Load orchestration method based on load balancing

Abstract

This invention provides a load balancing-based load orchestration method, relating to the field of data processing technology. The invention acquires initial business data corresponding to the target to be processed at each business node, dynamically processes the initial business data based on abnormal characteristics to obtain target task data, generates load estimates for each target task based on the identification task volume in the target task data, and matches the load estimates with the real-time occupancy rate of each candidate server to determine the target server for each target task data. This invention achieves load estimation and orchestration of identification tasks based on the actual content of the business data, improving server resource utilization and business processing efficiency.

Description

Technical Field

[0001] This invention relates to data processing technology, and more particularly to a load orchestration method based on load balancing. Background Technology

[0002] In the Industrial Internet of Things (IIoT) and smart manufacturing systems, the front end continuously collects a large amount of business data, such as sensor signals and industrial vision data, and sends the data to the back end server cluster for anomaly object identification and anomaly status determination. Since a single server cannot handle all the business data, multiple candidate servers are usually deployed, and a load balancing scheme is used to distribute various types of business data to ensure the efficient flow of the overall anomaly identification business.

[0003] Currently, load balancing solutions in industrial and IoT systems typically distribute business data based on real-time metrics such as server CPU utilization and connection count using algorithms like round-robin, or simply by allocating data in a tiered manner based on the size of transmitted data packets. Existing solutions lack in-depth analysis of the actual content of the business data and cannot differentiate the processing complexity of different data types. In practical applications, a small volume of business data containing multiple anomalies may require more server processing resources than a large volume of business data without anomalies. Existing load balancing solutions are prone to load prediction errors, leading to some servers being overloaded while others remain idle, resulting in wasted resources and reduced efficiency in anomaly detection.

[0004] Therefore, how to estimate and orchestrate the load of recognition tasks based on the actual content of business data, and improve server resource utilization and business processing efficiency, has become a key issue that urgently needs to be addressed. Summary of the Invention

[0005] This invention provides a load balancing-based load orchestration method that can estimate and orchestrate the load of identification tasks based on the actual content of business data, thereby improving server resource utilization and business processing efficiency.

[0006] A first aspect of the present invention provides a load orchestration method based on load balancing, comprising: Obtain the initial business data corresponding to the target to be processed at each business node, and dynamically process the initial business data according to the abnormal characteristics in the initial business data to obtain the target task data; Based on the identification task volume in the target task data, generate the load estimate of each target task data. The target server for each target task data is determined by matching the estimated load with the real-time occupancy rate of each candidate server, and the corresponding target task data is distributed to the target server for processing.

[0007] Optionally, in one possible implementation of the first aspect, the step of dynamically processing the initial business data based on the abnormal characteristics in the initial business data to obtain the target task data includes: Identify the target outline of the target to be processed in the initial business data, as well as the abnormal characteristics of the abnormal objects corresponding to the target to be processed; Based on the boundary position relationship between the aforementioned abnormal features and the target contour, the initial service data is dynamically processed to obtain relay service data. Based on the abnormal characteristics in the relay service data, the relay service data is customized and extracted to obtain the target task data.

[0008] Optionally, in one possible implementation of the first aspect, the step of dynamically processing the initial service data based on the boundary positional relationship between the abnormal features and the target contour to obtain relay service data includes: When it is determined that the abnormal features intersect with the corresponding target contour, the corresponding initial business data is used as the business data to be processed, and the remaining initial business data is used as relay business data. Obtain the contour edges that intersect with the target contour in the business data to be processed as the intersection edges, and take the position of the intersection edges in the target contour as the intersection position. Use the initial business data located at the intersection of the business data to be processed as the judgment business data; Based on the comparison results between the abnormal intersection points at the identified business data locations and the abnormal intersection points at the corresponding pending business data locations, the relay business data is obtained.

[0009] Optionally, in one possible implementation of the first aspect, obtaining relay service data based on the comparison result of the abnormal intersection point at the determined service data and the abnormal intersection point at the corresponding pending service data includes: When it is determined that the judgment business data is the business data to be processed, and the intersection direction of the judgment business data is opposite to the intersection direction of the business data to be processed, the first position of the intersection point of the intersection edge and the abnormal feature corresponding to the abnormal point in the business data to be processed, and the second position of the intersection point of the intersection edge and the abnormal feature corresponding to the abnormal point in the judgment business data are obtained. When the first position and the second position are the same, a comparison pass result is generated, and the judgment business data is concatenated with the corresponding business data to be processed to obtain the relay business data; If the first position is determined to be different from any second position, a comparison failure result is generated, and the corresponding pending business data is used as relay business data.

[0010] Optionally, in one possible implementation of the first aspect, the step of dynamically processing the initial service data based on the boundary positional relationship between the abnormal features and the target contour to obtain relay service data includes: When it is determined that the abnormal features intersect with the corresponding target contour, a target model corresponding to the target to be processed is constructed, and the corresponding abnormal features are used as features to be extracted. The features to be extracted are updated to the target model. When it is determined that the surface of the target model has interconnected features to be extracted, the corresponding interconnected features to be extracted are taken as cross features. The initial service data at the location of the cross-feature is concatenated to obtain relay service data, and the remaining initial service data is used as relay service data.

