Potato processing intelligent quality control method and system based on multi-source data fusion

Abstract

The invention relates to a potato processing intelligent quality control method and system based on multi-source data fusion, and belongs to the technical field of visual inspection application, and the method comprises the steps: synchronously collecting three-dimensional visual data, X-ray detection data and processing state data in a potato processing process, inputting the data into a potato body defect fusion recognition model, and outputting a quality control pre-judgment result; determining a first qualified potato material, a first defective potato material and a first waste potato material based on the quality control pre-judgment result and the quality sorting condition; a three-dimensional virtual model is established based on the three-dimensional visual data, a cutting track for the first defective potato materials is simulated and generated in the three-dimensional virtual model according to a quality control pre-judgment result, and second defective potato materials and second waste potato materials in the first defective potato materials are determined through re-screening according to the cutting track; and the sorting mechanism is controlled to sort the first qualified potato material, the first waste potato material, the second defective potato material and the second waste potato material onto the corresponding conveyor belts for respective processing. The finished product yield and the product quality stability are improved.

Description

Technical Field

[0001] This application relates to the technical field of visual inspection applications, and in particular to an intelligent quality control method and system for potato processing based on multi-source data fusion. Background Technology

[0002] In the frozen potato processing steps, the cutting process is one of the core processes, and its quality control directly determines product quality and raw material yield. Raw material yield and defect removal efficiency are two core indicators that determine economic benefits and product quality. Among them, yield is directly related to raw material utilization and processing costs, while defect removal efficiency determines the safety and stability of the finished product.

[0003] In the existing processing mode, defects are mostly identified by a single vision or independent X-ray equipment, which makes it difficult to comprehensively and accurately identify defective areas on the surface and inside of the potato. This results in insufficient precision in defect location. Based on the fuzzy defect information, in order to avoid defect residue, over-cutting or complete removal of the potato is usually adopted, resulting in a large amount of qualified potato flesh being wasted and a low yield. On the other hand, if the cutting process is simplified in order to improve the yield, defect residue is likely to occur, leading to a decrease in the qualified rate of the finished product and even causing food safety risks.

[0004] Therefore, there is an urgent need for a potato processing quality control method that synergistically optimizes yield and defect removal to ensure finished product yield and stable product quality. Summary of the Invention

[0005] To improve the yield and stability of product quality, this application provides a method and system for intelligent quality control of potato processing based on multi-source data fusion.

[0006] Firstly, this application provides an intelligent quality control method for potato processing based on multi-source data fusion, employing the following technical solution:

[0007] A smart quality control method for potato processing based on multi-source data fusion is applied to a potato processing quality control system. The potato processing quality control system includes multiple laser vision sensors, X-ray detection equipment, and a three-stage sorting mechanism arranged sequentially above a front-end conveyor belt along the conveying direction. The front-end conveyor belt is divided into a qualified potato conveyor belt, a defective potato conveyor belt, and a waste potato conveyor belt after the three-stage sorting mechanism. A first cutting mechanism is installed on the qualified potato conveyor belt. A defective potato cutting mechanism and a residual potato removal mechanism are installed at one end of the defective potato conveyor belt, and a second cutting mechanism is installed at the other end. The defective potato conveyor belt is conveyed from the defective potato cutting mechanism to the second cutting mechanism. The method includes:

[0008] Simultaneously collect 3D visual data, X-ray inspection data, and processing status data during potato processing;

[0009] The three-dimensional visual data, the X-ray detection data, and the processing status data are input into the potato defect fusion recognition model, and the quality control prediction results are output; wherein, the quality control prediction results include qualified areas and / or defective areas;

[0010] Based on the quality control prediction results and quality sorting conditions, the first qualified potato material, the first defective potato material, and the first waste potato material are determined.

[0011] A three-dimensional virtual model is established based on the three-dimensional visual data. A cutting trajectory for the first defective potato material is simulated and generated in the three-dimensional virtual model according to the quality control prediction result. The second defective potato material and the second waste potato material in the first defective potato material are determined by re-screening according to the cutting trajectory.

[0012] The sorting mechanism is controlled to sort the first qualified potato material, the first waste potato material, the second defective potato material, and the second waste potato material onto the corresponding conveyor belts;

[0013] Control the first cutting mechanism to cut the first qualified potato material;

[0014] The defect cutting mechanism is controlled to cut the second defective potato material to obtain the second qualified potato material and residual potato waste. At the same time, the residual potato removal mechanism is controlled to remove the residual potato waste from the defective potato material conveyor belt.

[0015] The second cutting mechanism is controlled to cut the second qualified potato material.

[0016] By adopting the above technical solutions, through multi-source data fusion and intelligent image recognition, the defects on the surface, inside, and foreign objects of potatoes can be accurately located. Combined with three-dimensional virtual modeling to simulate the cutting trajectory, the potato material can be accurately graded, sorted, and cut, reducing the waste of qualified potato material due to misjudgment. By directional removal of defective areas, the utilization rate of defective potato material can be maximized, while reducing ineffective adjustments to the cutting mechanism, reducing equipment wear and energy consumption, and significantly improving the yield and quality stability of potato processing. The entire potato processing chain can be analyzed, modeled, and adjusted to achieve strategic optimization of potato processing technology.

[0017] Furthermore, the three-dimensional virtual model includes virtual models of each potato on the conveyor belt, and the step of simulating and generating a cutting trajectory for the first defective potato based on the quality control prediction result in the three-dimensional virtual model includes:

[0018] The coordinates of the center point of the potato are determined based on the virtual model of the potato.

[0019] Based on the initial quality control prediction results, determine the coordinate set corresponding to the qualified area and / or defective area of ​​each potato in the three-dimensional virtual model, and display the qualified virtual area and / or defective virtual area on the potato virtual model according to the coordinate set to obtain the potato analysis model.

[0020] The initial cutting trajectory is generated by simulation based on the initially determined potato analysis model;

[0021] Analyze the shape of each of the virtual defect regions and select multiple feature points;

[0022] Connect the center point to each of the feature points to obtain multiple lines, determine the azimuth angle of each line, and obtain an azimuth angle set;

[0023] As the first defective potato material approaches the cutting mechanism, the updated set of azimuth angles is acquired in real time.

[0024] If the updated azimuth set changes, the cutting trajectory of the first defective potato is re-simulated and generated according to the updated potato analysis model; if the updated azimuth set does not change, the initial cutting trajectory is used as the cutting trajectory of the first defective potato.

[0025] By adopting the above technical solution, the center of the potato and the defect feature points are located through three-dimensional virtual modeling, and the cutting trajectory is dynamically adjusted by tracking the changes in azimuth angle in real time. This reduces the cutting deviation caused by the potato's posture shift, accurately matches the defect area, and improves the cutting accuracy and the efficiency of removing residual potatoes.

[0026] Further, the step of simulating and generating the initial cutting trajectory based on the initially determined potato analysis model includes:

[0027] Based on each type of virtual defect region, multiple sets of different alternative cutting trajectories are generated;

[0028] For each set of candidate cutting trajectories, calculate the safe distance between the cutting trajectory and the surface defect area, as well as the proportion of the qualified area covered by the cutting trajectory.

[0029] The alternative cutting trajectories with a safety distance greater than a safety threshold and a ratio less than a ratio threshold are determined as feasible trajectories;

[0030] Based on each of the feasible trajectories, simulate the operational complexity of the defect cutting mechanism and the residual potato removal mechanism in completing the feasible trajectory;

[0031] For each feasible trajectory, simulate the expected qualified potato material after cutting, calculate the ratio of the qualified area to the finished product volume in the expected qualified potato material, and obtain the qualification rate;

[0032] For each of the feasible trajectories, an ascending first sequence is generated based on the operational complexity and a descending second sequence is generated based on the pass rate. The index of each feasible trajectory in the first and second sequences is calculated by weighting to obtain a weighted index. A third sequence is obtained by ascending order based on the weighted index of each feasible trajectory. The first feasible trajectory in the third sequence is used as the initial cutting trajectory.

