Industrial internet real-time cooperative control method based on edge computing

Abstract

The invention discloses an industrial internet real-time cooperative control method based on edge computing, and the method comprises the steps: obtaining multi-source heterogeneous data of industrial equipment, and generating a standardized data set; performing time sequence alignment processing on the standardized data set to generate a time sequence data set; segmenting the data to obtain a standardized segmented data set; removing abnormal values and redundant information according to the standardized segmented data set to obtain a cleaned data set; performing format conversion on the cleaned data set according to an equipment state and a control strategy requirement to generate intermediate data; performing feature extraction on the intermediate data to generate an equipment state feature vector; according to real-time monitoring information of a network load and a transmission path, a computing resource allocation strategy of an edge node is adjusted, data of high-priority equipment is processed preferentially, and a resource allocation result is generated; and deploying a cooperative control algorithm based on an ARIMA time sequence model at the edge node, and generating a real-time control instruction in combination with the equipment state feature vector.

Description

Technical Field

[0001] The present invention belongs to the technical field of industrial collaborative control, and particularly relates to a real-time collaborative control method for industrial Internet based on edge computing. Background Art

[0002] In the real-time collaborative control scenario of industrial Internet, the application of edge computing faces a core technical problem: how to effectively process the collaboration and consistency of multi-source heterogeneous data while ensuring real-time performance. The data collected by industrial devices through sensors and actuators often have different sampling frequencies, precisions, and formats. When these data are preprocessed at the edge layer, it is necessary to solve the timing alignment problem of data cleaning and format conversion. Due to the rapid changes in the operating state of industrial devices, the delay in data collection may lead to misjudgment of the device state by the edge node, thereby affecting the correctness of the collaborative control strategy. For example, when the data of multiple devices converge at the edge node, due to the differences in transmission paths and network loads, the timing of data arrival at the edge node may be disrupted. At this time, if the data is directly cleaned and converted, it may lead to the distortion of device state information, further affecting the real-time performance and accuracy of the control strategy. In addition, the computing resources of edge nodes are limited. How to optimize the allocation of computing resources while ensuring the data processing accuracy is also a difficult problem to be solved urgently. If the data processing is too complex, it will occupy too much computing resources, resulting in a delay in the adjustment of the control strategy; while if the processing is too simplified, key information may be missed, affecting the control effect. Therefore, how to find a balance between the real-time performance and consistency of multi-source heterogeneous data is the core challenge of edge computing in industrial Internet collaborative control. Summary of the Invention

[0003] To solve the above technical problems, the present invention proposes a real-time collaborative control method for industrial Internet based on edge computing, which realizes the efficient processing and real-time intelligent control of industrial device data, and improves the reliability and control accuracy of the system.

[0004] To achieve the above object, the present invention provides a real-time collaborative control method for industrial Internet based on edge computing, including:

[0005] Obtain multi-source heterogeneous data of industrial devices, extract the sampling frequency, data precision, and format information of the multi-source heterogeneous data, and generate a standardized data set;

[0006] Perform timing alignment processing on the standardized data set at the edge node to generate a time series data set;

[0007] If there is a situation where the data arrival timing in the time series data set is disrupted, segment the data to obtain a standardized segmented data set;

[0008] Remove outliers and redundant information according to the standardized segmented dataset to obtain a cleaned dataset;

[0009] According to the device status and control strategy requirements, perform format conversion on the cleaned dataset to generate intermediate data in a unified format;

[0010] Extract features from the intermediate data in the unified format to generate a device status feature vector;

[0011] According to the real-time monitoring information of network load and transmission path, dynamically adjust the computing resource allocation strategy of the edge node, give priority to processing data of high-priority devices, and generate a resource allocation result, where the high-priority devices are determined according to the operating status and failure risk level of the devices;

[0012] Deploy a collaborative control algorithm based on the ARIMA time series model at the edge node, combine it with the device status feature vector to generate real-time control instructions, and complete real-time collaborative control of the industrial Internet.