[0011] Optionally, in one possible implementation of the first aspect, the step of customizing the interception of the relay service data based on abnormal characteristics in the relay service data to obtain the target task data includes: Obtain the logical partitions in each relay service data, wherein the logical partitions include a core processing area and a secondary data area; Identify abnormal features in relay service data, obtain the intersection of the abnormal features with the core processing area to obtain the first feature, and obtain the second feature based on the intersection of the abnormal features with the secondary data area; The relay service data is processed by coordinate conversion, and a reserved processing area corresponding to the second feature is constructed based on the coordinate extreme values ​​of the second feature; The reserved processing area and core processing area in the relay service data are extracted and processed to obtain the target task data.

[0012] Optionally, in one possible implementation of the first aspect, generating the load estimate of each target task data based on the identification task volume in the target task data includes: Obtain the first occupancy of the core processing area and the second occupancy of the secondary data area in the target task data; The number of data units corresponding to the first feature is counted to obtain the first number, and the number of data units corresponding to the second feature is counted to obtain the second number; The identification task quantity is obtained based on the first occupancy, the second occupancy, the first quantity, and the second quantity; The workload of the identification task is calculated to obtain the estimated load of each target task data.

[0013] Optionally, in one possible implementation of the first aspect, calculating the identification task volume to obtain the estimated load of each target task data includes: Retrieve the first baseline occupancy and the first baseline quantity corresponding to the core processing area, and the second baseline occupancy and the second baseline quantity corresponding to the secondary data area; The first occupancy coefficient is obtained based on the ratio of the first occupancy to the first baseline occupancy, and the second occupancy coefficient is obtained based on the ratio of the second occupancy to the second baseline occupancy. The first quantity coefficient is obtained based on the ratio of the first quantity to the first benchmark quantity, and the second quantity coefficient is obtained based on the ratio of the second quantity to the second benchmark quantity. The load estimate of each target task data is obtained based on the sum of the first occupancy coefficient, the second occupancy coefficient, the first quantity coefficient, and the second quantity coefficient.

[0014] Optionally, in one possible implementation of the first aspect, it also includes: When the core processing area in the target task data is determined to be a high-load business feature area, the number of high-load features of the corresponding high-load business entity in the high-load business feature area is retrieved. Based on the number of high-load features and the preset additional lookup table, the additional load value corresponding to the high-load service feature area is obtained. The preset additional lookup table has a one-to-one correspondence between the preset feature number range and the preset additional value.

[0015] Optionally, in one possible implementation of the first aspect, the step of matching the load estimate with the real-time occupancy rate of each candidate server to determine the target server for each target task data includes: The candidate servers are sorted in ascending order based on their real-time occupancy rate to obtain a matching sequence; The target task data is sorted in descending order based on the load estimate to obtain the load sequence. The candidate servers in the matching sequence are sequentially configured to the target task data in the load sequence to obtain the target server for each target task data.

[0016] The beneficial effects of this invention are as follows: 1. This invention acquires initial business data corresponding to the target to be processed at each business node, dynamically processes the initial business data based on abnormal characteristics to obtain target task data, generates load estimates for each target task based on the identified task volume in the target task data, and matches the load estimates with the real-time occupancy rate of each candidate server to determine the target server for each target task data. This invention realizes the identification of abnormal objects in business data and the dynamic orchestration of servers, avoiding server resource mismatch caused by load orchestration based solely on the overall data size of the business data.

[0017] 2. This invention identifies the target contour of the target to be processed and the abnormal features of abnormal objects in the initial business data, and dynamically processes the initial business data based on the boundary position relationship between the abnormal features and the target contour. This invention provides two processing methods. The first method involves determining the business data to be processed, obtaining the intersection edge and intersection direction, and using the initial business data located at the intersection direction as the judgment business data. The first and second positions of the abnormal intersection points in the business data to be processed and the judgment business data are compared. If the comparison passes, the business data to be processed and the judgment business data are concatenated to obtain relay business data. The second method involves constructing a target model corresponding to the target to be processed, updating the abnormal features on the target model, and using the interconnected features to be extracted as cross features. The initial business data located at the cross features are then concatenated to obtain relay business data. This invention merges the abnormal features collected from different business nodes for abnormal objects, avoiding multiple identification and processing of the same abnormal object, thus preventing wasted computing power.