[0033] By adopting the above technical solution, feasible trajectories are determined through dual screening of the safe distance and qualified area coverage ratio of multiple sets of alternative trajectories. Combined with the weighted ranking of operation complexity and qualification rate, the optimal initial cutting trajectory is accurately selected, which ensures that defects are completely removed while maximizing the retention of qualified potatoes, reducing the difficulty of cutting operations, and improving processing efficiency and finished product quality.

[0034] Furthermore, based on each virtual defect region, multiple sets of different alternative cutting trajectories are generated, including:

[0035] A safe cutting area is obtained by extending a predetermined length outward from the boundary of the defective area;

[0036] Determine the distance between two adjacent safe cutting regions, simulate them with different distance thresholds, and cluster at least two safe cutting regions whose distance is less than the distance threshold to obtain the integrated cutting region corresponding to each selected distance threshold.

[0037] The appropriate excision method is determined based on the shape of each of the safety cutting regions and the shape of each of the integrated cutting regions;

[0038] Multiple sets of alternative cutting trajectories are generated based on the resection methods of individual excision and integrated cutting.

[0039] By adopting the above technical solution, by setting a safe cutting area and clustering adjacent areas, multiple sets of alternative trajectories are generated to adapt to individual or integrated cutting methods. This ensures that defective areas are completely removed, reduces the number of cuttings, optimizes the cutting path, and improves cutting efficiency. At the same time, it maximizes the retention of qualified potato volume and reduces processing losses.

[0040] Furthermore, the operational complexity of simulating the defect cutting mechanism and the residual potato removal mechanism to complete the feasible trajectory includes:

[0041] Determine the number of cuts, the number of direction changes, and the number of removals when the defective cutting mechanism and the residual potato removal mechanism complete the feasible trajectory;

[0042] The initial complexity is determined based on the number of cuts and the number of removals.

[0043] The additional complexity of each cut is determined based on the number of directional changes during each cut;

[0044] Determine the correlation between the cutting mechanisms in two adjacent cuts, including:

[0045] Calculate the angle difference Δθ between two adjacent cuts, and calculate the correlation R based on the angle difference Δθ. Among them, R i,i+1 Let be the correlation between the (i+1)th cut and the ith cut, i = 1, 2, ..., n, n be the total number of cuts, n > 1 and n is an integer, R ∈ [0, 1], and C be the attenuation coefficient;

[0046] Based on the additional complexity K of each cut i The additional complexity is adjusted using the correlation R to obtain the mean additional complexity K. ' :

[0047]

[0048] In the formula, K i Let K1 be the additional complexity of the i-th cut, i = 1, 2, ..., n, where n is the total number of cuts, n > 1 and n is an integer; K1 is the additional complexity of the first cut.

[0049] The operational complexity is calculated by adding the initial complexity to the average of the additional complexity.

[0050] By adopting the above technical solution, an operation complexity calculation model is constructed by quantifying the number of cuts, the number of direction changes, the number of rejections, and the correlation between adjacent cutting angles. This model accurately assesses the execution difficulty of the cutting trajectory, selects low-complexity trajectories, reduces frequent equipment adjustments, lowers energy consumption and losses, and improves the continuity and efficiency of cutting operations.

[0051] Further, the defective area includes at least one of a surface defective area, an internal defective area, and a high-density foreign matter area. The step of determining the first qualified potato material, the first defective potato material, and the first discarded potato material based on the quality control prediction results and quality sorting conditions includes:

[0052] Get the total surface area and total volume of the current potato;

[0053] Based on the defective area, the total surface area, and the total volume, at least one of the following is calculated: the total percentage of defective area, the percentage of surface defective area, the percentage of internal defective area, and the percentage of high-density foreign matter area.

[0054] Determine whether the total percentage of the defective areas is greater than the first discard threshold; if so, determine that the current potato is the first discarded potato.

[0055] Otherwise, determine whether at least one of the surface defect area ratio, the internal defect area ratio, and the high-density foreign matter area ratio is greater than the corresponding second discard threshold.

[0056] If the value is greater than the second discard threshold, the current potato is determined to be the first defective potato.

[0057] If the value is less than or equal to the second discard threshold, then the current potato is determined to be the first qualified potato feed.

[0058] By adopting the above technical solution, the proportion of different types of defective areas is quantified, and the dual threshold judgment rule is combined to achieve accurate grading of potato material. This not only avoids potato material with an excessively high proportion of defects from flowing into the processing stage and affecting quality, but also distinguishes repairable defective potato material, reduces misjudgment of qualified potato material, and improves the utilization rate of potato material and the scientific nature of processing quality control.

[0059] Further, the step of determining the second defective potato and the second waste potato in the first defective potato material by re-screening according to the cutting trajectory includes:

[0060] The simulated cut potatoes are compared with the expected qualified potatoes and the state of residual defects with the preset qualified standards. Additionally, the cutting complexity of the cutting trajectory is compared with a preset complexity threshold.

[0061] If the size of the expected qualified potato material meets the requirements and the proportion of defective residue is not greater than the first preset proportion, and the cutting complexity is not greater than the complexity threshold, then it is determined to be the second defective potato material.

[0062] If the size of the expected qualified potato material meets the requirements, but the proportion of defective residue is greater than the first preset proportion and / or the cutting complexity is greater than the complexity threshold, then it is determined to be the second waste potato material;

[0063] If the size of the expected qualified potato material does not meet the requirements, it is determined to be the second type of waste potato material.

[0064] By adopting the above technical solution, through multi-dimensional comparison and screening of simulated cutting results with preset standards, and combining the expected finished product size, the proportion of defective residues and the cutting complexity, the second defective potato material that can be processed is accurately distinguished from the second waste potato material that has no value. This avoids ineffective cutting losses and ensures the quality and efficiency of the finished product in subsequent processing.

[0065] Secondly, this application provides an intelligent quality control system for potato processing based on multi-source data fusion, employing the following technical solution:

[0066] The detection data acquisition module is used to simultaneously acquire three-dimensional visual data, X-ray detection data, and processing status data during the potato processing process;

[0067] The quality control prediction module is used to input the three-dimensional visual data, the X-ray detection data, and the processing status data into the potato defect fusion recognition model and output the quality control prediction result; wherein, the quality control prediction result includes qualified areas and / or defective areas;

[0068] The initial potato quality grading module is used to determine the first qualified potato, the first defective potato, and the first discarded potato based on the quality control prediction results and cutting restrictions.

[0069] The potato quality grading and verification module is used to establish a three-dimensional virtual model based on the three-dimensional visual data, simulate and generate a cutting trajectory for the first defective potato in the three-dimensional virtual model according to the quality control prediction result, and screen and determine the second defective potato and the second waste potato in the first defective potato based on the cutting trajectory.

[0070] The potato sorting module is used to control the sorting mechanism to sort the first qualified potato, the first waste potato, the second defective potato and the second waste potato onto the corresponding conveyor belt;

[0071] The first qualified potato cutting module is used to control the first cutting mechanism to cut the first qualified potato.

[0072] The defective potato processing module is used to control the defective cutting mechanism and cut the second defective potato to obtain the second qualified potato and residual potato waste, and at the same time control the residual potato removal mechanism to remove the residual potato waste from the defective potato conveyor belt.

[0073] The second qualified potato cutting module is used to control the second cutting mechanism to cut the second qualified potato.

[0074] Thirdly, this application provides an electronic device that adopts the following technical solution:

[0075] An electronic device, comprising:

[0076] At least one processor;

[0077] Memory;

[0078] At least one computer program, wherein the at least one computer program is stored in the memory and configured to be executed by the at least one processor, the at least one computer program being configured to: perform the method as described in any one of the first aspects.

