According to the method, the aggregate image is acquired in real time through the image acquisition equipment, the image features in the aggregate image are extracted, the aggregate particle size proportion is identified based on the image features, the aggregate category of the unidentified aggregate is identified and analyzed according to the image features, and the aggregate quantity corresponding to the aggregate category is counted to synthesize the gradation. Setting an adjustment threshold according to historical gradation data in the database, obtaining corresponding equipment gradation data based on the difference between the synthesis gradation and the target gradation, and performing gradation adjustment on the gradation equipment through the equipment gradation data; and generating an alarm signal based on a comparison result of the synthetic gradation and a preset alarm threshold, and controlling an emergency switch of the equipment based on the alarm signal.

Description

Technical Field

[0001] The invention belongs to the field of equipment automation and relates to image analysis technology, in particular to an automatic grading control system and method based on aggregate image recognition. Background Art

[0002] When mixing asphalt or cement, it is necessary to mix a variety of different aggregates such as (coarse aggregate, fine aggregate, mineral powder and asphalt) in precise proportions. When mixing, the demand for different aggregates is different, and different aggregates correspond to different gears, and different aggregates belong to different gradations. The reasonable distribution of gradations affects the quality of the final product. Ensuring good gradations can improve the shear strength, compressive strength and durability of the mixed aggregates. When mixing, if the aggregate feeding speed of a certain gear exceeds the normal range, the feeding speed needs to be adjusted to maintain normal feeding and ensure the final product quality. Real-time automatic control of the feeding process plays a vital role in the production of asphalt or cement mixtures.

[0003] The invention patent with application number CN2023116226844 discloses an aggregate grading detection method, a silo discharge control method, system and device. The invention predicts the current grading of aggregates in the silo and the current discharge speed through the pressure generated by the falling aggregates hitting the pressure sensor, thereby calculating the total grading of aggregates in all silos under the current discharge state. When obtaining the required grading parameters, different categories of aggregates have different corresponding grading parameters. If only the pressure of the falling aggregate is considered, the pressure caused by the distribution or superposition of aggregates may not correspond to the actual transported aggregate, resulting in deviations in the set grading parameters.

[0004] The present invention provides an automatic grading control system and method based on aggregate image recognition to solve the above technical problems. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes an automatic grading control system and method based on aggregate image recognition, which is used to solve the technical problem in the prior art that the actual transported aggregate is not identified when adjusting the grading, resulting in deviations in setting the grading parameters.

[0006] To achieve the above-mentioned object, the first aspect of the present invention provides an automatic grading control system based on aggregate image recognition, comprising: an image collection module, a grading acquisition module, a grading control module and an early warning module;

[0007] Image collection module: used to collect aggregate images in different aggregate bins in real time;

[0008] Grading acquisition module: used to obtain the synthetic gradation of different aggregate bins according to the aggregate image, and match the synthetic gradation with the target gradation in the database to obtain the control data;

[0009] Grading control module: used to control the feeding equipment based on the control data to complete the grading adjustment;

[0010] Early warning module: used to control the emergency switch of equipment based on synthetic grading.

[0011] Preferably, the obtaining of synthetic gradations of different aggregate bins according to the aggregate image includes:

[0012] Extracting aggregate images;

[0013] The aggregate features of different aggregates in the aggregate image are extracted through the Python library and input into the aggregate recognition model as input data to obtain the corresponding aggregate category; wherein the aggregate recognition model is constructed based on the neural network model;

[0014] The aggregate whose aggregate category is not successfully identified is marked as unidentified aggregate, and the aggregate whose aggregate category is successfully identified is marked as identified aggregate;

[0015] The shape features of the identified aggregates are extracted and the aggregate particle size is obtained. The aggregate particle sizes of all identified aggregates in the aggregate image are counted, and the grading gears are determined. The number of aggregates of the identified aggregates is counted to obtain the number of gears. The number of gears of all aggregate bins is counted separately. The ratio of the number of gears in the aggregate image of the aggregate bin to the sum of the number of gears is taken as the single-gear sub-gradation of the aggregate bin. The sub-gradations of all aggregate bins are integrated to obtain the synthetic gradation.