[0018] 3. This invention obtains the core processing area and secondary data area from relay service data. It obtains a first feature by finding the intersection of anomaly features and the core processing area, and a second feature by identifying the intersection of anomaly features and the secondary data area. Based on the second feature, a reserved processing area is constructed. The reserved processing area and the core processing area are then processed to obtain the target task data. The invention also obtains the first occupancy of the core processing area and the second occupancy of the secondary data area from the target task data. The number of data units corresponding to the first and second features is counted to obtain the first and second quantities. The corresponding baseline occupancy and quantity are retrieved to calculate the first occupancy coefficient, the second occupancy coefficient, the first quantity coefficient, and the second quantity coefficient. These coefficients are then summed to obtain the estimated load of the target task data. This invention achieves the estimation of the target task data identification requirements and matches the corresponding estimated load with the real-time server occupancy rate. Attached Figure Description

[0019] Figure 1 A flowchart of a load orchestration method based on load balancing provided by the present invention; Figure 2 This is a schematic diagram of the target to be processed and the abnormal features in this invention; Figure 3 This is a schematic diagram of the intersection edges in this invention; Figure 4 This is a schematic diagram of abnormal intersections in this invention; Figure 5 This is a schematic diagram of the hardware structure of an electronic device provided by the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein.

[0022] It should be understood that in the various embodiments of the present invention, the sequence number of each process does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0023] It should be understood that in this invention, "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product, or device.

[0024] It should be understood that in this invention, "multiple" refers to two or more. "And / or" 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 alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "Contains A, B, and C", "Contains A, B, and C" means that all three A, B, and C are contained; "Contains A, B, or C" means that one of A, B, and C is contained; "Contains A, B, and / or C" means that any one, two, or three of A, B, and C are contained.

[0025] It should be understood that in this invention, "B corresponding to A", "B corresponding to A", "A and B correspond", or "B and A correspond" means that B is associated with A, and B can be determined based on A. Determining B based on A does not mean determining B solely based on A; B can also be determined based on A and / or other information. Matching A and B is defined as a similarity between A and B that is greater than or equal to a preset threshold.

[0026] Depending on the context, "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection."

[0027] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0028] This invention provides a load orchestration method based on load balancing, such as... Figure 1 As shown, it includes: S1, obtain the initial business data corresponding to the target to be processed at each business node, and dynamically process the initial business data according to the abnormal characteristics in the initial business data to obtain the target task data.

[0029] It should be noted that this invention takes industrial visual quality inspection as an example of a business scenario in the Industrial Internet of Things (IIoT) for detailed explanation.

[0030] Among them, the target to be processed refers to the workpiece currently entering the quality inspection process; the business node refers to the workstation pre-arranged around the target to be processed, i.e., the workpiece, for image acquisition; the initial business data refers to the image acquired from the target to be processed at the business node; and the abnormal features refer to the traces of abnormal objects such as scratches and cracks on the surface of the target to be processed in the initial business data.

[0031] Understandably, in a factory, workpieces typically require image acquisition from multiple angles during quality inspection, with each angle corresponding to a business node. The complete trace of an abnormal object may span different surfaces of the workpiece. The abnormal features acquired by adjacent business nodes only represent a portion of the abnormal object. If images acquired by adjacent business nodes are individually transmitted to the server cluster for identification, it becomes difficult to accurately determine the true length and extended shape of the abnormal object, and the same abnormal object may be subjected to repeated identification and analysis. Furthermore, a single image usually includes areas of the workpiece that require quality inspection as well as areas that do not. Identifying areas without quality inspection requirements would waste server computing power.

[0032] Therefore, in this step, when the target to be processed is sent to the quality inspection station, the cameras at each business node around the target to be processed are retrieved to take pictures of the target. The pictures are used as the initial business data. The initial business data is then dynamically processed by stitching together the images of the same abnormal object from multiple perspectives and removing the parts of the images that do not need to be identified, so as to obtain the target task data.

[0033] In some embodiments, step S1 (dynamically processing the initial business data based on the abnormal characteristics in the initial business data to obtain the target task data) includes S11-S13: S11, identify the target outline of the target to be processed in the initial business data, as well as the abnormal characteristics of the abnormal objects corresponding to the target to be processed.

[0034] The target contour refers to the boundary contour of the target to be processed in the initial business data, that is, the boundary contour of the workpiece in the image.

[0035] Understandably, the process involves identifying the target outline of the target to be processed in the initial business data, and then further identifying abnormal features such as scratches and cracks in the surface area of ​​the target to be processed enclosed by the target outline.

[0036] S12, based on the boundary position relationship between the abnormal features and the target contour, the initial service data is dynamically processed to obtain relay service data.

[0037] It should be noted that if the abnormal feature and the target contour do not intersect, it indicates that the abnormal object corresponding to the abnormal feature is fully presented in the single initial business data and does not need to be merged with other initial business data, i.e., image stitching is not required. If the abnormal feature and the target contour intersect, it indicates that the abnormal object corresponding to the abnormal feature may extend to adjacent surfaces, and it is necessary to determine whether image stitching is required based on the initial business data of adjacent business nodes. Therefore, this step determines whether it is necessary to stitch adjacent initial business data based on the boundary position relationship between the abnormal feature and the target contour, thereby obtaining relay business data.

[0038] Among them, boundary positional relationship refers to the positional relationship between abnormal features and target contours; relay service data refers to the image obtained after dynamic processing of initial service data.