[0079] Fourthly, this application provides a computer-readable storage medium, which adopts the following technical solution:

[0080] A computer-readable storage medium storing a computer program that can be loaded by a processor and execute the method as described in any one of the first aspects.

[0081] For a detailed description of the second to fourth aspects of the present invention and their various implementations, please refer to the detailed description in the first aspect and its various implementations; and for a detailed description of the beneficial effects of the second to fourth aspects and their various implementations, please refer to the beneficial effect analysis in the first aspect and its various implementations, which will not be repeated here.

[0082] In summary, this application includes at least one of the following beneficial technical effects:

[0083] 1. By integrating multi-source data and intelligent image recognition, the system accurately locates surface, internal, and foreign object defects in potatoes. Combined with 3D virtual modeling to simulate the cutting trajectory, it enables precise sorting and cutting of potato materials, maximizing the utilization rate of defective potato materials, reducing equipment wear and energy consumption, and significantly improving the yield and quality stability of potato processing. The system also analyzes, models, and adjusts the entire potato processing chain to optimize potato processing strategies.

[0084] 2. By locating the center of the potato and the defect feature points through 3D virtual modeling, the cutting trajectory is dynamically adjusted by tracking the changes in azimuth in real time, reducing the cutting deviation caused by the potato's posture shift, accurately matching the defect area, and improving the cutting accuracy and the efficiency of removing residual potatoes.

[0085] 3. By comparing the simulated cutting results with the preset standards in multiple dimensions, and combining the expected finished product size, the proportion of defective residues and the cutting complexity, the potato material is accurately graded to avoid ineffective cutting losses and ensure the quality and efficiency of the finished product in subsequent processing. Attached Figure Description

[0086] Figure 1 This is a schematic diagram of the potato processing quality control system of this application.

[0087] Figure 2 This is a flowchart illustrating the intelligent quality control method for potato processing based on multi-source data fusion in the embodiments of this application.

[0088] Figure 3 This is a structural block diagram of the intelligent quality control system for potato processing based on multi-source data fusion in the embodiments of this application.

[0089] Figure 4 This is a structural block diagram of the electronic device in the embodiments of this application. Detailed Implementation

[0090] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0091] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article, unless otherwise specified, generally indicates that the preceding and following related objects have an "or" relationship.

[0092] This application discloses an intelligent quality control method for potato processing based on multi-source data fusion. It is applied to a potato processing quality control system. (Refer to...) Figure 1 The potato processing quality control system includes multiple laser vision sensors, X-ray detection equipment, and a three-level diversion and sorting mechanism located above the front conveyor belt and arranged sequentially along the conveying direction. After the three-level diversion and sorting mechanism, the front conveyor belt is divided into a qualified potato conveyor belt, a defective potato conveyor belt, and a waste potato conveyor belt.

[0093] Laser vision sensors collect three-dimensional visual data of potatoes, while X-ray inspection equipment detects internal defects and high-density foreign objects in potatoes.

[0094] After being inspected by laser vision sensors and X-ray detection equipment, the quality grade of each potato is determined through analysis: qualified potatoes, defective potatoes, and discarded potatoes. Qualified potatoes are of superior quality and can be directly cut into strips or slices; defective potatoes have minor surface or internal defects, but after cutting and processing, they can yield intact potatoes that meet the size requirements, thus allowing for further cutting into strips or slices; discarded potatoes have too many surface or internal defects, making cutting difficult or resulting in potatoes that do not meet the size requirements, and are therefore discarded.

[0095] The three-level diversion and sorting mechanism can adopt any of the existing pneumatic push rod type, swing arm guide type, baffle reversing diversion type or vision positioning robot, etc. This application does not limit it.

[0096] Furthermore, the qualified potato conveyor belt is equipped with a first cutting mechanism, which is used to cut the potato into strips or slices.

[0097] A defective potato conveyor belt has a defect cutting mechanism and a residual potato removal mechanism at one end, which can be respectively set on both sides of the conveyor belt. A second cutting mechanism is set at the other end of the conveyor belt, and the conveyor belt conveys potatoes from the defective potato cutting mechanism to the second cutting mechanism. As the defective potatoes are conveyed on the conveyor belt, the location of the defects is determined based on the detection results of laser vision sensors and X-ray detection equipment. The defect cutting mechanism cuts at the defect location, and the residual potato removal mechanism removes the cut-off residual potatoes from the defective potato conveyor belt. The two mechanisms work together to completely remove the defects from the potatoes, resulting in qualified potatoes that can be sliced / diced, and then processed by the second cutting mechanism.

[0098] The method is executed by an electronic device, which can be a server or a terminal device. The server can be a standalone physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server providing cloud computing services. The terminal device can be a smartphone, tablet, desktop computer, etc., but is not limited to these. (See reference...) Figure 2 This includes steps S101 to S108:

[0099] Step S101: Synchronously collect 3D visual data, X-ray detection data and processing status data during potato processing.

[0100] Specifically, a laser vision sensor scans the potatoes on the front conveyor belt, collecting data on their shape, size, surface contour, and surface defects such as scratches, dents, and mold spots. An X-ray inspection device penetrates the potato to collect internal structural data, including internal holes, rotten areas, insect-infested passages, metal debris, and stones, while also recording the location coordinates and dimensions of defective areas. Furthermore, sensors mounted on the conveyor belt collect data on the conveyor belt's speed, sensor sampling frequency, and cutting mechanism rotation speed, thus obtaining processing status data. The electronic equipment can adjust the sampling frequency of the laser vision sensor and the relevant parameters of the X-ray inspection device based on this processing status data, resulting in more accurate data.

[0101] Electronic devices synchronize and integrate data to ensure that relevant test data for the same potato corresponds.

[0102] Step S102: Input the 3D visual data, X-ray inspection data and processing status data into the potato defect fusion recognition model and output the quality control prediction results; wherein, the quality control prediction results include qualified areas and / or defective areas.

[0103] Specifically, the electronic device pre-constructs a potato defect fusion and recognition model based on deep learning, employing a hybrid network structure of CNN deep neural network and Transformer. To train the model, a large amount of data, including 3D visual point cloud data and X-ray grayscale images, is acquired and channel fusion is performed to achieve multimodal image recognition, serving as the training dataset. The training dataset includes potato sample data of various varieties and defect types under different processing conditions. The training samples are input into the network for training, and the network parameters are adjusted during the training process, iterating until the model's test accuracy reaches the preset value.

[0104] The electronic device inputs synchronized 3D visual data, X-ray inspection data, and processing status data into the trained model. The model outputs the quality control prediction results for each potato, marking the location and type of qualified areas and / or defective areas.

[0105] Step S103: Based on the quality control prediction results and quality sorting conditions, determine the first qualified potato material, the first defective potato material, and the first discarded potato material, specifically including steps S1031 to S1035.

[0106] Step S1031: Obtain the total surface area and volume of the current potato.

[0107] Specifically, the electronic device obtains the total surface area and total volume of the potato through three-dimensional visual data.

[0108] Step S1032: Based on the defective area, total surface area, and total volume, calculate at least one of the following: total defective area percentage, surface defective area percentage, internal defective area percentage, and high-density foreign matter area percentage.

[0109] Specifically, defective areas include surface defective areas, internal defective areas, and high-density foreign matter areas. Electronic devices determine the type of defective area. If it is a surface defect, the ratio of the area of ​​the surface defective area to the total surface area is used as the percentage of the surface defective area. If it is an internal defect or a high-density foreign matter, the ratio of the volume of the internal defective area to the total volume is used as the percentage of the internal defective area, and the ratio of the volume of the high-density foreign matter area to the total volume is used as the percentage of the high-density foreign matter area. The sum of the percentages of surface defective areas, internal defective areas, and high-density foreign matter areas is used as the total percentage of defective areas.

[0110] Step S1033: Determine whether the total percentage of defective areas is greater than the first discard threshold; if so, proceed to step S1034: Determine that the current potato is the first discarded potato.

