Industrial Internet of Things Edge Computing Data Processing System Based on Multimodal Large Model

Abstract

The present invention provides an industrial Internet of Things edge computing data processing system based on a multi-modal large model, belonging to the technical field of data processing. Specifically, it includes: a node screening module responsible for determining available edge nodes of edge nodes, a group division module responsible for obtaining effective node groups in the node groups, a time period evaluation module responsible for determining the collaborative processing time periods of the effective node groups, and a data processing module responsible for determining a strategy for collaborative processing of large model data by using the collaborative processing types in different time periods to select preferred groups and edge nodes, thereby improving the reliability of data processing.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and particularly relates to an industrial Internet of Things edge computing data processing system based on a multimodal large model. Background Art

[0002] Multimodal large models have great advantages in aspects such as fault diagnosis and data analysis of industrial equipment. Therefore, in order to combine industrial Internet of Things edge computing nodes to realize the analysis and processing of large model data, in CN202411059341.6 "A Method and System for Optimizing Edge Computing Resource Scheduling Based on a Large Model", an evaluation function is introduced to obtain a scheduling index. When the index reaches a threshold, an instruction is issued and the scheduler is activated to bind the target POD to the target node, solving the technical problems of unreasonable resource allocation and poor scheduling effect in the prior art, and achieving the technical effect of ensuring the stable operation of the system and high-performance services.

[0003] However, in the process of data processing using edge nodes, the load data of the Internet of Things gateways between different edge nodes is ignored. When the load rate of the Internet of Things gateways between edge nodes is relatively high, using multiple edge nodes for collaborative computing analysis and processing will not only lead to slow data processing efficiency, but also cause the load rate of the Internet of Things gateways to be too high, thus affecting the transmission reliability of the Internet of Things transmission system to a certain extent.

[0004] In view of the above technical problems, specifically, the present application provides an industrial Internet of Things edge computing data processing system based on a multimodal large model. Summary of the Invention

[0005] To achieve the object of the present invention, the present invention adopts the following technical solutions:

[0006] Specifically, the present application provides an industrial Internet of Things edge computing data processing system based on a multimodal large model, which specifically includes:

[0007] A node screening module, a group division module, a time period evaluation module, and a data processing module;

[0008] The node screening module is responsible for determining the available edge nodes of the edge nodes based on the historical processing data of the multimodal large model of the edge nodes under different data processing types and the processing accuracy under different data processing types;

[0009] The group division module is responsible for constructing multiple node groups using the available edge nodes, and obtaining the effective node groups in the node groups based on the recognition accuracy of the historical processing data of the edge nodes of different node groups;

[0010] The above-mentioned time period evaluation module is responsible for determining the collaborative processing time period of the effective node group by using the historical load conditions of the Internet of Things gateways between the available edge nodes in the effective node group and the edge nodes at different time periods.

[0011] The data processing module is responsible for determining the preferred group in the effective node group and the collaborative processing types at different time periods based on the changes in the collaborative processing time periods of the effective node group on different dates, and using the collaborative processing types at different time periods to determine the strategy for the preferred group and the edge nodes to perform collaborative processing of large model data.

[0012] The beneficial effects of the present invention are as follows:

[0013] Based on the recognition accuracy rates of the historical processing data of different node groups at the edge node, the effective node group in the node group is obtained, realizing the screening of low recognition accuracy rates, avoiding the influence on the accuracy of the data processing results of large model data due to low recognition accuracy rates, and at the same time, through the screening of the effective node group, it also lays a foundation for further screening the preferred group in combination with the load conditions of the physical network gateway, improving the efficiency of the screening process of the preferred group.

[0014] Using the collaborative processing types at different time periods to determine the strategy for the preferred group and the edge nodes to perform collaborative processing of large model data, thereby realizing the determination of the collaborative processing strategy of large model data from the historical load conditions of the Internet of Things gateways in different time periods, avoiding the influence on the load of the Internet of Things gateway caused by performing collaborative processing during the time period when the usage load of the Internet of Things gateway is too large, and at the same time, improving the reliability of the data processing results of large model data.