[0039] In some embodiments, for workpieces with regular shapes, such as cubes, step S12 (dynamically processing the initial service data based on the boundary position relationship between the abnormal features and the target contour to obtain relay service data) includes A1-A4: A1. When it is determined that the abnormal features intersect with the corresponding target contour, the corresponding initial business data is used as the business data to be processed, and the remaining initial business data is used as relay business data.

[0040] It is understandable that if the abnormal features in the initial business data intersect with the target contour, the corresponding initial business data will be used as the business data to be processed; if the abnormal features in the initial business data do not intersect with the target contour, the corresponding initial business data will be used as the relay business data.

[0041] A2, obtain the contour edges that intersect at the target contour in the business data to be processed as the intersection edges, and take the position of the intersection edges at the target contour as the intersection position.

[0042] It should be noted that for targets with regular shapes, the initial business data collected by the business nodes typically corresponds to each face of the target. The target outline in each initial business data sheet is the boundary outline of each face. For example, ... Figure 2 As shown, the target object is a cube, and its outline is a square. If the abnormal feature intersects with the target outline, the intersection occurs on one of the contour edges of the target outline. Therefore, this step extracts the contour edge where the abnormal feature intersects with the target outline, and then determines the adjacent initial business data based on this contour edge.

[0043] It is understandable that, such as Figure 3 As shown, the contour edge on the target contour in the data to be processed that intersects with the abnormal feature is taken as the intersection edge, and the location of the intersection edge in the target contour is taken as the intersection orientation. For example, if the target to be processed is a cube, and in the data to be processed acquired from the main view angle, the abnormal feature intersects with the contour edge on the right side of the target contour, then this contour edge on the right is the intersection edge, and the intersection orientation is to the right.

[0044] A3 uses the initial business data located at the intersection of the business data to be processed as the judgment business data.

[0045] It should be noted that in a regular-shaped target, if an abnormal feature on a certain face extends out of the target outline from the intersection direction, then the abnormal object corresponding to that abnormal feature may extend to another adjacent face corresponding to the intersection direction.

[0046] For example, in the business data to be processed collected from the main viewpoint, if the intersection is on the right, then the initial business data collected by the business node on the right is the judgment business data.

[0047] A4. Based on the comparison results of the abnormal intersection points at the judgment business data and the abnormal intersection points at the corresponding pending business data, the relay business data is obtained.

[0048] Among them, the abnormal intersection point refers to the intersection point between the abnormal feature and the intersection edge.

[0049] It should be noted that after determining the business data to be judged, it is also necessary to compare whether the abnormal features in the business data to be judged and the corresponding business data to be processed correspond to the same abnormal object. If the intersection points of the abnormalities in the business data to be judged and the corresponding business data to be processed can be aligned, it indicates that the abnormal features in the two images are connected and belong to the same abnormal object. After stitching, the complete trace of the abnormal object can be obtained. If the intersection points of the abnormalities in the business data to be judged and the corresponding business data to be processed cannot be aligned, it indicates that the abnormal features in the two images correspond to different abnormal objects.

[0050] In some embodiments, step A4 (obtaining relay service data based on the comparison result of abnormal intersection points at the determined service data and the corresponding abnormal intersection points at the pending service data) includes A41-A43: A41, when it is determined that the judgment business data is the business data to be processed, and the intersection direction of the judgment business data is opposite to the intersection direction of the business data to be processed, the first position of the intersection point of the intersection edge and the abnormal feature corresponding to the abnormal point in the business data to be processed, and the second position of the intersection point of the intersection edge and the abnormal feature corresponding to the abnormal point in the judgment business data are obtained.

[0051] Understandably, in a target with a regular shape, if the judgment data also contains abnormal features that intersect with the target contour, and the intersection direction of the judgment data is opposite to that of the judgment data, it indicates that the intersection edge of the two images is the same edge, and anomaly intersection point comparison can be performed. For example, if the intersection direction of the judgment data is to the right, and the intersection direction of the judgment data is to the left, it indicates that the intersection edge of the judgment data and the judgment data is the same edge of the target.

[0052] Furthermore, if the above conditions are met, such as Figure 4 As shown, on the intersection edge of the business data to be processed, the position of the abnormal intersection point of the abnormal feature and the intersection edge is taken as the first position, and on the intersection edge of the business data to be judged, the position of the abnormal intersection point of the abnormal feature and the intersection edge is taken as the second position.

[0053] A42, when it is determined that the first position and the second position are the same, a comparison pass result is generated, and the judged business data is concatenated with the corresponding business data to be processed to obtain relay business data.

[0054] It should be noted that in objects with regular shapes, adjacent faces share the same edge, and the intersection point of an anomaly object crossing two adjacent faces with this edge remains unchanged. Therefore, one endpoint of the intersection edge is selected as a reference point, and the pixel distance between the first position and the reference point should be equal to the pixel distance between the second position and the reference point.

[0055] It is understandable that if two pixels are equidistant, it indicates that the abnormal intersection points in the two images are at the same location. The abnormal features in the business data to be processed and the business data to be judged belong to the same abnormal object. A comparison pass result is generated, and the two images are stitched together along the intersection edge to obtain the relay business data.

