A semantic extraction method, device, medium, and product based on edge devices.

By directly mapping and encoding raw data into capability semantics within the edge device, the problems of insufficient computing power and slow response of cloud servers are solved, enabling rapid business decision-making and convenient anomaly troubleshooting.

CN123115395APending Publication Date: 2026-11-17天津新智感知科技有限公司
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611531486.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-30
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In existing technologies, semantic extraction schemes for edge devices result in insufficient computing power and slow response speeds for cloud servers, and make it difficult to trace back and troubleshoot in abnormal situations.

Method used

Within the edge device, through data collection, trusted processing, judgment rule mapping, and self-evaluation, the raw data is directly mapped into capability semantics containing business intent, execution path, and effect evaluation, and then encoded and sent to the business intelligence agent for decision-making.

Benefits of technology

It improves the business decision response speed of cloud servers, simplifies the process of anomaly backtracking and troubleshooting, and enables edge devices to become capability nodes that can be invoked by business intelligence agents.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123115395A_ABST
    Figure CN123115395A_ABST
Patent Text Reader

Abstract

This invention relates to a semantic extraction method, device, medium, and product based on edge devices. The method includes: periodically collecting raw data according to data acquisition parameters; performing trust processing on the collected raw data to obtain trustworthy raw data; determining the business intent, execution path, and effect evaluation mapped to the trustworthy raw data, and defining the business intent, execution path, and effect evaluation as the capability semantics of the trustworthy raw data; determining a self-evaluation result; if the self-evaluation result is a pass, encoding the capability semantics to obtain encoded capability semantics; sending the encoded capability semantics to a business agent, so that the business agent can decode it to obtain the capability semantics, and make business decisions based on the capability semantics. Embodiments of this invention can directly map raw data into capability semantics containing business intent, execution path, and effect evaluation within the edge device, enabling the business agent to directly use the capability semantics for business decisions.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the fields of industrial Internet of Things (IoT) technology, embedded artificial intelligence technology, and business intelligent agent application technology, and in particular to a semantic extraction method, device, medium, and product based on edge devices. Background Technology

[0002] Industrial sites typically contain numerous edge devices. Edge devices are electronic devices installed in industrial sites to collect data and execute operations based on control commands. They execute operations according to control commands sent by a cloud server that controls the industrial site. Edge devices in industrial sites include pressure sensors, concentration sensors, temperature and humidity sensors, and various actuators installed in the aisles. These devices collect data from the industrial site at a fixed frequency, obtaining raw data. After collecting the raw data, semantic extraction is performed, mapping the raw data into capability semantics that contain business intent. This allows the business agent in the cloud server to use the capability semantics to make business decisions. Capability semantics is information used to help confirm the business operations to be performed after acquiring the raw data. Business intent is text describing the business operations to be performed after acquiring the raw data.

[0003] In related technologies, a common semantic extraction scheme involves the edge device uploading the collected raw data to a cloud server via a communication network. After acquiring the raw data, the business agent on the cloud server maps it to capability semantics containing business intent, thus transforming the raw data into capability semantics with business meaning. This entire semantic extraction process is lengthy, slow in decision-making, and prone to errors. The semantic extraction schemes in related technologies cannot provide directly usable capability semantics to the cloud server; the cloud server must convert the raw data into directly usable capability semantics. This results in most of the cloud server's computing power being used for basic tasks such as parsing, transformation, and feature extraction required to convert raw data into capability semantics, leaving relatively little computing power for actual business decision-making. Furthermore, the raw data is transmitted from the edge device to the cloud server and undergoes multiple rounds of processing before a decision is reached, resulting in a very slow response time. In normal scenarios, it may take hundreds of milliseconds or even minutes to reach a decision, and in high-risk scenarios, it may be impossible to reach a decision in time. In addition, capability semantics lacks execution paths and effect evaluation, making it difficult to trace and investigate anomalies. Execution paths are information used to record the process of determining business intent. Effectiveness evaluation is information used to help confirm the effectiveness of business operations performed based on business intent mapped to the original data. Summary of the Invention

[0004] This invention provides a semantic extraction method, device, medium, and product based on edge devices to solve the problems in related technologies where edge device-based semantic extraction schemes result in most cloud server computing power not being used for business decisions, slow response times, and difficulty in backtracking and troubleshooting when anomalies occur.

