Data processing method and system based on embedded industrial control screen

Abstract

The invention relates to a data processing method and system based on an embedded industrial control screen, and the method comprises the steps: firstly carrying out the segmentation processing of multi-source heterogeneous sensing data, then calculating the statistical feature data of the segmented data, determining the data processing priority according to the statistical feature data, carrying out the priority processing of part of data, and carrying out the data processing. And the effect of real-time data processing can be achieved by fully utilizing the limited computing power of the embedded industrial control screen. Moreover, a virtual processing unit suitable for an embedded industrial control screen application scene is used to process to-be-processed data into a multi-dimensional feature vector, then component data of the multi-dimensional feature vector is advanced, and then a visual rendering engine is used to process the component data into real-time trend data which can be visually displayed. The data related to the change trend of the sensor data can be quickly extracted from the to-be-processed data and converted into the data capable of being visually displayed, so that the real-time visual display requirement of an embedded industrial control screen on the change trend of the multi-source heterogeneous data is met.

Description

Technical Field

[0001] The present invention relates to the field of industrial automation technology, and particularly to a data processing method and system based on an embedded industrial control screen. Background Art

[0002] With the development of the industrial automation field, the embedded industrial control screen technology has emerged. An embedded industrial control screen is a computer device dedicated to industrial control systems. It is usually embedded in machine equipment for monitoring and managing the production process.

[0003] Due to the constraints of software and hardware, the computing power of traditional embedded industrial control screens is limited and it is difficult to meet the real-time analysis and visualization display requirements of multi-source data. And the efficient processing and analysis of data are crucial for improving production efficiency and decision-making quality in the field. Therefore, a method is needed to break this dilemma. Summary of the Invention

[0004] Based on this, in view of the above technical problems, it is necessary to provide a data processing method and system that can meet the real-time analysis requirements of multi-source data of embedded industrial control screens.

[0005] In a first aspect, the present invention provides a data processing method based on an embedded industrial control screen, the method comprising: Obtaining multi-source heterogeneous sensing data; Processing the multi-source heterogeneous sensing data through a sliding window technique to obtain segmented data; calculating statistical feature data of the segmented data, the statistical feature data constituting first statistical data; Establishing a task priority list according to the first statistical data and a preset priority rule; Converting the segmented data into a standard format data set; Performing feature point annotation on the standard format data set to obtain an annotated data set; Processing the annotated data set using a virtual processing unit according to the task priority list to obtain a multi-dimensional feature vector; Extracting component data of the multi-dimensional feature vector, the component data including a trend component and a seasonal component; Using a visualization rendering engine to process the component data in real time to obtain real-time trend data.

[0006] In one embodiment, the processing the annotated data set using a virtual processing unit according to the task priority list to obtain a multi-dimensional feature vector includes the following steps: Judging whether the data volume of the annotated data set exceeds a first preset threshold to obtain a judgment result; When the judgment result is that the data volume exceeds the first preset threshold, divide the labeled data set according to the task priority list and the preset partitioning rule to obtain partitioned data sets; Allocate multiple virtual processing units to extract the feature data of each partitioned data set in parallel; According to the feature data, use the principal component analysis algorithm to process the feature data to obtain a multi-dimensional projection; Use a preset projection template to extract the key dimension information of the multi-dimensional projection to obtain a multi-dimensional feature set; Use data integration and vector splicing to process the multi-dimensional feature set to obtain a multi-dimensional feature vector.

[0007] In one embodiment, the extraction of the component data of the multi-dimensional feature vector includes: According to the preset sharding rule, use a streaming computing framework to perform real-time sharding processing on the multi-dimensional feature vector to obtain data segments; Perform parallel computing on each data segment to obtain second statistical data, where the second statistical data includes the data change amplitude; When the data change amplitude exceeds the second preset threshold, use a time series decomposition algorithm to decompose the second statistical data to obtain decomposed data; Use the moving average method to extract the component data from the decomposed data.

[0008] In one embodiment, the use of the visualization rendering engine to process the component data in real time includes: According to the preset visualization template, convert the component data into visualization feature data through data mapping technology; Use the rendering engine to draw the visualization feature data in real time and dynamically refresh the rendering result to obtain real-time trend data.

[0009] In one embodiment, the method further includes: Obtain historical trend data; Calculate the cosine similarity between the real-time trend data and the historical trend data to obtain a first cosine similarity; When the first cosine similarity is lower than the third preset threshold, update the historical trend data set according to the real-time trend data to obtain an updated historical trend data set.

[0010] In one embodiment, the method further includes: Construct a real-time trend feature vector set according to the real-time trend data; Use the K-means algorithm to perform clustering processing on the historical trend data set to obtain a clustering result, where the clustering result includes a dynamic trend model and clustering center point data; Construct a historical trend feature vector set according to the clustering result; Calculate the Euclidean distance between the real-time trend feature vector set and the historical trend feature vector set to obtain the first Euclidean distance; When the first Euclidean distance is greater than the fourth preset threshold, calculate the difference vector between the real-time trend feature vector and the historical trend feature vector, and the difference vector constitutes a trend comparison data set; Process the trend comparison data set using vector graphics acceleration technology, a preset rendering frequency, and a preset scaling ratio parameter to obtain a first dynamic trend chart.

[0011] In one embodiment, the method further includes: Obtain real-time interaction data; Extract feature point data and metadata from the real-time interaction data through event listening technology, and the feature point data and the metadata constitute an interaction event feature set; Perform clustering processing on the feature point data using the K-means algorithm to obtain zoom focus data and annotation clarity data, and the zoom focus data and the annotation clarity data constitute an interaction response parameter set; When the interaction response parameter set does not match the preset rendering parameter threshold, adjust the zoom focus data and the annotation clarity data through a preset algorithm, and the adjusted zoom focus data and the adjusted annotation clarity data constitute an interaction configuration data set; Perform parameter mapping on the interaction configuration data set and the first dynamic trend chart to obtain interaction-enhanced trend data, and the interaction-enhanced trend data contains trend information of the interaction configuration.

[0012] In one embodiment, the method further includes: Obtain preset style configuration data and resolution adaptation requirement data; Adjust the style configuration data according to the resolution adaptation requirement data through a CSS preprocessor to obtain a multi-resolution adaptation style sheet, and the multi-resolution adaptation style sheet constitutes a dynamic style configuration set; According to the dynamic style configuration set and the rendering frequency, perform multi-resolution rendering on the trend comparison data set through a WebGL renderer to obtain rendering data including icon dynamic information; the multi-resolution rendering includes: when the rendering frequency is lower than the fifth preset threshold, adjust the frequency synchronization parameter; Use the K-means algorithm to cluster the feature points of the rendering data to obtain feature point annotation data; Generate a second dynamic trend chart including feature point annotations according to the scaling ratio parameter and the feature point annotation data through vector graphics drawing technology, and the second dynamic trend chart constitutes visual analysis report data.

[0013] In one embodiment, the method further includes: Extracting core data from the visual analysis report data; Compressing the core data using the LZ77 compression algorithm to obtain compressed data; Storing the compressed data in a specified database; Judging the storage stability of the compressed data in the specified database to obtain a judgment result, where the judgment result includes whether the storage stability reaches a sixth preset threshold; Generating a storage report, where the storage report includes a write success rate, a time consumption, the core data, and the judgment result.

[0014] In one embodiment, the method further includes: Receiving a data query instruction and a resolution parameter of a target terminal; Extracting the core data in the storage report through an interactive query interface according to the data query instruction; Performing structured processing on the core data using JSON parsing technology to obtain structured query data; Adjusting the output format style parameters of the structured query data through CSS media query according to the resolution parameter to obtain first display data adapted to the terminal; Calculating zoom focus coordinates using an interactive zoom technology; Generating a dynamic trend chart including the zoom focus coordinates through Canvas drawing technology to obtain interactive enhanced chart data; If the frame rate of the interactive enhanced chart data is lower than a seventh preset threshold, adjusting the rendering frequency parameter of the interactive enhanced chart data through a WebGL renderer to obtain terminal output data; Outputting the terminal output data to the target terminal.

