Multi-electronic control compatible sewing equipment internet of things data analysis optimization method

Abstract

The application relates to the technical field of data analysis and specifically discloses a sewing equipment Internet of Things data analysis optimization method for multi-electric control compatibility, which comprises the following steps: generating a plurality of associated characteristic vectors based on the protocol core information of multi-factory electric control systems and the signal characteristics of sewing equipment and performing clustering to obtain an electric control characteristic vector library of the sewing equipment; generating a to-be-matched electric control characteristic vector based on a current initial protocol signal and a real-time signal of a sewing equipment sensor, and performing two-dimensional matching with the electric control characteristic vector library of the sewing equipment to obtain a target protocol template and a sensor signal check rule; constructing an analysis efficiency optimization plane function based on historical analysis data of the target protocol template and the sensor signal check rule, and distributing a current cooperative analysis task to each distributed analysis node for data analysis based on the distance between the plane function and the three-dimensional load affine coordinates of each node; and the analysis efficiency and node load balancing are considered to improve the sewing equipment Internet of Things data analysis efficiency.

Description

Technical Field

[0001] This invention relates to the field of data parsing technology, and in particular to an optimization method for IoT data parsing of sewing equipment compatible with multiple electronic controls. Background Technology

[0002] In the development of the Internet of Things (IoT) for sewing equipment, achieving data parsing that ensures compatibility across multiple electronic control systems is a crucial step. With the development of the sewing industry, electronic control systems from different manufacturers are widely used in various types of sewing equipment; however, their data protocols differ significantly. Effectively solving the data parsing problem for compatibility across multiple electronic control systems will greatly promote the intelligent and networked upgrades of sewing equipment, improve production efficiency and management levels, and has broad application prospects in the field of Industrial IoT. By optimizing data parsing methods, sewing equipment with different electronic control systems can be better integrated into the IoT, achieving efficient data collection and processing, and providing enterprises with more accurate production decision-making support.

[0003] However, existing technologies have significant shortcomings. When dealing with multi-vendor electronic control systems, the lack of standardized data protocols makes it difficult to quickly and accurately identify and parse data from different systems. The absence of an effective protocol matching mechanism often necessitates manual configuration, which is inefficient, error-prone, and time-consuming. Furthermore, the technology lacks the capacity to process real-time data from multiple devices in high-concurrency scenarios, resulting in significant data parsing latency and failing to meet the needs of large-scale sewing workshops and similar environments. Simultaneously, existing technologies exhibit poor compatibility with new machine models' electronic control systems, making it difficult to quickly achieve protocol compatibility and data parsing when adding new models.

[0004] Therefore, this invention proposes an IoT data parsing and optimization method for sewing equipment that is compatible with multiple electronic controls. Summary of the Invention

[0005] This invention provides an optimization method for IoT data parsing of sewing equipment compatible with multiple electronic control systems. Based on the core information of multi-manufacturer electronic control system protocols and the physical signal characteristics of external sensors of the sewing equipment, it generates and clusters standardized protocol-sensor signal correlation feature vectors, constructing a sewing equipment electronic control feature vector library. This achieves the integration and standardization of information from multiple electronic control systems and sensors, providing a unified basis for data parsing. By generating matching electronic control feature vectors through the initial protocol signals of the current sewing equipment electronic control system and the real-time signals of the current sewing equipment sensors, a two-dimensional matching process is performed with the sewing equipment electronic control feature vector library to accurately screen target protocol templates and sensor signal verification rules, ensuring the accuracy of the parsing basis. Historical parsing data of the target protocol templates and sensor signal verification rules is used to construct a parsing efficiency optimization plane function, and the multi-dimensional load state of distributed parsing nodes is mapped to three-dimensional load affine coordinates, providing quantitative indicators for task allocation to optimize parsing efficiency. Based on the distance between the parsing efficiency optimization plane function and the three-dimensional load affine coordinates of each distributed parsing node, the protocol-sensor signal collaborative parsing tasks are rationally allocated, balancing parsing efficiency and node load balance, efficiently obtaining accurate IoT data parsing results for sewing equipment, and comprehensively improving the data parsing performance of IoT for sewing equipment.

[0006] This invention provides a method for optimizing IoT data parsing in sewing equipment compatible with multiple electronic controls, comprising:

[0007] Based on the core protocol information of multiple manufacturers' electronic control systems and the physical signal characteristics of external sensors of sewing equipment, multiple standardized protocol-sensor signal association feature vectors are generated. All standardized protocol-sensor signal association feature vectors are clustered to construct a sewing equipment electronic control feature vector library.

[0008] Based on the initial protocol signal of the current sewing equipment electronic control system and the real-time signal of the current sewing equipment sensor, a matching electronic control feature vector is generated. The matching electronic control feature vector is then matched with the sewing equipment electronic control feature vector library in two dimensions to select the target protocol template and sensor signal verification rules.

[0009] Based on the historical parsing data of the target protocol template and sensor signal verification rules, a parsing efficiency optimization plane function is constructed, and the current multidimensional load state of all distributed parsing nodes is mapped to three-dimensional load affine coordinates;

[0010] Based on the distance between the plane function and the three-dimensional load affine coordinates of each distributed parsing node, the current protocol-sensor signal collaborative parsing task is assigned to each distributed parsing node for data parsing to obtain the IoT data parsing results of the sewing equipment.

[0011] Preferably, based on the core protocol information of multi-manufacturer electronic control systems and the physical signal characteristics of external sensors of the sewing equipment, multiple standardized protocol-sensor signal association feature vectors are generated. All standardized protocol-sensor signal association feature vectors are clustered to construct a sewing equipment electronic control feature vector library, including:

[0012] Simultaneously collect core protocol information from multiple manufacturers' electronic control systems and physical signal characteristics from external sensors of the sewing equipment, and generate multi-dimensional feature parameters;

[0013] The multidimensional feature parameters are normalized to generate a standardized protocol-sensor signal correlation feature vector;

[0014] The K-means clustering algorithm is used to cluster all standardized protocol-sensor signal associated feature vectors to obtain multiple clusters. The cluster center vector of each cluster is calculated and the cluster identifier and typical electronic control model are labeled.

[0015] A feature vector library is built based on the cluster center vectors of all clusters, the standardized protocol-sensor signal association feature vectors, and the corresponding standardized protocol templates and sensor signal verification rules.

[0016] Preferably, a matching electronic control feature vector is generated based on the initial protocol signal of the current sewing equipment electronic control system and the real-time signal of the current sewing equipment sensor. This matching feature vector is then performed in a two-dimensional manner against the sewing equipment electronic control feature vector library to filter out the target protocol template and sensor signal verification rules, including:

[0017] When the sewing equipment is powered on, the initial protocol signal of the current sewing equipment electronic control system and the real-time signal of the current sewing equipment sensor are collected, and the multi-dimensional feature parameters of the initial protocol signal and the real-time signal of the current sewing equipment sensor are extracted to generate the electronic control feature vector to be matched.

[0018] Calculate the cosine similarity between the electronic control feature vector to be matched and the center vector of each cluster in the sewing equipment electronic control feature vector library, and select candidate clusters with similarity not less than a preset similarity threshold from the sewing equipment electronic control feature vector library;

[0019] Calculate the Euclidean distance between the feature vector to be matched and each standardized protocol-sensor signal associated feature vector in the candidate cluster, and select the standardized protocol template and sensor signal verification rule corresponding to the standardized protocol-sensor signal associated feature vector with the smallest Euclidean distance as the target protocol template and sensor signal verification rule.

[0020] Preferably, a plane function for optimizing parsing efficiency is constructed based on historical parsing data of the target protocol template and sensor signal verification rules, including:

[0021] Obtain historical parsing data corresponding to the target parsing protocol template and sensor signal verification rules. The historical parsing data corresponding to the target parsing protocol template and sensor signal verification rules includes: historical load parameters and historical parsing performance indicators of a large number of historical distributed parsing nodes that conform to the target parsing protocol template and sensor signal verification rules.

[0022] Based on the historical load parameters and historical parsing performance indicators of a large number of distributed parsing nodes, multi-dimensional historical load feature vectors and multi-dimensional historical parsing performance indicator vectors are generated for each historical distributed parsing node. A parsing load-performance correlation dataset is then constructed based on the multi-dimensional historical load feature vectors and multi-dimensional historical parsing performance indicator vectors of each historical distributed parsing node.

[0023] Cluster the multidimensional historical load feature vectors in the parsed load-performance correlation dataset to obtain multiple parsed load-performance correlation clusters;

[0024] Based on the multi-dimensional historical parsing performance index vector of each historical distributed parsing node in each parsing load-performance correlation cluster, the comprehensive performance value of each parsing load-performance correlation cluster is calculated.

[0025] Among all the parsing load-performance correlation clusters, the parsing load-performance correlation cluster corresponding to the maximum comprehensive performance value is regarded as the optimal performance cluster;

[0026] The multidimensional historical parsing performance index vectors of all historical distributed parsing nodes in the optimal performance cluster are uniformly reduced in dimensionality to obtain the three-dimensional key load quantification indexes of each historical distributed parsing node in the optimal performance cluster.