[0013] Optionally, obtain multi-source heterogeneous data of industrial devices, and extract the sampling frequency, data accuracy, and format information of the multi-source heterogeneous data, including:

[0014] Extract the sampling frequency, data accuracy, and format information of the data according to the obtained multi-source heterogeneous data;

[0015] According to the sampling frequency of the data, judge the time interval of data acquisition. If the time interval exceeds the preset threshold, adjust the sampling frequency;

[0016] Analyze the accuracy of the data through the data accuracy. If the accuracy is lower than the preset standard, use a data correction algorithm for correction;

[0017] According to the format information, determine the storage method and encoding specification of the data. If the formats are inconsistent, perform format conversion processing;

[0018] Integrate the processed multi-source heterogeneous data according to the unified sampling frequency, data accuracy, and format information to generate a standardized dataset.

[0019] Optionally, perform time series alignment processing on the standardized dataset at the edge node to generate a time series dataset, including:

[0020] According to the preset time series alignment rule, add timestamp marks to the standardized dataset at the edge node to generate a time series dataset.

[0021] Optionally, if there is a situation where the arrival time sequence of the data in the time series dataset is disrupted, perform segmentation processing on the data to obtain a standardized segmented dataset, including:

[0022] Obtain a time series data set, identify the timing state of data arrival. If the timing is disrupted, use a sliding window mechanism to segment the data.

[0023] Determine the timing consistency within each window based on the segmented data, and generate a segmented time series data set.

[0024] Extract the timestamp, sampling frequency, and data accuracy of the data according to the segmented time series data set, and determine whether the data meets the preset standard. If the data accuracy is lower than the preset standard, use a data correction algorithm for correction.

[0025] Integrate the processed segmented data to generate a standardized segmented data set.

[0026] Optionally, remove outliers and redundant information according to the standardized segmented data set, and the obtained cleaned data set includes:

[0027] Perform outlier detection on the standardized segmented data set. If the data value exceeds the preset threshold, mark it as an outlier and remove it.

[0028] For the redundant values in the standardized segmented data set, judge whether there are duplicate data through the matching relationship between the sampling rate and the timestamp. If so, perform deduplication processing to obtain a cleaned data set.

[0029] Optionally, perform format conversion on the cleaned data set according to the device status and control strategy requirements to generate intermediate data in a unified format, including:

[0030] Extract key feature values from the cleaned data set, obtain the feature information of the data, and generate a feature data set.

[0031] Analyze the feature data set to identify the regular patterns in the data and generate a pattern recognition result.

[0032] According to the pattern recognition result, use a standardization algorithm to normalize the data and generate intermediate data in a unified format.

[0033] Optionally, perform feature extraction on the intermediate data in the unified format to generate a device status feature vector, including:

[0034] Perform feature extraction on the intermediate data in the unified format to obtain the key features and regular information of the data.

[0035] According to the key features and regular information of the data, use the support vector machine algorithm to optimize and analyze the device status and control strategy, and generate a device status feature vector.

[0036] Optionally, according to the real-time monitoring information of network load and transmission path, dynamically adjust the computing resource allocation strategy of edge nodes, and give priority to processing data of high-priority devices. The generated resource allocation results include:

[0037] Adopt a preset monitoring frequency to obtain real-time monitoring data from network load and transmission path, and generate monitoring information of network load and transmission path;

[0038] According to the load value and path delay in the monitoring information, determine whether the network is in a high-load state. If it is in a high-load state, trigger the dynamic adjustment mechanism;

[0039] For data transmission requests of high-priority devices, reserve computing resources of edge nodes for high-priority devices;

[0040] According to the real-time changes of network load and transmission path, adjust the computing resource allocation ratio of edge nodes to obtain an optimized resource allocation strategy;

[0041] Through the optimized resource allocation strategy, give priority to processing data of high-priority devices and generate resource allocation results.

[0042] Optionally, the high-priority devices are determined according to the operating status and failure risk level of the devices, including:

[0043] Obtain the operating status and failure history data of all devices, and calculate the failure risk level of each device;

[0044] According to the failure risk level and device operating status, determine the device priority list, and the devices at the top of the list are high-priority devices.