[0111] Otherwise, proceed to step S1035: determine whether at least one of the following is greater than the corresponding second discard threshold: the proportion of surface defect area, the proportion of internal defect area, and the proportion of high-density foreign matter area.

[0112] If the value exceeds the second discard threshold, the current potato is determined to be the first defective potato.

[0113] If the value is less than or equal to the second discard threshold, then the current potato is determined to be the first qualified potato feed.

[0114] Specifically, the electronic device presets a first discard threshold and a second discard threshold. When the total proportion of defective areas is greater than the first discard threshold, the potato has no processing value and is determined to be the first discarded potato. If the proportion of any type of defective area is not greater than the first discard threshold but greater than the second discard threshold, the potato can be processed into qualified potato and is determined to be the first defective potato. If the proportion of all types of defective areas is less than or equal to the second discard threshold, it is determined to be the first qualified potato.

[0115] For example, the total surface area of ​​a certain potato is 180 cm². 2 The total volume is 250cm. 3 Surface defect area 9cm 2 The volume of the internal defect area is 25cm. 3 There are no high-density foreign objects. Therefore, the percentage of surface defective areas is 9 / 180 = 5%, the percentage of internal defective areas is 25 / 250 = 10%, and the total percentage of defective areas is 15%. The first rejection threshold is set at 25%, and the second rejection thresholds are: 3% for surface defective areas, 2% for internal defective areas, and 1% for high-density foreign object areas.

[0116] Therefore, the electronic device first determines that the total percentage of defective areas is 15% < 25%. Then, it further compares and determines that the percentage of surface defective areas is 5% > 3% and the percentage of internal defective areas is 10% > 2%, thus determining that the current potato is the first defective potato.

[0117] Step S104: Establish a three-dimensional virtual model based on three-dimensional visual data. In the three-dimensional virtual model, simulate and generate a cutting trajectory for the first defective potato material according to the quality control prediction results. Based on the cutting trajectory, re-screen to determine the second defective potato material and the second waste potato material in the first defective potato material.

[0118] Specifically, the electronic device uses relevant software to perform point cloud stitching, noise reduction, and meshing processing based on the collected 3D visual point cloud data to construct a 3D virtual model with the same proportion as the actual potato. The model restores the potato's outline, surface features, and internal defects.

[0119] In the 3D virtual model, the electronic equipment simulates the cutting process of the defect cutting mechanism based on the defective areas in the quality control prediction results, generating a cutting trajectory. By simulating the cutting, it is determined whether a qualified finished product can be obtained after cutting and the feasibility of the cutting operation. Then, the first defective potato material is re-screened into the second defective potato material and the second waste potato material.

[0120] When the electronic device simulates and generates the cutting trajectory for the first defective potato material in the three-dimensional virtual model based on the quality control prediction results, it includes steps S11 to S17:

[0121] Step S11: Determine the coordinates of the center point of the potato based on the virtual model of the potato.

[0122] Specifically, a spatial rectangular coordinate system is established in the virtual model, and then the average of the coordinates of all vertices of the potato is calculated to obtain the coordinates of the center point.

[0123] Step S12: Based on the initial quality control prediction results, determine the coordinate set corresponding to the qualified area and / or defective area of ​​each potato in the three-dimensional virtual model, and display the qualified virtual area and / or defective virtual area on the potato virtual model according to the coordinate set to obtain the potato analysis model.

[0124] Specifically, after the electronic device determines the qualified and / or defective areas of the potato, it can mark the qualified and defective virtual areas with different colors in the virtual model to form an intuitive potato analysis model.

[0125] Step S13: Simulate and generate the initial cutting trajectory based on the initially determined potato analysis model. This includes steps Sa to Sf.

[0126] Step Sa: Based on each type of defective virtual region, generate multiple sets of different alternative cutting trajectories.

[0127] Specifically, for each virtual defect area in the potato model, the electronic device generates multiple sets of alternative cutting trajectories based on the movement range and cutting method of the cutting mechanism. For example, for a long strip-shaped surface defect, several alternative cutting trajectories can be generated, such as "straight cutting along the length of the defect," "segmented cutting perpendicular to the direction of the defect," and "cutting around the curve of the defect edge." This includes steps Sa1 to Sa4.

[0128] Step Sa1: Extend the defective area outward by a preset length to obtain a safe cutting area.

[0129] Specifically, the safe cutting area includes the defective area, and when cutting along the safe area, the defect can be removed without any residue.

[0130] Step Sa2: Determine the distance between two adjacent safe cutting regions, simulate them with different distance thresholds, and cluster at least two safe cutting regions whose distance is less than the distance threshold to obtain the integrated cutting region corresponding to each selected distance threshold.

[0131] For example, three different distance thresholds can be set, such as 0.3cm, 0.5cm, and 0.8cm, to determine whether the distance between adjacent safe cutting regions is less than the threshold. If the distance is less than the threshold, the two safe cutting regions are clustered to form a unified cutting region. For example, if two safe cutting regions are 0.5cm apart, they are clustered to form a unified cutting region when the threshold is 0.5cm or 0.8cm; when the threshold is 0.3cm, they are not clustered, and the two safe cutting regions remain independent.

[0132] Step Sa3: Determine the appropriate excision method based on the shape of each safe cutting area and the shape of each integrated cutting area.

[0133] Specifically, for individual safe cutting regions that are not clustered, the cutting method is determined based on their shape. For example, circular safe cutting regions use a circular cutting method, rectangular regions use a rectangular cutting method, and irregular regions use a cutting method that follows the boundary. For clustered integrated cutting regions, an appropriate integrated cutting method is determined based on their overall shape, such as elliptical cutting or polygonal cutting.

[0134] Step Sa4: Generate multiple sets of alternative cutting trajectories based on the individual excision method and the integrated excision method respectively.

[0135] Specifically, the electronic device first determines a set of candidate cutting trajectories by individually removing each defective area. Then, it performs integrated cutting on the corresponding integrated area at each distance threshold to obtain multiple other sets of candidate cutting trajectories. For example, if there are two adjacent safe cutting areas, removing them individually generates two circular cutting trajectories, which is one set of candidate cutting trajectories. When the two safe cutting areas are at distance thresholds of 0.5cm or 0.8cm respectively, they can be integrated into one integrated cutting area. Cutting on each integrated cutting area generates one elliptical cutting trajectory, resulting in two other sets of candidate cutting trajectories. Therefore, for the same potato, a total of three candidate cutting trajectories can be obtained.

[0136] Step Sb: For each set of candidate cutting trajectories, calculate the safe distance between the cutting trajectory and the surface defect area, as well as the proportion of the qualified area covered by the cutting trajectory.

[0137] Step Sc: Determine the alternative cutting trajectories with a safe distance greater than the safe threshold and a proportion less than the proportion threshold as feasible trajectories.

[0138] The safety distance refers to the minimum distance between the cutting trajectory and the boundary of the surface defect area, with a camera safety threshold of 0.2cm. One candidate cutting trajectory has a minimum distance of 0.3cm > 0.2cm, meeting the safety distance requirement; the other trajectory has a minimum distance of 0.15cm < 0.2cm, failing to meet the requirement.

[0139] The qualified area coverage ratio refers to the ratio of the volume of the qualified area covered by the cutting trajectory to the total qualified area of ​​the potato. The threshold ratio is set at 10%. If the qualified area coverage ratio of the alternative cutting trajectory is 12% > 10%, too many defective potatoes need to be removed, which is wasteful and does not meet the requirements.

[0140] Step Sd: Based on each feasible trajectory, simulate the operational complexity of the defect cutting mechanism and the residual potato removal mechanism to complete the feasible trajectory. Specifically, this includes steps Sd1 to Sd6.

[0141] Step Sd1: Determine the number of cuts, the number of direction changes, and the number of rejections when the defect cutting mechanism and the residual potato removal mechanism complete the feasible trajectory.