[0015] A further technical solution is that the data processing types include voice, text, file, image, and video.

[0016] A further technical solution is that the historical processing data includes the processed data volumes of different data processing types.

[0017] A further technical solution is that the method for determining the available edge nodes of the edge node is as follows:

[0018] Based on the historical processing data of the multi-modal large model of the edge node under different data processing types, determine the processed data volumes of the edge node under different data processing types, and use the proportion of the processed data volumes under different data processing types in the processed data volume of the edge node to determine the weight coefficients of different data processing types;

[0019] Determine the sum of the weight coefficients of the processing accuracies of other edge nodes based on the processing accuracies of other edge nodes under different data processing types and in combination with the weight coefficients of the edge nodes under different data processing types, and use it as the corrected accuracy;

[0020] Based on the corrected accuracy, determine whether the other edge nodes are available edge nodes.

[0021] A further technical solution is that when the corrected accuracy of the other edge nodes is greater than a preset accuracy threshold, it is determined that the other edge nodes are available edge nodes.

[0022] A further technical solution is that the method for determining the preferred group in the effective node group is as follows:

[0023] Based on the changes in the collaborative processing periods of the effective node group on different dates, determine the dates when different periods are recognized as collaborative processing periods, and use them as collaborative matching dates;

[0024] According to the proportion of the number of collaborative matching dates of different periods within a preset time period, determine the collaborative matching coefficients of different periods;

[0025] Based on the collaborative matching coefficients of different periods, determine the number of periods with collaborative matching coefficients greater than the preset matching coefficient, and use the number of periods with collaborative matching coefficients greater than the preset matching coefficient to determine the preferred group in the effective node group.

[0026] A further technical solution is that the preferred group is the effective node group with the largest number of periods with collaborative matching coefficients greater than the preset matching coefficient.

[0027] A further technical solution is that the collaborative processing types of the periods include a first type of collaborative type, a second type of collaborative type, and a third type of collaborative type.

[0028] A further technical solution is to use the collaborative processing types of different periods to determine the strategy for the preferred group and edge nodes to perform collaborative processing of large model data, specifically including:

[0029] When the collaborative processing type of the period is the first type of collaborative type, use the preferred group and edge nodes to jointly perform collaborative processing of large model data;

[0030] When the collaborative processing type in the time period is a type-II collaborative type, it is also necessary to determine whether the average load rate of the Internet of Things gateways between the available edge nodes of the preferred group in the most recent unit time period and the edge node is less than a preset load rate threshold. If so, the preferred group and the edge node are used to jointly perform collaborative processing of large model data. If not, the edge node is used to process the large model data.

[0031] When the collaborative processing type in the time period is a type-III collaborative type, the edge node is used to process the large model data.

[0032] Other features and advantages will be described in the subsequent specification. The objectives and other advantages of the present invention are achieved and obtained by the structures specifically pointed out in the specification and the drawings.

[0033] To make the above objectives, features, and advantages of the present invention more obvious and understandable, the following specific preferred embodiments are given in conjunction with the accompanying drawings and are described in detail as follows. Description of the Drawings

[0034] By referring to the drawings and describing its exemplary embodiments in detail, the above and other features and advantages of the present invention will become more obvious.

[0035] Figure 1 is a flowchart of an industrial Internet of Things edge computing data processing system based on a multi-modal large model;

[0036] Figure 2 is a flowchart of a method for determining available edge nodes of an edge node;

[0037] Figure 3 is a flowchart of a method for determining an effective node group in a node group;

[0038] Figure 4 is a flowchart of a method for determining a collaborative processing time period of an effective node group;

[0039] Figure 5 is a flowchart of a method for determining a preferred group in an effective node group. Detailed Embodiments

[0040] To enable those skilled in the art of this technology to better understand the technical solutions in this specification, the following will clearly and completely describe the technical solutions in the embodiments of this specification in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of this specification, rather than all the embodiments. Based on the embodiments of this specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of this specification.

[0041] In this application, a differentiated collaborative processing strategy is generated according to the idle situation of the Internet of Things gateway devices between edge nodes and other edge nodes, thereby reducing the load impact on the Internet of Things gateway devices and at the same time improving the reliability of data processing of large model data by edge nodes.