[0045] Optionally, deploy a collaborative control algorithm based on the ARIMA time series model at the edge node, and combine the device status feature vector to generate real-time control instructions, including:

[0046] Obtain device status data from the edge node and extract the device status feature vector;

[0047] According to the device status feature vector, calculate the device operation trend;

[0048] Adopt the ARIMA time series model to predict the future operating status of the device and obtain the prediction result;

[0049] Combine the prediction result and the device status feature vector to generate control instruction parameters. If the device operation trend deviates from the preset range, adjust the control instruction parameters;

[0050] According to the adjusted control instruction parameters, generate real-time control instructions;

[0051] Send real-time control instructions to the edge nodes for execution.

[0052] Technical effects of the present invention: The present invention discloses a real-time collaborative control method for industrial Internet based on edge computing. By collecting multi-source heterogeneous data of industrial equipment, performing time series alignment and timestamp marking, a time series data set is generated. In view of the disordered arrival time series of data, a sliding window mechanism is adopted for segmented processing to ensure the time series consistency of data within the window. Subsequently, data cleaning and format conversion are carried out to generate intermediate data in a unified format. The lightweight K-means clustering algorithm is used to extract features and generate device state feature vectors. Based on the real-time monitoring information of network load and transmission path, the computing resource allocation strategy of edge nodes is dynamically adjusted to preferentially process data of high-priority devices. Finally, a collaborative control algorithm based on the ARIMA model is deployed at the edge nodes, and real-time control instructions are generated in combination with the device state feature vectors. The present invention realizes the efficient processing and real-time intelligent control of industrial equipment data, and improves the reliability and control accuracy of the system. Description of the Drawings

[0053] The drawings forming a part of this application are used to provide a further understanding of this application. The schematic embodiments of this application and their descriptions are used to explain this application and do not constitute an improper limitation to this application. In the drawings:

[0054] Figure 1 It is a schematic flowchart of a real-time collaborative control method for industrial Internet based on edge computing according to an embodiment of the present invention. Detailed Embodiments

[0055] It should be noted that, without conflict, the embodiments in this application and the features in the embodiments can be combined with each other. The following will refer to the drawings and combine the embodiments to detail this application.

[0056] It should be noted that the steps shown in the flowchart of the drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0057] As Figure 1 shown, in this embodiment, a real-time collaborative control method for industrial Internet based on edge computing is provided, including:

[0058] Obtain multi-source heterogeneous data of industrial equipment, extract the sampling frequency, data accuracy and format information of the multi-source heterogeneous data, and generate a standardized data set;

[0059] Perform time series alignment processing on the standardized data set at the edge node to generate a time series data set;

[0060] If there is a situation where the arrival time sequence of data in the time series dataset is disrupted, segment the data to obtain a standardized segmented dataset;

[0061] Remove outliers and redundant information based on the standardized segmented dataset to obtain a cleaned dataset;

[0062] According to the device status and control strategy requirements, perform format conversion on the cleaned dataset to generate intermediate data in a unified format;

[0063] Extract features from the intermediate data in the unified format to generate a device status feature vector;

[0064] According to the real-time monitoring information of network load and transmission path, dynamically adjust the computing resource allocation strategy of edge nodes, prioritize processing data of high-priority devices, and generate a resource allocation result, where the high-priority devices are determined according to the operating status and failure risk level of the devices;

[0065] Deploy a collaborative control algorithm based on the ARIMA time series model at the edge node, combine it with the device status feature vector, generate real-time control instructions, and complete real-time collaborative control of the industrial Internet.