[0056] For example, the upper endpoint of the intersection edge is selected as the reference point. The pixel distance between the first position in the business data to be processed and the reference point is 150 pixels. It is determined that the pixel distance between the second position in the business data and the reference point is also 150 pixels. A comparison pass result is generated, and the two images are stitched together to obtain the relay business data.

[0057] A43, if the first position is not the same as any second position, generate a comparison failure result and use the corresponding pending business data as relay business data.

[0058] Understandably, referring to step A42, if the pixel distance between the first position and the reference point on the intersection edge is not equal to the pixel distance between the second position and the reference point, a comparison failure result is generated, and the corresponding pending business data is used as relay business data.

[0059] In some embodiments, for workpieces with irregular shapes, step S12 (dynamically processing the initial service data based on the boundary position relationship between the abnormal features and the target contour to obtain relay service data) includes B1-B3: B1. When it is determined that the abnormal features intersect with the corresponding target contour, a target model corresponding to the target to be processed is constructed, and the corresponding abnormal features are used as features to be extracted.

[0060] Understandably, for targets with irregular shapes, a 3D model corresponding to the target (i.e., the target model) can be constructed first. If the abnormal features in the initial business data intersect with the target contour, the corresponding abnormal features are used as features to be extracted.

[0061] B2, update the features to be extracted to the target model, and when it is determined that the surface of the target model has interconnected features to be extracted, use the corresponding interconnected features to be extracted as cross features.

[0062] It is understandable that the features to be extracted from each initial business data are updated to the corresponding positions in the target model. If multiple features to be extracted are connected end to end, forming a continuous trace, it indicates that these features to be extracted belong to the same abnormal object, and the corresponding interconnected features to be extracted are taken as cross features.

[0063] B3. The initial service data of the cross-feature is spliced ​​together to obtain relay service data, and the remaining initial service data is used as relay service data.

[0064] It is understandable that the initial business data corresponding to the cross-features are stitched together, and the resulting image is used as relay business data. The remaining initial business data that do not have cross-features are used as relay business data separately.

[0065] S13, Based on the abnormal characteristics in the relay service data, the relay service data is customized and extracted to obtain the target task data.

[0066] It should be noted that the surface of the workpiece can be divided into areas with quality inspection requirements and areas without quality inspection requirements. If the relay business data is directly distributed to the server, the areas that do not need to be identified will increase unnecessary calculations for the server.

[0067] Therefore, this step extracts the relay service data, identifies the parts that need to be identified, and removes invalid data.

[0068] In some embodiments, step S13 (customized interception of relay service data based on abnormal characteristics in the relay service data to obtain target task data) includes S131-S134: S131, Obtain the logical partitions in each relay service data, wherein the logical partitions include the core processing area and the secondary data area.

[0069] Logical partitioning refers to the division of the target surface to be processed according to quality inspection requirements, including core processing area and secondary data area. Core processing area refers to the area of ​​the workpiece surface that has quality inspection requirements, such as mating surfaces and welding joints related to product functions. Secondary data area refers to the area of ​​the workpiece surface that does not have quality inspection requirements, such as patterns and labels that only serve a decorative purpose.

[0070] Understandably, the surface of the target data to be processed is pre-divided into two logical partitions: a core processing area with quality inspection requirements and a secondary data area without quality inspection requirements. Furthermore, the logical partition categories within the relay business data are identified.

[0071] S132, identify abnormal features in relay service data, obtain the intersection of the abnormal features and the core processing area to obtain the first feature, and obtain the second feature based on the intersection of the abnormal features and the secondary data area.

[0072] It should be noted that the secondary data area itself does not require quality inspection. However, if the abnormal features intersect with the secondary data area, then it is necessary to extract the area corresponding to the abnormal features in the secondary data area for subsequent identification.

[0073] It is understandable that, based on the core processing area and secondary data area in the relay service data identified in step S131, the intersection of the abnormal features and the core processing area is taken as the first feature, and the intersection of the abnormal features and the secondary data area is taken as the second feature.

[0074] S133, the relay service data is processed into coordinates, and a reserved processing area corresponding to the second feature is constructed based on the coordinate extreme values ​​of the second feature.

[0075] The reserved processing area refers to the smallest rectangular region that surrounds the second feature.

[0076] It is understandable that a Cartesian coordinate system is established for the relay service data, the minimum and maximum values ​​of the second feature on the horizontal axis are obtained, and the minimum and maximum values ​​of the second feature on the vertical axis are obtained. A minimum rectangular region enclosing the second feature is constructed using these four coordinate extreme values ​​as boundaries, and this minimum rectangular region is used as the reserved processing area.

[0077] S134, the reserved processing area and the core processing area in the relay service data are extracted and processed to obtain the target task data.

[0078] Understandably, the core processing area is the area that must be quality inspected, while the reserved processing area is the area that needs to be quality inspected extracted from the secondary data area. The core processing area and the reserved processing area in the relay business data are retained, and other secondary data areas in the relay business data are partially removed to obtain a new image that includes only the core processing area and the reserved processing area, which is the target task data.