[0042] Available edge nodes are other edge nodes whose average processing accuracy under the data processing type of the edge node is greater than 0.7.

[0043] Using the average processing accuracy of available edge nodes in the node group under different data processing types, determine the recognition accuracy under different data processing types. When the recognition accuracy under different data processing types is greater than 0.8, then determine the node group as an effective node group.

[0044] The collaborative processing period is the period when the historical load rate of the Internet of Things gateway at different times is less than 0.5.

[0045] Based on the change situation of the collaborative processing period of the effective node group on different dates, determine the dates when different periods are recognized as the collaborative processing period, and use them as the collaborative matching dates. According to the proportion of the number of collaborative matching dates of different periods in the preset time period, determine the collaborative matching coefficient of different periods, and use the effective node group with the largest number of periods whose collaborative matching coefficient is greater than the preset matching coefficient as the preferred group.

[0046] The collaborative processing types of the period include type I collaborative type, type II collaborative type, and type III collaborative type.

[0047] Embodiment 1

[0048] As Figure 1 shown, this application provides an industrial Internet of Things edge computing data processing system based on a multimodal large model, specifically including:

[0049] A node screening module, a group division module, a period evaluation module, and a data processing module;

[0050] The node screening module is responsible for determining the available edge nodes of the edge node based on the historical processing data of the multimodal large model of the edge node under different data processing types and the processing accuracy under different data processing types;

[0051] The group division module is responsible for constructing multiple node groups using the available edge nodes, and obtaining the effective node groups in the node group based on the recognition accuracy of the historical processing data of the edge node of different node groups;

[0052] The time period evaluation module is responsible for determining the collaborative processing time period of the effective node group by using the historical load conditions of the IoT gateway between the available edge nodes in the effective node group and the edge nodes at different time periods.

[0053] The data processing module is responsible for determining the preferred group in the effective node group and the collaborative processing types at different time periods based on the changes in the collaborative processing time periods of the effective node group on different dates, and determining the strategy for the preferred group and the edge nodes to perform collaborative processing of large model data by using the collaborative processing types at different time periods.

[0054] Further, the data processing types include voice, text, file, image, and video.

[0055] Specifically, the historical processing data includes the processed data volumes of different data processing types.

[0056] Specifically, as Figure 2 shown, the method for determining the available edge nodes of the edge node is:

[0057] Based on the historical processing data of the multi-modal large model of the edge node under different data processing types, determine the processed data volumes of the edge node under different data processing types, and use the proportion of the processed data volumes under different data processing types in the processed data volume of the edge node to determine the weight coefficients of different data processing types;

[0058] According to the processing accuracies of other edge nodes under different data processing types, and combining the weight coefficients of the edge node under different data processing types, determine the sum of the weight coefficients of the processing accuracies of other edge nodes, and use it as the corrected accuracy;

[0059] Based on the corrected accuracy, determine whether the other edge nodes are available edge nodes.

[0060] Further, when the corrected accuracy of the other edge nodes is greater than the preset accuracy threshold, determine that the other edge nodes are available edge nodes.

[0061] In another possible embodiment, the method for determining the available edge nodes of the edge node is:

[0062] Based on the historical processing data of the multi-modal large model of the edge node under different data processing types, determine the processed data volumes of the edge node under different data processing types, and use the data processing types with processed data volumes greater than the preset data volume threshold for processing data volumes, and use it as the matching data type;

[0063] Determine the average value of the processing accuracy rates of the other edge nodes under different matching data processing types as the average accuracy rate, where the average value is calculated based on the processing accuracy rates of the other edge nodes under different matching data processing types.

[0064] Based on the average accuracy rate, determine whether the other edge nodes are available edge nodes.

[0065] Further, when the average accuracy rate of the other edge nodes is greater than a preset accuracy rate threshold, determine that the other edge nodes are available edge nodes.

[0066] It should be noted that constructing multiple node groups using the available edge nodes specifically includes:

[0067] Freely combine different available edge nodes to obtain multiple node groups.