[0066] Further, obtain multi-source heterogeneous data of industrial devices, and extract the sampling frequency, data accuracy, and format information of the multi-source heterogeneous data, including:

[0067] Extract the sampling frequency, data accuracy, and format information of the data according to the obtained multi-source heterogeneous data;

[0068] Judge the time interval of data acquisition according to the sampling frequency of the data. If the time interval exceeds the preset threshold, adjust the sampling frequency;

[0069] Analyze the accuracy of the data through the data accuracy. If the accuracy is lower than the preset standard, use a data correction algorithm for correction;

[0070] Determine the storage method and encoding specification of the data according to the format information. If the formats are inconsistent, perform format conversion processing;

[0071] Integrate the processed multi-source heterogeneous data according to the unified sampling frequency, data accuracy, and format information to generate a standardized dataset.

[0072] Specifically, in industrial sites, there is a wide variety of sensors and actuators. For example, the data obtained by pressure sensors is in Pascal units, and the data obtained by temperature sensors is in degrees Celsius. The data formats of actuators from different equipment manufacturers also vary. By identifying the source device type of the data, subsequent processing can be carried out in a targeted manner. Taking a cement production line as an example, the temperature sensor in the kiln tail flue gas chamber collects data once per second, and the pressure sensor collects data once per minute. This difference in sampling rates will affect the integrated analysis of the data. If the sampling interval of the pressure data exceeds the preset threshold of 30 seconds, the sampling frequency needs to be adjusted to once every 30 seconds to ensure the continuity of the data. In terms of data accuracy, taking the temperature measurement in the steel smelting process as an example, if the sensor shows a temperature of 1500 degrees, but the actual measurement accuracy is only plus or minus 10 degrees, lower than the preset standard of plus or minus 5 degrees, data correction needs to be carried out through algorithms such as Kalman filtering to improve the measurement accuracy. In data format processing, the data storage formats of different devices vary greatly. For example, the liquid level sensor in a chemical plant outputs decimal values, while the flow meter outputs hexadecimal data, and they need to be uniformly converted to the floating-point format for storage. Through normalization processing, ensure that all data uses a unified timestamp and numerical format. In petroleum refining units, the processed multi-source data such as temperature, pressure, and flow are integrated into a standardized data set. The principal component analysis method is used to extract data features, and the clustering algorithm is used to identify the equipment operation mode. For example, it is found that there is a positive correlation between temperature and pressure, and there is a lag effect between flow and pressure. These findings have important guiding significance for optimizing the production process. Through data processing and analysis, equipment anomalies can be detected in a timely manner. For example, in an air compressor unit, when the data of the vibration sensor suddenly increases, combined with the change trends of temperature and current data, it can be predicted whether there is a bearing fault. This comprehensive analysis ability based on multi-source data provides strong support for predictive maintenance. Each link of data processing is interrelated. The unification of the sampling rate ensures the timeliness of the data, the accuracy correction guarantees the reliability of the data, the format conversion realizes the interoperability of the data, and finally through feature extraction and pattern recognition, the valuable information contained in the data is mined. This systematic data processing method lays a foundation for the intelligent monitoring and optimized decision-making of industrial equipment.

[0073] Furthermore, perform time series alignment processing on the standardized data set at the edge node, and the generated time series data set includes:

[0074] According to the preset time series alignment rules, add timestamp marks to the standardized data set at the edge node to generate a time series data set.

[0075] Specifically, the data formats output by different devices vary. For example, temperature data may adopt a floating-point format, pressure data an integer format, and flow data a string format. To facilitate unified processing, all data needs to be converted into a unified format. For example, all numerical data is converted into double-precision floating-point numbers, and time information is converted into a unified timestamp format. The generation process of the standardized dataset needs to consider the time alignment of the data. When adding timestamp markers at the edge node, millisecond-level precision can be selected to ensure the timing of the data. For example, marking a temperature data point as "20240410123000" indicates the data collected at 12:30:00 on April 10, 2024. When extracting features from time series data, statistical features such as mean, standard deviation, and peak value can be calculated. Taking temperature data as an example, the temperature change trend and periodic fluctuation characteristics can be extracted. Through pattern recognition algorithms such as clustering analysis, typical patterns of temperature changes can be discovered, such as the rapid increase in temperature when the device starts up and the stable fluctuations during normal operation. These features and patterns can be used for device status monitoring and fault warning. By analyzing the correlation between the data of multiple sensors, such as the correlation between temperature and pressure, the operating mechanism of the device can be deeply understood, providing a basis for industrial process optimization.