[0005] According to one aspect of the present invention, a semantic extraction method based on an edge device is provided, comprising: Collect raw data periodically according to the data collection parameters; The collected raw data is processed to obtain reliable raw data; Based on the set of judgment rules, the business intent, execution path, and effect evaluation mapped to the trusted original data are determined, and the business intent, execution path, and effect evaluation are determined as the capability semantics of the trusted original data. Perform a self-evaluation on the capability semantics and determine the self-evaluation result of the capability semantics; If the self-assessment result is that the assessment is passed, then the capability semantics are encoded according to the preset encoding rules to obtain the encoded capability semantics; The encoded capability semantics are sent to the business intelligence agent, so that the business intelligence agent can decode the encoded capability semantics to obtain the capability semantics, and make business decisions based on the capability semantics.

[0006] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising: At least one processor; And a memory corresponding to the at least one processor, wherein the memory is either built-in memory in the at least one processor or external memory communicatively connected to the at least one processor; The memory stores a computer program that is executed by the at least one processor, which enables the at least one processor to perform the semantic extraction method based on the edge device according to any embodiment of the present invention.

[0007] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the semantic extraction method based on an edge device as described in any embodiment of the present invention.

[0008] According to another aspect of the present invention, a computer program product is provided, the computer program product comprising a computer program that, when executed by a processor, implements the semantic extraction method based on an edge device as described in any embodiment of the present invention.

[0009] The technical solution of this invention involves periodically collecting raw data based on data acquisition parameters. For each piece of raw data collected, the following operations are performed: the collected raw data undergoes reliable processing to obtain reliable raw data; then, based on a set of judgment rules, the business intent, execution path, and effect evaluation mapped to the reliable raw data are determined, and the business intent, execution path, and effect evaluation are defined as the capability semantics of the reliable raw data; the capability semantics are then self-evaluated to determine the self-evaluation result; if the self-evaluation result is a pass, the capability semantics are encoded according to preset encoding rules to obtain encoded capability semantics; finally, the encoded capability semantics are sent to the business intelligence agent so that the business intelligence agent can decode the encoded capability semantics to obtain the capability semantics, and make business decisions based on the capability semantics. The technical solution of this invention addresses the problem in related technologies where semantic extraction schemes based on edge devices result in most cloud server computing power not being used for business decision-making, leading to slow response times and difficulty in backtracking and troubleshooting when anomalies occur. This invention allows for direct semantic extraction of raw data within the edge device, mapping the raw data into capability semantics containing business intent, execution path, and effect evaluation. This semantics is then transmitted to the business agent in the cloud server, eliminating the need for the business agent to infer the meaning of the raw data and allowing it to directly use the capability semantics for business decision-making. This ensures that most of the cloud server's computing power is used for business decision-making rather than basic tasks, improving response speed and facilitating backtracking and troubleshooting based on execution paths and effect evaluations. The edge device truly becomes a capability node that the business agent can invoke.

[0010] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 This is a flowchart of a semantic extraction method based on an end-side device provided in Embodiment 1 of the present invention.

[0013] Figure 2 This is a flowchart of a semantic extraction method based on an end-side device provided in Embodiment 2 of the present invention.

[0014] Figure 3A schematic diagram of the structure of an electronic device for implementing the semantic extraction method based on the end-side device in this embodiment of the invention. Detailed Implementation

[0015] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0016] It should be noted that the terms "target," "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising," "including," and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0017] Example 1 Figure 1 This is a flowchart illustrating a semantic extraction method based on an edge device, as provided in Embodiment 1 of the present invention. This embodiment is applicable to semantic extraction of raw data collected by an edge device, mapping the raw data into capability semantics containing business intent, so that a business agent in a cloud server can use the capability semantics to make business decisions. This method can be executed by an edge-device-based semantic extraction device, which can be implemented in hardware and / or software and configured within the edge device. An edge device is an electronic device installed in an industrial site for data acquisition and operation execution according to control commands. For example, an edge device may be a pressure sensor, concentration sensor, temperature and humidity sensor, or actuator installed in a channel in the industrial site. The channel in the industrial site may be a channel for transmitting a specified gas or liquid, including but not limited to gas pipelines. Figure 1 As shown, the method includes: Step 101: Collect raw data periodically according to the data collection parameters.

[0018] Specifically, the raw data that the end-side device needs to collect can be the pressure in the channel where the end-side device is located, the gas concentration in the channel where the end-side device is located, the temperature at the location of the end-side device, or the humidity at the location of the end-side device.

[0019] Optionally, the data acquisition parameters include a desired sampling rate and a desired analog-to-digital converter resolution that match the original data; according to the data acquisition parameters, the original data is acquired periodically, including: acquiring the original data periodically according to the desired sampling rate and the desired analog-to-digital converter resolution, and storing the acquired original data in a preset circular buffer.