[0015] In a second aspect, the present invention provides a data processing system based on an embedded industrial control screen, and the system includes: A first data acquisition module, configured to acquire multi-source heterogeneous sensing data; A segmented calculation module, configured to process the multi-source heterogeneous sensing data through a sliding window technology to obtain segmented data; calculating statistical feature data of the segmented data, where the statistical feature data constitutes first statistical data; A priority module, configured to establish a task priority list according to the first statistical data and a preset priority rule; A format conversion module, configured to convert the segmented data into a standard format data set; A data annotation module, used to perform feature point annotation on the standard format data set to obtain an annotated data set; A virtual unit module, used to process the annotated data set using virtual processing units according to the task priority list to obtain multi-dimensional feature vectors; A component extraction module, used to extract the component data of the multi-dimensional feature vector, and the component data includes a trend component and a seasonal component; A data rendering module, used to use a visualization rendering engine to process the component data in real time to obtain real-time trend data.

[0016] In the above data processing method based on an embedded industrial control screen, the multi-source heterogeneous sensing data is first segmented, and then the statistical feature data of the segmented data is calculated. According to the statistical feature data, the data processing priority is determined, and part of the data is processed preferentially, which helps to make full use of the limited computing power of the embedded industrial control screen to achieve the effect of real-time data processing. Moreover, by using virtual processing units suitable for the application scenario of the embedded industrial control screen, the data to be processed is processed into multi-dimensional feature vectors, and then the component data of the multi-dimensional feature vectors is extracted in advance. Then, the visualization rendering engine is used to process the component data into real-time trend data that can be visually displayed, which can quickly extract the data related to the change trend of the sensor data from the data to be processed and convert it into data that can be visually displayed, thus meeting the real-time visualization display requirements of the embedded industrial control screen for the change trend of multi-source heterogeneous data. Description of the Drawings

[0017] The drawings here are incorporated into the specification and form a part of this specification, showing the embodiments in line with the present invention, and are used together with the specification to explain the principles of the present invention, and do not constitute an improper limitation to the present invention.

[0018] Figure 1 It is a schematic flowchart of the data processing method of the embedded industrial control screen in an embodiment; Figure 2 It is a schematic flowchart of step S106 in an embodiment; Figure 3 It is a schematic flowchart of step S107 in an embodiment; Figure 4 It is a schematic structural diagram of the data processing system of the embedded industrial control screen in an embodiment. Detailed Embodiments

[0019] In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention and are not used to limit the present invention.

[0020] It should be noted that the terms "first", "second", etc. in the description, claims and above-mentioned drawings of the present invention are used to distinguish similar objects, and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present invention described here can be implemented in an order other than those illustrated or described here. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. On the contrary, they are merely examples of systems and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0021] In one embodiment, as Figure 1 shown, a data processing method based on an embedded industrial control screen is provided, and the method includes the following steps: S101, obtaining multi-source heterogeneous sensing data.

[0022] Among them, multi-source heterogeneous sensing data may refer to data from different sensors with different structures and formats.

[0023] Specifically, the system may have a data receiving end, which can receive data from a variety of sensors. There are many types of sensors, such as temperature sensors, humidity sensors, pressure sensors, etc. The data formats of sensors are often not unified. For example, a temperature sensor outputs JSON format (JSON is the English abbreviation of JavaScript Object Notation, indicating a lightweight data exchange format), and a pressure sensor outputs XML format (XML is the English abbreviation of Extensible Markup Language, indicating a plain text file).

[0024] S102, processing the multi-source heterogeneous sensing data through a sliding window technique to obtain segmented data; calculating statistical feature data of the segmented data, and the statistical feature data constitutes first statistical data.

[0025] Specifically, the sliding window technique is an existing flow control technique that moves a fixed-size window over an array or string, enabling a series of data to be processed in a relatively short time, which is particularly efficient. Using the sliding window technique to process the multi-source heterogeneous sensing data can improve the data processing speed and meet the real-time processing requirements of the industrial control screen data as much as possible. The multi-source heterogeneous sensing data is segmented using the sliding window technique to obtain segmented multi-source heterogeneous sensing data. The statistical features of the segmented data are calculated, and the statistical features can include at least one of data statistical values such as the average value, variance, maximum value, minimum value, median value, etc. Specifically, the size of the sliding window can be set as needed, for example, it can be 3 seconds, 5 seconds, 8 seconds, 10 seconds, etc., and the step size can be 1 second, 2 seconds, etc.

[0026] It should be noted that before processing the multi-source heterogeneous sensing data through the sliding window technique, some common data preprocessing can be performed on the multi-source heterogeneous sensing data. For example, when the formats of the multi-source heterogeneous sensing data are inconsistent, a preset standardization protocol can be used to unify the formats of the multi-source heterogeneous sensing data. For example, the MQTT (MQTT is the English abbreviation of Message Queuing Telemetry Transport, representing Message Queuing Telemetry Transport) protocol can be used as the standardization protocol to convert different format data into a unified JSON format. For example, when the multi-source heterogeneous sensing data is relatively chaotic, the timestamp synchronization method can be used to align the multi-source heterogeneous sensing data. Specifically, the NTP protocol (NTP is the English abbreviation of Network Time Protocol, representing Network Time Protocol) can be selected to add a unified timestamp to each data point to ensure data alignment and obtain a synchronized data stream, thus facilitating the use of the sliding window technique.

[0027] S103. Establish a task priority list according to the first statistical data and the preset priority rules.

[0028] Among them, the priority rules can refer to the rules that can be used to judge the priority of data processing.

[0029] Specifically, since the first statistical data is segmented, obviously the data content of different segments is different, and the processing priorities are different. Even some of the data in the segments do not need to be processed. Therefore, certain priority rules can be preset so that the first statistical data of different segments has different processing priorities, forming a priority list for data processing tasks. For example, the processing priority can be determined according to the urgency or importance of data processing. For instance, the segmented data with abnormal temperature fluctuations has a higher priority, and a task priority list can be generated based on the abnormal temperature fluctuation situation. This method can effectively identify critical tasks, improve the system response efficiency, and help to quickly meet the data processing requirements by invoking limited computing resources.

[0030] S104, convert the segmented data into a standard format data set.

[0031] Specifically, the segmented data may include structured data and unstructured data. For example, the structured data may be the JSON format output of a temperature sensor, including fields such as "temperature value" and "timestamp"; the unstructured data may be the text description of a log record, including natural language information about the device operating status. Preset parsing rules can be used to perform field mapping on the structured data. For the unstructured data, semantic units can be extracted through text segmentation technology to obtain a preliminary extraction data set. For the preliminary extraction data set, a unified data standardization protocol is used to perform format conversion on the heterogeneous data therein. If the format of the heterogeneous data does not conform to the preset specification, the data structure is adjusted through field completion technology to obtain the standard format data set.

[0032] In a possible implementation, the parsing rules are defined based on JSONSchema (which represents a JSON-based declarative language), mapping the "temperature value" field to the unified field name "temp_value", and the timestamp field to "timestamp". If the data lacks key fields, such as the humidity data not including "unit", it is completed by the rules and a default value such as "%" is added. This method ensures field consistency and facilitates subsequent processing. For unstructured data, semantic units can be extracted through text segmentation technology. For example, the log text "Device A has abnormal temperature for 10 minutes" can be segmented by natural language processing tools to extract semantic units such as "Device A", "abnormal temperature", and "10 minutes".

[0033] In one embodiment, a dictionary-based word segmentation method is adopted, combined with a preset keyword library, to identify key entities and actions, and generate a set of structured semantic units. This method effectively converts unstructured data into a processable format. According to the preliminary extracted dataset, a unified data standardization protocol is adopted for format conversion. For example, Protobuf is selected as the standardization protocol to convert heterogeneous data in JSON and XML formats into Protobuf format. If the data format does not conform to the specification, such as the XML data missing the required field "sensor ID", then through the field completion technology, the ID value is inferred and added based on the device configuration file. This process ensures the uniformity of the data format and reduces the complexity of subsequent processing. There are many similar data format conversion methods in the prior art, and those skilled in the art can choose according to actual needs, which will not be elaborated here.

[0034] S105, perform feature point annotation on the standard format dataset to obtain an annotated dataset.