[0027] Based on the three-dimensional key load quantification index of all historical distributed parsing nodes of the optimal performance cluster, a plane function for optimizing parsing efficiency is fitted.

[0028] Preferably, the multi-dimensional historical parsing performance index vectors of all historical distributed parsing nodes in the optimal performance cluster are uniformly reduced in dimensionality to obtain the three-dimensional key load quantification indexes of each historical distributed parsing node in the optimal performance cluster, including:

[0029] A historical parsing performance index matrix is ​​constructed based on the multi-dimensional historical parsing performance index vector of all historical distributed parsing nodes in the optimal performance cluster.

[0030] Principal component analysis is used to reduce the dimensionality of the historical parsing performance index matrix, thereby obtaining three-dimensional key load quantification indicators for all historical distributed parsing nodes.

[0031] Preferably, the current multidimensional load state of all distributed parsing nodes is mapped to three-dimensional load affine coordinates, including:

[0032] Collect the current multi-dimensional load status of all distributed parsing nodes;

[0033] The current multidimensional load status of all distributed parsing nodes is uniformly reduced to obtain three-dimensional key load quantification indicators.

[0034] Three-dimensional load affine coordinates are generated based on the three-dimensional key load quantization metrics of all distributed parsing nodes.

[0035] Preferably, based on the distance between the plane function and the three-dimensional load affine coordinates of each distributed parsing node, the current protocol-sensor signal collaborative parsing task is assigned to each distributed parsing node for data parsing to obtain the IoT data parsing results for the sewing equipment, including:

[0036] The distance between the analytic efficiency optimization plane function and the 3D load affine coordinates of each distributed analytic node is calculated.

[0037] Based on the historical number of times each distributed parsing node processes the target protocol template and the average verification pass rate, the parsing adaptability score of each distributed parsing node is calculated.

[0038] The distance between the plane function for optimizing parsing efficiency and the 3D load affine coordinates of each distributed parsing node is normalized to obtain the optimal parsing load fit for each distributed parsing node.

[0039] Based on the parsing adaptability score and optimal parsing load adaptability of each distributed parsing node, the comprehensive adaptability score of each distributed parsing node is calculated.

[0040] Based on the total data volume of the current protocol-sensor signal collaborative parsing task, the remaining processing capacity parameters of each distributed parsing node, and the correlation between node allocation data and parsing efficiency of similar historical tasks, the current protocol-sensor signal collaborative parsing task is divided into sub-tasks and allocated to each distributed parsing node.

[0041] Each distributed parsing node is controlled to call the target protocol template to parse the electronic control protocol data, extract key data, and call the sensor signal verification rules to verify the consistency between the key data and the real-time signals of the current sewing equipment sensors. The parsing results that pass the verification are summarized as the sewing equipment IoT data parsing results.

[0042] Preferably, based on the total data volume of the current protocol-sensor signal collaborative parsing task, the remaining processing capacity parameters of each distributed parsing node, and the correlation between node allocation data and parsing efficiency in historical similar tasks, the current protocol-sensor signal collaborative parsing task is divided into sub-tasks and allocated to each distributed parsing node, including:

[0043] Collect the remaining processing capacity parameters of each distributed parsing node, and generate a remaining processing capacity vector based on the remaining processing capacity parameters of each distributed parsing node. The remaining processing capacity parameters include the number of remaining tasks, the percentage of remaining CPU resources, the percentage of remaining memory resources, and the historical processing rate of similar tasks.

[0044] The remaining processing capacity vector of each distributed parsing node is evaluated to obtain the resource sufficiency and rate adaptability of each distributed parsing node, and the comprehensive score of the remaining processing capacity of each distributed parsing node is determined based on the resource sufficiency and rate adaptability of each distributed parsing node.

[0045] Based on the comprehensive score of the remaining processing capacity of each distributed parsing node, the initial subtask splitting ratio of each distributed parsing node is determined.

[0046] Based on the correlation matrix between the subtask allocation ratio and the parsing latency reduction rate, the initial subtask splitting ratio of each distributed parsing node is corrected to obtain the final subtask splitting ratio of each distributed parsing node.

[0047] Based on the final subtask splitting ratio of each distributed parsing node, the current protocol-sensor signal collaborative parsing task is split into subtasks and assigned to each distributed parsing node.

[0048] Preferably, the process of determining the correlation matrix between the subtask allocation ratio and the parsing latency reduction rate includes:

[0049] Obtain node allocation data and parsing efficiency data of historical similar protocol-sensor signal collaborative parsing tasks within a preset historical period, and construct a time series dataset. The node allocation data includes node ID and the number of allocated subtasks, and the parsing efficiency data includes parsing latency and verification pass rate.

[0050] Correlation analysis is performed on the upward, stationary, and downward trends of each sequence data in the time series dataset. The correlation between the subtask allocation ratio of each distributed parsing node and the parsing latency reduction rate is calculated, and a correlation matrix is ​​generated.

[0051] Preferably, the initial subtask splitting ratio of each distributed parsing node is corrected based on the correlation matrix to obtain the final subtask splitting ratio of each distributed parsing node, including:

[0052] The weighted sum of all matrix elements corresponding to the initial subtask splitting ratio and the allowable parsing latency reduction rate of each distributed parsing node in the correlation matrix is ​​taken as the strong correlation coefficient of the corresponding distributed parsing node.

[0053] Based on the strong correlation coefficient of each distributed resolution node, nodes to be optimized for allocation are selected from all distributed resolution nodes;

[0054] Based on the strong correlation coefficients of all nodes to be optimized, the initial subtask splitting ratios of all nodes to be optimized are corrected to obtain the final subtask splitting ratios of each distributed parsing node.

[0055] The beneficial effects of this invention compared to the prior art are as follows:

[0056] First, to address the issue of inconsistent data protocols across multiple manufacturers' electronic control systems, making rapid and accurate identification and parsing difficult, a standardized protocol-sensor signal correlation feature vector is generated to construct a sewing equipment electronic control feature vector library. This library is then matched in two dimensions with the electronic control feature vectors to be matched, quickly filtering out target protocol templates and verification rules without the need for manual configuration, thus improving identification and matching efficiency and accuracy.

[0057] Second, to address the issue of high data parsing latency in high-concurrency scenarios, a parsing efficiency optimization planar function is constructed based on historical parsing data. Combined with the three-dimensional load affine coordinates of distributed parsing nodes, parsing tasks are rationally allocated according to distance, effectively controlling parsing latency and meeting the needs of large-scale scenarios.

[0058] Third, given the current situation where it is difficult to quickly achieve compatibility of the electronic control system for new sewing equipment, this invention constructs a universal feature vector library for the electronic control system of sewing equipment, so that new models can be integrated into the system simply by importing a new template, thereby achieving protocol compatibility and data parsing and enhancing the system's compatibility with new models.

[0059] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in this application.

[0060] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0061] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0062] Figure 1 This invention relates to an IoT data parsing and optimization method for sewing equipment with multi-electronic control compatibility.

[0063] Figure 2 This is a flowchart of the target filtering protocol and rules in an embodiment of the present invention;

[0064] Figure 3 This is a flowchart illustrating the task allocation and data parsing process in an embodiment of the present invention. Detailed Implementation

[0065] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0066] like Figure 1 As shown, this invention provides an implementation method for optimizing IoT data parsing in sewing equipment compatible with multiple electronic controls, comprising:

[0067] Based on the core protocol information of multiple manufacturers' electronic control systems and the physical signal characteristics of external sensors of sewing equipment, multiple standardized protocol-sensor signal association feature vectors are generated. All standardized protocol-sensor signal association feature vectors are clustered to construct a sewing equipment electronic control feature vector library.

[0068] Based on the initial protocol signal of the current sewing equipment electronic control system and the real-time signal of the current sewing equipment sensor, a matching electronic control feature vector is generated. The matching electronic control feature vector is then matched with the sewing equipment electronic control feature vector library in two dimensions to select the target protocol template and sensor signal verification rules.

[0069] Based on the historical parsing data of the target protocol template and sensor signal verification rules, a parsing efficiency optimization plane function is constructed, and the current multidimensional load state of all distributed parsing nodes is mapped to three-dimensional load affine coordinates;

[0070] Based on the distance between the plane function and the three-dimensional load affine coordinates of each distributed parsing node, the current protocol-sensor signal collaborative parsing task is assigned to each distributed parsing node for data parsing to obtain the IoT data parsing results of the sewing equipment.

[0071] In this embodiment, multi-manufacturer electronic control systems refer to electronic control systems in sewing equipment manufactured by different manufacturers, such as those for sewing machines and template machines. The inconsistency in protocols between these different manufacturers' electronic control systems poses difficulties for data parsing, which is the key problem this optimization method aims to solve.

[0072] In this embodiment, the core protocol information includes frame header identifiers, baud rates, data frame lengths, and verification rules for multiple manufacturers' electronic control systems. This information is crucial for data communication within the electronic control system. Since the core protocol information differs between manufacturers, it needs to be collected synchronously via an upgraded data acquisition cloud box to construct feature vectors and facilitate subsequent protocol adaptation.