[0016] Preferably, the aggregate identification model is constructed based on a neural network model, including:

[0017] Historical aggregate image features and corresponding aggregate categories are extracted from the database; the historical aggregate image features and the corresponding aggregate categories are used as training data and test data to train the neural network model, and after the training, the training effect is tested through the test data, and the neural network model is adjusted according to the test results to obtain an aggregate recognition model whose input data is the aggregate image features and output data is the aggregate category.

[0018] Preferably, the step of matching the synthetic gradation with the target gradation in the database to obtain the control data includes:

[0019] Extracting the synthetic gradation, and extracting the target gradation from the database; the target gradation includes the target category gradation corresponding to each aggregate, the aggregate category contained in each sub-gradation, and the target sub-gradation corresponding to each sub-gradation;

[0020] Setting an adjustment threshold; determining whether the difference between the sub-gradation in the synthetic gradation and the corresponding target sub-gradation in the target gradation is greater than the adjustment threshold; if yes, marking the corresponding sub-gradation as a sub-gradation requiring adjustment; if no, marking the corresponding sub-gradation as a normal sub-gradation; wherein the adjustment threshold is set according to the working quality of the grading equipment;

[0021] The aggregates in the sub-gradation containing the aggregate category are numbered as i, and the aggregate category is divided into sub-gradation categories of each level. The sub-gradation and the target sub-gradation are statistically obtained, and the sub-gradation of each level is numbered as j; through the formula

[0022] Calculate the similarity ZXji of aggregate numbered i in sub-gradation category numbered j; where JPi represents the number of aggregates numbered i, ZJPi represents the number of sub-gradation aggregates in sub-gradation category numbered j, MJPi represents the target number of aggregates numbered i, and MZJPj represents the target sub-gradation in sub-gradation category numbered j;

[0023] Determine whether the aggregate similarity of the aggregate category is greater than 0; if yes, mark the adjustment state of the corresponding aggregate category as a deceleration state; if no, mark the adjustment state of the corresponding aggregate category as a speed-up state;

[0024] Calculate the normalized average value of aggregate similarity of all aggregate categories in each sub-gradation category and multiply it by the proportion of the corresponding sub-gradation category in the total gradation to obtain the gradation coefficient; multiply all sub-gradations by the corresponding gradation coefficient to obtain the total gradation value;

[0025] The equipment grading data is obtained by integrating the sub-grading that needs to be adjusted, the adjustment status of each aggregate category and the total grading value.

[0026] Preferably, the adjustment threshold is set according to the working quality of the grading equipment, including:

[0027] Extract the historical gradation data corresponding to each level of distribution equipment from the database; wherein the historical gradation data includes the historical synthetic gradation of each level of distribution equipment and the corresponding historical product quality; the historical product quality is evaluated by the staff, and the historical product quality includes excellent products, good products and inferior products;

[0028] The historical composite gradation corresponding to the superior and good products is statistically obtained to obtain the superior historical gradation, and the historical composite gradation corresponding to the inferior products is statistically obtained to obtain the inferior historical gradation;

[0029] The mode in the historical composite gradation is counted and marked as the standard gradation, and the adjustment threshold is obtained by calculating the ratio of the good historical gradation to the inferior historical gradation plus one and multiplying it by the standard gradation.

[0030] Preferably, the step of regulating the feeding device based on the regulation data to complete the grading adjustment includes:

[0031] Extract regulatory data;

[0032] Adjust the feeding equipment parameters of the corresponding sub-gradation according to the control data; take the number of sub-gradations in the composite gradation as the original feeding number, combine the control data with the composite gradation to obtain the regulated composite gradation, take the corresponding regulated sub-gradation rate in the regulated composite gradation as the feeding rate, and adjust the corresponding feeding equipment parameters to the equipment parameters that meet the feeding rate.