[0068] In one possible embodiment, as Figure 3 shown, the method for determining the valid node groups in the node groups is as follows:

[0069] Based on the composition data of the available edge nodes in the node group, determine the matching edge nodes in the node group.

[0070] Based on the average value of the processing accuracy rates of different matching edge nodes under different data processing types and different data volume intervals, determine the group processing accuracy rates of the node group under different data processing types and different data volume intervals.

[0071] Based on the proportion of the historical processing times of the edge nodes under different data processing types and different data volume intervals, determine the weight coefficients for different data volume intervals, and in combination with the group processing accuracy rates of the node group under different data processing types and different data volume intervals, determine the recognition accuracy rate of the node group, and use the recognition accuracy rate to determine whether the node group is a valid node group.

[0072] Further, the valid node groups are the node groups whose recognition accuracy rates meet the requirements.

[0073] Optionally, the method for determining the valid node groups in the node group is as follows:

[0074] Based on the composition data of the available edge nodes in the node group, determine the matching edge nodes in the node group.

[0075] Determine the group processing accuracy of the node group under different data processing types and different data volume ranges based on the average of the processing accuracies of different matching edge nodes under different data processing types and different data volume ranges;

[0076] Determine the recognition accuracy of the node group based on the average of the group processing accuracies of the node group under different data processing types and different data volume ranges, and use the recognition accuracy to determine whether the node group is a valid node group.

[0077] Optionally, the method for determining the valid node group in the node group is as follows:

[0078] Determine the matching edge nodes in the node group based on the constituent data of the available edge nodes in the node group;

[0079] Determine the group processing accuracy of the node group under different data processing types and different data volume ranges based on the average of the processing accuracies of different matching edge nodes under different data processing types and different data volume ranges. When the group processing accuracies of the node group under different data processing types and different data volume ranges all meet the requirements, then determine that the node group is a valid node group;

[0080] When there is a data volume range in the node group where the group recognition accuracy does not meet the requirements:

[0081] Obtain the number of data volume ranges in the node group where the group recognition accuracy does not meet the requirements under different data processing types. When the number of data processing types for which the number of data volume ranges in the node group where the group recognition accuracy does not meet the requirements does not meet the requirements, then determine that the node does not belong to the valid node group;

[0082] When there is no data processing type in the node group for which the number of data volume ranges where the group recognition accuracy does not meet the requirements does not meet the requirements:

[0083] Based on the number of data volume ranges where the group recognition accuracy does not meet the requirements under different data processing types, calculate the total number of data volume ranges where the group recognition accuracy does not meet the requirements. When the total number of data volume ranges where the group recognition accuracy does not meet the requirements is greater than the preset interval number, then determine that the node does not belong to the valid node group;

[0084] When the total number of data volume ranges where the group recognition accuracy does not meet the requirements is not greater than the preset interval number:

[0085] Determine the recognition reliability coefficient for different data processing types based on the number of data volume intervals where the group recognition accuracy under different data processing types does not meet the requirements and the group recognition accuracy under different data volume intervals. When there is a data processing type with a recognition reliability coefficient that does not meet the requirements, determine that the node does not belong to the effective node group;

[0086] When there is no data processing type with a recognition reliability coefficient that does not meet the requirements:

[0087] Determine the recognition accuracy of the node group based on the average value of the recognition reliability coefficients of the node group under different data processing types, and use the recognition accuracy to determine whether the node group is an effective node group.

[0088] Specifically, as Figure 4 shown, the method for determining the collaborative processing period of the effective node group is:

[0089] Take the Internet of Things gateway between the available edge nodes and the edge nodes as the node connection gateway;

[0090] Determine the average gateway load rate at different times based on the average value of the load rates of the node connection gateways of different available edge nodes at different times in the period;

[0091] Based on the average gateway load rate, determine the high-load moments in the period, and use the proportion of the number of high-load moments in the period to determine whether the period is the collaborative processing period of the effective node group.

[0092] Furthermore, the high-load moment is the moment when the average gateway load rate does not meet the requirements.