[0073] In this embodiment, the external sensors of the sewing equipment are sensors installed outside the sewing equipment, such as photoelectric sensors, Hall effect sensors, and proximity switches. They are used to collect physical signals during the operation of the sewing equipment, providing important real-time information for data analysis and co-processing with the electronic control system protocol data.

[0074] In this embodiment, physical signal characteristics refer to the signal characteristics collected by external sensors of the sewing equipment, including the sewing speed fluctuation range, stitch count statistics cycle, signal sampling frequency, etc.

[0075] In this embodiment, the current sewing equipment electronic control system refers to the electronic control system of the sewing equipment that is currently running and requires data analysis.

[0076] In this embodiment, the initial protocol signal is the protocol signal of the electrical control system in its initial state, which is collected by the upgraded data acquisition cloud box when the current sewing equipment electrical control system is powered on.

[0077] In this embodiment, the real-time signal of the current sewing equipment sensor refers to the signal collected in real time by the external sensor of the sewing equipment at the current moment.

[0078] In this embodiment, the target protocol template and sensor signal verification rules are the results of a two-dimensional matching process between the electronic control feature vector to be matched and the electronic control feature vector library. The target protocol template is used to parse the electronic control protocol data, and the sensor signal verification rules are used to verify the authenticity of the parsed data. Both provide a basis for subsequent data parsing and verification.

[0079] In this embodiment, the analytical efficiency optimization plane function is a three-dimensional linear plane function generated by least squares fitting based on historical analytical data of the target protocol template and sensor signal verification rules. , ( For the number of tasks, CPU utilization, For memory usage, , , These are the weighting coefficients, and This function characterizes the load parameter ratio when the parsing efficiency is optimal. The coordinate points on and near the plane represent the historically optimal parsing state in terms of the ratio of task count, CPU, and memory, which can then be used to guide the allocation of parsing resources.

[0080] In this embodiment, all distributed parsing nodes refer to edge computing modules equipped with data parsing programs and communicating with the upgraded data acquisition cloud box. They receive electrical control protocol data and sensor signals transmitted by the cloud box and perform parsing. Their current multidimensional load state is mapped to three-dimensional load affine coordinates for subsequent parsing resource allocation.

[0081] In this embodiment, the current multidimensional load status refers to the current task load of all distributed parsing nodes, including parameters such as the current number of tasks, current CPU utilization, current memory utilization, and current sensor data processing volume.

[0082] In this embodiment, the IoT data parsing results for the sewing equipment are obtained by each distributed parsing node calling the target protocol template to parse the electronic control protocol data, extracting key data such as sewing speed and stitch count, and simultaneously calling sensor signal verification rules to perform consistency verification between the key data and the real-time sensor signals (verification is considered passed if the deviation is ≤5%). The parsing results that pass the verification are then summarized. This result is uploaded to the monitoring platform of the sewing equipment IoT system, providing relevant personnel with accurate data support for equipment operation and other aspects.

[0083] To achieve the integration and standardized storage of multi-control information and provide a foundation for subsequent matching, this paper proposes to generate multiple standardized protocol-sensor signal association feature vectors based on the core protocol information of multi-manufacturer control systems and the physical signal characteristics of external sensors of sewing equipment. All standardized protocol-sensor signal association feature vectors are then clustered to construct a sewing equipment control feature vector library, including:

[0084] Simultaneously collect core protocol information from multiple manufacturers' electronic control systems and physical signal characteristics from external sensors of the sewing equipment, and generate multi-dimensional feature parameters;

[0085] The multidimensional feature parameters are normalized to generate a standardized protocol-sensor signal correlation feature vector;

[0086] The K-means clustering algorithm is used to cluster all standardized protocol-sensor signal associated feature vectors to obtain multiple clusters. The cluster center vector of each cluster is calculated and the cluster identifier and typical electronic control model are labeled.

[0087] A feature vector library is built based on the cluster center vectors of all clusters, the standardized protocol-sensor signal association feature vectors, and the corresponding standardized protocol templates and sensor signal verification rules.

[0088] In this embodiment, the simultaneous acquisition of core protocol information from multiple manufacturers' electronic control systems and physical signal characteristics from external sensors of the sewing equipment to generate multi-dimensional feature parameters refers to using an upgraded data acquisition cloud box to simultaneously collect core protocol information such as frame header identifiers, baud rates, data frame lengths, and verification rules from different manufacturers' electronic control systems, as well as physical signal characteristics such as sewing speed fluctuation range, stitch count counting period, and signal sampling frequency collected by external sensors such as photoelectric / Hall sensors. This information from both the protocol and sensor sources is integrated to form feature parameters with six dimensions. For example, a multi-dimensional feature parameter might consist of a specific baud rate and data frame length from a particular manufacturer's electronic control system, combined with the sewing speed fluctuation range and stitch count counting period collected by the corresponding sensor, preparing for the subsequent generation of associated feature vectors.

[0089] In this embodiment, the multidimensional feature parameters are normalized to generate a standardized protocol-sensor signal correlation feature vector. This involves mathematically transforming the previously obtained multidimensional feature parameters, mapping the values ​​of each dimension to the [0,1] interval. This eliminates the impact of large differences in the numerical ranges of different dimensions on subsequent analysis, making all parameters comparable on the same scale.

[0090] In this embodiment, the K-means clustering algorithm is used to cluster all standardized protocol-sensor signal associated feature vectors, resulting in multiple clusters. The cluster center vector of each cluster is calculated, and a cluster identifier and typical electronic control unit (ECU) model are labeled. The K-means clustering algorithm groups similar standardized protocol-sensor signal associated feature vectors into one category, forming multiple clusters. The number of clusters is typically set to the number of mainstream ECU brands, such as 3-5 categories. For each cluster, the center position of all feature vectors is calculated to obtain the cluster center vector. Simultaneously, each cluster is labeled with a unique cluster identifier and associated with a typical ECU model and the corresponding sensor signal baseline value. For example, cluster 1 might correspond to a Panasonic A6 ECU with speed fluctuation ≤5%. Through this clustering and labeling, ECU systems with similar protocols and signal characteristics can be grouped together, reducing the dimensionality of subsequent adaptation and improving adaptation efficiency.

[0091] In this embodiment, a feature vector library is built based on the cluster center vectors of all clusters, the standardized protocol-sensor signal association feature vectors, and the corresponding standardized protocol templates and sensor signal verification rules. The cluster center vectors of each cluster, all standardized protocol-sensor signal association feature vectors, and their respective standardized protocol templates (used for parsing electronic control protocol data) and sensor signal verification rules (used for verifying the authenticity of the parsed data) are integrated to construct a database-style feature vector library. This library stores rich information, providing data support for quickly matching and determining the appropriate protocol parsing and verification method when the sewing equipment is powered on.

[0092] like Figure 2 As shown, in order to select the target protocol template and sensor signal verification rules that are most suitable for the current equipment and to ensure the accuracy of the data parsing basis, a method is proposed to generate a matching electronic control feature vector based on the initial protocol signal of the current sewing equipment's electronic control system and the real-time signal of the current sewing equipment's sensors. This matching feature vector is then performed in a two-dimensional match with the sewing equipment's electronic control feature vector library to select the target protocol template and sensor signal verification rules, including:

[0093] When the sewing equipment is powered on, the initial protocol signal of the current sewing equipment electronic control system and the real-time signal of the current sewing equipment sensor are collected, and the multi-dimensional feature parameters of the initial protocol signal and the real-time signal of the current sewing equipment sensor are extracted to generate the electronic control feature vector to be matched.

[0094] Calculate the cosine similarity between the electronic control feature vector to be matched and the center vector of each cluster in the sewing equipment electronic control feature vector library, and select candidate clusters with similarity not less than a preset similarity threshold from the sewing equipment electronic control feature vector library;

[0095] Calculate the Euclidean distance between the feature vector to be matched and each standardized protocol-sensor signal associated feature vector in the candidate cluster, and select the standardized protocol template and sensor signal verification rule corresponding to the standardized protocol-sensor signal associated feature vector with the smallest Euclidean distance as the target protocol template and sensor signal verification rule.

[0096] In this embodiment, "powering on the sewing equipment" means that the sewing equipment is connected to a power source and starts operating. At this time, the equipment's electronic control system and external sensors enter a working state, preparing for data acquisition and subsequent analysis processes. This is the trigger point for the entire data analysis process; only when the equipment is powered on can the various signal data required for the current operation of the equipment be obtained.

[0097] In this embodiment, the initial protocol signal of the current sewing equipment's electronic control system and the real-time signal of the current sewing equipment's sensors are collected, and multi-dimensional feature parameters of the initial protocol signal and the real-time signal of the current sewing equipment's sensors are extracted to generate a matching electronic control feature vector. After the sewing equipment is powered on, the upgraded data acquisition cloud box simultaneously collects the initial protocol signal emitted by the electronic control system and the signals collected in real time by external sensors. Protocol-related features such as the protocol's frame header identifier and baud rate, as well as physical signal features such as the sewing speed fluctuation range and stitch count period collected by the sensors, are extracted from these signals to form multi-dimensional feature parameters. Then, these multi-dimensional feature parameters are subjected to normalization processing similar to the previous one (mapped to the [0,1] interval) to generate a 1×6 order matching electronic control feature vector. This vector represents the comprehensive features of the protocol and sensor signals in the initial state of the current equipment operation and is used to match with the established electronic control feature vector library.