[0033] Preferably, the emergency switch of the synthetic grading control device includes:

[0034] Extract the synthetic gradation and adjustment difference; set the alarm adjustment threshold and alarm type threshold; wherein the alarm adjustment threshold is set according to the maximum adjustment threshold that can be tolerated during production; the alarm type threshold is set according to the gradation level corresponding to the aggregate image;

[0035] Determine whether the aggregate particle size of the identified aggregate meets the alarm type threshold; if yes, mark the corresponding grading device as a normal feeding device; if no, mark the corresponding grading device as an error feeding device and generate a feeding alarm signal;

[0036] Determine whether the adjustment difference of the grading equipment is greater than the alarm adjustment threshold; if yes, generate an adjustment alarm signal; if no, continue monitoring;

[0037] The feed alarm signal and the adjustment alarm signal are integrated into an alarm signal. When the alarm signal is detected, the emergency switch of the equipment is controlled based on the alarm signal.

[0038] Preferably, the emergency switch of the device based on the alarm signal control comprises:

[0039] The alarm signal is sent to the management system, which controls the power switch of the grading equipment corresponding to the sub-grading to be adjusted to be cut off and records the alarm time;

[0040] The grading device that detects the feeding alarm signal is marked as a feeding alarm device, the aggregate in the feeding alarm device is adjusted, and the detection and identification are re-performed. If the feeding alarm device is a normal feeding device after re-feeding, the feeding alarm device is restarted;

[0041] The grading device that detects the adjustment alarm signal is marked as an adjustment alarm device, and the difference between the sub-gradation to be adjusted and the target sub-gradation of the adjustment alarm device is marked as an adjustment aggregate rate; it is determined whether the adjustment state of the aggregate category corresponding to the alarm device is a speed-up state; if yes, the aggregate rate is adjusted as an adjustment amount to increase the aggregate feeding speed; if no, the aggregate rate is adjusted as a reduction aggregate rate to reduce the aggregate feeding speed;

[0042] During the period when the grading equipment is shut down, the workload of the grading equipment of the unified grading combination category is regulated according to the target grading of the alarm device; after the aggregate adjustment is completed, the equipment switch is turned on again.

[0043] Preferably, the step of regulating the workload of the unified gradation combination category gradation equipment according to the target gradation of the alarm equipment includes:

[0044] Extract target gradation and gradation combination categories of alarm equipment;

[0045] Match the grading equipment with the same grading combination category as the alarm equipment and mark it as the matching grading equipment; divide the target gradation of the alarm equipment into several emergency working gradations according to the ratio of the target gradation of the matching grading equipment, and add the emergency working gradation to the target gradation of the corresponding matching grading equipment to obtain the emergency target gradation.

[0046] The second aspect of the present invention provides an automatic grading control method based on aggregate image recognition, which is applied to the automatic grading control system based on aggregate image recognition, and is characterized by comprising:

[0047] Step 1: Collect aggregate images in different aggregate bins in real time;

[0048] Step 2: Obtain the composite gradation of different aggregate bins based on the aggregate image;

[0049] Step 3: Match the synthetic gradation with the target gradation in the database to obtain control data;

[0050] Step 4: Regulate the feeding equipment based on the control data to complete the grading adjustment;

[0051] Step 5: Control the emergency switch of the equipment based on the synthetic grading.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] 1. The present invention collects aggregate images on a conveyor belt, extracts aggregate features of different aggregates in the aggregate images, and inputs the aggregate features into an aggregate recognition model constructed based on a neural network model to obtain aggregate categories, thereby more accurately classifying different types of aggregates, ensuring the consistency of product quality, and improving the qualified rate of products; the gradation combination categories are divided according to the aggregate data, and the synthetic gradation is obtained by integrating the corresponding aggregate quantity, and the adjustment threshold is set according to the historical gradation data in the database, and the corresponding equipment gradation data is obtained based on the difference between the synthetic gradation and the target gradation, thereby more accurately controlling the gradation state of the aggregate, improving work efficiency, and ensuring The quality of each batch of products is consistent. The equipment is adjusted in time according to the equipment grading data. The automatic adjustment of grading parameters simplifies the operating process and improves the operator's work efficiency. The automatic adjustment of grading parameters reduces the number of manual interventions, improves the degree of automation of the production line, and improves production quality. The difference between the synthetic grading and the target grading is compared with the preset alarm threshold to generate an alarm signal. The emergency switch of the equipment is controlled based on the alarm signal, which reduces the risk of affecting the entire production process due to abnormalities in a certain area, improves the stability and reliability of the system, and ensures the smooth completion of production tasks and avoids waste of resources.