[0035] Specifically, for the standard format dataset, there are various methods to achieve feature point annotation. For example, a key point detection algorithm can be used to annotate key dimension features. For example, in temperature data, the key dimensions are "temperature value" and "timestamp". In one possible implementation, a threshold-based detection algorithm is used to annotate the data points with a temperature value exceeding 50°C as "abnormal", and the annotation information is recorded through a preset template, such as {"point_id":1,"type":"abnormal","value":52.3,"timestamp":"2025-05-09 10:00:00"}. This method facilitates the subsequent analysis of key data points. For the feature annotated dataset, data integration techniques can also be used for clustering. For example, based on the K-means clustering algorithm, the annotated data is classified into three categories: "slight", "moderate", and "severe" according to the degree of temperature abnormality. In one embodiment, the preset clustering rule is: the temperature value is slight at 50 - 60°C, moderate at 60 - 70°C, and severe above 70°C. Finally, an annotated dataset is generated, including the clustered annotation information, such as {"cluster_id":1,"type":"severe","points":[point_id1,point_id2]}. This method effectively integrates the data and facilitates the system to identify and process key events. It can be understood that the above methods form a complete processing chain from data extraction to clustering. Each link supports each other to ensure the efficient and accurate conversion of data from the original stream to the final annotated dataset, which is applicable to the real-time data processing scenario of embedded systems.

[0036] S106, according to the task priority list, use a virtual processing unit to process the annotated dataset to obtain a multi-dimensional feature vector.

[0037] Among them, the virtual processing unit can refer to the logical unit representing processing capabilities in a virtual environment.

[0038] Specifically, according to the task priority list, one or more virtual processing units can be used to process the corresponding labeled data sets in order of priority. According to the task priority list, a heuristic algorithm can be used for task allocation. In one embodiment, a genetic algorithm is used to optimize the allocation, and high-priority tasks are preferentially allocated to idle virtual processing units. For example, the temperature anomaly processing task is allocated to Unit 1. If the resource occupancy of a certain unit exceeds the threshold, such as 80%, the low-priority task is migrated to Unit 2 to generate a task allocation table. It can be understood that this dynamic adjustment mechanism avoids resource bottlenecks and improves system stability. For example, in an industrial monitoring scenario, an embedded system needs to process sensor data in real time. Through the above method, the system can efficiently allocate resources, preferentially process critical anomalies, and ensure monitoring reliability. The implementation of each technical theme revolves around efficient resource utilization and real-time data processing, supporting each other to form a complete solution, significantly improving the performance of the embedded system.

[0039] In addition, regarding the construction of virtual processing units, a preset lightweight virtual framework can be used to deploy a virtualization environment in the embedded industrial control screen system to obtain multiple virtual processing units for efficient resource management. For example, the LXC (Linux Container) container technology is adopted, combined with a preset configuration file. The configuration file contains the number of virtual processing units, such as 4, and resource limit parameters, such as 2GB of memory and 1 CPU core are allocated to each unit. Through LXC, 4 virtual processing units are quickly created to form a set of virtual processing units. Selecting this method can ensure resource isolation, reduce system overhead, and improve the resource utilization rate of embedded devices. It is widely used in the prior art and will not be elaborated here.

[0040] S107, extract the component data of the multi-dimensional feature vector, and the component data includes a trend component and a seasonal component.

[0041] Among them, the trend component can refer to the component representing the long-term increase or decrease of data, which can be linear or non-linear. The seasonal component can refer to the component representing the periodic fluctuation of data.

[0042] Specifically, depending on the type and change characteristics of the sensor data, there can be multiple types of trend components. For example, it can be a temperature rising trend, a temperature falling trend, a humidity rising trend, a humidity falling trend, a pressure rising trend, a pressure falling trend, and so on. Similarly, there can be multiple types of seasonal components. For example, daily temperature cyclic changes, daily humidity cyclic changes, annual pressure cyclic changes, and so on. For a multi-dimensional feature vector, multiple methods can be used to extract the component data therein. For example, the multi-dimensional feature vector can be first decomposed and then the data can be extracted to obtain the component data.

[0043] S108, Use a visualization rendering engine to process the component data in real time to obtain real-time trend data.

[0044] Among them, the visualization rendering engine can refer to a software tool used to create and present three-dimensional graphics and scenes in a computer. The real-time trend data can refer to data that can be visually displayed and contains data change trend information.

[0045] Specifically, the component data can be first converted into a visualization data structure, and then a visualization rendering engine applicable to the industrial control screen application scenario can be used to perform real-time drawing on the visualization data structure, dynamically refresh the rendering result, and obtain real-time trend data.

[0046] In the above data processing method based on an embedded industrial control screen, the multi-source heterogeneous sensing data is first segmented, and then the statistical feature data of the segmented data is calculated. According to the statistical feature data, the data processing priority is determined, and some data is preferentially processed, which helps to make full use of the limited computing power of the embedded industrial control screen to achieve the effect of real-time data processing. Moreover, by using a virtual processing unit suitable for the embedded industrial control screen application scenario, the data to be processed is processed into a multi-dimensional feature vector, and then the component data of the multi-dimensional feature vector is extracted in advance. Then, a visualization rendering engine is used to process the component data into real-time trend data that can be visually displayed, which can quickly extract data related to the change trend of the sensor data from the data to be processed and convert it into data that can be visually displayed, thus meeting the real-time visualization display requirements of the embedded industrial control screen for the change trend of multi-source heterogeneous data.

[0047] In one embodiment, as Figure 2 shown, step S106 includes the following steps: S201, Determine whether the data volume of the labeled data set exceeds a first preset threshold to obtain a judgment result.

[0048] S202, When the judgment result is that the data volume exceeds the first preset threshold, split the labeled data set according to the task priority list and the preset partitioning rule to obtain a partitioned data set.

[0049] S203, allocate multiple virtual processing units to extract the feature data of each partition data set in parallel.

[0050] S204, according to the feature data, use the principal component analysis algorithm to process the feature data to obtain a multi-dimensional projection.

[0051] S205, use a preset projection template to extract the key dimension information of the multi-dimensional projection to obtain a multi-dimensional feature set.

[0052] S206, use data integration and vector splicing to process the multi-dimensional feature set to obtain a multi-dimensional feature vector.

[0053] Among them, the first preset threshold may refer to a preset threshold representing the data volume size. The principal component analysis algorithm is a commonly used dimensionality reduction algorithm, which maps high-dimensional data to a low-dimensional space through multi-dimensional projection and retains the main information.

[0054] Specifically, the first preset threshold can be set according to actual needs, and the present invention does not make specific limitations. For example, the first preset threshold can be 10 4 ~10 5A certain value in the data. For the case where the data volume of the labeled dataset exceeds a preset threshold, the labeled dataset can be split into multiple data partitions according to the task assignment table, and the partitioned dataset, that is, the partitioned dataset, can be generated through a preset partitioning rule. The preset partitioning rule can be set according to the specific data situation. For example, a partition can be generated for each hour of data. For the partitioned dataset, virtualization technology is used to allocate each data partition to the corresponding virtual processing unit, and the virtual processing unit can be synchronously processed through a parallel computing framework to extract the feature data therein. According to the feature data, the principal components of the principal component analysis algorithm are determined, and thus the principal component analysis algorithm is used to perform multi-dimensional projection processing on the data of the virtual processing unit, and the key dimension information is extracted through a preset projection template to obtain a multi-dimensional feature set. For the multi-dimensional feature set, the multi-dimensional feature set is merged through data integration technology, and the vector splicing technology is used to generate the final feature vector to obtain the feature vector dataset. For example, in the data processing scenario of an embedded system, the processing of the labeled dataset needs to handle large-scale data and high-concurrency requirements. If the data volume of the labeled dataset exceeds the preset threshold, such as 1 million records, it needs to be split through a partitioning strategy. In one possible implementation, based on the task priority list, the dataset is split according to the sensor type. For example, the temperature and humidity sensor data form sub-datasets respectively. The partitioning rule can be based on the timestamp range. For example, a partition is generated for each hour of data to obtain a partitioned dataset, such as the "temperature_2025-05-09_10:00" subset. This method facilitates subsequent parallel processing. For the partitioned dataset, virtualization technology is used to allocate resources. Specifically, each data partition is mapped to an independent virtual processing unit, such as a virtual machine or a container. In one embodiment, Docker containers (Docker containers represent an open-source application container engine) are used to allocate computing resources for each partition, and the Spark (Spark represents a memory-based parallel computing framework) parallel computing framework is used for synchronous processing. For example, the temperature sub-dataset is processed in container A, and the humidity sub-dataset is processed in container B. Each container extracts feature data in parallel, such as the average temperature value. According to the feature data, the principal component analysis algorithm is used for dimensionality reduction and key information extraction. In one possible implementation, the temperature and humidity data are projected into a two-dimensional space, and key dimensions such as "temperature mean" and "humidity variance" are extracted. The projection template can be preset to retain the dimensions with 90% variance to generate a multi-dimensional feature set. This method effectively simplifies the data complexity and helps to improve the processing efficiency.