[0098] In this embodiment, the preset similarity threshold is a pre-set value used to determine whether the similarity between the electronic control feature vector to be matched and the cluster center vector is high enough. After calculating the cosine similarity between the electronic control feature vector to be matched and each cluster center vector, these similarity values ​​are compared with the preset similarity threshold. For example, if the preset similarity threshold is set to 90% (0.9), then only clusters with a cosine similarity greater than or equal to 0.9 will be considered to have a high similarity to the current vector to be matched and will enter the subsequent screening process; otherwise, they will be excluded. Setting this threshold helps to quickly narrow down the matching range and improve matching efficiency and accuracy.

[0099] In this embodiment, the Euclidean distance between the feature vector to be matched and each standardized protocol-sensor signal associated feature vector within the candidate cluster is calculated. After selecting candidate clusters whose similarity meets a preset threshold using cosine similarity, the Euclidean distance between the feature vector to be matched and each standardized protocol-sensor signal associated feature vector within the candidate cluster is further calculated. Euclidean distance is a way to measure the distance between two points in multidimensional space, and here it is used to more accurately evaluate the similarity between the feature vector to be matched and each vector within the candidate cluster. The smaller the distance, the more similar the feature vector to be matched is to the standardized protocol-sensor signal associated feature vector. By comparing these distances, the standardized protocol template and sensor signal verification rule corresponding to the vector with the smallest distance are selected as the final matching result to achieve accurate adaptation between the protocol and the sensor signal.

[0100] To fit a plane function that reflects the relationship between load and performance and optimizes parsing efficiency, a plane function for optimizing parsing efficiency is proposed, based on historical parsing data of the target protocol template and sensor signal verification rules. This includes:

[0101] Obtain historical parsing data corresponding to the target parsing protocol template and sensor signal verification rules. The historical parsing data corresponding to the target parsing protocol template and sensor signal verification rules includes: historical load parameters and historical parsing performance indicators of a large number of historical distributed parsing nodes that conform to the target parsing protocol template and sensor signal verification rules.

[0102] Based on the historical load parameters and historical parsing performance indicators of a large number of distributed parsing nodes, multi-dimensional historical load feature vectors and multi-dimensional historical parsing performance indicator vectors are generated for each historical distributed parsing node. A parsing load-performance correlation dataset is then constructed based on the multi-dimensional historical load feature vectors and multi-dimensional historical parsing performance indicator vectors of each historical distributed parsing node.

[0103] Cluster the multidimensional historical load feature vectors in the parsed load-performance correlation dataset to obtain multiple parsed load-performance correlation clusters;

[0104] Based on the multi-dimensional historical parsing performance index vector of each historical distributed parsing node in each parsing load-performance correlation cluster, the comprehensive performance value of each parsing load-performance correlation cluster is calculated.

[0105] Among all the parsing load-performance correlation clusters, the parsing load-performance correlation cluster corresponding to the maximum comprehensive performance value is regarded as the optimal performance cluster;

[0106] The multidimensional historical parsing performance index vectors of all historical distributed parsing nodes in the optimal performance cluster are uniformly reduced in dimensionality to obtain the three-dimensional key load quantification indexes of each historical distributed parsing node in the optimal performance cluster.

[0107] Based on the three-dimensional key load quantification index of all historical distributed parsing nodes of the optimal performance cluster, a plane function for optimizing parsing efficiency is fitted.

[0108] In this embodiment, the large number of historical distributed parsing nodes that conform to the target parsing protocol template and sensor signal verification rules refer to numerous distributed parsing nodes that used the same rules for data parsing in the past as the currently selected target protocol template and sensor signal verification rules. The relevant data from these nodes is collected and analyzed to determine the relationship between parsing efficiency and load, providing a basis for optimizing parsing resource allocation.

[0109] In this embodiment, historical load parameters include the number of historical tasks of the distributed parsing node, historical CPU utilization, historical memory utilization, and historical sensor data processing volume. The number of historical tasks represents the total amount of collaborative parsing tasks involving electronic control protocol data and sensor signals processed by the node in the past; historical CPU utilization reflects the proportion of CPU resources used by the node when processing historical collaborative parsing tasks; historical memory utilization reflects the memory resource usage of the node when processing historical tasks; and historical sensor data processing volume is the total amount of data collected by external sensors processed by the node in the past. These parameters comprehensively reflect the load status of the parsing node during the historical parsing process.

[0110] In this embodiment, historical parsing performance metrics include historical parsing latency and historical data verification pass rate. Historical parsing latency refers to the time spent from receiving data to completing parsing in the past parsing process; historical data verification pass rate is the proportion of data that passes the consistency verification of the parsed electronic control data based on sensor signal verification rules. These two metrics are used to measure the performance of the parsing node in historical parsing tasks.

[0111] In this embodiment, based on the historical load parameters and historical parsing performance indicators of a large number of distributed parsing nodes, multi-dimensional historical load feature vectors and multi-dimensional historical parsing performance indicator vectors are generated for each historical distributed parsing node. A parsing load-performance correlation dataset is then constructed based on these vectors. Specifically, the historical load parameters (historical task count, historical CPU utilization, historical memory utilization, historical sensor data processing volume) of each distributed parsing node are normalized and mapped to the [0,1] interval, generating a 4-dimensional historical load feature vector. Simultaneously, the historical parsing performance indicators (historical parsing latency, historical data verification pass rate) are also standardized, with parsing latency and verification pass rate both standardized to the [0,1] interval, forming a multi-dimensional historical parsing performance indicator vector. Then, these two vectors corresponding to each node are combined to construct a parsing load-performance correlation dataset. This dataset records the relationship between the load characteristics and parsing performance of each node, providing a data foundation for subsequent analysis of the correlation between parsing efficiency and load.

[0112] In this embodiment, clustering the multidimensional historical load feature vectors in the parsing load-performance correlation dataset to obtain multiple parsing load-performance correlation clusters refers to using the K-means clustering algorithm to group similar multidimensional historical load feature vectors in the dataset into one class, forming multiple different clusters. The number of clusters is generally set to 3-5, based on the mainstream parsing load scenarios. Through clustering, nodes with similar load characteristics can be grouped together, facilitating the analysis of the relationship between different load types and parsing performance.

[0113] In this embodiment, based on the multi-dimensional historical parsing performance index vector of each historical distributed parsing node in each parsing load-performance associated cluster, the comprehensive performance value of each parsing load-performance associated cluster is calculated. The calculation method may comprehensively consider indicators such as historical parsing latency and historical data verification pass rate. For example, a formula might be set, such as Comprehensive Performance Value = a × Historical Data Verification Pass Rate - b × Historical Parsing Latency (where a and b are weighting coefficients set according to actual conditions; assuming that in the current scenario, more emphasis is placed on the impact of historical data verification pass rate on overall parsing performance, and it is desirable to highlight the inverse effect of parsing latency on performance, then a = 0.7 and b = 0.3 can be set). This quantifies the overall parsing performance of nodes within each cluster, facilitating comparison of the performance of different clusters and identifying the cluster with the best performance.

[0114] In this embodiment, a parsing efficiency optimization plane function is fitted based on the three-dimensional key load quantification indicators of all historical distributed parsing nodes of the optimal performance cluster. This is achieved by using the historical task count, historical CPU utilization, and historical memory utilization extracted from the optimal performance cluster as the three-dimensional key load quantification indicators (corresponding to the values ​​in the plane function). , , Using the least squares method and other fitting techniques, a three-dimensional linear plane function is obtained (variable). This function represents the ratio between load parameters (number of tasks, CPU utilization, memory utilization) when parsing efficiency is optimal. It is used to guide the subsequent optimization of parsing resource allocation, so that the node load is closer to the optimal state and parsing efficiency is improved.

[0115] To effectively simplify multidimensional indicators and extract key information, a method is proposed to uniformly reduce the dimensionality of the multidimensional historical parsing performance indicator vectors of all historical distributed parsing nodes in the optimal performance cluster. This yields three-dimensional key load quantification indicators for each historical distributed parsing node in the optimal performance cluster, including:

[0116] A historical parsing performance index matrix is ​​constructed based on the multi-dimensional historical parsing performance index vector of all historical distributed parsing nodes in the optimal performance cluster.

[0117] Principal component analysis is used to reduce the dimensionality of the historical parsing performance index matrix, thereby obtaining three-dimensional key load quantification indicators for all historical distributed parsing nodes.

[0118] In this embodiment, a historical parsing performance index matrix is ​​constructed based on the multi-dimensional historical parsing performance index vectors of all historical distributed parsing nodes in the optimal performance cluster. This involves integrating the multi-dimensional historical parsing performance index vectors of each historical distributed parsing node in the optimal performance cluster. Each vector contains dimensional features such as the number of historical tasks, historical CPU utilization, historical memory utilization, historical sensor data processing volume, and historical verification time. These vectors form the rows, and the dimensional features form the columns, forming a matrix. For example, assuming there are 100 historical distributed parsing nodes belonging to the optimal performance cluster, and each node's multi-dimensional historical parsing performance index vector is 5-dimensional, then the constructed historical parsing performance index matrix would be 100 rows and 5 columns.