[0076] Furthermore, if there is a situation where the arrival timing of the data in the time series dataset is disrupted, the data is segmented to obtain a standardized segmented dataset, including:

[0077] Obtain the time series dataset, identify the timing status of data arrival. If the timing is disrupted, use the sliding window mechanism to segment the data.

[0078] Based on the segmented data, determine the timing consistency within each window and generate the segmented time series dataset.

[0079] According to the segmented time series dataset, extract the timestamp, sampling frequency, and data precision of the data, and determine whether the data meets the preset standard; if the data precision is lower than the preset standard, use the data correction algorithm for correction.

[0080] Integrate the processed segmented data to generate a standardized segmented dataset.

[0081] Specifically, in a certain smart city project, the data formats of different device manufacturers may vary. Some use the format of timestamp plus numerical value, while others use the date-time string format. By uniformly converting to a standard format, such as uniformly using the format of timestamp plus numerical value, the data processing efficiency can be improved. In the feature extraction and pattern recognition links, machine learning algorithms can mine the deep rules in the data. Taking the operation data of a certain wind farm as an example, by analyzing time series data such as wind speed, wind direction, and power generation, the impact mode of weather changes on power generation efficiency can be identified. Through time series decomposition, the data can be decomposed into three components: trend, seasonality, and random fluctuations, which helps to predict future changes in power generation. In practical applications, the implementation of these steps can significantly improve data quality and analysis effects. For example, in the energy consumption monitoring system of an industrial park, by performing time series alignment and feature extraction on the power consumption data, the power consumption patterns of different enterprises are successfully identified, providing a basis for optimizing energy scheduling. Through data standardization processing and feature analysis, the prediction accuracy of the system has been increased from 85% to 93%, saving significant energy costs for enterprises.

[0082] Furthermore, according to the standardized segmented data set, outliers and redundant information are removed, and the obtained cleaned data set includes:

[0083] Perform outlier detection on the standardized segmented data set. If the data value exceeds the preset threshold, it is marked as an outlier and removed;

[0084] For the redundant values in the standardized segmented data set, by the matching relationship between the sampling rate and the timestamp, it is judged whether there is duplicate data. If so, duplicate data removal processing is performed to obtain the cleaned data set.

[0085] Specifically, using the rule method for outlier detection is a key step in time series data processing. Taking the temperature sensor data as an example, the normal temperature range is between minus 20 degrees and 40 degrees. When an outlier such as 60 degrees appears, it needs to be marked and removed. Such outliers are usually caused by sensor failures or external interferences. Removing outliers can improve data quality. On an industrial production line, there may be redundant values in the vibration data collected by sensors.

[0086] Furthermore, according to the device status and control strategy requirements, the format of the cleaned data set is converted to generate intermediate data in a unified format, including:

[0087] Extract key feature values from the cleaned data set, obtain the feature information of the data, and generate a feature data set;

[0088] Analyze the feature data set, identify the regular patterns in the data, and generate a pattern recognition result;

[0089] According to the pattern recognition result, a standardization algorithm is used to normalize the data to generate intermediate data in a unified format.

[0090] Furthermore, feature extraction is performed on the intermediate data in the unified format to generate a device status feature vector, including:

[0091] Feature extraction is performed on the intermediate data in the unified format to obtain the key features and regular information of the data;

[0092] According to the key features and regular information of the data, a support vector machine algorithm is used to optimize and analyze the device status and control strategy to generate a device status feature vector.