[0055] For a multi-dimensional feature set, data integration techniques are used for combined processing. Preferably, vector splicing techniques can be employed to integrate each multi-dimensional feature set into a multi-dimensional feature vector. For example, the temperature feature vector [25.5, 0.8] and the humidity feature vector [60, 1.2] are spliced into [25.5, 0.8, 60, 1.2]. In one embodiment, the features are normalized before splicing to ensure a consistent numerical range, such as from 0 to 1. This process facilitates subsequent analysis. An efficient processing chain is formed during the process, with each link supporting each other to ensure the rapid processing of large-scale data in an embedded system. Partitioning reduces the pressure on a single node, virtualization and parallel computing improve efficiency, principal component analysis streamlines the data, and feature splicing ensures information integrity. This process is applicable to real-time data analysis scenarios.

[0056] In one embodiment, as Figure 3 shown, step S107 includes: S301, according to a preset sharding rule, use a streaming computing framework to perform real-time sharding processing on the multi-dimensional feature vector to obtain data segments.

[0057] S302, perform parallel computing on each data segment to obtain second statistical data, where the second statistical data includes the data change amplitude.

[0058] S303, when the data change amplitude exceeds a second preset threshold, use a time series decomposition algorithm to decompose the second statistical data to obtain decomposition data.

[0059] S304, use the moving average method to extract component data from the decomposition data.

[0060] Among them, the streaming computing framework can refer to a computing architecture capable of processing real-time data streams. The time series decomposition algorithm can refer to an algorithm capable of analyzing and predicting time series data. The moving average method is a common statistical method. The second preset threshold can refer to a preset threshold representing the data change amplitude.

[0061] Specifically, for the multi-dimensional feature vector, the real-time data stream is fragmented by a streaming computing framework and divided into multiple data segments according to a preset fragmentation rule. The parallel processing technology is used to synchronously calculate the data segments to calculate the data change amplitude and obtain the second statistical data. The second statistical data may also include mean data. If the data change amplitude exceeds the second preset threshold, the time series decomposition algorithm is used to decompose the second statistical data to obtain the decomposed data. The second preset threshold may be a percentage or a specific change value, which can be set according to actual needs and is not specifically limited in the present invention. For example, for the change in temperature data, the second preset threshold may be a 20% increase in temperature or a 2°C increase in temperature. The component data, such as the trend component and the seasonal component, is extracted by the moving average method, and the component data may form a component data set.

[0062] Exemplarily, in the real-time data processing scenario of an embedded system, the processing of multi-dimensional feature vectors needs to cope with the requirements of high concurrency and dynamic changes. Exemplarily, by fragmenting the real-time data stream through a streaming computing framework, the high speed and continuity of the data stream can be effectively coped with. A streaming computing framework, such as Apache Flink (the English name of an open-source stream processing framework), can be used to split the sensor data stream by time window. For example, a data segment is generated every 5 seconds. Specifically, the temperature sensor data stream is divided into segments according to timestamps, such as "temperature_2025-05-09_10:00:00-10:00:05". This fragmentation method facilitates subsequent parallel processing. It should be noted that the synchronous calculation of the data segments by the parallel processing technology can improve the processing efficiency. In one possible implementation, the multi-threading technology is used to allocate independent computing threads for each data segment. For example, segment A is processed by thread 1, and segment B is processed by thread 2. Each thread synchronously calculates the statistical features of the segment, such as the temperature mean. Preferably, the thread pool management technology can dynamically adjust the number of threads to ensure resource utilization. For the dynamic change of the second statistical data, if the change amplitude exceeds the preset dynamic threshold, such as the temperature mean fluctuates by more than 2 degrees Celsius, further processing is required. It can be understood that the time series decomposition algorithm can effectively extract the trends and laws of the data. In one embodiment, the moving average method is used to decompose the data, the window size is set to 10 seconds, and the trend component, such as the temperature rising trend, and the seasonal component, such as the daily temperature cycle change, are extracted. This decomposition method facilitates the analysis of the long-term change pattern of the data.

[0063] In one embodiment, step S108 includes: S401, according to a preset visualization template, convert the component data into visualization feature data through data mapping technology.

[0064] S402. Use the rendering engine to draw the visualized feature data in real time, dynamically refresh the rendering result, and obtain the real-time trend data.

[0065] Specifically, according to the component data set, project the component data onto a preset visualization template through data mapping technology, and use a vector conversion algorithm to generate a visualization data structure to obtain the visualized feature data. For the visualized feature data, use the rendering engine to draw the visualized feature data in real time, and use a dynamic refresh technology to update the rendering result to obtain the real-time trend data. For example, the trend component is mapped to the Y-axis of a line chart, and the seasonal component is mapped to the color intensity. The vector conversion algorithm can convert the data into a visualization data structure, such as a coordinate point set in JSON format, which is convenient for rendering. Exemplarily, the temperature trend data is converted into [{time: 10:00:05, value: 25.5}, {time: 10:00:10, value: 25.7}]. Specifically, the rendering engine draws the visualization data structure in real time. The rendering engine can use WebGL technology to efficiently render a dynamic line chart. The dynamic refresh technology can include updating the image at a frequency of once per second. The temperature trend line chart extends in real time as new data flows in, and the color changes with the seasonal component. This way intuitively presents the data changes and is convenient for users to monitor the device status. It can be understood that the above process forms a complete chain from sharding to visualization. Sharding and parallel processing handle high concurrency, time series decomposition extracts key patterns, and mapping and rendering achieve intuitive presentation. The logic of each link is connected to jointly support the real-time data analysis requirements.

[0066] In one embodiment, the method further includes: S501. Obtain historical trend data.

[0067] S502. Calculate the cosine similarity between the real-time trend data and the historical trend data to obtain the first cosine similarity.

[0068] S503. When the first cosine similarity is lower than the third preset threshold, update the historical trend data set according to the real-time trend data to obtain the updated historical trend data set.

[0069] Among them, the third preset threshold may refer to a preset cosine similarity threshold.

[0070] Specifically, the third preset threshold can be set according to actual needs, and the present invention does not make specific limitations. For example, it can be ≥0.8 (normalized data). Historical trend data can be obtained from the historical trend dataset. The specific method can be: locating the target dataset through data indexing, and retrieving using a query statement according to the preset index fields to obtain the historical trend data. The query statement can be an SQL (SQL is the English abbreviation of Structured Query Language, representing Structured Query Language) query statement, which helps to efficiently filter and reduce query time. For the historical trend data and real-time trend data, vectorization technology is used to convert both into multi-dimensional feature vectors, and then the cosine similarity of the vectors is calculated to obtain the first cosine similarity. If the first cosine similarity is lower than the third preset threshold, the historical trend data is locally replaced, and the real-time trend data is written into the historical trend dataset through transaction processing to obtain the updated historical trend dataset. The calculation of cosine similarity is based on the vector angle and can intuitively reflect the closeness of data trends. The updated historical trend dataset can be used as the basis for subsequent analysis, such as for predicting the operating state of the device. The third preset threshold needs to be adjusted according to business requirements. For example, a higher threshold can be set for scenarios sensitive to temperature fluctuations. The process in this embodiment is clear and definite, which helps to improve the accuracy and real-time performance of data processing.

[0071] In one embodiment, the method further includes: S601, constructing a real-time trend feature vector set according to the real-time trend data.

[0072] S602, performing clustering processing on the historical trend dataset using the K-means algorithm to obtain a clustering result, where the clustering result includes a dynamic trend model and clustering center point data.

[0073] S603, constructing a historical trend feature vector set according to the clustering result.

[0074] S604, calculating the Euclidean distance between the real-time trend feature vector set and the historical trend feature vector set to obtain the first Euclidean distance.

[0075] S605, when the first Euclidean distance is greater than the fourth preset threshold, calculating the difference vector between the real-time trend feature vector and the historical trend feature vector, where the difference vector constitutes a trend comparison dataset.

[0076] S606, processing the trend comparison dataset using vector graphics acceleration technology, a preset rendering frequency, and a preset scaling ratio parameter to obtain a first dynamic trend chart.