[0119] In this embodiment, the historical parsing performance index matrix is ​​dimensionality-reduced based on principal component analysis to obtain three-dimensional key load quantification indicators for all historical distributed parsing nodes. The specific steps are as follows: First, the covariance matrix of the historical parsing performance index matrix is ​​calculated. The covariance matrix reflects the correlation between the features of each dimension in the matrix. Next, the eigenvalues ​​and eigenvectors of the covariance matrix are solved. The eigenvalues ​​represent the variance explained by each eigenvector. The top three principal components with the highest eigenvalues ​​are selected. These three principal components can guarantee a cumulative variance contribution of ≥92%, meaning that the data information loss rate after dimensionality reduction is ≤8%, which can better preserve the main information of the original data. For example, suppose the first principal component is mainly associated with the number of historical tasks and the amount of historical sensor data processed (mapping coefficients are 0.65 and 0.58, respectively), the second principal component is mainly associated with the historical CPU utilization rate (mapping coefficient 0.82), and the third principal component is mainly associated with the historical memory utilization rate (mapping coefficient 0.79). Then, the scores of each historical distributed parsing node on these three principal components are quantized using the linear transformation formula = (principal component score - principal component minimum value) / (principal component maximum value - principal component minimum value) and mapped to the [0,1] interval. Finally, three-dimensional key load quantification indicators of task-CPU-memory are generated. These indicators are used to construct the parsing efficiency optimization plane function and so on to achieve optimized allocation of parsing resources.

[0120] To provide intuitive quantified coordinates for subsequent distance-based allocation tasks, a method is proposed to map the current multidimensional load state of all distributed parsing nodes to three-dimensional load affine coordinates, including:

[0121] Collect the current multi-dimensional load status of all distributed parsing nodes;

[0122] The current multidimensional load status of all distributed parsing nodes is uniformly reduced to obtain three-dimensional key load quantification indicators.

[0123] Three-dimensional load affine coordinates are generated based on the three-dimensional key load quantization metrics of all distributed parsing nodes.

[0124] In this embodiment, collecting the current multidimensional load status of all distributed parsing nodes refers to obtaining multiple load-related parameters of each edge computing module (i.e., distributed parsing node) that is equipped with a data parsing program and communicates with the upgraded data acquisition cloud box. These parameters include the current number of tasks, current CPU utilization, current memory utilization, and current sensor data processing volume, which comprehensively reflect the current workload of each node and provide raw data for subsequent analysis of node load and task allocation.

[0125] In this embodiment, the current multidimensional load status of all distributed parsing nodes is uniformly reduced in dimensionality to obtain three-dimensional key load quantification indicators. The purpose is to simplify complex multidimensional data, highlight key information, and thus better analyze and process node load. The specific approach is similar to the processing of historical data, and principal component analysis may also be used. First, the covariance matrix of the multidimensional data containing current load parameters is calculated to understand the correlation between parameters. Then, the eigenvalues ​​and eigenvectors of the covariance matrix are solved, and the top three principal components that can explain most of the data variance are selected (for example, ensuring that the cumulative variance contribution of these three principal components meets certain requirements, such as retaining a sufficient proportion of the original data information). Assuming that these three principal components are mainly related to the current number of tasks, current CPU utilization, and current memory utilization, the scores of each node on these three principal components are then mapped to the [0,1] interval through a specific linear transformation (e.g., quantification value = (principal component score - principal component minimum value) / (principal component maximum value - principal component minimum value)), thereby obtaining the three-dimensional key load quantification indicators of task-CPU-memory for each node. These indicators represent the key characteristics of the node's current load in a more concise form.

[0126] In this embodiment, generating 3D load affine coordinates based on the 3D key load quantization indicators of all distributed parsing nodes involves mapping the previously obtained 3D key load quantization indicators of each node onto the coordinate axes of 3D space to form 3D load affine coordinates. For example, if a node's task-CPU-memory 3D key load quantization indicators are 0.6, 0.4, and 0.5 respectively, then its corresponding 3D load affine coordinates might be (0.6, 0.4, 0.5). In this way, each distributed parsing node has a corresponding coordinate position in 3D space, which facilitates subsequent evaluation of the node load's suitability for the optimal parsing state by calculating the distance between these coordinate points and the parsing efficiency optimization plane function. This allows for the rational allocation of protocol-signal collaborative parsing tasks, achieving parsing load balancing in multi-device concurrent scenarios.

[0127] like Figure 3 As shown, to achieve reasonable task allocation and efficient parsing, a method is proposed that optimizes the distance between the planar function and the three-dimensional load affine coordinates of each distributed parsing node based on parsing efficiency. This method distributes the current protocol-sensor signal collaborative parsing task to each distributed parsing node for data parsing, obtaining the IoT data parsing results for the sewing equipment, including:

[0128] The distance between the analytic efficiency optimization plane function and the 3D load affine coordinates of each distributed analytic node is calculated.

[0129] Based on the historical number of times each distributed parsing node processes the target protocol template and the average verification pass rate, the parsing adaptability score of each distributed parsing node is calculated.

[0130] The distance between the plane function for optimizing parsing efficiency and the 3D load affine coordinates of each distributed parsing node is normalized to obtain the optimal parsing load fit for each distributed parsing node.

[0131] Based on the parsing adaptability score and optimal parsing load adaptability of each distributed parsing node, the comprehensive adaptability score of each distributed parsing node is calculated.

[0132] Based on the total data volume of the current protocol-sensor signal collaborative parsing task, the remaining processing capacity parameters of each distributed parsing node, and the correlation between node allocation data and parsing efficiency of similar historical tasks, the current protocol-sensor signal collaborative parsing task is divided into sub-tasks and allocated to each distributed parsing node.

[0133] Each distributed parsing node is controlled to call the target protocol template to parse the electronic control protocol data, extract key data, and call the sensor signal verification rules to verify the consistency between the key data and the real-time signals of the current sewing equipment sensors. The parsing results that pass the verification are summarized as the sewing equipment IoT data parsing results.

[0134] In this embodiment, the distance between the resolution efficiency optimization plane function and the three-dimensional load affine coordinates of each distributed resolution node is calculated using the distance formula from a spatial point to a plane. This distance reflects the degree of deviation between the current load state of the node and the optimal resolution efficiency state. The smaller the distance, the closer the node load is to the optimal resolution load ratio.

[0135] In this embodiment, the historical processing count and average verification pass rate of each distributed parsing node for the target protocol template are used. The historical processing count refers to the number of times each distributed parsing node has used the target protocol template to parse electronic control protocol data in the past. The average verification pass rate refers to the average proportion of data that passes the consistency verification based on sensor signal verification rules after each data parsing using the target protocol template. These two data points reflect the node's familiarity with the target protocol template and the reliability of the parsed data.

[0136] In this embodiment, a parsing adaptation score is calculated for each distributed parsing node based on its historical processing count and average verification pass rate for the target protocol template. This may be achieved using a weighted calculation method. For example, assuming the parsing adaptation score is based on the historical processing count and average verification pass rate, where 'a' and 'b' are weighting coefficients set according to actual conditions, a score ranging from 0 to 1 is obtained. This score measures the node's adaptation to the target protocol template; a higher score indicates better adaptation.

[0137] In this embodiment, the distance between the plane function for optimizing parsing efficiency and the affine coordinates of the three-dimensional load of each distributed parsing node is normalized to obtain the optimal parsing load fit of each distributed parsing node. The purpose of normalization is to unify the distance data to the range of 0 to 1, which facilitates comprehensive comparison with other indicators.

[0138] In this embodiment, a comprehensive suitability score for each distributed parsing node is calculated based on its parsing suitability score and optimal parsing load suitability, typically using a weighted summation method. For example, weights of 0.6 and 0.4 can be used, determined by the degree to which load suitability and protocol suitability affect parsing efficiency. This formula combines the node's load suitability and protocol template suitability to obtain a more comprehensive score reflecting the node's overall suitability for the current parsing task.

[0139] In this embodiment, the remaining processing capacity parameters of each distributed parsing node include the number of remaining tasks, the percentage of remaining CPU resources, the percentage of remaining memory resources, and the historical processing rate of similar tasks. The number of remaining tasks refers to the number of parsing tasks a node can currently handle, calculated by subtracting the currently allocated tasks from the node's maximum task count. The percentage of remaining CPU resources is the proportion of currently unused CPU resources in the total CPU resources. Similarly, the percentage of remaining memory resources is the proportion of unused memory resources in the total memory resources. The historical processing rate of similar tasks is the average rate at which the node historically processes collaborative parsing tasks involving electronic control protocol data and sensor signals. These parameters are used to assess whether a node has sufficient capacity to handle new parsing tasks.