[0079] S2, Based on the number of identified tasks in the target task data, generate the load estimate of each target task data.

[0080] The load estimate is a numerical value that represents the workload of image processing, calculated based on the amount of recognition tasks. The higher the load estimate, the more resources the server needs for image processing.

[0081] Understandably, this step obtains the occupancy of the core processing area and secondary data area in the target task data, as well as the number of data units of the first feature and the second feature, as the identification task volume of the target task data, and then generates the load estimate of each target task data based on the identification task volume.

[0082] It is worth mentioning that the amount of space occupied by the core processing area and the secondary data area in the target task data in this step is the area of ​​the core processing area and the secondary data area in the target task data.

[0083] In some embodiments, step S2 (generating the load estimate of each target task data based on the identified task volume in the target task data) includes S21-S24: S21, obtain the first occupancy of the core processing area and the second occupancy of the secondary data area in the target task data.

[0084] It is understandable that the area occupied by the core processing area in the target task data is taken as the first occupancy, and the area occupied by the secondary data area is taken as the second occupancy.

[0085] S22, count the number of data units corresponding to the first feature to obtain the first number, and count the number of data units corresponding to the second feature to obtain the second number.

[0086] It should be noted that the size and length of abnormal features in the target task data will also affect the workload of identification and processing.

[0087] The number of data units refers to the number of pixels in the image.

[0088] Therefore, the number of data units corresponding to the first feature is taken as the first quantity, and the number of data units corresponding to the second feature is taken as the second quantity.

[0089] S23, based on the first occupancy, the second occupancy, the first quantity, and the second quantity, the identification task quantity is obtained.

[0090] It is understandable that the identification task volume includes the first occupancy, the second occupancy, the first quantity, and the second quantity.

[0091] S24, calculate the amount of identification tasks to obtain the estimated load of each target task data.

[0092] In some embodiments, step S24 (calculating the identification task volume to obtain the load estimate of each target task data) includes S241-S244: S241, retrieve the first reference occupancy and the first reference quantity corresponding to the core processing area, and the second reference occupancy and the second reference quantity corresponding to the secondary data area.

[0093] The first baseline occupancy refers to a pre-set baseline occupancy used to calculate the ratio with the first occupancy of the core processing area; the second baseline occupancy refers to a pre-set baseline occupancy used to calculate the ratio with the second occupancy of the secondary data area.

[0094] In addition, the first benchmark quantity refers to a pre-set benchmark quantity used for ratio calculation with the first quantity; the second benchmark quantity refers to a pre-set benchmark quantity used for ratio calculation with the second quantity.

[0095] S242, based on the ratio of the first occupancy amount to the first baseline occupancy amount, a first occupancy amount coefficient is obtained, and based on the ratio of the second occupancy amount to the second baseline occupancy amount, a second occupancy amount coefficient is obtained.

[0096] Understandably, the ratio of the first occupancy to the first baseline occupancy is calculated to obtain the first occupancy coefficient; the ratio of the second occupancy to the second baseline occupancy is calculated to obtain the second occupancy coefficient.

[0097] S243, based on the ratio of the first quantity to the first benchmark quantity, a first quantity coefficient is obtained, and based on the ratio of the second quantity to the second benchmark quantity, a second quantity coefficient is obtained.

[0098] Understandably, the ratio of the first quantity to the first benchmark quantity is calculated to obtain the first quantity coefficient; the ratio of the second quantity to the second benchmark quantity is calculated to obtain the second quantity coefficient.

[0099] S244. Based on the sum of the first occupancy coefficient, the second occupancy coefficient, the first quantity coefficient, and the second quantity coefficient, the load estimate of each target task data is obtained.

[0100] Understandably, the sum of the first occupancy coefficient, the second occupancy coefficient, the first quantity coefficient, and the second quantity coefficient is calculated and used as the load estimate.

[0101] In some embodiments, S25-S26 are also included: S25, when it is determined that the core processing area in the target task data is a high-load business feature area, retrieve the number of high-load features of the high-load business entity corresponding to the high-load business feature area.

[0102] Among them, high-load business entities refer to fastening parts, such as bolts and screws, which match the high-load business feature area; the high-load business feature area refers to the area used to support the high-load business entity and to achieve tightening and fixation with the high-load business entity; the number of high-load features refers to the number of textures on the high-load business entity used for fastening, such as the number of threads.

[0103] It should be noted that the quality inspection difficulty of high-load service feature areas is higher than that of other core processing areas, and the recognition and processing volume of high-load service feature areas is related to the number of textures on the high-load service entities, i.e., the number of threads. The more threads there are, the greater the computational load during recognition. Therefore, in this embodiment, when the core processing area is determined to be a high-load service feature area, an additional load value is added to the original load estimate.

[0104] It is understandable that when the core processing area in the target task data is determined to be a high-load business feature area, the number of high-load features on the corresponding high-load business entity is retrieved.