[0077] Among them, the K-means algorithm is an existing distance-based unsupervised clustering algorithm. The fourth preset threshold may refer to a preset Euclidean distance value. The vector graphics acceleration technology refers to a technology that improves the processing speed and efficiency of vector graphics through hardware or software means. The rendering frequency may refer to the number of frames rendered per second when performing graphics rendering.

[0078] Specifically, the fourth preset threshold can be set according to actual needs, and the present invention does not make specific limitations. For example, it can be 2, 3, 5, or 10. The real-time trend data is extracted and represented in vector form to obtain a real-time trend feature vector, and the real-time trend feature vectors form a real-time trend feature vector set. For the updated historical trend data set, the K-means algorithm is used to perform clustering processing on the historical trend data to generate a dynamic trend model. Each dynamic trend model will have a corresponding clustering center point. The clustering center point data is converted into vector form, that is, a historical trend feature vector is obtained, and multiple historical trend feature vectors form a historical trend feature vector set. If the Euclidean distance between the real-time trend feature vector set and the historical trend feature vector set is greater than the fourth preset threshold, the difference vector between the two is calculated through matrix operations, and the difference vector forms a trend comparison data set. According to the trend comparison data set, the vector graphics acceleration technology is used to generate a first dynamic trend chart through a preset rendering frequency and a preset scaling ratio parameter. The first dynamic trend chart is a visual trend analysis result.

[0079] Regarding the use of the K-means algorithm for clustering processing, for example, the historical data contains the temperature sequence of the past 30 days. K-means can classify the data into 2 categories according to features such as the average daily temperature and the daily fluctuation range, such as a stable trend and an abnormal fluctuation trend. The dynamic trend model can be represented by the clustering center point data. For example, the clustering center point [24.5, 0.5] is used to represent the mean and fluctuation of the stable trend. This method is convenient for summarizing long-term patterns.

[0080] When comparing the real-time and historical trend feature vector sets, the Euclidean distance is used to measure the difference. For example, the distance between the real-time vector [25.3, 25.8] and the historical vector [24.5, 0.5] is greater than the threshold 2.0, indicating a trend shift. In a possible implementation, the difference vector is calculated through matrix operations, such as [0.8, 25.3], and the difference vector data constitutes the trend comparison data set. This method intuitively reflects the deviation direction and magnitude, facilitating subsequent analysis. According to the trend comparison result data set, the vector graphics acceleration technology is used for visualization. The preset rendering frequency can be set according to actual needs, for example, it can be once or twice per second. The preset scaling ratio can be set according to actual needs, and can be 1:100 or 1:150. The difference vector can be displayed through a line chart. For example, the horizontal axis is time and the vertical axis is the temperature deviation, and the graph is dynamically updated. This visualization method intuitively presents the trend changes and assists in the evaluation of the device status. Each link in the process of this embodiment supports each other, contributing to the improvement of the real-time analysis ability.

[0081] In one embodiment, the method further includes: S701, obtaining real-time interaction data.

[0082] S702, extracting feature point data and metadata from the real-time interaction data through event listening technology, and the feature point data and the metadata constitute an interaction event feature set.

[0083] S703, performing clustering processing on the feature point data by using the K-means algorithm to obtain zoom focus data and annotation clarity data, and the zoom focus data and the annotation clarity data constitute an interaction response parameter set.

[0084] S704, when the interaction response parameter set does not match the preset rendering parameter threshold, adjusting the zoom focus data and the annotation clarity data through a preset algorithm, and the adjusted zoom focus data and the adjusted annotation clarity data constitute an interaction configuration data set.

[0085] S705, performing parameter mapping on the interaction configuration data set and the first dynamic trend chart to obtain interaction enhanced trend data, and the interaction enhanced trend data includes the trend information of the interaction configuration.

[0086] Specifically, in the scenario of real-time interactive data processing, obtaining interactive event data from the data stream is a key link. Event listening technology is used to capture user operations on the embedded system interface, such as touch screen clicks or swipes. In one possible implementation, the listener records 100 events per second, extracts feature points such as click coordinates and swipe speeds, and metadata such as event timestamps and operation types to form an interactive event feature set. For example, a click event may generate a feature vector [120, 150, 0.5, click], indicating coordinates (120, 150), speed 0.5, and type click. This approach facilitates capturing user behavior patterns. It should be noted that based on the interactive event feature set, the K-means algorithm clusters the feature points to determine the zoom focus and annotation clarity. Specifically, cluster analysis is performed on the dense areas of user operations. For example, the coordinate data is divided into three categories: high-frequency interaction area, low-frequency interaction area, and abnormal interaction area. In one embodiment, the center point of the high-frequency interaction area is [130, 160], indicating that users often click on this area. The zoom focus is set to the center point, and the annotation clarity is set to high to highlight it. This clustering method effectively identifies interactive hotspots. When the interactive response parameter set does not match the preset rendering parameter threshold, the zoom focus data and the annotation clarity data are adjusted so that the adjusted data is as close as possible to the data before adjustment, while keeping the adjusted data within the preset rendering parameter threshold. There can be various specific preset algorithms, which are technically easy to implement. The present invention does not limit this. Those skilled in the art can select existing algorithms according to needs or summarize based on experience. For example, the following method can be adopted: The preset focus closest to the zoom focus A is B, and the preset deviation distance is d. If the distance between A and B is greater than d, then among the multiple pixel points whose distance from B is less than d, find the pixel point closest to A. For example, it is pixel point C. Then the coordinate data corresponding to C is the adjusted zoom focus data, that is, a new zoom center point is obtained. Then the annotation clarity corresponding to the new zoom center point is set to high. This ensures that the display effect meets expectations, helps save computing resources, and improves the rendering efficiency while satisfying user needs as much as possible. It should be noted that the high or low of the aforementioned annotation clarity represents the relative difference in clarity, aiming to emphasize the relative amount of corresponding resource calls. Its specific meaning is related to the specific device. The specific embedded industrial control screen can define the specific meanings of high, medium, and low annotation clarity based on its own configuration and display requirements. Regarding the preset rendering parameter threshold, the present invention does not make specific limitations and can be set according to the specific application scenario. For example, the rendering parameter threshold can be specified as the zoom focus deviation being less than a certain pixel value, such as 5 pixels, 10 pixels, 20 pixels, etc., or the clarity level difference being less than a certain value, such as 2, 3, or 5. The dynamic linkage technology maps the interactive configuration data set to a trend chart to generate interactive enhanced trend data.For example, a trend chart shows temperature changes. The interactive configuration dataset adjusts the chart zoom, magnifying the display in the high-frequency interaction area and clearly marking key points such as temperature peaks. In one embodiment, when the user clicks on [130, 160], the chart is magnified to show the details of temperature fluctuations within 10 seconds. This way enhances the user's interaction experience with the chart and intuitively presents data changes. The steps in this embodiment support each other, contributing to the seamless connection between interactive data and trend analysis and providing a smooth operation experience for users.

[0087] In one embodiment, the method further includes: S801, obtaining preset style configuration data and resolution adaptation requirement data.

[0088] S802, according to the resolution adaptation requirement data, adjusting the style configuration data through a CSS preprocessor to obtain a multi-resolution adaptation style sheet, and the multi-resolution adaptation style sheet constitutes a dynamic style configuration set.

[0089] S803, according to the dynamic style configuration set and the rendering frequency, rendering the trend comparison dataset in multiple resolutions through a WebGL renderer to obtain rendering data including icon dynamic information; the multi-resolution rendering includes: when the rendering frequency is lower than a fifth preset threshold, adjusting the frequency synchronization parameter.

[0090] S804, clustering the feature points of the rendering data using the K-means algorithm to obtain feature point annotation data.

[0091] S805, according to the zoom ratio parameter and the feature point annotation data, generating a second dynamic trend chart including feature point annotations through vector graphics drawing technology, and the second dynamic trend chart constitutes visualization analysis report data.

[0092] Among them, a CSS preprocessor is a tool in the prior art to extend the native CSS function. WebGL represents a 3D drawing protocol, the full English name is Web Graphics Library, and a WebGL renderer is a renderer based on WebGL technology. A style sheet is also called a CSS style sheet. CSS is the abbreviation of Cascading Style Sheet, and the Chinese name is Cascading Style Sheets, which can represent a web page content formatting technology used to decorate the display content of a web page. The fifth preset threshold may refer to a preset rendering frequency threshold.