[0142] For example, if there is a square trajectory and a circular trajectory in the feasible trajectory, then the number of cutting times is 2. The first direction changes 4 times, and the second direction changes p times, where p is a preset value. When the defect cutting mechanism needs to cut the circular area, the number of direction changes is determined to be p. Furthermore, the number of rejections is 2, that is, the square waste potatoes and the circular waste potatoes that are cut off need to be rejected separately.

[0143] Step Sd2: Determine the initial complexity based on the number of cuts and removals.

[0144] Specifically, the electronic device presets the weights corresponding to the number of cuts and the number of rejections, and then multiplies the number of cuts and the number of rejections by their respective weights and adds them together to calculate the initial complexity.

[0145] Step Sd3: Determine the additional complexity of each cut based on the number of orientation changes during each cut.

[0146] Specifically, an electronic device has a preset complexity comparison table, where each complexity corresponds to a preset range of directional changes. Therefore, the additional complexity can be determined by comparing the number of directional changes.

[0147] Step Sd4: Determine the correlation between the cutting mechanisms in two adjacent cuts, including:

[0148] Calculate the angle difference Δθ between two adjacent cuts, and then calculate the correlation R based on the angle difference Δθ. Among them, R i,i+1 Let be the correlation between the (i+1)th cut and the ith cut, i = 1, 2, ..., n, n be the total number of cuts, n > 1 and n is an integer, R ∈ [0, 1], and C be the attenuation coefficient.

[0149] Specifically, the smaller the change in the cutting blade angle between two adjacent cuts, the stronger the correlation, the better the operational continuity of the defect cutting mechanism, and the lower the additional complexity should be; conversely, the larger the angle change, the worse the operational continuity, and the higher the additional complexity should be.

[0150] Furthermore, in order to limit the correlation value range to [0, 1], and to ensure that the correlation is negatively correlated with the angle difference Δθ, the exponential function... The conditions are met. The attenuation count C can be set flexibly.

[0151] For example, when the angle difference between the first and second cuts is Δθ = 0°, meaning the two cuts are exactly the same, then R 1,2 =1, the highest relevance, indicating that the operation is completely consistent and no additional complexity needs to be corrected;

[0152] When Δθ = 10°, C = 10, The correlation is moderate, indicating that the operation has a certain degree of continuity, and the additional complexity needs to be adjusted appropriately.

[0153] When Δθ≥30°, C=10, The relevance is extremely low, the operational consistency is poor, and the additional complexity needs to be significantly modified.

[0154] Step Sd5: Based on the additional complexity K of each cut i The additional complexity is adjusted by the correlation R, resulting in the mean additional complexity K. ' :

[0155]

[0156] In the formula, K i Let K1 be the additional complexity of the i-th cut, i = 1, 2, ..., n, where n is the total number of cuts, n > 1 and n is an integer; K1 is the additional complexity of the first cut.

[0157] Specifically, K1 represents the additional complexity of the first step of the cut. Starting from the second step, the correlation between the second and first steps is introduced. Since a lower correlation requires a higher cutting complexity, the additional complexity of the second step is K2 × (1 - R). 1,2 Similarly, by summing the additional complexity of each cut and dividing by the number of cuts n, we obtain the average additional complexity K. ' .

[0158] Step Sd6: Add the average of the initial complexity and the additional complexity to calculate the operation complexity.

[0159] Specifically, by adding the initial complexity to the additional complexity, the number of cuts and the direction of change of the defect cutting mechanism and the number of rejections of the waste potato rejection mechanism were measured, thus quantifying the execution difficulty of the cutting trajectory.

[0160] Step Se: For each feasible trajectory, simulate the expected qualified potato material after cutting, calculate the ratio of the qualified area to the finished product volume in the expected qualified potato material, and obtain the qualification rate.

[0161] For example: After simulating cutting along a feasible trajectory, the expected volume of the qualified potato material's qualified area is 80 cm³. 3 The total volume of the finished product is 85cm². 3 The pass rate was approximately 94.1% (80 / 85 ≈ 94.1%); after another trajectory simulation cutting, the volume of the qualified area was 70 cm³. 3 The total volume of the finished product is 80cm. 3 The pass rate was 87.5%.

[0162] Step Sf: For each feasible trajectory, generate an ascending first sequence based on the operation complexity and a descending second sequence based on the pass rate. Calculate the weighted index of each feasible trajectory in the first and second sequences to obtain a weighted index. Obtain a third sequence in ascending order based on the weighted indexes of each feasible trajectory. Use the first feasible trajectory in the third sequence as the initial cutting trajectory.

[0163] Specifically, the electronic device sorts all feasible trajectories in ascending order of operational complexity to obtain the first sequence. In the first sequence, the smaller the sequence number, the simpler the operation. It then sorts them in descending order of pass rate to obtain the second sequence. In the second sequence, the smaller the sequence number, the higher the pass rate.

[0164] The weight of the first sequence is set to 0.4, and the weight of the second sequence is set to 0.6. The weighted index is calculated as: first sequence index × 0.4 + second sequence index × 0.6. Therefore, the first feasible trajectory in the resulting third sequence is simple to operate and has a high success rate.

[0165] Step S14: Analyze the shape of each defective virtual area and select multiple feature points.

[0166] Specifically, for each virtual defect region, the electronic device selects feature points that characterize its shape and location, including boundary vertices, center points, and the endpoint of the longest axis. For example, a rectangular defect region selects 5 feature points, including 4 vertices and the center point; an irregular curved defect region selects 11 feature points, including 10 vertices and the center point evenly distributed on the boundary.

[0167] Step S15: Connect the center point to each feature point to obtain multiple lines, determine the azimuth angle of each line, and obtain the azimuth angle set.

[0168] Specifically, the electronic device establishes a local polar coordinate system with the center point of the potato as the origin, the X-axis as the polar axis, and the azimuth angle as the angle between the line connecting the feature point and the center point and the polar axis.

[0169] Step S16: As the first defective potato material approaches the cutting mechanism, the updated azimuth angle set is obtained in real time.

[0170] Specifically, as the first defective potato moves towards the cutting mechanism on the conveyor belt, its posture in the 3D virtual model may change due to factors such as conveyor belt vibration and slight rolling of the potato. The system acquires an updated potato analysis model every preset time interval, recalculates the azimuth angles of feature points, and obtains an updated set of azimuth angles.

[0171] Step S17: If the updated azimuth set changes, the cutting trajectory of the first defective potato is re-simulated and generated according to the updated potato analysis model; if the updated azimuth set does not change, the initial cutting trajectory is used as the cutting trajectory of the first defective potato.

[0172] Specifically, the electronic device compares the updated azimuth set with the initial azimuth set. If the change in all azimuth angles is ≤5°, it is determined that the azimuth set has not changed, and the initial cutting trajectory is used. If the change in any azimuth angle is >5°, it is determined that a change has occurred, and the cutting trajectory is re-simulated and generated based on the updated potato analysis model.

[0173] For example, if a potato rolls backward 30° along the x-axis during transport, the azimuth angles of the feature points in the defective area will change. In this case, a new cutting trajectory adapted to the new posture needs to be generated to ensure cutting accuracy.

[0174] Furthermore, when the electronic device re-screens the first defective potato material to determine the second defective potato material and the second waste potato material based on the cutting trajectory, it includes:

[0175] The simulated cut potatoes were compared with the expected qualified potatoes and the state of residual defects with the preset qualified standards. Additionally, the cutting complexity of the cutting trajectory was compared with the preset complexity threshold.

[0176] Among the preset qualification standards, the finished size of the qualified potato material must meet the process settings. The first preset percentage of defect residue can be 5%. The ratio of the volume of the defective area remaining after cutting to the total volume of the expected qualified potato material should be less than or equal to 5%. The preset complexity threshold can be set according to the actual operating capacity of the defect cutting mechanism. If the threshold is exceeded, the cutting efficiency will be too low or the equipment will not be able to complete the cutting.