[0020] Data acquisition parameters can be a desired sampling rate and a desired analog-to-digital converter (ADC) resolution that match the raw data. The desired sampling rate can be a pre-selected sampling rate suitable for use when periodically acquiring the raw data. The desired sampling rate can be selected based on the frequency characteristics of the raw data. For example, if the main frequency of the raw data is 5Hz, the desired sampling rate matching the raw data is 20Hz, which is a 4x frequency, capturing the raw data completely while keeping the data volume small. The desired ADC resolution can be a pre-selected ADC resolution suitable for use when periodically acquiring the raw data. The desired ADC resolution can be selected based on the accuracy requirements of the raw data. For example, the desired ADC resolution is 16-bit or 12-bit.

[0021] Raw data can be collected periodically according to the desired sampling rate and desired analog-to-digital converter resolution, and the collected raw data can be stored in a preset ring buffer. The preset ring buffer can be a ring buffer set in the end-side device for storing the collected raw data.

[0022] Under normal circumstances, the default ring buffer retains only a small amount of raw data. Once an anomaly is detected, the default ring buffer can automatically retain the raw data from 10 seconds before and after the anomaly, approximately 400 raw data entries, occupying only 800 bytes of RAM. Thus, it does not affect the normal operation of the edge device, while providing context for subsequent analysis.

[0023] Step 102: Perform credibility processing on the collected raw data to obtain credible raw data.

[0024] Specifically, reliable raw data can refer to clean, reliable, and traceable raw data. Reliable processing of collected raw data can refer to detecting and processing the collected raw data to determine whether it is clean, reliable, and traceable, or to determine whether the collected raw data is not clean, reliable, or traceable and therefore does not require further processing.

[0025] Optionally, the collected raw data is subjected to trust processing to obtain trustworthy raw data, including: detecting whether there are any abnormalities in the collected raw data according to the check code, and detecting whether the end-side device is in a fault state; if there are no abnormalities in the collected raw data and the end-side device is not in a fault state, then the collected raw data is subjected to temperature drift compensation processing and clock synchronization processing, and the processed raw data is determined as trustworthy raw data.

[0026] Anomalies in the raw data can refer to communication tampering or transmission errors. A 16-bit Cyclic Redundancy Check (CRC) code is automatically generated after each acquisition of raw data. The CRC check is used to detect anomalies in the acquired raw data, including: recalculating the raw data to generate a new 16-bit CRC code; comparing the recalculated CRC code with the automatically generated CRC code after data acquisition; if the recalculated CRC code matches, the acquired raw data is considered normal; otherwise, if the recalculated CRC code does not match, the acquired raw data is considered abnormal.

[0027] The detection of whether the end-side device is in a fault state includes: detecting whether the end-side device has short circuits, open circuits, or abnormal noise according to preset fault detection rules; if short circuits, open circuits, or abnormal noise are detected, the end-side device is determined to be in a fault state; if no short circuits, open circuits, or abnormal noise are detected, the end-side device is determined to be not in a fault state. The preset fault detection rules can be pre-set rules for detecting short circuits, open circuits, and abnormal noise in the end-side device. Therefore, if the end-side device itself has a problem, it can identify it first, avoiding mistaking false anomalies for real leaks.

[0028] Temperature variations can cause drift in edge devices, leading to errors in the acquired raw data. Temperature drift compensation processing corrects the acquired raw data, reducing the errors caused by temperature changes. A preset temperature drift compensation algorithm is a pre-set algorithm used to correct the raw data acquired by the edge device and reduce errors caused by temperature variations. Therefore, a preset temperature drift compensation algorithm can be used to correct the acquired raw data and reduce errors caused by temperature changes.

[0029] The raw data acquisition time can be the recorded time when the raw data was acquired. Clock synchronization processing of the acquired raw data can refer to unifying the acquisition time of the raw data with the acquisition time of raw data acquired by other end devices within 5 milliseconds, so that the raw data of each end device corresponds to the same time.

[0030] Under normal circumstances, if the collected raw data is free of anomalies and the end-side device is not faulty, the raw data after temperature drift compensation and clock synchronization is considered clean, reliable, and traceable. This processed raw data can be deemed reliable and further processing can proceed. Conversely, if the collected raw data is anomaly-prone or the end-side device is faulty, it can be determined that the collected raw data is not clean, reliable, or traceable, and further processing is unnecessary. Processing should wait for newly collected raw data.