[0054] 2. The present invention sets an adjustment threshold according to historical grading data, obtains a standard gradation by counting historical synthetic gradations of production quality of superior products, good products and inferior products in the historical grading data, and counts the mode of the historical synthetic gradations, and obtains an adjustment threshold by combining the standard gradation, the historical synthetic gradation and the corresponding historical product quality. By adjusting the threshold, it is determined whether the corresponding grading equipment needs to be adjusted, and the capacity status of the production equipment can be more accurately controlled, thereby improving production efficiency, reasonably allocating production resources, avoiding waste of resources caused by excessive use of certain equipment, timely discovering and correcting deviations, reducing product defects caused by uneven grading, improving the qualified rate of products, and improving the flexibility and adaptability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] 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, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0056] Figure 1 The following is a flowchart of an embodiment of the present invention.

[0057] Figure 2 It is a system composition diagram of the present invention.

[0058] Figure 3A complete flow chart for obtaining control data in one embodiment of the present invention. DETAILED DESCRIPTION

[0059] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than 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.

[0060] See also Figure 1-Figure 3 , the first aspect of the present invention provides an automatic grading control system based on aggregate image recognition, including: an image collection module, a grading acquisition module, a grading control module and an early warning module;

[0061] Image collection module: used to collect aggregate images in different aggregate bins in real time.

[0062] Exemplarily, several CCD cameras of grading devices are set to collect images of each grading device when feeding, and the collected images are marked as aggregate images.

[0063] Grading acquisition module: used to obtain the synthetic gradation based on the aggregate image, and match the synthetic gradation with the target gradation in the database to obtain the control data.

[0064] Exemplarily, historical aggregate image features and corresponding aggregate categories are extracted from a database; the historical aggregate image features and corresponding aggregate categories are used as training data and test data to train a neural network model, and after the training, the training effect is tested through the test data, and the neural network model is adjusted according to the test results to obtain an aggregate recognition model whose input data is aggregate image features and output data is aggregate categories.

[0065] In this embodiment, an aggregate image is collected at the feed port of the grading device A, and the aggregate features of all aggregates in the aggregate image are extracted through a Python library (such as the Scikit-Image library). The grayscale co-occurrence matrix of the aggregate in the aggregate image is extracted to obtain the texture features of the aggregate. The shape features of the aggregate are obtained through an edge detection algorithm, and shape parameters such as the diameter and circumference of the aggregate are calculated. Different aggregates have different texture features, and the same aggregates but different particle sizes can be distinguished by shape parameters. The aggregate features of the aggregates are input as input data into an aggregate recognition model to obtain the corresponding aggregate category; the aggregates whose aggregate categories are not successfully identified are marked as unidentified aggregates, and the aggregates whose aggregate categories are successfully identified are marked as identified aggregates.

[0066] The aggregate particle size of the aggregate is obtained according to the shape characteristics of the aggregate through the Python library. In this embodiment, there are five grading gears, which are respectively divided into a first gear, a second gear, a third gear, a fourth gear and a fifth gear. Different grading gears correspond to different particle size ranges, the first gear is 0-3 mm, the second gear is 3-6 mm, the third gear is 6-10 mm, the fourth gear is 10-15 mm, and the fifth gear is above 15 mm; in this embodiment, the aggregate particle size of the aggregate in the grading device A is 5 mm, and the corresponding grading gear is a third gear. The number of all aggregate particle sizes of 5 mm in the aggregate image of the grading device A is 200, and the ratio of the number of gears corresponding to the third gear to the sum of the number of all gears is 20%, and the sub-gradation of the grading device A is 20%; the sub-gradation of the five-grade grading gears is integrated to obtain a synthetic gradation.