[0177] (1) If the size of the expected qualified potato material meets the requirements and the proportion of defective residue is not greater than the first preset proportion, and the cutting complexity is not greater than the complexity threshold, then it is determined to be the second defective potato material.

[0178] Specifically, since the potatoes are easy to cut and produce good finished products, they are considered as the second type of defective potato material and are subject to subsequent actual defect removal and processing.

[0179] (2) If the size of the expected qualified potato material meets the requirements, but the proportion of defective residue is greater than the first preset proportion and / or the cutting complexity is greater than the complexity threshold, it is determined to be the second waste potato material.

[0180] Specifically, if the potatoes cannot be completely removed after being cut according to the planned trajectory, or if the cutting is too difficult and the economic benefits are low, then the potatoes are directly identified as second-stage waste potatoes and discarded in subsequent processes.

[0181] (3) If the size of the expected qualified potato material does not meet the requirements, it shall be judged as the second waste potato material.

[0182] Specifically, if the potatoes cannot be cut into qualified potatoes according to the planned rules, they will be directly identified as second waste potatoes and discarded in subsequent processes.

[0183] Step S105: Control the sorting mechanism to sort the first qualified potato material, the first waste potato material, the second defective potato material, and the second waste potato material onto the corresponding conveyor belt.

[0184] Specifically, the electronic equipment sends control signals to the three-stage sorting and distribution mechanism based on the initial grading results and the re-screening results, including the position coordinates of the potatoes and the direction of sorting.

[0185] The first qualified potato material is diverted to the qualified potato material conveyor belt, the first and second waste potato materials are diverted to the waste potato material conveyor belt, and the second defective potato material is diverted to the defective potato material conveyor belt.

[0186] Step S106: Control the first cutting mechanism to cut the first qualified potato material.

[0187] Step S107: Control the defect cutting mechanism and cut the second defective potato material to obtain the second qualified potato material and residual potato waste. At the same time, control the residual potato removal mechanism to remove the residual potato waste from the defective potato material conveyor belt.

[0188] Specifically, the electronic device sets the cutting parameters of the defect cutting mechanism based on the cutting trajectory determined in the 3D virtual model. When the second defective potato material arrives at the defect cutting mechanism, the mechanism cuts along the set trajectory to remove the defective area, obtaining the second qualified potato material and residual potato waste. After cutting, the residual potato removal mechanism is activated. Based on the position coordinates of the residual potato waste, it adsorbs the residual potatoes and removes them from the defective potato material conveyor belt to the waste collection box, ensuring that the second qualified potato material continues to be transported along the conveyor belt to the second cutting mechanism.

[0189] Step S108: Control the second cutting mechanism to cut the second qualified potato material.

[0190] Specifically, the potato residues from the first and second cutting mechanisms are kept consistent to ensure uniform finished product specifications.

[0191] To better implement the above method, this application also provides an intelligent quality control system for potato processing based on multi-source data fusion, referring to... Figure 3 The intelligent quality control system for potato processing based on multi-source data fusion includes:

[0192] The detection data acquisition module 201 is used to simultaneously acquire three-dimensional visual data, X-ray detection data and processing status data during the potato processing process;

[0193] The quality control prediction module 202 is used to input three-dimensional visual data, X-ray inspection data and processing status data into the potato defect fusion recognition model and output the quality control prediction result; wherein, the quality control prediction result includes qualified areas and / or defective areas;

[0194] The initial potato quality grading module 203 is used to determine the first qualified potato, the first defective potato, and the first discarded potato based on the quality control prediction results and cutting constraints.

[0195] The potato quality grading and verification module 204 is used to establish a three-dimensional virtual model based on three-dimensional visual data. In the three-dimensional virtual model, a cutting trajectory for the first defective potato is simulated and generated according to the quality control prediction results. The second defective potato and the second waste potato in the first defective potato are determined by re-screening according to the cutting trajectory.

[0196] The potato sorting module 205 is used to control the sorting mechanism to sort the first qualified potato, the first waste potato, the second defective potato and the second waste potato onto the corresponding conveyor belt;

[0197] The first qualified potato cutting module 206 is used to control the first cutting mechanism to cut the first qualified potato.

[0198] The defective potato processing module 207 is used to control the defect cutting mechanism and cut the second defective potato to obtain the second qualified potato and residual potato waste. At the same time, it controls the residual potato removal mechanism to remove the residual potato waste from the defective potato conveyor belt.

[0199] The second qualified potato cutting module 208 is used to control the second cutting mechanism to cut the second qualified potato.

[0200] Furthermore, when the potato quality grading verification module 204 simulates and generates the cutting trajectory for the first defective potato in the three-dimensional virtual model based on the quality control prediction results, it is specifically used for:

[0201] Determine the coordinates of the center point of the potato based on a virtual model of the potato;

[0202] Based on the initial quality control prediction results, determine the coordinate set corresponding to the qualified area and / or defective area of ​​each potato in the three-dimensional virtual model, and display the qualified virtual area and / or defective virtual area on the potato virtual model according to the coordinate set to obtain the potato analysis model.

[0203] The initial cutting trajectory is generated by simulation based on the initially determined potato analysis model;

[0204] Analyze the shape of each virtual defect area and select multiple feature points;

[0205] Connect the center point to each feature point to obtain multiple lines, determine the azimuth angle of each line, and obtain the azimuth angle set;

[0206] As the first defective potato approaches the cutting mechanism, the updated set of azimuth angles is acquired in real time.

[0207] If the updated azimuth set changes, the cutting trajectory of the first defective potato is re-simulated and generated according to the updated potato analysis model; if the updated azimuth set does not change, the initial cutting trajectory is used as the cutting trajectory of the first defective potato.

[0208] The initial cutting trajectory is generated based on the initially determined potato analysis model, including:

[0209] Based on each type of virtual defect region, multiple sets of different alternative cutting trajectories are generated;

[0210] For each set of alternative cutting trajectories, calculate the safe distance between the cutting trajectory and the surface defect area, as well as the proportion of the qualified area covered by the cutting trajectory.

[0211] Candidate cutting trajectories with a safe distance greater than the safety threshold and a proportion less than the proportion threshold are identified as feasible trajectories;

[0212] Based on each feasible trajectory, simulate the operational complexity of the defect cutting mechanism and the residual potato removal mechanism to complete the feasible trajectory;

[0213] For each feasible trajectory, simulate the expected qualified potato material after cutting, calculate the ratio of qualified area to finished product volume in the expected qualified potato material, and obtain the qualification rate;

[0214] For each feasible trajectory, an ascending first sequence is generated based on the operational complexity, and a descending second sequence is generated based on the pass rate. The weighted sequence number of each feasible trajectory in the first and second sequences is calculated. Based on the weighted sequence numbers of each feasible trajectory, a third sequence is obtained in ascending order. The first feasible trajectory in the third sequence is used as the initial cutting trajectory.

[0215] Based on each type of virtual defect region, multiple sets of different alternative cutting trajectories are generated, including:

[0216] The safe cutting area is obtained by extending a preset length outward from the boundary of the defective area;

[0217] Determine the distance between two adjacent safe cutting regions, simulate them with different distance thresholds, and cluster at least two safe cutting regions whose distance is less than the distance threshold to obtain the integrated cutting region corresponding to each selected distance threshold.

[0218] The appropriate excision method is determined based on the shape of each safe cutting area and the shape of each integrated cutting area;

[0219] Multiple sets of alternative cutting trajectories are generated based on the resection methods of individual excision and integrated cutting.

[0220] The operational complexity of simulating the defect cutting mechanism and the residual potato removal mechanism to complete a feasible trajectory includes:

[0221] Determine the number of cuts, the number of direction changes, and the number of removals required when the defective cutting mechanism and the residual potato removal mechanism complete a feasible trajectory;

[0222] The initial complexity is determined based on the number of cuts and removals.