[0031] Step 103: Based on the set of judgment rules, determine the business intent, execution path, and effect evaluation mapped to the trusted original data, and determine the business intent, execution path, and effect evaluation as the capability semantics of the trusted original data.

[0032] Specifically, the semantic capabilities of raw data can be information used to help confirm the business operations to be performed after the raw data is acquired. The business intent mapped to the raw data can be text describing the business operations to be performed after the raw data is acquired. The execution path mapped to the raw data can be information recording the process of determining the business intent mapped to the raw data. The effect evaluation of the mapping to the raw data can be information used to help confirm the effect of performing business operations based on the business intent mapped to the raw data.

[0033] Operational tasks can be those used to ensure the normal operation of an industrial site. For example, possible operational tasks include closing designated valves in the industrial site, opening designated valves in the industrial site, controlling the operation of ventilation devices in the industrial site, and sending abnormal notification messages to maintenance personnel. Each operational task is assigned identification information. The identification information for an operational task can be text used to uniquely identify the task.

[0034] A set of decision rules is a collection of multiple decision rules. Each decision rule can be a rule used to determine the business operation to be performed after obtaining raw data of a specified type, based on the numerical value of the raw data. The decision rules are not the same for each type of raw data. Typically, for each piece of reliable raw data, the set of decision rules contains decision rules that can be used to determine the business operation to be performed after obtaining the raw data.

[0035] Optionally, based on the set of judgment rules, the business intent, execution path, and effect evaluation mapped to the trusted original data are determined, including: determining the business intent mapped to the trusted original data through a business intent identification model and a set of judgment rules; determining the original data associated with the business intent, processing description information, and hit judgment rules as the execution path mapped to the trusted original data; calculating the confidence level and assessing the risk level of the business intent to determine the confidence level and risk level of the business intent; and determining the confidence level and risk level of the business intent as the effect evaluation of the mapping to the trusted original data.

[0036] Specifically, a business intent recognition model can be a model pre-trained by technicians and set up on edge devices to analyze and detect raw data and a set of judgment rules to determine the business intent mapped to the raw data. The input to the business intent recognition model is the set of judgment rules and the raw data for which the mapped business intent needs to be determined. The output of the business intent recognition model is the business intent mapped to the raw data and the hit judgment rules associated with the business intent. The hit judgment rules associated with the business intent can be the judgment rules used when determining the business intent. The set of judgment rules and the raw data for which the mapped business intent needs to be determined can be input into the business intent recognition model. The business intent recognition model analyzes and tests each judgment rule in the judgment rule set and the original data. It queries the rule set for judgment rules that can determine the business operation to be performed after acquiring the original data. Once a judgment rule is found, the model determines the business operation to be performed based on that rule. The identifier information of the business operation to be performed after acquiring the original data is identified as the business intent mapped to the original data. This judgment rule is then identified as the hit judgment rule associated with the business intent. The model outputs the business intent mapped to the original data and the hit judgment rule associated with that business intent. The model can retrieve the business intent mapped to the original data and the hit judgment rule associated with that business intent.

[0037] The business intent mapped to the trusted original data is determined by using a business intent recognition model and a set of judgment rules. This includes: inputting the trusted original data and the set of judgment rules into the business intent recognition model to obtain the business intent mapped to the trusted original data and the hit judgment rules associated with the business intent, which are output by the business intent recognition model.

[0038] The raw data associated with the business intent can refer to the initially collected raw data corresponding to trusted raw data. The processing description information associated with the business intent can be information describing the processing procedure of the initially collected raw data corresponding to trusted raw data. The processing description information may include trusted processing identification information. Trusted processing identification information can be pre-set text used to characterize the trusted processing procedure. The raw data associated with the business intent, the processing description information, and the hit determination rules are information that can be used to record the process of determining the business intent mapped to trusted raw data. The raw data associated with the business intent, the processing description information, and the hit determination rules can be used to determine the execution path mapped to trusted raw data.

[0039] The confidence level of a business intent can be a numerical value used to characterize the credibility of that intent. A higher confidence level indicates greater credibility. A pre-set confidence level algorithm can be used to calculate the confidence level of a business intent. The risk level of a business intent can be a numerical value used to characterize the probability of risk occurring after executing the business operation described by the intent. A higher risk level indicates a higher probability of risk occurring. A preset risk level assessment algorithm can be used to calculate the risk level of a business intent. Both the confidence level and risk level of a business intent can be used to help confirm the effects of executing business operations based on the intent. The confidence level and risk level of a business intent can be used as an effect assessment mapped to credible raw data.