[0093] Specifically, the fifth preset threshold can be set according to actual needs, and the present invention does not make specific limitations. For example, it can be 30fps, 50fps, 80fps, or 100fps. Receive the style configuration data obtained from a preset templated style library. The preset templated style library can contain a variety of predefined CSS style templates, such as font size, color theme, and margin configuration, which are applicable to different device resolutions. For the resolution adaptation requirements of industrial control screens, the CSS preprocessor adjusts the style parameters of the style configuration data through variable calculation, and uses the adjusted configuration data to generate a style sheet adapted to multiple resolutions. Multiple style sheets constitute a dynamic style configuration set. According to the dynamic style configuration set, combined with the shared rendering frequency parameter, the WebGL renderer performs a multi-resolution rendering task on the trend comparison data set, specifically including: if the rendering frame rate is lower than the fifth preset threshold, adjust the frequency synchronization parameter to obtain rendering data containing chart dynamics. For the rendering data, use the K-means algorithm to cluster the feature points of the rendering data to obtain feature point annotation data. Then, combined with the scaling ratio and the feature point annotation data, use vector graphics drawing technology to generate a second dynamic trend chart containing feature point annotations, that is, obtain the visual analysis report data.

[0094] In a possible implementation, there are many types of CSS preprocessors. For example, it can be Sass (Sass is the English abbreviation of Syntactically Awesome Style Sheets, representing a CSS preprocessor). For the resolution adaptation task, Sass can be used to dynamically adjust parameters. For example, for a 1080p resolution, the font size is set to 16px, while for a 4K resolution, it is adjusted to 20px to ensure visual consistency.

[0095] Regarding the WebGL renderer performing a multi-resolution rendering task, WebGL uses the GPU (GPU represents the graphics processing unit) to accelerate the drawing of complex charts, such as line charts or heat maps. In one embodiment, the rendering frequency is set to 60fps. If it is detected that the rendering frequency drops below 50fps, adjust the frequency synchronization parameter, such as reducing the anti-aliasing level, to ensure smoothness. The adjusted rendering data contains chart dynamics information, such as animation transition effects, enhancing the user experience. For the rendering data, the K-means algorithm clusters the feature points to optimize the annotation positions. For example, if a chart contains 1000 data points, it can be divided into 3 categories by the clustering algorithm: high-density area, low-density area, and isolated points. Combine the scaling accuracy parameter to adjust the annotation positions to avoid overlap. For example, the annotation font in the high-density area is bolded and offset by 5px to ensure clear readability.

[0096] There are various vector graphics drawing techniques. For example, it can be the SVG technology (SVG is the English abbreviation of Scalable Vector Graphics, which means scalable vector graphics).

[0097] In one embodiment, the method further includes: S901, extracting core data from the visual analysis report data.

[0098] S902, compressing the core data using the LZ77 compression algorithm to obtain compressed data.

[0099] S903, storing the compressed data in a specified database.

[0100] S904, judging the storage stability of the compressed data in the specified database to obtain a judgment result, where the judgment result includes whether the storage stability reaches a sixth preset threshold.

[0101] S905, generating a storage report, where the storage report includes a write success rate, a time consumption, the core data, and the judgment result.

[0102] Among them, the LZ77 compression algorithm is an existing dictionary compression algorithm based on a sliding window, and compression is achieved by replacing repeated strings with a triple of distance, length, and subsequent characters. The sixth preset threshold can refer to a preset storage stability threshold.

[0103] Specifically, the sixth preset threshold can be set according to actual needs, and the present invention does not make specific limitations. For example, it can be 90%, 99%, or 99.9%. Generally, the generated visual analysis report data needs to be stored efficiently to support subsequent queries. Extracting core data from the visual analysis report data, performing data compression processing using the LZ77 compression algorithm, storing the compressed data in a specified database through a database interface, judging whether the storage stability reaches the sixth preset threshold, and obtaining a storage report. The core data can be reduced in storage space through the LZ77 compression algorithm. For example, 1MB of original data is compressed to 300KB, improving the transmission efficiency. The compressed data can be stored in a MySQL database (MySQL represents a relational database management system) through a database interface, and the storage stability needs to reach a preset threshold of 99.9%. The database interface generally supports batch writing to reduce latency. For example, 1000 records are written per second to ensure data integrity. The storage report records the write success rate and the time consumption, providing a basis for system optimization. The process in this embodiment in a real-time monitoring scenario, style adaptation ensures consistent display on different devices, WebGL rendering ensures smooth charts, clustering optimization improves annotation clarity, and compression and storage improve data management efficiency. Each link supports each other to form an efficient closed-loop for trend chart generation and data processing.

[0104] In one embodiment, the method further includes: S1001, receiving a data query instruction and the resolution parameter of the target terminal.

[0105] S1002, according to the data query instruction, extracting the core data in the storage report through the interactive query interface.

[0106] S1003, performing structured processing on the core data by using JSON parsing technology to obtain structured query data.

[0107] S1004, according to the resolution parameter, adjusting the output format style parameter of the structured query data through CSS media query to obtain the first display data adapted to the terminal.

[0108] S1005, calculating the zoom focus coordinates by using interactive zoom technology.

[0109] S1006, generating a dynamic trend chart containing the zoom focus coordinates by using Canvas drawing technology to obtain interactive enhanced chart data.

[0110] S1007, if the frame rate of the interactive enhanced chart data is lower than the seventh preset threshold, adjusting the rendering frequency parameter of the interactive enhanced chart data through the WebGL renderer to obtain the terminal output data.

[0111] S1008, outputting the terminal output data to the target terminal.

[0112] Among them, the JSON parsing technology may refer to the technology of converting JSON - formatted data into a data structure that can be processed by a program. CSS media query is a query tool in CSS. Canvas is the name of an existing drawing technology. The seventh preset threshold may refer to a preset frame rate threshold.

[0113] Specifically, the seventh preset threshold can be set according to actual needs, and the present invention does not make specific limitations. For example, it can be 30fps, 50fps, 80fps, or 100fps. Obtain query parameters from an external query instruction, extract the core data in the storage report through an interactive query interface, and perform structured processing on the core data using JSON parsing technology to obtain structured query data. The query parameters may include a time range, data type, or specific metrics, such as traffic data in the past 24 hours. The query instruction is transmitted in JSON format through a RESTful API (RESTful API represents a design specification for application programming interfaces based on the HTTP protocol and following the REST architectural style. The HTTP protocol represents the Hypertext Transfer Protocol, and the REST architecture is the name of an existing software architecture), including fields such as "start_time": "2025-05-08 00:00:00" and "metric": "traffic". The interface needs to support high-concurrency access, for example, processing 1000 queries per second, ensuring that the response time is less than 200 milliseconds. According to the structured query data, combine the terminal adaptation protocol to detect the resolution parameters of the target terminal, and adjust the style parameters of the output format through CSS media queries to obtain the first display data adapted to the terminal. For the display data of the adapted terminal, use interactive zooming technology to calculate the zoom focus coordinates, and then generate a dynamic trend chart including the focus annotation through Canvas drawing technology to obtain interactive enhanced chart data. If the frame rate of the interactive enhanced chart data is lower than the seventh preset threshold, adjust the rendering frequency parameters through a WebGL renderer to achieve optimization processing of the chart data. The obtained data is the terminal output data, and the terminal output data is output to the target terminal for visual display on the target terminal. In a possible implementation, the interactive query interface is designed based on GraphQL (GraphQL is the name of a query language) and supports dynamic field selection. After the interface parses the query parameters, it extracts the core data of the storage report from a MySQL database, such as the original data set of the trend chart. Using JSON parsing technology to perform structured processing on the core data is an important step in data normalization. The core data may include timestamps and values, such as {"timestamp": "2025-05-08 12:00:00", "value": 500}. In one embodiment, the JSON parser converts the data into a key-value pair structure to generate structured query data for subsequent processing. Preferably, the parsing process supports error checking, such as detecting missing fields to ensure data integrity. For example, when parsing 1000 records, if 5 records are found to be missing timestamps, log the records and skip the invalid data. Combining the terminal adaptation protocol to detect the resolution parameters of the target terminal according to the structured query data is the core of achieving display optimization.For example, the terminal adaptation protocol obtains device information through the HTTP request header, such as a resolution of 1920x1080. CSS media queries adjust style parameters according to the resolution. For example, the chart font is set to 14px at 1080p and adjusted to 18px at 2560x1440. It can be understood that the media queries dynamically load style sheets through the min-width condition to ensure consistent display on different devices.