[0105] S26. Based on the number of high-load features and the preset additional lookup table, the additional load value corresponding to the high-load service feature area is obtained. The preset additional lookup table has a one-to-one correspondence between the preset feature number range and the preset additional value.

[0106] Among them, the preset additional reference table refers to a pre-set reference table with a one-to-one correspondence between preset feature quantity ranges and preset additional values; the preset feature quantity range refers to a pre-divided range of high load feature quantities; the preset additional value refers to a pre-set additional value corresponding to each preset feature quantity range; and the additional load value refers to the extra load amount used to superimpose on the original load estimate.

[0107] Understandably, the number of high-load features is compared with a preset additional comparison table. Based on the preset feature number range into which the number of high-load features falls, the corresponding preset additional value is retrieved as the additional load value. The retrieved additional load value is then added to the corresponding load estimate to obtain the final load estimate of the corresponding target task data.

[0108] S3. Match the estimated load with the real-time occupancy rate of each candidate server to determine the target server for each target task data, and distribute the corresponding target task data to the target server for processing.

[0109] Among them, candidate servers refer to servers used for identifying and processing target task data; real-time utilization rate refers to the proportion of computing resources currently used by candidate servers to the total computing resources.

[0110] Understandably, server occupancy is dynamic and needs to be allocated based on current occupancy. This step distributes the target task data based on the real-time occupancy of each candidate server.

[0111] In some embodiments, step S3 (matching the load estimate with the real-time occupancy rate of each candidate server to determine the target server for each target task data) includes S31-S33: S31. Based on the real-time occupancy rate of each candidate server, the candidate servers are sorted in ascending order to obtain the matching sequence.

[0112] Understandably, the real-time occupancy rate of each candidate server is obtained, and the candidate servers are sorted in ascending order of real-time occupancy rate to obtain the matching sequence.

[0113] S32, sort the target task data in descending order according to the load estimate to obtain the load sequence.

[0114] It is understandable that the target task data is sorted in descending order of load estimates to obtain the load sequence.

[0115] S33, the candidate servers in the coordination sequence are sequentially configured to the target task data in the load sequence to obtain the target server for each target task data.

[0116] It is understandable that candidate servers in the coordination sequence are sequentially configured to target task data in the load sequence. For example, the first candidate server in the coordination sequence is configured to the first target task data in the load sequence, the second candidate server in the coordination sequence is configured to the second target task data in the load sequence, and so on, until all target task data are assigned to candidate servers, and the candidate server assigned to the target task data is used as the target server.

[0117] See Figure 5 This is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present invention. The electronic device 50 includes: a processor 51, a memory 52, and a computer program; wherein... The memory 52 is used to store the computer program, and the memory may also be flash memory. The computer program is, for example, an application program or functional module that implements the above method.

[0118] The processor 51 is configured to execute the computer program stored in the memory to implement the various steps performed by the device in the above method. For details, please refer to the relevant descriptions in the preceding method embodiments.

[0119] Alternatively, the memory 52 can be either standalone or integrated with the processor 51.

[0120] When the memory 52 is a device independent of the processor 51, the device may further include: Bus 53 is used to connect the memory 52 and the processor 51.

[0121] The present invention also provides a readable storage medium storing a computer program, which, when executed by a processor, is used to implement the methods provided in the various embodiments described above.

[0122] The readable storage medium can be a computer storage medium or a communication medium. A communication medium includes any medium that facilitates the transfer of computer programs from one location to another. A computer storage medium can be any available medium accessible to a general-purpose or special-purpose computer. For example, a readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application-Specific Integrated Circuit (ASIC). Alternatively, the ASIC can be located in a user equipment. Of course, the processor and the readable storage medium can also exist as discrete components in a communication device. The readable storage medium can be a read-only memory (ROM), random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0123] The present invention also provides a program product including executable instructions stored in a readable storage medium. At least one processor of the device can read the executable instructions from the readable storage medium, and the at least one processor executes the executable instructions to cause the device to implement the methods provided in the various embodiments described above.

[0124] In the embodiments of the above-described device, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly manifested as execution by a hardware processor, or execution by a combination of hardware and software modules within the processor.