[0040] The business intent, execution path, and effect evaluation mapped to trusted raw data are information that can help identify the business operations that need to be performed after the raw data is acquired. The business intent, execution path, and effect evaluation mapped to trusted raw data can be defined as the capability semantics of trusted raw data.

[0041] Optionally, if the confidence level of the business intent is less than the target value, the business intent can be reviewed. The target value can be a pre-set value. For example, the target value is 70%. Reviewing the business intent can mean determining that the business intent will no longer be issued, does not need to be processed further, and awaits processing of newly collected raw data.

[0042] Step 104: Perform a self-evaluation of the capability semantics and determine the self-evaluation result of the capability semantics.

[0043] Specifically, self-evaluation of capability semantics refers to assessing the accuracy, real-time performance, and resource consumption of capability semantics within the device itself. The self-evaluation result can be text used to characterize whether there are any anomalies in the accuracy, real-time performance, or resource consumption of the capability semantics. The self-evaluation result is either "evaluation passed" or "evaluation failed." A "passed" result indicates that there are no anomalies in the accuracy, real-time performance, or resource consumption of the capability semantics. A "failed" result indicates that there are anomalies in the accuracy, real-time performance, or resource consumption of the capability semantics.

[0044] Optionally, the capability semantics are self-evaluated to determine the self-evaluation result of the capability semantics, including: self-evaluating the capability semantics based on accuracy statistics, real-time statistics, and resource usage statistics to determine the self-evaluation result of the capability semantics.

[0045] Specifically, accuracy statistics include classification accuracy, false negative rate, and risk level deviation ratio. Classification accuracy can be the ratio of the number of correct capability semantics counted by the edge device to the total number of capability semantics. The total number of capability semantics can be the total number of capability semantics sent by the edge device to the business intelligence agent. The number of correct capability semantics can be the total number of capability semantics sent by the edge device to the business intelligence agent that has been confirmed as correct by the cloud server. False negative rate can be the ratio of the number of incorrect capability semantics counted by the edge device to the total number of capability semantics. The number of incorrect capability semantics can be the total number of capability semantics sent by the edge device to the business intelligence agent that has been confirmed as incorrect by the cloud server. Risk level deviation ratio can be the ratio of the number of risk level deviations counted by the edge device to the total number of capability semantics. The number of risk level deviations can be the total number of capability semantics containing risk levels that have been confirmed as having deviations by the cloud server, which have been included in the capability semantics sent by the edge device to the business intelligence agent.

[0046] Real-time statistics include response time. Response time can be the total time from data acquisition to the determination of capability semantics, as recorded by the edge device.

[0047] Resource usage statistics include computing power usage ratio, semantic random access usage, and semantic read-only usage. The computing power usage ratio can be the ratio of the semantic computing power of the endpoint device to the total computing power. Semantic computing power can be the computing power of the endpoint device used from the acquisition of raw data to the determination of the semantic capabilities. Total computing power can be the maximum computing power that the endpoint device can provide. Semantic random access usage can be the capacity of random access memory in the endpoint device used from the acquisition of raw data to the determination of the semantic capabilities. Semantic read-only usage can be the capacity of read-only memory in the endpoint device used from the acquisition of raw data to the determination of the semantic capabilities.

[0048] Optionally, based on accuracy statistics, real-time statistics, and resource usage statistics, a self-evaluation of the capability semantics is performed to determine the self-evaluation result, including: whether the classification accuracy is greater than a first value, whether the false negative rate is less than or equal to a second value, whether the risk level deviation ratio is greater than a third value, whether the response time is less than or equal to the target time, whether the computing power usage ratio is less than a fourth value, whether the semantic random usage is less than or equal to a first capacity threshold, and whether the semantic read-only usage is less than or equal to a second capacity threshold; if the classification accuracy is greater than the first value, the false negative rate is less than or equal to the second value, the risk level deviation ratio is greater than the third value, and the response time is less than or equal to the target value, then... If the normal value of the length, computing power utilization ratio is less than the fourth value, the peak value of the computing power utilization ratio is less than the fifth value, the semantic random utilization is less than or equal to the first capacity threshold, and the semantic read-only utilization is less than or equal to the second capacity threshold, then the self-assessment result of the capability semantics is determined to be an assessment pass. If the following are detected: classification accuracy is less than or equal to the first value, false negative rate is greater than the second value, risk level deviation ratio is less than or equal to the third value, response time is greater than the target time, the normal value of the computing power utilization ratio is greater than or equal to the fourth value, the peak value of the computing power utilization ratio is greater than or equal to the fifth value, the semantic random utilization is greater than the first capacity threshold, or the semantic read-only utilization is greater than the second capacity threshold, then the self-assessment result of the capability semantics is determined to be an assessment fail. The first, second, third, fourth, and fifth values ​​are four pre-set values. The target time is a pre-set time. The first and second capacity thresholds are two pre-set capacity thresholds. For example, the first value is 95%, the second value is 0.5%, the third value is 92%, the fourth value is 20%, and the fifth value is 28%. The target duration is 100 milliseconds or 10 milliseconds, the first capacity threshold is 32KB, and the second capacity threshold is 64KB. The normal value of the computing power utilization ratio refers to the long-term stable baseline value of the computing power utilization ratio, while the peak value of the computing power utilization ratio refers to the highest extreme value that the computing power utilization ratio can reach at a specific moment.