[0223] The additional complexity of each cut is determined by the number of times the direction is changed during each cut;

[0224] Determine the correlation between the cutting mechanisms in two adjacent cuts, including:

[0225] Calculate the angle difference Δθ between two adjacent cuts, and then calculate the correlation R based on the angle difference Δθ. Among them, R i,i+1 Let be the correlation between the (i+1)th cut and the ith cut, i = 1, 2, ..., n, n be the total number of cuts, n > 1 and n is an integer, R ∈ [0, 1], and C be the attenuation coefficient;

[0226] Based on the additional complexity K of each cut i The additional complexity is adjusted based on the correlation R, resulting in the mean additional complexity K′:

[0227]

[0228] In the formula, K i Let K1 be the additional complexity of the i-th cut, i = 1, 2, ..., n, where n is the total number of cuts, n > 1 and n is an integer; K1 is the additional complexity of the first cut.

[0229] The operational complexity is calculated by adding the average of the initial complexity and the additional complexity.

[0230] Furthermore, the initial potato feed quality grading module 203 is specifically used for:

[0231] Get the total surface area and total volume of the current potato;

[0232] Based on the defective area, total surface area, and total volume, calculate at least one of the following: the total percentage of defective areas, the percentage of surface defective areas, the percentage of internal defective areas, and the percentage of high-density foreign matter areas.

[0233] Determine if the total percentage of defective areas is greater than the first discard threshold; if so, determine that the current potato is the first type of discarded potato.

[0234] Otherwise, determine whether at least one of the following is greater than the corresponding second discard threshold: the proportion of surface defect area, the proportion of internal defect area, and the proportion of high-density foreign matter area.

[0235] If the value exceeds the second discard threshold, the current potato is determined to be the first defective potato.

[0236] If the value is less than or equal to the second discard threshold, then the current potato is determined to be the first qualified potato feed.

[0237] Furthermore, the potato quality grading verification module 204, when determining the second defective potato and the second discarded potato in the first defective potato category based on the cutting trajectory through re-screening, is specifically used for:

[0238] The simulated cut potatoes were compared with the expected qualified potatoes and the state of residual defects with the preset qualified standards. Additionally, the cutting complexity of the cutting trajectory was compared with the preset complexity threshold.

[0239] If the size of the expected qualified potato material meets the requirements and the proportion of defective residue is not greater than the first preset proportion, and the cutting complexity is not greater than the complexity threshold, then it is judged as the second defective potato material.

[0240] If the size of the expected qualified potato material meets the requirements, but the proportion of defective residue is greater than the first preset proportion and / or the cutting complexity is greater than the complexity threshold, it is determined to be the second type of waste potato material.

[0241] If the size of the expected qualified potato material does not meet the requirements, it will be judged as the second type of waste potato material.

[0242] The various variations and specific examples in the methods of the foregoing embodiments are also applicable to the intelligent quality control system for potato processing based on multi-source data fusion in this embodiment. Through the foregoing detailed description of the intelligent quality control method for potato processing based on multi-source data fusion, those skilled in the art can clearly understand the implementation method of the intelligent quality control system for potato processing based on multi-source data fusion in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0243] To better implement the above methods, embodiments of this application provide an electronic device, referring to... Figure 4 The electronic device 300 includes a processor 301, a memory 303, and a display screen 305. The memory 303 and the display screen 305 are both connected to the processor 301, such as via a bus 302. Optionally, the electronic device 300 may also include a transceiver 304. It should be noted that in practical applications, the transceiver 304 is not limited to one type, and the structure of this electronic device 300 does not constitute a limitation on the embodiments of this application.

[0244] Processor 301 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 301 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0245] Bus 302 may include a pathway for transmitting information between the aforementioned components. Bus 302 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 302 may be divided into address bus, data bus, control bus, etc.

[0246] The memory 303 may be a ROM (Read Only Memory) or other type of static storage device capable of storing static information and instructions, RAM (Random Access Memory) or other type of dynamic storage device capable of storing information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0247] The memory 303 is used to store application code that executes the solution of this application, and its execution is controlled by the processor 301. The processor 301 is used to execute the application code stored in the memory 303 to implement the content shown in the foregoing method embodiments.

[0248] Figure 4 The electronic device 300 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0249] This application also provides a computer-readable storage medium storing a computer program. When executed by a processor, the program implements the intelligent quality control method for potato processing based on multi-source data fusion provided in the above embodiments. The processor executes the computer program in the memory, and through multi-source data fusion and intelligent image recognition, accurately locates defects on the surface, inside, and in the form of foreign objects in potatoes. Combined with three-dimensional virtual modeling to simulate the cutting trajectory, it achieves precise sorting and cutting of potato materials, reducing the waste of qualified potato materials due to misjudgment. Furthermore, by directional removal of defective areas, it maximizes the utilization rate of defective potato materials, while reducing ineffective adjustments to the cutting mechanism, lowering equipment wear and energy consumption, and significantly improving the yield and quality stability of potato processing products.

[0250] In this embodiment, the computer-readable storage medium can be a tangible device that holds and stores instructions used by an instruction execution device. The computer-readable storage medium can be, but is not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any combination thereof. Specifically, the computer-readable storage medium can be a portable computer disk, a hard disk, a USB flash drive, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory stick, floppy disk, optical disk, magnetic disk, mechanical encoding device, or any combination thereof.

[0251] The computer program in this embodiment includes program code for performing all the aforementioned methods. The program code may include instructions corresponding to the method steps provided in the above embodiments. The computer program can be downloaded from a computer-readable storage medium to various computing / processing devices, or downloaded to an external computer or external storage device via a network (e.g., the Internet, a local area network, a wide area network, and / or a wireless network). The computer program can be executed entirely on the user's computer as a standalone software package.

[0252] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

[0253] Additionally, it should be understood that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. The terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

Claims

1. A smart quality control method for potato processing based on multi-source data fusion, characterized in that, An application is made in a potato processing quality control system. The system includes multiple laser vision sensors, X-ray inspection equipment, and a three-stage sorting mechanism arranged sequentially above a front-end conveyor belt along the conveying direction. The front-end conveyor belt is divided into a qualified potato conveyor belt, a defective potato conveyor belt, and a waste potato conveyor belt after the three-stage sorting mechanism. A first cutting mechanism is installed on the qualified potato conveyor belt. A defective potato cutting mechanism and a residual potato removal mechanism are installed at one end of the defective potato conveyor belt, and a second cutting mechanism is installed at the other end. The defective potato conveyor belt is conveyed from the defective potato cutting mechanism to the second cutting mechanism. The method includes: Simultaneously collect 3D visual data, X-ray inspection data, and processing status data during potato processing; The three-dimensional visual data, the X-ray detection data, and the processing status data are input into the potato defect fusion recognition model, and the quality control prediction results are output; wherein, the quality control prediction results include qualified areas and / or defective areas; Based on the quality control prediction results and quality sorting conditions, the first qualified potato material, the first defective potato material, and the first waste potato material are determined. A three-dimensional virtual model is established based on the three-dimensional visual data. A cutting trajectory for the first defective potato material is simulated and generated in the three-dimensional virtual model according to the quality control prediction result. The second defective potato material and the second waste potato material in the first defective potato material are determined by re-screening according to the cutting trajectory. The sorting mechanism is controlled to sort the first qualified potato material, the first waste potato material, the second defective potato material, and the second waste potato material onto the corresponding conveyor belts; Control the first cutting mechanism to cut the first qualified potato material; The defect cutting mechanism is controlled to cut the second defective potato material to obtain the second qualified potato material and residual potato waste. At the same time, the residual potato removal mechanism is controlled to remove the residual potato waste from the defective potato material conveyor belt. The second cutting mechanism is controlled to cut the second qualified potato material.