[0114] In one embodiment, the style parameters further include chart margins and color themes. For example, the background color is #333 in dark mode. The user zooms in on the chart using the mouse wheel, and the system calculates the focus coordinates based on the mouse position, such as (500, 300). The Canvas drawing technique draws the magnified area based on the coordinates. For example, the data point details are displayed magnified by 2 times. Canvas supports drawing dynamic line charts and annotating the data values at the focus, such as "Traffic: 600". For example, the chart contains 500 data points, and only 50 points near the focus are displayed after zooming to improve readability. If the frame rate of the interactive enhanced chart data is lower than the seventh preset threshold, such as 60fps, the rendering frequency parameter needs to be adjusted to optimize performance. For example, when it is detected that the frame rate drops to 45fps, the WebGL renderer reduces the rendering precision, such as turning off some anti-aliasing effects. Exemplarily, the optimized chart still maintains a smooth animation transition, such as the smooth movement of data points. WebGL uses GPU acceleration to draw complex graphics, such as heat maps, reducing the CPU load.

[0115] The process in this embodiment forms a closed loop from query to display. The query parameters drive data extraction, JSON parsing ensures data structuring, terminal adaptation optimizes the display effect, and interactive zooming and WebGL rendering enhance the user experience. Each link supports each other to ensure efficient data visualization presentation.

[0116] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are sequentially shown according to the indications of the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0117] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present invention are all information and data authorized by the user or fully authorized by all parties.

[0118] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided by the present invention can include at least one of non-volatile and volatile memories. Non-volatile memories can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memories can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided by the present invention can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided by the present invention can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., without limitation.

[0119] Based on the same inventive concept, an embodiment of the present invention further provides a data processing system based on an embedded industrial control screen for implementing the data processing method based on an embedded industrial control screen involved above. The implementation solutions provided by this system for solving problems are similar to the implementation solutions described in the above method. Therefore, the specific limitations in one or more embodiments of the data processing system based on an embedded industrial control screen provided below can refer to the limitations on the data processing method based on an embedded industrial control screen in the above text, and will not be repeated here.

[0120] In one embodiment, as Figure 4 shown, a data processing system based on an embedded industrial control screen is provided, including: a first data acquisition module, a segmentation calculation module, a priority module, a format conversion module, a data annotation module, a virtual unit module, a component extraction module, and a data rendering module, where: The first data acquisition module 11 is used to acquire multi-source heterogeneous sensing data; The segmentation calculation module 12 is used to process the multi-source heterogeneous sensing data through a sliding window technique to obtain segmented data; calculate the statistical feature data of the segmented data, and the statistical feature data constitutes the first statistical data; The priority module 13 is used to establish a task priority list according to the first statistical data and a preset priority rule; The format conversion module 14 is used to convert the segmented data into a standard format data set; The data annotation module 15 is used to perform feature point annotation on the standard format data set to obtain an annotated data set; The virtual unit module 16 is used to process the annotated data set using virtual processing units according to the task priority list to obtain multi-dimensional feature vectors; The component extraction module 17 is used to extract the component data of the multi-dimensional feature vector, and the component data includes a trend component and a seasonal component; The first data rendering module 18 is used to use a visualization rendering engine to process the component data in real time to obtain real-time trend data.

[0121] In one embodiment, the virtual unit module is used to implement the following steps: Judge whether the data volume of the annotated data set exceeds a first preset threshold to obtain a judgment result; When the judgment result is that the data volume exceeds the first preset threshold, split the annotated data set according to the task priority list and a preset partitioning rule to obtain a partitioned data set; Allocate multiple virtual processing units to extract the feature data of each partitioned data set in parallel; According to the feature data, use a principal component analysis algorithm to process the feature data to obtain a multi-dimensional projection; Extract the key dimension information of the multi-dimensional projection using a preset projection template to obtain a multi-dimensional feature set; Process the multi-dimensional feature set using data integration and vector splicing to obtain a multi-dimensional feature vector.

[0122] In one embodiment, the first data rendering module is used to implement the following steps: Convert the component data into visual feature data through data mapping technology according to a preset visualization template; Use a rendering engine to draw the visual feature data in real time, dynamically refresh the rendering result, and obtain real-time trend data.

[0123] In one embodiment, the system further includes: A second data acquisition module for acquiring historical trend data; A cosine calculation module for calculating the cosine similarity between the real-time trend data and the historical trend data to obtain a first cosine similarity; A data update module for updating the historical trend data set according to the real-time trend data when the first cosine similarity is lower than a third preset threshold to obtain an updated historical trend data set.

[0124] In one embodiment, the system further includes: A first vector module for constructing a real-time trend feature vector set according to the real-time trend data; A first clustering module for performing clustering processing on the historical trend data set using the K-means algorithm to obtain a clustering result, where the clustering result includes a dynamic trend model and clustering center point data; A second vector module for constructing a historical trend feature vector set according to the clustering result; An Euclidean distance module for calculating the Euclidean distance between the real-time trend feature vector set and the historical trend feature vector set to obtain a first Euclidean distance; A third vector module for calculating a difference vector between the real-time trend feature vector and the historical trend feature vector when the first Euclidean distance is greater than a fourth preset threshold, and the difference vector constitutes a trend comparison data set; A trend comparison module for processing the trend comparison data set using vector graphics acceleration technology, a preset rendering frequency, and a preset scaling ratio parameter to obtain a first dynamic trend chart.

[0125] In one embodiment, the system further includes: A third data acquisition module for acquiring real-time interaction data; A feature extraction module, which is used to extract feature point data and metadata from the real-time interaction data through event listening technology, and the feature point data and the metadata constitute an interaction event feature set; A second clustering module, which is used to perform clustering processing on the feature point data by using the K-means algorithm to obtain zoom focus data and annotation clarity data, and the zoom focus data and the annotation clarity data constitute an interaction response parameter set; A data adjustment module, which is used to adjust the zoom focus data and the annotation clarity data through a preset algorithm when the interaction response parameter set does not match a preset rendering parameter threshold, and the adjusted zoom focus data and the adjusted annotation clarity data constitute an interaction configuration data set; A first mapping module, which is used to perform parameter mapping on the interaction configuration data set and the first dynamic trend chart to obtain interaction enhanced trend data, and the interaction enhanced trend data contains trend information of the interaction configuration.

[0126] In one embodiment, the system further includes: A fourth data acquisition module, which is used to acquire preset style configuration data and resolution adaptation requirement data; A resolution adaptation module, which is used to adjust the style configuration data through a CSS preprocessor according to the resolution adaptation requirement data to obtain a multi-resolution adaptation style sheet, and the multi-resolution adaptation style sheet constitutes a dynamic style configuration set; A second data rendering module, which is used to multi-resolution render the trend comparison data set according to the dynamic style configuration set and the rendering frequency through a WebGL renderer to obtain rendering data containing icon dynamic information; the multi-resolution rendering includes: when the rendering frequency is lower than a fifth preset threshold, adjusting the frequency synchronization parameter; A third clustering module, which is used to cluster the feature points of the rendering data by using the K-means algorithm to obtain feature point annotation data; A report data module, which is used to generate a second dynamic trend chart containing feature point annotations through vector graphics drawing technology according to the zoom ratio parameter and the feature point annotation data, and the second dynamic trend chart constitutes visual analysis report data.

[0127] In one embodiment, the system further includes: A first data extraction module, which is used to extract core data from the visual analysis report data; A data compression module, which is used to compress the core data by using the LZ77 compression algorithm to obtain compressed data; A data storage module, which is used to store the compressed data in a specified database; A data judgment module, configured to judge the storage stability of the compressed data in the specified database, and obtain a judgment result, where the judgment result includes whether the storage stability reaches a sixth preset threshold; A storage report module, configured to generate a storage report, where the storage report includes a write success rate, a time consumption, the core data, and the judgment result.