[0049] Step 105: If the self-evaluation result is that the evaluation is passed, then the capability semantics are encoded according to the preset encoding rules to obtain the encoded capability semantics.

[0050] Specifically, the preset encoding rules can be pre-set rules used to encode capability semantics and compress the amount of capability semantic data. Encoding capability semantics refers to the data obtained after encoding capability semantics according to the preset encoding rules. Encoding capability semantics includes, but is not limited to: encoding capability semantics into 16-bit binary data and encoding capability semantics into 8-bit binary data.

[0051] Optionally, the capability semantics are encoded according to a preset encoding rule to obtain encoded capability semantics, including: determining the type of the capability semantics; if the type of the capability semantics is basic semantics, then the capability semantics are encoded according to a first encoding rule to obtain encoded capability semantics; if the type of the capability semantics is enhanced semantics, then the capability semantics are encoded according to a second encoding rule to obtain encoded capability semantics.

[0052] The type of capability semantics can be information used to characterize the importance of capability semantics. The type of capability semantics can be basic semantics or enhanced semantics. A basic semantics type indicates a lower importance of the capability semantics. An enhanced semantics type indicates a higher importance of the capability semantics. A semantic type recognition model is set up in the edge device. The semantic type recognition model can be a model pre-trained by technicians and set up in the edge device to determine the type of capability semantics. The input of the semantic type recognition model is capability semantics. The output of the semantic type recognition model is the type of capability semantics. Capability semantics can be input into the semantic type recognition model. The semantic type recognition model analyzes and detects the capability semantics, selects the type of capability semantics from basic and enhanced semantics, and outputs the type of capability semantics. The type of capability semantics output by the semantic type recognition model can be obtained. Preset encoding rules can include a first encoding rule and a second encoding rule. The first encoding rule can be a rule suitable for encoding capability semantics of type basic semantics, compressing the data volume of capability semantics. The second encoding rule can be a rule suitable for encoding capability semantics of type enhanced semantics, compressing the data volume of capability semantics.

[0053] Step 106: Send the encoded capability semantics to the business intelligence agent so that the business intelligence agent can decode the encoded capability semantics to obtain the capability semantics and make business decisions based on the capability semantics.

[0054] Specifically, a business intelligence agent can be an agent configured on a cloud server used to control an industrial site and make business decisions. Making business decisions means determining the business operation that needs to be performed, and then controlling the edge device capable of performing that operation to execute it by sending control commands. Encoded capability semantics can be sent to the business intelligence agent. After receiving the encoded capability semantics, the business intelligence agent can decode it to obtain the capability semantics, and then directly make business decisions based on the capability semantics.

[0055] The decision-making results of the business intelligence agent are fed back to the edge devices. If the actual business operation executed by the decision is inconsistent with the business operation expected by the capability semantics, the edge devices can adjust parameters and update rules based on the feedback information, forming a closed loop of continuous optimization.

[0056] Optionally, after performing a self-evaluation of the capability semantics and determining the self-evaluation result, the method further includes: if the self-evaluation result is a failure, then reviewing the capability semantics. Reviewing the capability semantics may mean determining that the capability semantics will no longer be issued, does not require further processing, and awaits processing of newly collected raw data.