[0093] In addition, it should be noted that when the proportion of the number of high-load moments in the period is greater than the preset proportion of the number of load moments, determine that the period is the collaborative processing period of the effective node group.

[0094] It can be understood that, as Figure 5 shown, the method for determining the preferred group in the effective node group is:

[0095] Based on the change situation of the collaborative processing periods of the effective node group on different dates, determine the dates when different periods are recognized as collaborative processing periods, and use them as collaborative matching dates;

[0096] Determine the collaborative matching coefficient of different periods based on the proportion of the number of collaborative matching dates of different periods within a preset time period;

[0097] Based on the collaborative matching coefficients in different time periods, determine the number of time periods in which the collaborative matching coefficient is greater than the preset matching coefficient, and use the number of time periods in which the collaborative matching coefficient is greater than the preset matching coefficient to determine the preferred group in the effective node group.

[0098] Further, the preferred group is the effective node group with the largest number of time periods in which the collaborative matching coefficient is greater than the preset matching coefficient.

[0099] In addition, it should be noted that the collaborative processing types of the time periods include type-one collaborative type, type-two collaborative type, and type-three collaborative type.

[0100] It should be noted that the method for determining the collaborative processing type of the time period is as follows:

[0101] When the collaborative matching coefficient of the time period is greater than the preset matching coefficient threshold, then determine that the collaborative processing type of the time period is type-one collaborative type;

[0102] When the collaborative matching coefficient of the time period is not greater than the preset matching coefficient threshold, it is further necessary to determine whether the collaborative matching coefficient of the time period is within the preset collaborative matching coefficient interval. If so, determine that the collaborative processing type of the time period is type-two collaborative type, and if not, determine that the collaborative processing type of the time period is type-three collaborative type.

[0103] Optionally, the method for determining the preferred group in the effective node group is as follows:

[0104] S41 Determine the gateway idle coefficient of the effective node group based on the number of collaborative processing time periods of the effective node group on different dates within a preset time period;

[0105] S42 Determine the dates when different time periods are recognized as collaborative processing time periods based on the change situation of the collaborative processing time periods of the effective node group on different dates, and use them as collaborative matching dates. Determine the collaborative matching coefficients of different time periods according to the proportion of the number of collaborative matching dates of different time periods within a preset time period, and determine the idle time period change coefficient of the effective node group according to the collaborative matching coefficients of different time periods;

[0106] S43 Based on the idle time period change coefficient and the gateway idle coefficient, determine the gateway availability coefficient of the effective node group, and use the gateway availability coefficient to determine the preferred group in the effective node group.

[0107] Further, the preferred group is the effective node group with the largest gateway availability coefficient.

[0108] Specifically, using the collaborative processing types of different time periods to determine the strategy for the preferred group and the edge nodes to perform collaborative processing of large model data specifically includes:

[0109] When the collaborative processing type of the time period is a first type of collaborative type, the preferred group and the edge node are used to jointly perform collaborative processing of large model data;

[0110] When the collaborative processing type of the time period is a second type of collaborative type, it is also necessary to determine whether the average load rate of the Internet of Things gateways between the available edge nodes of the preferred group and the edge node within the nearest unit time period is less than the preset load rate threshold. If so, the preferred group and the edge node are used to jointly perform collaborative processing of large model data. If not, the edge node is used to process the large model data.

[0111] When the collaborative processing type of the time period is a third type of collaborative type, the edge node is used to process the large model data.

[0112] Embodiment 2

[0113] Optionally, the above step S41 includes the following content:

[0114] S411 Determine whether the number of collaborative processing time periods of the effective node group on different dates within the preset time period is less than the preset number of processing time periods. If so, determine that the effective node group does not belong to the preferred group. When there is a date when the number of collaborative processing time periods of the effective node group is not less than the preset number of processing time periods, go to step S412;

[0115] S412 Obtain the proportion of the number of dates when the number of collaborative processing time periods of the effective node group is not less than the preset number of processing time periods. When the proportion of the number of dates does not meet the requirements, determine that the effective node group does not belong to the preferred group. When the proportion of the number of dates meets the requirements, go to step S413;

[0116] S413 Determine the gateway idle coefficient of the effective node group based on the number of collaborative processing time periods of the effective node group on different dates within the preset time period. When the gateway idle coefficient of the effective node group is less than the preset idle coefficient threshold, determine that the effective node group does not belong to the preferred group. When the gateway idle coefficient of the effective node group is less than the preset idle coefficient threshold, go to step S42.