[0067] In some other embodiments, during actual mixing, aggregates from other grading levels may be mixed into the aggregate bin of a grading level, for example, a small amount of 8 mm aggregate is mixed into the fourth level. When performing mixed identification, the aggregates from other grading levels are identified and recorded as mixed aggregates, and the feed speed of the grading equipment that matches the mixed aggregates is reduced.

[0068] In some other preferred embodiments, a CCD camera may also be provided at the discharge port of the grading equipment to collect the discharge image, compare the discharge image with the image during feeding, mark the sub-gradation when the discharge image is identified as the discharge sub-gradation, mark the sub-gradation when the feeding image is identified as the feeding sub-gradation, compare the discharge sub-gradation with the feeding sub-gradation, if the discharge sub-gradation is greater than the feeding sub-gradation, it means that the corresponding aggregate bin is mixed with mixed aggregates with a lower grading level, and the feeding rate of the grading equipment corresponding to the mixed aggregate should be reduced; if the discharge sub-gradation is less than the feeding sub-gradation, it means that the corresponding aggregate bin is mixed with mixed aggregates with a higher grading level, and the feeding rate of the grading equipment corresponding to the mixed aggregate should be reduced.

[0069] The standard aggregate quantity corresponding to each grading combination category in this production plan is extracted from the database, and the historical synthetic grading and the corresponding historical production quality of each level of distribution equipment are extracted, specifically the aggregate quantity of each device used for distributing aggregates at a certain moment in the historical data image during the same production plan and the product quality of the final product produced; the mode in the historical synthetic grading is counted as the standard grading, the historical synthetic grading with historical product quality of excellent and good products is counted and marked as excellent historical grading, the historical synthetic grading with historical production quality of inferior products is counted and marked as inferior historical grading, and the adjustment threshold TZS is calculated by the formula TZS=BJS×(YLP / LJP+1), wherein BJS represents the standard grading, YLP represents the excellent historical grading, and LJP represents the inferior historical grading.

[0070] Determine whether the difference between the sub-gradation in the synthetic gradation and the corresponding target sub-gradation in the target gradation is greater than the adjustment threshold. In this embodiment, the sub-gradation corresponding to the secondary gradation gear of the synthetic gradation is 2000 kg per minute, and the corresponding target sub-gradation in the target gradation is 2200 kg per minute. In this embodiment, the adjustment threshold is calculated to be 300 kg per minute. It is determined that the difference between the sub-gradation and the target sub-gradation is greater than the adjustment threshold, and the corresponding secondary gradation gear is marked as the sub-gradation to be adjusted, and the adjustment rate of the sub-gradation to be adjusted is recorded.

[0071] By formula The aggregate similarity ZX11 of aggregate No. 1 in the first-level grading is calculated; the final aggregate similarity value is between [-1, 1]. The closer the aggregate similarity is to 0, the greater the difference between the sub-grading and the target sub-grading; the closer the aggregate similarity is to 1, the smaller the difference between the sub-grading and the target sub-grading, and the feeding speed of the sub-grading is faster than that of the target sub-grading, and the feeding speed needs to be reduced, so the adjustment state of the sub-grading to be adjusted is marked as the deceleration state; the closer the aggregate similarity is to -1, the smaller the difference between the sub-grading and the target sub-grading, and the feeding speed of the sub-grading is slower than that of the target sub-grading, and the feeding speed needs to be increased, and the adjustment state of the corresponding aggregate category is marked as the speed-up state.

[0072] The aggregate similarity of all aggregate categories in the five-level sub-grading is calculated and normalized, and the average value after summation is multiplied by the proportion of the corresponding sub-grading in the sum of the sub-gradings to obtain the grading coefficient; all sub-gradings are multiplied by the corresponding grading coefficient to obtain the total grading value, and the sub-grading that needs to be adjusted, the adjustment status of each aggregate category and the total grading value are integrated to obtain the equipment grading data; the capacity status of the production equipment can be controlled more accurately, thereby improving production efficiency and reasonably allocating production resources.