[0057] The technical solution of this invention involves periodically collecting raw data based on data acquisition parameters. For each piece of raw data collected, the following operations are performed: the collected raw data undergoes reliable processing to obtain reliable raw data; then, based on a set of judgment rules, the business intent, execution path, and effect evaluation mapped to the reliable raw data are determined, and the business intent, execution path, and effect evaluation are defined as the capability semantics of the reliable raw data; the capability semantics are then self-evaluated to determine the self-evaluation result; if the self-evaluation result is a pass, the capability semantics are encoded according to preset encoding rules to obtain encoded capability semantics; finally, the encoded capability semantics are sent to the business intelligence agent so that the business intelligence agent can decode the encoded capability semantics to obtain the capability semantics, and make business decisions based on the capability semantics. The technical solution of this invention addresses the problem in related technologies where semantic extraction schemes based on edge devices result in most cloud server computing power not being used for business decision-making, leading to slow response times and difficulty in backtracking and troubleshooting when anomalies occur. This invention allows for direct semantic extraction of raw data within the edge device, mapping the raw data into capability semantics containing business intent, execution path, and effect evaluation. This semantics is then transmitted to the business agent in the cloud server, eliminating the need for the business agent to infer the meaning of the raw data and allowing it to directly use the capability semantics for business decision-making. This ensures that most of the cloud server's computing power is used for business decision-making rather than basic tasks, improving response speed and facilitating backtracking and troubleshooting based on execution paths and effect evaluations. The edge device truly becomes a capability node that the business agent can invoke.

[0058] In one optional embodiment of the present invention, the capability semantics including business intent, execution path and effect evaluation can be packaged into one frame, or split into multiple frames for time-division reporting, or differential encoding can be used to report only the changed parts.

[0059] In one optional embodiment of the present invention, the preset encoding rules used can be set according to business requirements.

[0060] In one optional embodiment of the present invention, in addition to basic semantics and enhanced semantics, the types of capability semantics may also include emergency semantics, aggregation semantics, multi-device collaborative semantics, etc.

[0061] In one optional embodiment of the present invention, the method for detecting whether the collected raw data is abnormal, the method for detecting whether the end-side device is in a fault state, the temperature drift compensation method, and the clock synchronization method can be set according to business requirements.

[0062] In one optional embodiment of the present invention, the hardware platform for implementing the semantic extraction method based on the edge device can be configured according to business requirements.

[0063] Example 2 Figure 2 This is a flowchart illustrating a semantic extraction method based on an edge device, as provided in Embodiment 2 of the present invention. This embodiment of the present invention can be combined with various optional solutions from one or more of the above embodiments. For example... Figure 2 As shown, the method includes: Step 201: Collect raw data periodically according to the data collection parameters.

[0064] Step 202: Perform credibility processing on the collected raw data to obtain credible raw data.

[0065] Step 203: Determine the business intent mapped to the trusted original data through the business intent identification model and the set of judgment rules.

[0066] Step 204: Determine the original data associated with the business intent, the processing description information, and the hit judgment rules as the execution path mapped to the trusted original data.

[0067] Step 205: Calculate the confidence level and assess the risk level of the business intent, determine the confidence level and risk level of the business intent, and use the confidence level and risk level of the business intent as an evaluation of the mapping effect with the trusted original data.

[0068] Step 206: Determine the business intent, the execution path, and the effect evaluation as the capability semantics of the trusted raw data.

[0069] Step 207: Perform a self-evaluation of the capability semantics and determine the self-evaluation result of the capability semantics.

[0070] Step 208: If the self-evaluation result is that the evaluation is passed, then the capability semantics are encoded according to the preset encoding rules to obtain the encoded capability semantics.

[0071] Step 209: Send the encoded capability semantics to the business intelligence agent so that the business intelligence agent can decode the encoded capability semantics to obtain the capability semantics and make business decisions based on the capability semantics.

[0072] The technical solution of this invention can perform semantic extraction on raw data within the edge device, determine the business intent mapped to the raw data, and identify execution paths and effect evaluations that can be used for backtracking and investigation. The raw data is mapped into capability semantics containing business intent, execution paths, and effect evaluations, and then transmitted to the business agent in the cloud server. This allows the business agent to make business decisions directly using capability semantics without having to infer the meaning of the raw data. This means that most of the cloud server's computing power is used for business decision-making rather than basic tasks, improving response speed and facilitating backtracking and investigation based on execution paths and effect evaluations. The edge device truly becomes a capability node that the business agent can invoke.

[0073] Example 3 Figure 3 A schematic diagram of an electronic device 10, which can be used to implement the semantic extraction method based on an edge device according to embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, electronic devices, blade electronic devices, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0074] like Figure 3As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory 12 or a random access memory 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the read-only memory 12 or loaded from storage unit 18 into the random access memory 13. The random access memory 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, read-only memory 12, and random access memory 13 are interconnected via a bus 14. An input / output interface 15 is also connected to the bus 14.

[0075] Multiple components in electronic device 10 are connected to input / output interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of monitors, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0076] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, central processing units, graphics processing units, various special-purpose artificial intelligence computing chips, various processors running machine learning model algorithms, digital signal processors, and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as semantic extraction methods based on edge devices.