[0140] In this embodiment, the correlation between node allocation information and parsing efficiency for historical tasks of the same type refers to the relationship between different node allocation methods (e.g., which nodes are responsible for which tasks) and the final parsing efficiency (e.g., parsing latency, verification pass rate, etc.) when handling similar protocol-signal collaborative parsing tasks in the past. By analyzing this correlation, we can understand what kind of node allocation can improve parsing efficiency and provide a reference for the allocation of current protocol-signal collaborative parsing tasks.

[0141] In this embodiment, controlling each distributed parsing node to call the target protocol template to parse the electronic control protocol data and extract key data means that each distributed parsing node uses the previously selected target protocol template to interpret and analyze the electronic control protocol data according to the rules set in the template. From this, data such as sewing speed and stitch count, which are of great significance for understanding the operating status of the sewing equipment, are extracted. This data is the key data, which provides a basis for subsequent judgment and analysis of the equipment's operating status.

[0142] In this embodiment, sensor signal verification rules are invoked to verify the consistency between key data and the real-time signals of the current sewing equipment sensors. The verification results that pass are summarized as the sewing equipment IoT data parsing results. Specifically, using the previously determined sensor signal verification rules, key data extracted from the electronic control protocol data is compared with the signals collected in real-time by the external sensors of the sewing equipment. For example, it checks whether the sewing speed in the key data and the sewing speed detected by the sensor in real time are within the allowable deviation range (e.g., deviation ≤ 5%). Only when the key data and the real-time signal conform to the verification rules is that part of the parsing result considered valid. Summarizing all the verified parsing results forms the final sewing equipment IoT data parsing result. This result is uploaded to the monitoring platform to provide relevant personnel with accurate and reliable equipment operation information.

[0143] To achieve accurate decomposition and allocation of the current protocol-sensor signal collaborative parsing task, a method is proposed that, based on the total data volume of the current protocol-sensor signal collaborative parsing task, the remaining processing capacity parameters of each distributed parsing node, and the correlation between node allocation data and parsing efficiency in historical similar tasks, the current protocol-sensor signal collaborative parsing task is decomposed into subtasks and allocated to each distributed parsing node, including:

[0144] Collect the remaining processing capacity parameters of each distributed parsing node, and generate a remaining processing capacity vector based on the remaining processing capacity parameters of each distributed parsing node. The remaining processing capacity parameters include the number of remaining tasks, the percentage of remaining CPU resources, the percentage of remaining memory resources, and the historical processing rate of similar tasks.

[0145] The remaining processing capacity vector of each distributed parsing node is evaluated to obtain the resource sufficiency and rate adaptability of each distributed parsing node, and the comprehensive score of the remaining processing capacity of each distributed parsing node is determined based on the resource sufficiency and rate adaptability of each distributed parsing node.

[0146] Based on the comprehensive score of the remaining processing capacity of each distributed parsing node, the initial subtask splitting ratio of each distributed parsing node is determined.

[0147] Based on the correlation matrix between the subtask allocation ratio and the parsing latency reduction rate, the initial subtask splitting ratio of each distributed parsing node is corrected to obtain the final subtask splitting ratio of each distributed parsing node.

[0148] Based on the final subtask splitting ratio of each distributed parsing node, the current protocol-sensor signal collaborative parsing task is split into subtasks and assigned to each distributed parsing node.

[0149] In this embodiment, the remaining processing capacity vector is generated based on the remaining processing capacity parameters of each distributed parsing node. This is achieved by integrating the remaining number of tasks, the remaining CPU resource ratio, the remaining memory resource ratio, and the historical processing rate of similar tasks for each distributed parsing node into a single vector.

[0150] In this embodiment, the remaining processing capacity vector of each distributed parsing node is evaluated to obtain the resource sufficiency and rate adaptability of each distributed parsing node. For the remaining processing capacity vector, two evaluation dimensions, resource sufficiency and rate adaptability, are first determined. In the resource sufficiency dimension, the parameters of the remaining number of tasks, the remaining CPU resource ratio, and the remaining memory resource ratio are weighted according to certain weights (e.g., the remaining number of tasks has a weight of 0.3, the remaining CPU resource ratio has a weight of 0.4, and the remaining memory resource ratio has a weight of 0.3) to obtain the evaluation value of resource sufficiency. In the rate adaptability dimension, the historical processing rate of similar tasks is directly used as a basis, and the rate adaptability is obtained by comparing it with a set standard rate (e.g., the standard rate is set to 40MB / minute, and if the actual rate is greater than the standard rate, a higher score is given according to a certain proportion). In this way, the corresponding resource sufficiency and rate adaptability evaluation results are obtained for the remaining processing capacity vector of each node.

[0151] In this embodiment, a comprehensive score for the remaining processing capacity of each distributed resolution node is determined based on its resource sufficiency and rate adaptability. A weighted summation method is generally used; for example, if resource sufficiency is weighted at 0.6 and rate adaptability at 0.4, then the comprehensive score for remaining processing capacity = resource sufficiency × 0.6 + rate adaptability × 0.4. This method integrates the evaluation results of resource sufficiency and rate adaptability into a single comprehensive score, fully reflecting the remaining processing capacity of each node. A higher score indicates stronger remaining processing capacity for the node.

[0152] In this embodiment, the initial subtask splitting ratio of each distributed parsing node is determined based on a comprehensive score of its remaining processing capacity. Assuming the total number of protocol-signal collaborative parsing tasks is T, and the comprehensive score of a node's remaining processing capacity is... The total score of the remaining processing capacity of all nodes is: .So That is, the proportion of the initial subtasks that each node should undertake is determined based on the proportion of the remaining processing capacity of each node to the total remaining processing capacity of all nodes. The stronger the processing capacity of a node, the higher the proportion of subtasks it is initially assigned.

[0153] In this embodiment, based on the correlation matrix between the subtask allocation ratio and the parsing latency reduction rate, the initial subtask splitting ratio of each distributed parsing node is corrected to obtain the final subtask splitting ratio of each distributed parsing node. The correlation matrix between the subtask allocation ratio and the parsing latency reduction rate records the changes in the parsing latency reduction rate under different subtask allocation conditions in the past. For example, the elements in the matrix... Indicates the first The allocation ratio of each node is: The correlation between the parsing latency reduction rate and the initial subtask split ratio is determined. For each node, the corresponding correlation value is found in the correlation matrix based on its initial subtask split ratio. If the correlation value indicates that the parsing latency reduction rate under the current allocation ratio is not ideal (e.g., below a certain threshold), the initial subtask split ratio is adjusted. If the correlation is high, the subtask ratio of the node may be appropriately increased; if the correlation is low, it may be appropriately decreased. In this way, the initial subtask split ratio is corrected based on historical experience to obtain a more reasonable final subtask split ratio, thereby improving the overall parsing efficiency.

[0154] In this embodiment, the current protocol-signal collaborative parsing task is divided into subtasks and allocated to each distributed parsing node based on the final subtask splitting ratio of each distributed parsing node. It is assumed that the current protocol-signal collaborative parsing task can be split into multiple subtasks according to certain rules (such as by data volume, by device type, etc.). Based on the final subtask splitting ratio of each node, the number or range of subtasks to be allocated to each node is determined. For example, if a node's final subtask splitting ratio is 0.2, and the total task can be divided into 100 subtasks, then that node is allocated 20 subtasks. In this way, tasks are rationally allocated to each distributed parsing node, achieving effective allocation of parsing tasks in multi-device concurrent scenarios and improving overall parsing efficiency.

[0155] To generate a correlation matrix for revising task allocation ratios, a process for determining the correlation matrix between subtask allocation ratios and parsing latency reduction rate is proposed, including:

[0156] Obtain node allocation data and parsing efficiency data of historical similar protocol-sensor signal collaborative parsing tasks within a preset historical period, and construct a time series dataset. The node allocation data includes node ID and the number of allocated subtasks, and the parsing efficiency data includes parsing latency and verification pass rate.

[0157] Correlation analysis is performed on the upward, stationary, and downward trends of each sequence data in the time series dataset. The correlation between the subtask allocation ratio of each distributed parsing node and the parsing latency reduction rate is calculated, and a correlation matrix is ​​generated.

[0158] In this embodiment, the preset historical period refers to a pre-defined time period used to collect and analyze historical data. The selection of this time period needs to be determined based on the actual situation, and it should be able to cover a sufficient amount of historical data of the same type of task to ensure the reliability and representativeness of the analysis results. For example, it may be set to the past 6 months or 1 year.

[0159] In this embodiment, historical protocol-sensor signal collaborative parsing tasks refer to tasks performed in the past that are similar to the current protocol-sensor signal collaborative parsing task in terms of type, data characteristics, and parsing requirements. The data from these historical tasks, including node allocation and parsing efficiency information, is of significant reference value for optimizing the allocation of the current task and improving parsing efficiency.

[0160] In this embodiment, node allocation data and parsing efficiency data of historical similar protocol-sensor signal collaborative parsing tasks within a preset historical period are acquired, and a time-series dataset is constructed. Specifically, within the set preset historical period, allocation information of each distributed parsing node is collected for each execution of the same type of task, such as which nodes participate in the task and the workload of each node; this is the node allocation data. Simultaneously, parsing efficiency data for each task is collected, such as parsing latency and data verification pass rate. Then, these data are organized into a time-series dataset according to chronological order. For example, using one week as a time unit, the node allocation and parsing efficiency data of the same type of task are recorded weekly, forming a dataset arranged by time, facilitating observation and analysis of the changing patterns between task allocation and parsing efficiency over time.