[0093] Specifically, for the data processing of the refrigeration system of a certain industrial device, the process of extracting the timestamp and data value needs to consider multiple dimensions. For example, the temperature change curves are basically similar after starting up every morning, and this kind of regular information is very helpful for predicting the device performance. The standardization process converts data with different dimensions to a unified scale. In addition to temperature data, it also includes various parameters such as pressure and current. Through standardization, these data can be uniformly mapped to the interval from zero to one, which is convenient for subsequent analysis. The decision tree model training can find out the key factors affecting the refrigeration effect. Through analysis, it is found that when the ambient temperature exceeds 30 degrees and the running time exceeds 4 hours, the refrigeration effect will decrease significantly. This kind of association rule can guide the optimization of device operation. The support vector machine algorithm is used to predict and optimize the control strategy. The model trained based on historical data can predict the best operating parameters under different working conditions. For example, the refrigeration power can be adjusted in advance according to the prediction result to reduce energy consumption while ensuring the refrigeration effect. When it is predicted that there is a peak electricity consumption period in the afternoon, the system can increase the refrigeration capacity in advance to avoid high-power operation during the peak period.

[0094] Furthermore, according to the real-time monitoring information of the network load and transmission path, the calculation resource allocation strategy of the edge node is dynamically adjusted to preferentially process the data of high-priority devices, and the generated resource allocation result includes:

[0095] Using a preset monitoring frequency, real-time monitoring data is obtained from the network load and transmission path to generate monitoring information of the network load and transmission path;

[0096] According to the load value and path delay in the monitoring information, it is judged whether the network is in a high-load state. If it is in a high-load state, the dynamic adjustment mechanism is triggered;

[0097] For the data transmission request of high-priority devices, computing resources of the edge node are reserved for high-priority devices;

[0098] Adjust the computing resource allocation ratio of edge nodes according to the real-time changes of network load and transmission path, and obtain an optimized resource allocation strategy;

[0099] Through the optimized resource allocation strategy, prioritize the processing of data from high-priority devices to generate a resource allocation result.

[0100] Specifically, during the dynamic resource allocation process, the system adjusts the allocation ratio according to the real-time load situation. When network congestion is detected, the resource quota for ordinary services may be reduced by 30%, and the released resources are allocated to high-priority services. This dynamic adjustment not only ensures the service quality of critical services but also improves resource utilization efficiency. Through the optimized resource allocation strategy, the system can still operate stably when the network load fluctuates. For example, during the morning business peak, even when the overall network load reaches 90%, the response time of the critical business system can still be kept within an acceptable range, reflecting the intelligent and refined management capabilities of resource scheduling.

[0101] Furthermore, the high-priority devices are determined based on the operating status and failure risk level of the devices, including:

[0102] Obtain the operating status and failure history data of all devices, and calculate the failure risk level of each device;

[0103] Determine the device priority list according to the failure risk level and device operating status, and the devices at the top of the list are high-priority devices.

[0104] Furthermore, deploy a collaborative control algorithm based on the ARIMA time series model at the edge node, and generate real-time control instructions in combination with the device status feature vector, including:

[0105] Obtain device status data from the edge node and extract the device status feature vector;

[0106] Calculate the device operation trend according to the device status feature vector;

[0107] Adopt the ARIMA time series model to predict the future operating status of the device and obtain the prediction result;

[0108] Combine the prediction result and the device status feature vector to generate control instruction parameters, and if the device operation trend deviates from the preset range, adjust the control instruction parameters;

[0109] Generate real-time control instructions according to the adjusted control instruction parameters;

[0110] Send the real-time control instructions to the edge node for execution.