[0128] In one embodiment, the system further includes: A fifth data acquisition module, configured to receive a data query instruction and a resolution parameter of a target terminal; A second data extraction module, configured to extract the core data in the storage report through an interactive query interface according to the data query instruction; A structure processing module, configured to perform structure processing on the core data by using JSON parsing technology to obtain structured query data; A format adjustment module, configured to adjust output format style parameters of the structured query data through CSS media query according to the resolution parameter to obtain first display data adapted to the terminal; A focus calculation module, configured to calculate zoom focus coordinates by using an interactive zoom technology; A chart generation module, configured to generate a dynamic trend chart including the zoom focus coordinates through Canvas drawing technology to obtain interactive enhanced chart data; A parameter adjustment module, configured to, when the frame rate of the interactive enhanced chart data is lower than a seventh preset threshold, adjust the rendering frequency parameter of the interactive enhanced chart data through a WebGL renderer to obtain terminal output data; A data output module, configured to output the terminal output data to a target terminal.

[0129] Each module in the above data processing system based on an embedded industrial control screen can be implemented in whole or in part by software, hardware, and their combination. The above modules can be embedded in the processor of the embedded industrial control screen in hardware form or be independent of it, or be stored in the memory of the embedded industrial control screen in software form, so that the processor can call and execute the operations corresponding to the above modules.

[0130] The technical features of the above embodiments can be combined arbitrarily. For the sake of brief description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0131] The above-described embodiments merely represent several implementation manners of the present invention. The description thereof is relatively specific and detailed, but it should not be construed as a limitation on the scope of the patent for the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all fall within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the appended claims.

Claims

1. A data processing method based on an embedded industrial control screen, characterized in that The method includes: Obtaining multi-source heterogeneous sensing data; Processing the multi-source heterogeneous sensing data through a sliding window technique to obtain segmented data; calculating statistical feature data of the segmented data, and the statistical feature data constitutes first statistical data; Establishing a task priority list according to the first statistical data and a preset priority rule; Converting the segmented data into a standard format data set; Performing feature point annotation on the standard format data set to obtain an annotated data set; Processing the annotated data set using a virtual processing unit according to the task priority list to obtain a multi-dimensional feature vector; Extracting component data of the multi-dimensional feature vector, and the component data includes a trend component and a seasonal component; Using a visualization rendering engine to process the component data in real time to obtain real-time trend data.

2. The data processing method based on an embedded industrial control screen according to claim 1, wherein The step of processing the annotated data set using a virtual processing unit according to the task priority list to obtain a multi-dimensional feature vector includes the following steps: Judging whether the data volume of the annotated data set exceeds a first preset threshold to obtain a judgment result; When the judgment result is that the data volume exceeds the first preset threshold, splitting the annotated data set according to the task priority list and a preset partitioning rule to obtain partitioned data sets; Allocating multiple virtual processing units to extract feature data of each partitioned data set in parallel; Processing the feature data using a principal component analysis algorithm according to the feature data to obtain a multi-dimensional projection; Extracting key dimension information of the multi-dimensional projection using a preset projection template to obtain a multi-dimensional feature set; Processing the multi-dimensional feature set using data integration and vector splicing to obtain a multi-dimensional feature vector.

3. The data processing method based on an embedded industrial control screen according to claim 1, wherein The step of using a visualization rendering engine to process the component data in real time includes: Converting the component data into visualization feature data through data mapping technology according to a preset visualization template; Using the rendering engine to draw the visualization feature data in real time and dynamically refreshing the rendering result to obtain real-time trend data.

4. The data processing method based on an embedded industrial control screen according to claim 3, wherein, The method further includes: Obtaining historical trend data; Calculating the cosine similarity between the real-time trend data and the historical trend data to obtain a first cosine similarity; When the first cosine similarity is lower than a third preset threshold, updating the historical trend data set according to the real-time trend data to obtain an updated historical trend data set.

5. The data processing method based on an embedded industrial control screen according to claim 4, wherein The method further includes: Constructing a real-time trend feature vector set according to the real-time trend data; Performing clustering processing on the historical trend data set using the K-means algorithm to obtain a clustering result, and the clustering result includes a dynamic trend model and clustering center point data; Constructing a historical trend feature vector set according to the clustering result; Calculating the Euclidean distance between the real-time trend feature vector set and the historical trend feature vector set to obtain a first Euclidean distance; When the first Euclidean distance is greater than a fourth preset threshold, calculating a difference vector between the real-time trend feature vector and the historical trend feature vector, and the difference vector constitutes a trend comparison data set; Processing the trend comparison data set using a vector graphics acceleration technology, a preset rendering frequency, and a preset scaling ratio parameter to obtain a first dynamic trend chart.

6. The data processing method based on an embedded industrial control screen according to claim 5, wherein The method further includes: Obtaining real-time interaction data; Extract the feature point data and metadata from the real-time interaction data through event listening technology, and the feature point data and the metadata constitute an interaction event feature set; Use the K-means algorithm to cluster the feature point data to obtain zoom focus data and annotation clarity data, and the zoom focus data and the annotation clarity data constitute an interaction response parameter set; When the interaction response parameter set does not match the preset rendering parameter threshold, adjust the zoom focus data and the annotation clarity data through a preset algorithm. The adjusted zoom focus data and the adjusted annotation clarity data constitute an interaction configuration data set; map the parameters of the interaction configuration data set to the first dynamic trend chart to obtain interaction enhanced trend data, and the interaction enhanced trend data contains the trend information of the interaction configuration.

7. The data processing method based on an embedded industrial control screen according to claim 5, wherein The method further includes: Obtain preset style configuration data and resolution adaptation requirement data; According to the resolution adaptation requirement data, adjust the style configuration data through a CSS preprocessor to obtain a multi-resolution adaptation style sheet, and the multi-resolution adaptation style sheet constitutes a dynamic style configuration set; According to the dynamic style configuration set and the rendering frequency, perform multi-resolution rendering on the trend comparison data set through a WebGL renderer to obtain rendering data containing icon dynamic information; the multi-resolution rendering includes: when the rendering frequency is lower than the fifth preset threshold, adjust the frequency synchronization parameter; Use the K-means algorithm to cluster the feature points of the rendering data to obtain feature point annotation data; According to the zoom ratio parameter and the feature point annotation data, generate a second dynamic trend chart containing feature point annotations through vector graphics drawing technology, and the second dynamic trend chart constitutes visual analysis report data.

8. The data processing method based on an embedded industrial control screen according to claim 7, wherein The method further includes: Extract core data from the visual analysis report data; Compress the core data using the LZ77 compression algorithm to obtain compressed data; Store the compressed data in a specified database; Judge the storage stability of the compressed data in the specified database to obtain a judgment result, and the judgment result includes whether the storage stability reaches the sixth preset threshold; Generate a storage report, and the storage report includes the write success rate, time consumption, the core data, and the judgment result.

9. The data processing method based on an embedded industrial control screen according to claim 8, wherein The method further includes: Receive a data query instruction and the resolution parameter of the target terminal; According to the data query instruction, extract the core data in the storage report through an interactive query interface; Use JSON parsing technology to structure the core data to obtain structured query data; According to the resolution parameter, adjust the output format style parameter of the structured query data through CSS media query to obtain the first display data adapted to the terminal; Use interactive zoom technology to calculate the zoom focus coordinates; Generate a dynamic trend chart containing the zoom focus coordinates through Canvas drawing technology to obtain interaction enhanced chart data; If the frame rate of the interactive enhanced chart data is lower than the seventh preset threshold, the rendering frequency parameter of the interactive enhanced chart data is adjusted through a WebGL renderer to obtain terminal output data; Output the terminal output data to the target terminal.

10. A data processing system based on an embedded industrial control screen, characterized in that, The system includes: A first data acquisition module, configured to acquire multi-source heterogeneous sensing data; A segmentation calculation module, configured to process the multi-source heterogeneous sensing data through a sliding window technique to obtain segmented data; calculate the statistical feature data of the segmented data, and the statistical feature data constitutes first statistical data; A priority module, configured to establish a task priority list according to the first statistical data and a preset priority rule; A format conversion module, configured to convert the segmented data into a standard format data set; A data annotation module, configured to perform feature point annotation on the standard format data set to obtain an annotated data set; A virtual unit module, configured to process the annotated data set using a virtual processing unit according to the task priority list to obtain a multi-dimensional feature vector; A component extraction module, configured to extract the component data of the multi-dimensional feature vector, and the component data includes a trend component and a seasonal component; A data rendering module, configured to use a visualization rendering engine to process the component data in real time to obtain real-time trend data.

Images