[0125] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A load orchestration method based on load balancing, characterized in that, include: Obtain the initial business data corresponding to the target to be processed at each business node, and dynamically process the initial business data according to the abnormal characteristics in the initial business data to obtain the target task data; Based on the identification task volume in the target task data, generate the load estimate of each target task data. The target server for each target task data is determined by matching the estimated load with the real-time occupancy rate of each candidate server, and the corresponding target task data is distributed to the target server for processing. The step of dynamically processing the initial business data based on the abnormal characteristics in the initial business data to obtain the target task data includes: Identify the target outline of the target to be processed in the initial business data, as well as the abnormal characteristics of the abnormal objects corresponding to the target to be processed; Based on the boundary position relationship between the aforementioned abnormal features and the target contour, the initial service data is dynamically processed to obtain relay service data. Based on the abnormal characteristics in the relay service data, the relay service data is customized and extracted to obtain the target task data; The step of generating a load estimate for each target task based on the identification task volume in the target task data includes: Obtain the first occupancy of the core processing area and the second occupancy of the secondary data area in the target task data; The number of data units corresponding to the first feature is counted to obtain the first number, and the number of data units corresponding to the second feature is counted to obtain the second number; The identification task quantity is obtained based on the first occupancy, the second occupancy, the first quantity, and the second quantity; The workload of the identification task is calculated to obtain the estimated load of each target task data; The calculation of the identification task volume to obtain the load estimate of each target task data includes: Retrieve the first baseline occupancy and the first baseline quantity corresponding to the core processing area, and the second baseline occupancy and the second baseline quantity corresponding to the secondary data area; The first occupancy coefficient is obtained based on the ratio of the first occupancy to the first baseline occupancy, and the second occupancy coefficient is obtained based on the ratio of the second occupancy to the second baseline occupancy. The first quantity coefficient is obtained based on the ratio of the first quantity to the first benchmark quantity, and the second quantity coefficient is obtained based on the ratio of the second quantity to the second benchmark quantity. The load estimate of each target task data is obtained based on the sum of the first occupancy coefficient, the second occupancy coefficient, the first quantity coefficient, and the second quantity coefficient.

2. The method according to claim 1, characterized in that, The initial service data is dynamically processed based on the boundary position relationship between the abnormal features and the target contour to obtain relay service data, including: When it is determined that the abnormal features intersect with the corresponding target contour, the corresponding initial business data is used as the business data to be processed, and the remaining initial business data is used as relay business data. Obtain the contour edges that intersect with the target contour in the business data to be processed as the intersection edges, and take the position of the intersection edges in the target contour as the intersection position. Use the initial business data located at the intersection of the business data to be processed as the judgment business data; Based on the comparison results between the abnormal intersection points at the identified business data locations and the abnormal intersection points at the corresponding pending business data locations, the relay business data is obtained.

3. The method according to claim 2, characterized in that, The relay service data is obtained based on the comparison results between the abnormal intersection points at the identified service data and the corresponding abnormal intersection points at the pending service data, including: When it is determined that the judgment business data is the business data to be processed, and the intersection direction of the judgment business data is opposite to the intersection direction of the business data to be processed, the first position of the intersection point of the intersection edge and the abnormal feature corresponding to the abnormal point in the business data to be processed, and the second position of the intersection point of the intersection edge and the abnormal feature corresponding to the abnormal point in the judgment business data are obtained. When the first position and the second position are the same, a comparison pass result is generated, and the judgment business data is concatenated with the corresponding business data to be processed to obtain the relay business data; If the first position is determined to be different from any second position, a comparison failure result is generated, and the corresponding pending business data is used as relay business data.

4. The method according to claim 1, characterized in that, The initial service data is dynamically processed based on the boundary position relationship between the abnormal features and the target contour to obtain relay service data, including: When it is determined that the abnormal features intersect with the corresponding target contour, a target model corresponding to the target to be processed is constructed, and the corresponding abnormal features are used as features to be extracted. The features to be extracted are updated to the target model. When it is determined that the surface of the target model has interconnected features to be extracted, the corresponding interconnected features to be extracted are taken as cross features. The initial service data at the location of the cross-feature is concatenated to obtain relay service data, and the remaining initial service data is used as relay service data.

5. The method according to claim 1, characterized in that, The calculation of the identification task volume to obtain the load estimate of each target task data includes: Retrieve the first baseline occupancy and the first baseline quantity corresponding to the core processing area, and the second baseline occupancy and the second baseline quantity corresponding to the secondary data area; The first occupancy coefficient is obtained based on the ratio of the first occupancy to the first baseline occupancy, and the second occupancy coefficient is obtained based on the ratio of the second occupancy to the second baseline occupancy. The first quantity coefficient is obtained based on the ratio of the first quantity to the first benchmark quantity, and the second quantity coefficient is obtained based on the ratio of the second quantity to the second benchmark quantity. The load estimate of each target task data is obtained based on the sum of the first occupancy coefficient, the second occupancy coefficient, the first quantity coefficient, and the second quantity coefficient.

6. The method according to claim 5, characterized in that, Also includes: When the core processing area in the target task data is determined to be a high-load business feature area, the number of high-load features of the corresponding high-load business entity in the high-load business feature area is retrieved. Based on the number of high-load features and the preset additional lookup table, the additional load value corresponding to the high-load service feature area is obtained. The preset additional lookup table has a one-to-one correspondence between the preset feature number range and the preset additional value.

7. The method according to claim 5, characterized in that, The step of matching the estimated load with the real-time occupancy rate of each candidate server to determine the target server for each target task data includes: The candidate servers are sorted in ascending order based on their real-time occupancy rate to obtain a matching sequence; The target task data is sorted in descending order based on the load estimate to obtain the load sequence. The candidate servers in the matching sequence are sequentially configured to the target task data in the load sequence to obtain the target server for each target task data.

Images