[0073] In some other preferred embodiments, the difference between the sub-gradation in the synthetic gradation and the corresponding target sub-gradation in the target gradation is less than the adjustment threshold, indicating that the corresponding generated gradation is within the normal range and no adjustment is required; if the number of aggregates that need to adjust the sub-gradation is greater than the target sub-gradation, it means that the current generated gradation is larger than the target gradation and the feed speed needs to be reduced, so the adjustment state of the adjusted sub-gradation is marked as a deceleration state.

[0074] Alarm module: used to control the emergency switch of the equipment based on the alarm signal.

[0075] Exemplarily, the alarm signal is detected in real time, and after the alarm signal is detected, the alarm signal is sent to the management system, the management system controls to cut off the power switch of the corresponding grading equipment, matches the grading equipment with the same grading combination category as the alarm equipment and marks it as a matching grading equipment. In this embodiment, the secondary grading range includes three grading equipment, which are marked as equipment A, equipment B and equipment C respectively. If equipment A detects an alarm signal, the difference between the sub-grading to be adjusted and the target sub-grading of equipment A is the adjusted feed rate, and the adjustment state of the corresponding aggregate category is the speed-up state, then the current feed rate of equipment A is increased; when equipment A stops working, the target grading of equipment A is 1500 kg Per minute, the target gradation of equipment B is 2000 kg per minute, and the target gradation of equipment C is 1000 kg per minute. The target gradation of equipment A is allocated according to the ratio of the target gradations of equipment B and equipment C. The emergency working gradation allocated to equipment B is 1000 kg per minute, and the emergency working gradation allocated to equipment C is 500 kg per minute. The emergency target gradation of equipment B is 2500 kg per minute, and the emergency target gradation of equipment C is 1500 kg per minute. Inputting the emergency target gradation into the management system to keep the system running normally can ensure that the quality of subsequent products will not be greatly affected, and ensure the normal operation of the system and the normal operating rate of each equipment.

[0076] In some other preferred embodiments, when the aggregate bin corresponding to the grading level is mixed with aggregates of other grading levels, the proportion of the mixed aggregate in the total number of levels is counted and added to the number of levels of the corresponding grading level to obtain the actual grading, and the aggregate rate is adjusted according to the actual grading calculation.

[0077] In some other preferred embodiments, if device A fails and cannot continue to work, devices B and C continue to maintain emergency working levels.

[0078] The second aspect of the present invention provides an automatic grading control method based on aggregate image recognition, which is applied to the automatic grading control system based on aggregate image recognition, and is characterized by comprising:

[0079] Step 1: Collect aggregate images in different aggregate bins in real time;

[0080] Step 2: Obtain the composite gradation of different aggregate bins based on the aggregate image;

[0081] Step 3: Match the synthetic gradation with the target gradation in the database to obtain control data;

[0082] Step 4: Regulate the feeding equipment based on the control data to complete the grading adjustment;

[0083] Step 5: Control the emergency switch of the equipment based on the synthetic grading.

[0084] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0085] Working principle of the present invention:

[0086] The present invention collects aggregate images through image acquisition equipment, extracts image features of the aggregate images, obtains aggregate categories of aggregates in regional images according to a neural network model, segments stacked aggregates and adhered aggregates based on the image features of the aggregates to identify corresponding aggregate categories, counts aggregate categories and corresponding aggregate quantities to obtain synthetic gradations, sets adjustment thresholds according to historical grading data in a database, calculates corresponding equipment grading data based on the difference between the synthetic gradation and the target gradation, generates an alarm signal based on a comparison result between the synthetic gradation and a preset alarm threshold, and controls an emergency switch of the equipment based on the alarm signal.

[0087] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An automatic grading control system based on aggregate image recognition, characterized in that: include: Image collection module, gradation acquisition module, gradation control module and early warning module; Image collection module: used to collect aggregate images in different aggregate bins in real time; Grading acquisition module: used to obtain the synthetic gradation of different aggregate bins according to the aggregate image, and match the synthetic gradation with the target gradation in the database to obtain the control data; Grading control module: used to control the feeding equipment based on the control data to complete the grading adjustment; Early warning module: used to control the emergency switch of equipment based on synthetic grading.