[0117] Optionally, the above step S42 includes the following content:

[0118] S421 determines the dates when different periods are recognized as collaborative processing periods based on the changes in the collaborative processing periods of the effective node group on different dates, and uses them as collaborative matching dates. According to the proportion of the number of collaborative matching dates of different periods within a preset time period, the collaborative matching coefficients of different periods are determined. When the collaborative matching coefficients of different periods are all less than the preset collaborative matching coefficient threshold, it is determined that the effective node group does not belong to the preferred group. When there is a period with a collaborative matching coefficient not less than the preset collaborative matching coefficient threshold, it proceeds to step S413;

[0119] S422 obtains the number of periods with a collaborative matching coefficient not less than the preset collaborative matching coefficient threshold. When the number of periods with a collaborative matching coefficient not less than the preset collaborative matching coefficient threshold is less than the preset number of collaborative periods, it proceeds to step S423. When the number of periods with a collaborative matching coefficient not less than the preset collaborative matching coefficient threshold is not less than the preset number of collaborative periods, it proceeds to step S424;

[0120] S423 When the gateway idle coefficient of the effective node group is within the preset idle coefficient range, it is determined that the effective node group does not belong to the preferred group. When the gateway idle coefficient of the effective node group is not within the preset idle coefficient range, it proceeds to step S424;

[0121] S424 determines the idle period change coefficient of the effective node group according to the collaborative matching coefficients of different periods. When the idle period change coefficient of the effective node group does not meet the requirements, it is determined that the effective node group does not belong to the preferred group. When the idle period change coefficient of the effective node group meets the requirements, it proceeds to step S43.

[0122] Embodiment 3

[0123] Optionally, the method for determining the available edge nodes of the edge node is as follows:

[0124] Based on the historical processing data of the multi-modal large model of the edge node under different data processing types, determine the data processing types for which the edge node has data processing, and use them as node processing types. When the average processing accuracy of other edge nodes under different node processing types does not meet the requirements, it is determined that the other edge nodes do not belong to the available edge nodes;

[0125] When the average processing accuracy of other edge nodes under different node processing types meets the requirements:

[0126] Based on the historical processed data of the multimodal large model of the edge node under different data processing types, determine the amount of processed data of the edge node under different data processing types, and use the amount of processed data under different data processing types to perform data processing types with a processed data amount greater than a preset data amount threshold, and use it as a matching data type. When there is a matching data type with a processing accuracy that does not meet the requirements in the other edge nodes, it is determined that the other edge nodes do not belong to available edge nodes;

[0127] When there is no matching data type with a processing accuracy that does not meet the requirements in the other edge nodes:

[0128] According to the processing accuracy of the other edge nodes under different matching data processing types, determine the average value of the processing accuracy of the other edge nodes under different processing accuracies, and use it as the average accuracy. When the average accuracy does not meet the requirements, it is determined that the other edge nodes do not belong to available edge nodes;

[0129] When the average accuracy meets the requirements:

[0130] Use the proportion of the amount of processed data under different data processing types in the amount of processed data of the edge node to determine the weight coefficient of different data processing types;

[0131] According to the processing accuracy of the other edge nodes under different data processing types, and combined with the weight coefficient of the edge node under different data processing types, determine the sum of the weight coefficients of the processing accuracy of the other edge nodes, and use it as the corrected accuracy, and determine whether the other edge nodes are available edge nodes based on the corrected accuracy.

[0132] Each embodiment in this specification is described in a progressive manner. The same or similar parts between each embodiment can be referred to each other. The key point of each embodiment is to illustrate the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0133] The above describes specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the embodiments and still achieve the desired result. Additionally, the processes depicted in the figures do not necessarily require the particular order or sequential order shown to achieve the desired result. In certain implementations, multitasking and parallel processing are also possible or may be advantageous.