[0161] In this embodiment, each sequence data refers to a record of node allocation data and parsing efficiency data arranged chronologically within a time-series dataset. Each set of data represents relevant information about historical collaborative parsing tasks of the same protocol and sensor signals at a specific point in time (or time period), such as the task allocation of each node within a week, along with corresponding parsing delays, verification pass rates, and other data, constituting a sequence data set. This sequence data forms the basis for subsequent trend analysis and correlation calculations.

[0162] In this embodiment, correlation analysis is performed on the upward, stable, and downward trends of each sequence data in the time series dataset. The correlation between the subtask allocation ratio of each distributed parsing node and the parsing latency reduction rate is calculated, and a correlation matrix is ​​generated. The specific steps are as follows: First, observe the changes in key indicators such as the parsing latency reduction rate over time for each sequence data to determine whether it shows an upward, stable, or downward trend. For example, if the parsing latency reduction rate gradually increases over time, it can be considered to show an upward trend. Then, for each distributed parsing node, analyze the relationship between the changes in its subtask allocation ratio and the changes in its parsing latency reduction rate at different time points. Quantify this correlation using specific statistical methods (such as calculating the correlation coefficient) to obtain the correlation degree between the subtask allocation ratio and the parsing latency reduction rate. Organize this correlation information for all nodes into a matrix, which is the correlation matrix. Each element in the matrix represents the degree of correlation between a certain node and the parsing latency reduction rate under a specific subtask allocation ratio. This matrix provides a basis for subsequent adjustments to the subtask allocation ratio, helping to optimize parsing efficiency.

[0163] In this embodiment, the parsing latency reduction rate refers to the percentage reduction in parsing latency compared to previous parsing scenarios during the parsing of historical similar protocol-sensor signal collaborative parsing tasks. The calculation method is as follows: For example, if parsing a task previously took 100 seconds, but now, with a certain task allocation method, parsing the same task only takes 80 seconds, then the parsing latency reduction rate = (100-80) ÷ 100 × 100% = 20%. The parsing latency reduction rate is an important indicator for measuring the improvement in parsing efficiency and plays a crucial role in analyzing the relationship between node allocation and parsing efficiency. A higher parsing latency reduction rate usually indicates a better task allocation method or parsing strategy.

[0164] To obtain a more reasonable final subtask splitting ratio, a method is proposed to correct the initial subtask splitting ratio of each distributed parsing node based on the correlation matrix, thereby obtaining the final subtask splitting ratio of each distributed parsing node, including:

[0165] The weighted sum of all matrix elements corresponding to the initial subtask splitting ratio and the allowable parsing latency reduction rate of each distributed parsing node in the correlation matrix is ​​taken as the strong correlation coefficient of the corresponding distributed parsing node.

[0166] Based on the strong correlation coefficient of each distributed resolution node, nodes to be optimized for allocation are selected from all distributed resolution nodes;

[0167] Based on the strong correlation coefficients of all nodes to be optimized, the initial subtask splitting ratios of all nodes to be optimized are corrected to obtain the final subtask splitting ratios of each distributed parsing node.

[0168] In this embodiment, the allowable parsing latency reduction rate is a pre-defined reference standard used to measure the acceptable degree of parsing latency reduction. It represents the expected or permissible percentage reduction in parsing latency under the current requirements and conditions of the parsing task. For example, the allowable parsing latency reduction rate might be set to ≥15%, meaning that during task allocation and parsing, a parsing latency reduction rate reaching or exceeding this value is considered to be in line with expectations. This standard provides an important metric for subsequent analysis and adjustment of task allocation.

[0169] In this embodiment, the weighted sum of all matrix elements in the correlation matrix corresponding to the initial subtask splitting ratio and the allowed parsing latency reduction rate of each distributed parsing node is used as the strong correlation coefficient of the corresponding distributed parsing node. Specifically, for each distributed parsing node, the row corresponding to the node's initial subtask splitting ratio in the correlation matrix is ​​found. Then, all matrix elements related to the allowed parsing latency reduction rate (i.e., the parsing latency reduction rate ≥ 15%) in that row are selected (these elements reflect the degree of correlation between the parsing latency reduction rate and task allocation under that initial subtask splitting ratio). These elements are assigned corresponding weights (the weights may be based on experience or judgment of the importance of different factors), and then they are summed to obtain the strong correlation coefficient of the distributed parsing node. For example, assuming that there are three matrix elements related to the allowed parsing latency reduction rate in the row corresponding to the initial subtask splitting ratio of a node, namely 0.6, 0.7, and 0.8, and the weights are set to 0.3, 0.4, and 0.3 respectively, then the strong correlation coefficient of the node = 0.6 × 0.3 + 0.7 × 0.4 + 0.8 × 0.3 = 0.7. This strong correlation coefficient reflects the overall correlation between the initial subtask splitting ratio of the node and the allowed resolution latency reduction rate. The higher the coefficient, the more beneficial the current subtask allocation method of the node is to achieving the allowed resolution latency reduction rate.

[0170] In this embodiment, nodes to be optimized for allocation are selected from all distributed parsing nodes based on the strong correlation coefficient of each node. Specifically, a threshold for the strong correlation coefficient is set (e.g., 0.6), and the strong correlation coefficient of each distributed parsing node is compared with this threshold. If a node's strong correlation coefficient is lower than the threshold, it indicates that the correlation between the node's current initial subtask splitting ratio and the allowed parsing latency reduction rate is not ideal, and its task allocation may need optimization. Such nodes are selected as nodes to be optimized for allocation. This selection method identifies which nodes' task allocation needs further adjustment to improve overall parsing efficiency.

[0171] In this embodiment, the initial subtask splitting ratio of all nodes to be optimized is adjusted based on the strong correlation coefficient of all nodes to be optimized, thus obtaining the final subtask splitting ratio of each distributed parsing node. For each selected node to be optimized, the initial subtask splitting ratio is adjusted according to its strong correlation coefficient.

[0172] Assume the pre-defined rules are as follows:

[0173] The ideal level for the strong correlation coefficient is set at 0.8, and the adjustment range coefficient is set at 0.1.

[0174] When the strong correlation coefficient is below 0.6:

[0175] ;

[0176] For example, if the strong correlation coefficient of a node to be optimized is 0.5, then the reduction in the initial subtask splitting ratio is (0.6-0.5)×0.1=0.01. If the initial subtask splitting ratio of this node is 0.2, then the corrected final subtask splitting ratio is 0.2-0.01=0.19.

[0177] When the strong correlation coefficient is between 0.6 (inclusive) and 0.8:

[0178] ;

[0179] For example, if the strong correlation coefficient of a node to be optimized is 0.7, then the increase in the initial subtask splitting ratio is (0.7-0.6)×0.1=0.01. If the initial subtask splitting ratio of this node is 0.15, then the corrected final subtask splitting ratio is 0.15+0.01=0.16.

[0180] In this way, based on the strong correlation coefficient of each node to be optimized, the initial subtask splitting ratio is adjusted according to the above rules, ultimately resulting in a more reasonable final subtask splitting ratio for each distributed parsing node. This helps improve overall parsing efficiency and achieves the allowed reduction in parsing latency. This rule, through a quantitative approach, adjusts the task allocation ratio with a fixed proportional coefficient based on the difference between the strong correlation coefficient and the ideal level, thereby optimizing the task allocation strategy.

[0181] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for optimizing IoT data parsing in sewing equipment compatible with multiple electronic controls, characterized in that: include: Based on the core protocol information of multiple manufacturers' electronic control systems and the physical signal characteristics of external sensors of sewing equipment, multiple standardized protocol-sensor signal association feature vectors are generated. All standardized protocol-sensor signal association feature vectors are clustered to construct a sewing equipment electronic control feature vector library. Among them, physical signal characteristics include sewing speed fluctuation range, stitch count statistics cycle, and signal sampling frequency. Based on the initial protocol signal of the current sewing equipment electronic control system and the real-time signal of the current sewing equipment sensor, a matching electronic control feature vector is generated. The matching electronic control feature vector is then matched with the sewing equipment electronic control feature vector library in two dimensions to select the target protocol template and sensor signal verification rules. Based on the historical parsing data of the target protocol template and sensor signal verification rules, a parsing efficiency optimization plane function is constructed, and the current multidimensional load state of all distributed parsing nodes is mapped to three-dimensional load affine coordinates; The distance between the analytic efficiency optimization plane function and the 3D load affine coordinates of each distributed analytic node is calculated. Based on the historical number of times each distributed parsing node processes the target protocol template and the average verification pass rate, the parsing adaptability score of each distributed parsing node is calculated. The distance between the plane function for optimizing parsing efficiency and the 3D load affine coordinates of each distributed parsing node is normalized to obtain the optimal parsing load fit for each distributed parsing node. Based on the parsing adaptability score and optimal parsing load adaptability of each distributed parsing node, the comprehensive adaptability score of each distributed parsing node is calculated. Based on the total data volume of the current protocol-sensor signal collaborative parsing task, the remaining processing capacity parameters of each distributed parsing node, and the correlation between node allocation data and parsing efficiency of similar historical tasks, the current protocol-sensor signal collaborative parsing task is divided into sub-tasks and allocated to each distributed parsing node. Each distributed parsing node is controlled to call the target protocol template to parse the electronic control protocol data, extract key data, and call the sensor signal verification rules to verify the consistency between the key data and the real-time signals of the current sewing equipment sensors. The parsing results that pass the verification are summarized as the sewing equipment IoT data parsing results. The distance between the analytical efficiency optimization planar function and the 3D load affine coordinates of each distributed analytical node is calculated using the distance formula from a spatial point to a plane.