2. The automatic grading control system based on aggregate image recognition according to claim 1 is characterized in that: The method of obtaining the composite gradation of different aggregate bins according to the aggregate image includes: Extract aggregate image; The aggregate features of different aggregates in the aggregate image are extracted through the Python library and input into the aggregate recognition model as input data to obtain the corresponding aggregate category; wherein the aggregate recognition model is constructed based on the neural network model; The aggregate whose aggregate category is not successfully identified is marked as unidentified aggregate, and the aggregate whose aggregate category is successfully identified is marked as identified aggregate; The shape features of the identified aggregates are extracted and the aggregate particle size is obtained. The aggregate particle sizes of all identified aggregates in the aggregate image are counted, and the grading gears are determined. The number of aggregate particle sizes of the identified aggregates is counted to obtain the number of gears. The number of gears of all aggregate bins is counted separately. The ratio of the number of gears in the aggregate image of the aggregate bin to the sum of the number of gears is taken as the single-gear sub-gradation of the aggregate bin. The sub-gradations of all aggregate bins are integrated to obtain the synthetic gradation.

3. The automatic grading control system based on aggregate image recognition according to claim 2 is characterized in that: The aggregate identification model is constructed based on a neural network model, and includes: Historical aggregate image features and corresponding aggregate categories are extracted from the database; the historical aggregate image features and the corresponding aggregate categories are used as training data and test data to train the neural network model, and after the training, the training effect is tested through the test data, and the neural network model is adjusted according to the test results to obtain an aggregate recognition model whose input data is the aggregate image features and output data is the aggregate category.

4. The automatic grading control system based on aggregate image recognition according to claim 1 is characterized in that: The step of matching the synthesized gradation with the target gradation in the database to obtain the control data includes: Extracting the synthetic gradation, and extracting the target gradation from the database; the target gradation includes the target category gradation corresponding to each aggregate, the aggregate category contained in each sub-gradation, and the target sub-gradation corresponding to each sub-gradation; Setting an adjustment threshold; determining whether the difference between the sub-gradation in the synthetic gradation and the corresponding target sub-gradation in the target gradation is greater than the adjustment threshold; if yes, marking the corresponding sub-gradation as a sub-gradation requiring adjustment; if no, marking the corresponding sub-gradation as a normal sub-gradation; wherein the adjustment threshold is set according to the working quality of the grading equipment; The aggregates in the sub-gradation containing the aggregate category are numbered as i, and the aggregate category is divided into sub-gradation categories of each level. The sub-gradation and the target sub-gradation are statistically obtained, and the sub-gradation of each level is numbered as j; through the formula Calculate the similarity ZXji of aggregate numbered i in sub-gradation category numbered j; where JPi represents the proportion of aggregate numbered i, ZJPi represents the proportion of sub-gradation aggregates in sub-gradation category numbered j, MJPi represents the proportion of target aggregates in aggregate numbered i, and MZJPj represents the target sub-gradation in sub-gradation category numbered j; Determine whether the aggregate similarity of the aggregate category is greater than 0; if yes, mark the adjustment state of the corresponding aggregate category as a deceleration state; if no, mark the adjustment state of the corresponding aggregate category as a speed-up state; Calculate the normalized average value of aggregate similarity of all aggregate categories in each sub-gradation category and multiply it by the proportion of the corresponding sub-gradation category in the total gradation to obtain the gradation coefficient; multiply all sub-gradations by the corresponding gradation coefficient to obtain the total gradation value; The equipment grading data is obtained by integrating the sub-grading that needs to be adjusted, the adjustment status of each aggregate category and the total grading value.

5. The automatic grading control system based on aggregate image recognition according to claim 4 is characterized in that: The adjustment threshold is set according to the working quality of the grading equipment, including: Extract the historical gradation data corresponding to each level of distribution equipment from the database; wherein the historical gradation data includes the historical synthetic gradation of each level of distribution equipment and the corresponding historical product quality; the historical product quality is evaluated by the staff, and the historical product quality includes excellent products, good products and inferior products; The historical composite gradation corresponding to the superior and good products is statistically obtained to obtain the superior historical gradation, and the historical composite gradation corresponding to the inferior products is statistically obtained to obtain the inferior historical gradation; The mode in the historical composite gradation is counted and marked as the standard gradation, and the adjustment threshold is obtained by calculating the ratio of the good historical gradation to the inferior historical gradation plus one and multiplying it by the standard gradation.