[0111] Specifically, in this embodiment, the edge node serves as the front end for data collection and processing, and it collects device status information in real time, such as operating parameters like temperature, pressure, and vibration. Taking a cooling tower as an example, by analyzing historical operation data, the future water temperature change trend can be predicted, and the cooling fan speed can be adjusted in advance. By comparing the prediction results of the model with the actual operation status, potential abnormal situations can be discovered. The generation of control instruction parameters needs to comprehensively consider the prediction results and the current status. Taking an air compressor unit as an example, according to the prediction result of gas consumption and the system pressure status, the number of operating units and the load distribution are dynamically adjusted. When the prediction result shows that a gas consumption peak is about to occur, standby units are started in advance. The deviation monitoring of the operation trend is based on a preset safety threshold. Taking a chemical reactor as an example, when the temperature rising trend exceeds the safety range, the heating power and the stirring speed need to be adjusted in time. By analyzing the material ratio trend, the feeding rate is adjusted appropriately to maintain the reaction stability. The real-time adjustment of control instructions needs to consider the device response characteristics. Taking a central air conditioning system as an example, according to the load change prediction, the cooling capacity is adjusted step by step to avoid increased energy consumption caused by frequent start-stop. The system also needs to consider the linkage relationship between devices to ensure the coordinated operation of multiple subsystems. When the edge node executes control instructions, the real-time and reliability of the instructions need to be ensured. Taking a three-dimensional warehousing system as an example, the motion control of the stacker crane needs to respond in real time, and based on the storage location status and the inbound and outbound tasks, the motion trajectory is dynamically planned. Through the feedback analysis of the execution results, the control strategy is continuously optimized.

[0112] The present invention discloses a real-time collaborative control method for an industrial Internet based on edge computing. By collecting multi-source heterogeneous data of industrial devices, time series alignment and timestamp marking are performed to generate a time series data set. In the case of chaotic data arrival time series, a sliding window mechanism is adopted for segmented processing to ensure the time series consistency of the data within the window. Subsequently, data cleaning and format conversion are carried out to generate intermediate data in a unified format. The lightweight K-means clustering algorithm is used to extract features and generate a device status feature vector. Based on the real-time monitoring information of network load and transmission path, the computing resource allocation strategy of the edge node is dynamically adjusted to preferentially process data of high-priority devices. Finally, a collaborative control algorithm based on the ARIMA model is deployed at the edge node, and real-time control instructions are generated by combining the device status feature vector. The present invention realizes the efficient processing and real-time intelligent control of industrial device data, and improves the reliability and control accuracy of the system.

[0113] The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed in the present application should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A real-time collaborative control method for industrial Internet based on edge computing, characterized in that, Including: Obtain multi-source heterogeneous data of industrial equipment, extract the sampling frequency, data accuracy, and format information of the multi-source heterogeneous data, and generate a standardized data set; Perform time series alignment processing on the standardized data set at the edge node to generate a time series data set; If there is a situation where the data arrival time series is disrupted in the time series data set, segment the data to obtain a standardized segmented data set; Remove outliers and redundant information according to the standardized segmented data set to obtain a cleaned data set; According to the device status and control strategy requirements, perform format conversion on the cleaned data set to generate intermediate data in a unified format; Extract features from the intermediate data in the unified format to generate a device status feature vector; According to the real-time monitoring information of network load and transmission path, dynamically adjust the computing resource allocation strategy of the edge node, give priority to processing data of high-priority devices, and generate a resource allocation result, where the high-priority devices are determined according to the operating status and failure risk level of the devices; Deploy a collaborative control algorithm based on the ARIMA time series model at the edge node, combine the device status feature vector, generate real-time control instructions, and complete real-time collaborative control of the industrial Internet.

2. The real-time collaborative control method for industrial Internet based on edge computing according to claim 1, wherein Obtaining multi-source heterogeneous data of industrial equipment and extracting the sampling frequency, data accuracy, and format information of the multi-source heterogeneous data includes: According to the obtained multi-source heterogeneous data, extract the sampling frequency, data accuracy, and format information of the data; According to the sampling frequency of the data, judge the time interval of data collection. If the time interval exceeds the preset threshold, adjust the sampling frequency; Analyze the accuracy of the data through the data accuracy. If the accuracy is lower than the preset standard, use a data correction algorithm for correction; According to the format information, determine the storage method and encoding specification of the data. If the formats are inconsistent, perform format conversion processing; Integrate the processed multi-source heterogeneous data according to the unified sampling frequency, data accuracy, and format information to generate a standardized data set.

3. The real-time collaborative control method for industrial Internet based on edge computing according to claim 1, characterized in that, Performing time series alignment processing on the standardized data set at the edge node to generate a time series data set includes: According to the preset time series alignment rule, add a timestamp mark to the standardized data set at the edge node to generate a time series data set.