[0134] The above description is only for one or more embodiments of this specification and is not intended to limit this specification. For those skilled in the art, various changes and modifications can be made to one or more embodiments of this specification. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification shall be included within the scope of the claims of this application.

Claims

1. An industrial Internet of Things edge computing data processing system based on a multi-modal large model, characterized in that, Specifically include: Node screening module, group division module, time period evaluation module, data processing module; The node screening module is responsible for determining the available edge nodes of the edge nodes based on the historical processing data of the multi-modal large model of the edge nodes under different data processing types and the processing accuracy under different data processing types; The group division module is responsible for using the available edge nodes to construct multiple node groups, and obtaining the effective node groups in the node groups based on the recognition accuracy of the historical processing data of the edge nodes in different node groups; The time period evaluation module is responsible for determining the collaborative processing time period of the effective node groups by using the historical load conditions of the Internet of Things gateways between the available edge nodes in the effective node groups and the edge nodes at different time periods; The data processing module is responsible for determining the preferred group in the effective node groups and the collaborative processing types at different time periods based on the change situation of the collaborative processing time periods of the effective node groups on different dates, and using the collaborative processing types at different time periods to determine the strategy for the preferred group and the edge nodes to perform collaborative processing of large model data; The available edge nodes are other edge nodes whose average processing accuracy under the data processing type of the edge nodes is greater than 0.7; Using the average value of the processing accuracy of the available edge nodes in the node group under different data processing types to determine the recognition accuracy under different data processing types. When the recognition accuracy under different data processing types is greater than 0.8, the node group is determined as an effective node group; The collaborative processing time period is the time period when the historical load rates of the Internet of Things gateways at different times are all less than 0.5; Based on the change situation of the collaborative processing time periods of the effective node groups on different dates, determine the dates when different time periods are recognized as collaborative processing time periods, and use them as collaborative matching dates. According to the proportion of the number of collaborative matching dates of different time periods in the preset time period, determine the collaborative matching coefficients of different time periods, and use the effective node group with the largest number of time periods with collaborative matching coefficients greater than the preset matching coefficient as the preferred group; Using the collaborative processing types at different time periods to determine the strategy for the preferred group and the edge nodes to perform collaborative processing of large model data, specifically including: When the collaborative processing type of the time period is a type I collaborative type, use the preferred group and the edge nodes to jointly perform collaborative processing of large model data; When the collaborative processing type of the time period is a type II collaborative type, it is also necessary to judge whether the average value of the load rates of the Internet of Things gateways between the available edge nodes of the preferred group and the edge nodes in the nearest unit time period is less than the preset load rate threshold. If so, use the preferred group and the edge nodes to jointly perform collaborative processing of large model data. If not, use the edge nodes to perform processing of large model data; When the collaborative processing type of the time period is a type III collaborative type, use the edge nodes to perform processing of large model data.

2. The industrial Internet of Things edge computing data processing system based on the multimodal large model according to claim 1, characterized in that, The data processing types include voice, text, file, image, and video.

3. The industrial Internet of Things edge computing data processing system based on a multimodal large model according to claim 1, wherein The historical processed data includes the processed data volumes of different data processing types.

4. The industrial Internet of Things edge computing data processing system based on a multimodal large model according to claim 1, wherein Constructing multiple node groups by using the available edge nodes specifically includes: Performing free combination of different available edge nodes to obtain multiple node groups.

5. The industrial Internet of Things edge computing data processing system based on a multimodal large model according to claim 1, characterized in that The method for determining the collaborative processing type of the time period is as follows: When the collaborative matching coefficient of the time period is greater than the preset matching coefficient threshold, it is determined that the collaborative processing type of the time period is a type I collaborative type; When the collaborative matching coefficient of the time period is not greater than the preset matching coefficient threshold, it is also necessary to further determine whether the collaborative matching coefficient of the time period is within the preset collaborative matching coefficient range. If so, it is determined that the collaborative processing type of the time period is a type II collaborative type; if not, it is determined that the collaborative processing type of the time period is a type III collaborative type.

Images