6. The automatic grading control system based on aggregate image recognition according to claim 1 is characterized in that: The step of adjusting the feeding device based on the control data to complete the grading adjustment includes: Extract regulatory data; Adjust the feeding equipment parameters of the corresponding sub-gradation according to the control data; take the number of sub-gradations in the composite gradation as the original feeding number, combine the control data with the composite gradation to obtain the regulated composite gradation, take the corresponding regulated sub-gradation rate in the regulated composite gradation as the feeding rate, and adjust the corresponding feeding equipment parameters to the equipment parameters that meet the feeding rate.

7. The automatic grading control system based on aggregate image recognition according to claim 1 is characterized in that: The emergency switch of the synthetic grading control device comprises: Extract the synthetic gradation and adjustment difference; set the alarm adjustment threshold and alarm type threshold; wherein the alarm adjustment threshold is set according to the maximum adjustment threshold that can be tolerated during production; the alarm type threshold is set according to the gradation level corresponding to the aggregate image; Determine whether the aggregate particle size of the identified aggregate meets the alarm type threshold; if yes, mark the corresponding grading device as a normal feeding device; if no, mark the corresponding grading device as an error feeding device and generate a feeding alarm signal; Determine whether the adjustment difference of the grading equipment is greater than the alarm adjustment threshold; if yes, generate an adjustment alarm signal; if no, continue monitoring; The feed alarm signal and the adjustment alarm signal are integrated into an alarm signal. When the alarm signal is detected, the emergency switch of the equipment is controlled based on the alarm signal.

8. The automatic grading control system based on aggregate image recognition according to claim 7 is characterized in that: The emergency switch of the control device based on the alarm signal includes: The alarm signal is sent to the management system, which controls the power switch of the grading equipment corresponding to the sub-grading to be adjusted to be cut off and records the alarm time; The grading device that detects the feeding alarm signal is marked as a feeding alarm device, the aggregate in the feeding alarm device is adjusted, and the detection and identification are re-performed. If the feeding alarm device is a normal feeding device after re-feeding, the feeding alarm device is restarted; The grading device that detects the adjustment alarm signal is marked as an adjustment alarm device, and the difference between the sub-gradation to be adjusted and the target sub-gradation of the adjustment alarm device is marked as an adjustment aggregate rate; it is determined whether the adjustment state of the aggregate category corresponding to the alarm device is a speed-up state; if yes, the aggregate rate is adjusted as an adjustment amount to increase the aggregate feeding speed; if no, the aggregate rate is adjusted as a reduction aggregate rate to reduce the aggregate feeding speed; During the shutdown period of the grading equipment, the workload of the grading equipment of the unified grading combination category is regulated according to the target grading of the alarm device; the equipment switch is turned on again after the aggregate adjustment is completed.

9. The automatic grading control system based on aggregate image recognition according to claim 8, characterized in that: The method of regulating the workload of the unified gradation combination category gradation equipment according to the target gradation of the alarm equipment includes: Extract target gradation and gradation combination categories of alarm equipment; Match the grading equipment with the same grading combination category as the alarm equipment and mark it as the matching grading equipment; divide the target gradation of the alarm equipment into several emergency working gradations according to the ratio of the target gradation of the matching grading equipment, and add the emergency working gradation to the target gradation of the corresponding matching grading equipment to obtain the emergency target gradation.

10. An automatic grading control method based on aggregate image recognition, applied to an automatic grading control system based on aggregate image recognition as claimed in any one of claims 1 to 9, characterized in that: include: Step 1: Collect aggregate images in different aggregate bins in real time; Step 2: Obtain the composite gradation of different aggregate bins based on the aggregate image; Step 3: Match the synthetic gradation with the target gradation in the database to obtain control data; Step 4: Regulate the feeding equipment based on the control data to complete the grading adjustment; Step 5: Control the emergency switch of the equipment based on the synthetic grading.

Images