2. The method for data parsing and optimization of IoT for sewing equipment compatible with multiple electronic controls as described in claim 1, characterized in that, Based on the core protocol information of multi-manufacturer electronic control systems and the physical signal characteristics of external sensors of sewing equipment, multiple standardized protocol-sensor signal association feature vectors are generated. All standardized protocol-sensor signal association feature vectors are clustered to construct a sewing equipment electronic control feature vector library, including: Simultaneously collect core protocol information from multiple manufacturers' electronic control systems and physical signal characteristics from external sensors of the sewing equipment, and generate multi-dimensional feature parameters; The multidimensional feature parameters are normalized to generate a standardized protocol-sensor signal correlation feature vector; The K-means clustering algorithm is used to cluster all standardized protocol-sensor signal associated feature vectors to obtain multiple clusters. The cluster center vector of each cluster is calculated and the cluster identifier and typical electronic control model are labeled. A feature vector library is built based on the cluster center vectors of all clusters, the standardized protocol-sensor signal association feature vectors, and the corresponding standardized protocol templates and sensor signal verification rules.

3. The method for data parsing and optimization of IoT for sewing equipment compatible with multiple electronic controls as described in claim 1, characterized in that, Based on the initial protocol signal of the current sewing equipment's electronic control system and the real-time signals of the current sewing equipment's sensors, a matching electronic control feature vector is generated. This feature vector is then matched in a two-dimensional manner against the sewing equipment's electronic control feature vector library to filter out target protocol templates and sensor signal verification rules, including: When the sewing equipment is powered on, the initial protocol signal of the current sewing equipment electronic control system and the real-time signal of the current sewing equipment sensor are collected, and the multi-dimensional feature parameters of the initial protocol signal and the real-time signal of the current sewing equipment sensor are extracted to generate the electronic control feature vector to be matched. Calculate the cosine similarity between the electronic control feature vector to be matched and the center vector of each cluster in the sewing equipment electronic control feature vector library, and select candidate clusters with similarity not less than a preset similarity threshold from the sewing equipment electronic control feature vector library; Calculate the Euclidean distance between the feature vector to be matched and each standardized protocol-sensor signal associated feature vector in the candidate cluster, and select the standardized protocol template and sensor signal verification rule corresponding to the standardized protocol-sensor signal associated feature vector with the smallest Euclidean distance as the target protocol template and sensor signal verification rule.

4. The method for data parsing and optimization of IoT for sewing equipment compatible with multiple electronic controls as described in claim 1, characterized in that, Based on historical parsing data of the target protocol template and sensor signal verification rules, a plane function for optimizing parsing efficiency is constructed, including: Obtain historical parsing data corresponding to the target parsing protocol template and sensor signal verification rules. The historical parsing data corresponding to the target parsing protocol template and sensor signal verification rules includes: historical load parameters and historical parsing performance indicators of a large number of historical distributed parsing nodes that conform to the target parsing protocol template and sensor signal verification rules. Based on the historical load parameters and historical parsing performance indicators of a large number of distributed parsing nodes, multi-dimensional historical load feature vectors and multi-dimensional historical parsing performance indicator vectors are generated for each historical distributed parsing node. A parsing load-performance correlation dataset is then constructed based on the multi-dimensional historical load feature vectors and multi-dimensional historical parsing performance indicator vectors of each historical distributed parsing node. Cluster the multidimensional historical load feature vectors in the parsed load-performance correlation dataset to obtain multiple parsed load-performance correlation clusters; Based on the multi-dimensional historical parsing performance index vector of each historical distributed parsing node in each parsing load-performance correlation cluster, the comprehensive performance value of each parsing load-performance correlation cluster is calculated. Among all the parsing load-performance correlation clusters, the parsing load-performance correlation cluster corresponding to the maximum comprehensive performance value is regarded as the optimal performance cluster; The multidimensional historical parsing performance index vectors of all historical distributed parsing nodes in the optimal performance cluster are uniformly reduced in dimensionality to obtain the three-dimensional key load quantification indexes of each historical distributed parsing node in the optimal performance cluster. Based on the three-dimensional key load quantification index of all historical distributed parsing nodes of the optimal performance cluster, a plane function for optimizing parsing efficiency is fitted.

5. The method for data parsing and optimization of IoT for sewing equipment compatible with multiple electronic controls as described in claim 4, characterized in that, The multidimensional historical parsing performance index vectors of all historical distributed parsing nodes in the optimal performance cluster are uniformly reduced in dimensionality to obtain the three-dimensional key load quantification indicators of each historical distributed parsing node in the optimal performance cluster, including: A historical parsing performance index matrix is ​​constructed based on the multi-dimensional historical parsing performance index vector of all historical distributed parsing nodes in the optimal performance cluster. Principal component analysis was used to reduce the dimensionality of the historical parsing performance index matrix, resulting in a three-dimensional key load quantification index for all historical distributed parsing nodes.

6. The method for data parsing and optimization of IoT for sewing equipment compatible with multiple electronic controls as described in claim 1, characterized in that, Map the current multidimensional load state of all distributed parsing nodes to three-dimensional load affine coordinates, including: Collect the current multi-dimensional load status of all distributed parsing nodes; The current multidimensional load status of all distributed parsing nodes is uniformly reduced to obtain three-dimensional key load quantification indicators. Three-dimensional load affine coordinates are generated based on the three-dimensional key load quantization metrics of all distributed parsing nodes.

7. The method for data parsing and optimization of IoT for sewing equipment compatible with multiple electronic controls as described in claim 1, characterized in that, Based on the total data volume of the current protocol-sensor signal collaborative parsing task, the remaining processing capacity parameters of each distributed parsing node, and the correlation between node allocation data and parsing efficiency in historical similar tasks, the current protocol-sensor signal collaborative parsing task is divided into sub-tasks and allocated to each distributed parsing node, including: Collect the remaining processing capacity parameters of each distributed parsing node, and generate a remaining processing capacity vector based on the remaining processing capacity parameters of each distributed parsing node. The remaining processing capacity parameters include the number of remaining tasks, the percentage of remaining CPU resources, the percentage of remaining memory resources, and the historical processing rate of similar tasks. The remaining processing capacity vector of each distributed parsing node is evaluated to obtain the resource sufficiency and rate adaptability of each distributed parsing node, and the comprehensive score of the remaining processing capacity of each distributed parsing node is determined based on the resource sufficiency and rate adaptability of each distributed parsing node. Based on the comprehensive score of the remaining processing capacity of each distributed parsing node, the initial subtask splitting ratio of each distributed parsing node is determined. Based on the correlation matrix between the subtask allocation ratio and the parsing latency reduction rate, the initial subtask splitting ratio of each distributed parsing node is corrected to obtain the final subtask splitting ratio of each distributed parsing node. Based on the final subtask splitting ratio of each distributed parsing node, the current protocol-sensor signal collaborative parsing task is split into subtasks and assigned to each distributed parsing node.

8. The method for data parsing and optimization of IoT for sewing equipment compatible with multiple electronic controls as described in claim 7, characterized in that, The process of determining the correlation matrix between the subtask allocation ratio and the parsing latency reduction rate includes: Obtain node allocation data and parsing efficiency data of historical similar protocol-sensor signal collaborative parsing tasks within a preset historical period, and construct a time series dataset. The node allocation data includes node ID and the number of allocated subtasks, and the parsing efficiency data includes parsing latency and verification pass rate. Correlation analysis is performed on the upward, stationary, and downward trends of each sequence data in the time series dataset. The correlation between the subtask allocation ratio of each distributed parsing node and the parsing latency reduction rate is calculated, and a correlation matrix is ​​generated.

9. The method for data parsing and optimization of IoT for sewing equipment compatible with multiple electronic controls as described in claim 7, characterized in that, The initial subtask splitting ratios of each distributed parsing node are adjusted based on the correlation matrix to obtain the final subtask splitting ratios of each distributed parsing node, including: The weighted sum of all matrix elements corresponding to the initial subtask splitting ratio and the allowable parsing latency reduction rate of each distributed parsing node in the correlation matrix is ​​taken as the strong correlation coefficient of the corresponding distributed parsing node. Based on the strong correlation coefficient of each distributed resolution node, nodes to be optimized for allocation are selected from all distributed resolution nodes; Based on the strong correlation coefficients of all nodes to be optimized, the initial subtask splitting ratios of all nodes to be optimized are corrected to obtain the final subtask splitting ratios of each distributed parsing node.

Images