2. The method according to claim 1, characterized in that, The three-dimensional virtual model includes virtual models of each potato on the conveyor belt. The step of simulating and generating a cutting trajectory for the first defective potato in the three-dimensional virtual model based on the quality control prediction results includes: The coordinates of the center point of the potato are determined based on the virtual model of the potato. Based on the initial quality control prediction results, determine the coordinate set corresponding to the qualified area and / or defective area of ​​each potato in the three-dimensional virtual model, and display the qualified virtual area and / or defective virtual area on the potato virtual model according to the coordinate set to obtain the potato analysis model. The initial cutting trajectory is generated by simulation based on the initially determined potato analysis model; Analyze the shape of each of the virtual defect regions and select multiple feature points; Connect the center point to each of the feature points to obtain multiple lines, determine the azimuth angle of each line, and obtain an azimuth angle set; As the first defective potato material approaches the cutting mechanism, the updated set of azimuth angles is acquired in real time. If the updated azimuth set changes, the cutting trajectory of the first defective potato is re-simulated and generated according to the updated potato analysis model; if the updated azimuth set does not change, the initial cutting trajectory is used as the cutting trajectory of the first defective potato.

3. The method according to claim 2, characterized in that, The step of simulating and generating the initial cutting trajectory based on the initially determined potato analysis model includes: Based on each type of virtual defect region, multiple sets of different alternative cutting trajectories are generated; For each set of candidate cutting trajectories, calculate the safe distance between the cutting trajectory and the surface defect area, as well as the proportion of the qualified area covered by the cutting trajectory. The alternative cutting trajectories with a safety distance greater than a safety threshold and a ratio less than a ratio threshold are determined as feasible trajectories; Based on each of the feasible trajectories, simulate the operational complexity of the defect cutting mechanism and the residual potato removal mechanism in completing the feasible trajectory; For each feasible trajectory, simulate the expected qualified potato material after cutting, calculate the ratio of the qualified area to the finished product volume in the expected qualified potato material, and obtain the qualification rate; For each of the feasible trajectories, an ascending first sequence is generated based on the operational complexity and a descending second sequence is generated based on the pass rate. The index of each feasible trajectory in the first and second sequences is calculated by weighting to obtain a weighted index. A third sequence is obtained by ascending order based on the weighted index of each feasible trajectory. The first feasible trajectory in the third sequence is used as the initial cutting trajectory.

4. The method according to claim 3, characterized in that, Based on each type of virtual defect region, multiple sets of different alternative cutting trajectories are generated, including: A safe cutting area is obtained by extending a predetermined length outward from the boundary of the defective area; Determine the distance between two adjacent safe cutting regions, simulate them with different distance thresholds, and cluster at least two safe cutting regions whose distance is less than the distance threshold to obtain the integrated cutting region corresponding to each selected distance threshold. The appropriate excision method is determined based on the shape of each of the safety cutting regions and the shape of each of the integrated cutting regions; Multiple sets of alternative cutting trajectories are generated based on the resection methods of individual excision and integrated cutting.

5. The method according to claim 4, characterized in that, The operational complexity of simulating the defect cutting mechanism and the residual potato removal mechanism to complete the feasible trajectory includes: Determine the number of cuts, the number of direction changes, and the number of removals when the defective cutting mechanism and the residual potato removal mechanism complete the feasible trajectory; The initial complexity is determined based on the number of cuts and the number of removals. The additional complexity of each cut is determined based on the number of directional changes during each cut; Determine the correlation between the cutting mechanisms in two adjacent cuts, including: Calculate the angle difference Δθ between two adjacent cuts, and calculate the correlation R based on the angle difference Δθ. Among them, R i,i+1 Let be the correlation between the (i+1)th cut and the ith cut, i = 1, 2, ..., n, n be the total number of cuts, n > 1 and n is an integer, R ∈ [0, 1], and C be the attenuation coefficient; Based on the additional complexity K of each cut i The additional complexity is adjusted using the correlation R to obtain the mean additional complexity K. ' : In the formula, K i Let K1 be the additional complexity of the i-th cut, i = 1, 2, ..., n, where n is the total number of cuts, n > 1 and n is an integer; K1 is the additional complexity of the first cut. The operational complexity is calculated by adding the initial complexity to the average of the additional complexity.

6. The method according to claim 1, characterized in that, The defective area includes at least one of surface defective areas, internal defective areas, and high-density foreign matter areas. The determination of the first qualified potato material, the first defective potato material, and the first discarded potato material based on the quality control prediction results and quality sorting conditions includes: Get the total surface area and total volume of the current potato; Based on the defective area, the total surface area, and the total volume, at least one of the following is calculated: the total percentage of defective area, the percentage of surface defective area, the percentage of internal defective area, and the percentage of high-density foreign matter area. Determine whether the total percentage of the defective areas is greater than the first discard threshold; if so, determine that the current potato is the first discarded potato. Otherwise, determine whether at least one of the surface defect area ratio, the internal defect area ratio, and the high-density foreign matter area ratio is greater than the corresponding second discard threshold. If the value is greater than the second discard threshold, the current potato is determined to be the first defective potato. If the value is less than or equal to the second discard threshold, then the current potato is determined to be the first qualified potato feed.

7. The method according to claim 5, characterized in that, The step of determining the second defective potato and the second waste potato in the first defective potato material by re-screening according to the cutting trajectory includes: The simulated cut potatoes are compared with the expected qualified potatoes and the state of residual defects with the preset qualified standards. Additionally, the cutting complexity of the cutting trajectory is compared with a preset complexity threshold. If the size of the expected qualified potato material meets the requirements and the proportion of defective residue is not greater than the first preset proportion, and the cutting complexity is not greater than the complexity threshold, then it is determined to be the second defective potato material. If the size of the expected qualified potato material meets the requirements, but the proportion of defective residue is greater than the first preset proportion and / or the cutting complexity is greater than the complexity threshold, then it is determined to be the second waste potato material; If the size of the expected qualified potato material does not meet the requirements, it is determined to be the second type of waste potato material.

8. A smart quality control system for potato processing based on multi-source data fusion, characterized in that, include: The detection data acquisition module is used to simultaneously acquire three-dimensional visual data, X-ray detection data, and processing status data during the potato processing process; The quality control prediction module is used to input the three-dimensional visual data, the X-ray detection data, and the processing status data into the potato defect fusion recognition model and output the quality control prediction result; wherein, the quality control prediction result includes qualified areas and / or defective areas; The initial potato quality grading module is used to determine the first qualified potato, the first defective potato, and the first discarded potato based on the quality control prediction results and cutting restrictions. The potato quality grading and verification module is used to establish a three-dimensional virtual model based on the three-dimensional visual data, simulate and generate a cutting trajectory for the first defective potato in the three-dimensional virtual model according to the quality control prediction result, and screen and determine the second defective potato and the second waste potato in the first defective potato based on the cutting trajectory. The potato sorting module is used to control the sorting mechanism to sort the first qualified potato, the first waste potato, the second defective potato and the second waste potato onto the corresponding conveyor belt; The first qualified potato cutting module is used to control the first cutting mechanism to cut the first qualified potato. The defective potato processing module is used to control the defective cutting mechanism and cut the second defective potato to obtain the second qualified potato and residual potato waste, and at the same time control the residual potato removal mechanism to remove the residual potato waste from the defective potato conveyor belt. The second qualified potato cutting module is used to control the second cutting mechanism to cut the second qualified potato.

9. An electronic device, characterized in that, include: At least one processor; Memory; At least one computer program, wherein the at least one computer program is stored in the memory and configured to be executed by the at least one processor, the at least one computer program being configured to: perform a smart quality control method for potato processing based on multi-source data fusion as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer program is stored and can be loaded by a processor and executed as described in any one of claims 1 to 7, which is a method for intelligent quality control of potato processing based on multi-source data fusion.

Images