4. The real-time collaborative control method for industrial Internet based on edge computing according to claim 1, wherein, If there is a situation where the data arrival time series is disrupted in the time series data set, segmenting the data to obtain a standardized segmented data set includes: Obtain the time series data set, identify the time series status of data arrival. If the time series is disrupted, use a sliding window mechanism to segment the data; According to the segmented data, determine the time series consistency within each window to generate a segmented time series data set; According to the segmented time series data set, extract the timestamp, sampling frequency, and data accuracy of the data, and judge whether the data meets the preset standard; if the data accuracy is lower than the preset standard, use a data correction algorithm for correction; Integrate the processed segmented data to generate a standardized segmented data set.

5. The real-time collaborative control method for industrial Internet based on edge computing according to claim 1, characterized in that Removing outliers and redundant information according to the standardized segmented data set to obtain a cleaned data set includes: Perform outlier detection on the standardized segmented dataset. If the data value exceeds the preset threshold, mark it as an outlier and remove it. For the redundant values in the standardized segmented dataset, determine whether there are duplicate data based on the matching relationship between the sampling rate and the timestamp. If there are duplicates, perform deduplication to obtain the cleaned dataset.

6. The real-time collaborative control method for industrial Internet based on edge computing according to claim 1, characterized in that, According to the device status and control strategy requirements, perform format conversion on the cleaned dataset to generate intermediate data in a unified format, including: Extract key feature values from the cleaned dataset, obtain the feature information of the data, and generate a feature dataset. Analyze the feature dataset to identify the regular patterns in the data and generate a pattern recognition result. According to the pattern recognition result, use a standardization algorithm to normalize the data and generate intermediate data in a unified format.

7. The real-time collaborative control method for industrial Internet based on edge computing according to claim 1, characterized in that Extract features from the intermediate data in the unified format to generate a device status feature vector, including: Extract features from the intermediate data in the unified format to obtain the key features and regular information of the data. According to the key features and regular information of the data, use the support vector machine algorithm to optimize the analysis of the device status and control strategy and generate a device status feature vector.

8. The real-time collaborative control method for industrial Internet based on edge computing according to claim 1, wherein According to the real-time monitoring information of the network load and transmission path, dynamically adjust the computing resource allocation strategy of the edge node, give priority to processing the data of high-priority devices, and generate a resource allocation result, including: Use a preset monitoring frequency to obtain real-time monitoring data from the network load and transmission path and generate monitoring information on the network load and transmission path. According to the load value and path delay in the monitoring information, determine whether the network is in a high-load state. If it is in a high-load state, trigger the dynamic adjustment mechanism. For the data transmission requests of high-priority devices, reserve the computing resources of the edge node for high-priority devices. According to the real-time changes in the network load and transmission path, adjust the computing resource allocation ratio of the edge node to obtain an optimized resource allocation strategy. Through the optimized resource allocation strategy, give priority to processing the data of high-priority devices and generate a resource allocation result.

9. The real-time collaborative control method for industrial Internet based on edge computing according to claim 1, characterized in that, The high-priority devices are determined according to the operating status and failure risk level of the devices, including: Obtain the operating status and failure history data of all devices and calculate the failure risk level of each device. According to the failure risk level and device operating status, determine the device priority list, and the devices at the top of the list are high-priority devices.

10. The real-time collaborative control method for industrial Internet based on edge computing according to claim 1, characterized in that, Deploy a collaborative control algorithm based on the ARIMA time series model at the edge node. Combine the device status feature vector to generate real-time control instructions, including: Obtain device status data from the edge node and extract the device status feature vector. According to the device status feature vector, calculate the device operation trend. Use the ARIMA time series model to predict the future operating status of the device and obtain a prediction result. Combine the prediction result and the device status feature vector to generate control instruction parameters. If the device operation trend deviates from the preset range, adjust the control instruction parameters. According to the adjusted control instruction parameters, generate real-time control instructions. Send the real-time control instructions to the edge node for execution.

Images