[0077] In some embodiments, the device-based semantic extraction method can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via read-only memory 12 and / or communication unit 19. When the computer program is loaded into random access memory 13 and executed by processor 11, one or more steps of the device-based semantic extraction method described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform the device-based semantic extraction method by any other suitable means (e.g., by means of firmware).

[0078] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays, application-specific integrated circuits (ASICs), application-specific standard products (ASICs), systems-on-a-chip (SoCs), payload programmable logic devices (PLCs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0079] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or electronic device.

[0080] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory, optical fibers, portable compact disk read-only memory, optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0081] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a cathode ray tube or liquid crystal display monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0082] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0083] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product within the cloud computing service system. This addresses the shortcomings of traditional physical hosts and virtual private servers, such as high management difficulty and weak business scalability.

[0084] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0085] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for extracting semantics based on end-side device, characterized in that, Applied to end-side devices, including: Collect raw data periodically according to the data collection parameters; The collected raw data is processed to obtain reliable raw data; Based on the set of judgment rules, the business intent, execution path, and effect evaluation mapped to the trusted original data are determined, and the business intent, execution path, and effect evaluation are determined as the capability semantics of the trusted original data. Perform a self-evaluation on the capability semantics and determine the self-evaluation result of the capability semantics; If the self-assessment result is that the assessment is passed, then the capability semantics are encoded according to the preset encoding rules to obtain the encoded capability semantics; The encoded capability semantics are sent to the business intelligence agent, so that the business intelligence agent can decode the encoded capability semantics to obtain the capability semantics, and make business decisions based on the capability semantics.

2. The end-side device based semantic extraction method of claim 1, wherein, The data acquisition parameters include the desired sampling rate and the desired analog-to-digital converter resolution that match the raw data; Based on the data acquisition parameters, raw data is collected periodically, including: According to the desired sampling rate and the desired analog-to-digital converter resolution, raw data is collected periodically and stored in a preset circular buffer.

3. The end-side device based semantic extraction method of claim 1, wherein, The collected raw data undergoes reliability processing to obtain reliable raw data, including: The system detects whether there are any abnormalities in the raw data collected based on the verification code, and also detects whether the end-side device is in a faulty state. If the raw data collected is not abnormal and the end device is not in a fault state, then the raw data collected is subjected to temperature drift compensation and clock synchronization processing, and the processed raw data is determined as reliable raw data.

4. The end-side device based semantic extraction method of claim 1, wherein, Based on the set of judgment rules, determine the business intent, execution path, and effect evaluation mapped to the trusted original data, including: The business intent mapped to the trusted original data is determined by using a business intent recognition model and a set of judgment rules. The raw data associated with the business intent, the processing description information, and the hit judgment rules are determined as the execution path mapped to the trusted raw data; The confidence level and risk level of the business intent are calculated and assessed to determine the confidence level and risk level of the business intent. The confidence level and risk level of the business intent are then used to evaluate the effectiveness of mapping with the trusted original data.

5. The semantic extraction method based on end-side devices according to claim 1, characterized in that, The self-evaluation of the capability semantics is performed, and the self-evaluation result of the capability semantics is determined, including: Based on accuracy statistics, real-time statistics, and resource usage statistics, the capability semantics are self-evaluated to determine the self-evaluation result of the capability semantics.

6. The end-side device based semantic extraction method of claim 1, wherein, After performing a self-evaluation of the capability semantics and determining the self-evaluation result of the capability semantics, the method further includes: If the self-assessment result is that the assessment fails, the semantics of the capability will be reviewed.

7. The end-side device based semantic extraction method of claim 1, wherein, The capability semantics are encoded according to a preset encoding rule to obtain encoded capability semantics, including: Determine the type of the capability semantics; If the type of the capability semantics is basic semantics, then the capability semantics are encoded according to the first encoding rule to obtain the encoded capability semantics; If the type of the capability semantics is enhanced semantics, then the capability semantics are encoded according to the second encoding rule to obtain the encoded capability semantics.

8. An electronic device, comprising: The electronic device includes: At least one processor; And a memory corresponding to the at least one processor, wherein the memory is either built-in memory in the at least one processor or external memory communicatively connected to the at least one processor; The memory stores a computer program that is executed by the at least one processor, which enables the at least one processor to perform the semantic extraction method based on the edge device as described in any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the semantic extraction method based on any one of claims 1-7.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the semantic extraction method based on any one of claims 1-7.