A new energy data acquisition device and method for multiple application scenarios

Abstract

This invention relates to the field of new energy data acquisition methods, specifically to a new energy data acquisition device and method for multiple application scenarios. The method includes: receiving acquisition intentions submitted by upper-layer applications; converting the acquisition intentions into data requirement tuples based on a pre-built resource dependency knowledge graph; when multiple acquisition intentions exist simultaneously, performing conflict detection on the data requirement tuples of each acquisition intention; taking the highest frequency and highest precision requirement of the same variable as basic acquisition parameters; obtaining the current available communication bandwidth, edge node computing power, and storage space resource status; downgrading and adjusting the acquisition parameters according to the priority of each acquisition intention; generating an acquisition strategy for each sensor or device; distributing the acquisition strategy to the edge acquisition gateway; and then performing data acquisition. This invention can dynamically adjust the acquisition strategy and rationally allocate resources according to the differentiated needs of multiple application scenarios when acquiring new energy data.

Description

Technical Field

[0001] This invention relates to the field of new energy data acquisition methods, and in particular to a new energy data acquisition device and method for multiple application scenarios. Background Technology

[0002] With the large-scale deployment of new energy power generation (such as photovoltaic and wind power), energy storage systems, and electric vehicle charging facilities, new energy data acquisition systems need to simultaneously serve multiple upper-level applications, including power prediction, equipment fault diagnosis, power quality analysis, electricity billing, and carbon footprint tracking. Different applications have significantly different requirements for the frequency, real-time performance, accuracy, and data dimensions of the collected data. For example, ultra-short-term power prediction requires irradiance sampling frequencies at the second or even sub-second level, with a latency of less than 200 milliseconds; while monthly electricity billing only requires active energy readings at the 15-minute level, with extremely low real-time requirements.

[0003] Traditional fixed-period data acquisition methods typically employ a "one-collection-multiple-use" or "configuration according to the highest requirements" strategy. This means that all sensors collect data at the highest frequency and with the highest precision, and all raw data is uploaded to the cloud. This approach results in a large amount of redundant data occupying communication bandwidth, edge node computing power, and cloud storage space. For example, to meet the high-frequency requirements of power prediction, all power data is collected at a frequency of seconds, but billing applications only require data at the minute level, resulting in a large amount of useless data consuming network and storage resources.

[0004] For the reasons mentioned above, there is an urgent need for a new energy data acquisition method that can dynamically adjust the acquisition strategy and rationally allocate resources according to the differentiated needs of multiple application scenarios. Summary of the Invention

[0005] The purpose of this invention is to provide a new energy data acquisition device and method for multiple application scenarios, which can dynamically adjust the acquisition strategy and rationally allocate resources according to the differentiated needs of multiple application scenarios when acquiring new energy data.

[0006] To achieve the above objectives, in a first aspect, the present invention provides a new energy data acquisition method for multiple application scenarios, comprising: S1 receives the collection intent submitted by the upper layer application and converts the collection intent into a data requirement tuple based on the pre-built resource dependency knowledge graph. When multiple acquisition intentions exist simultaneously, S2 performs conflict detection on the data requirement tuples of each acquisition intention, takes the highest frequency and highest precision requirement of the same variable as the basic acquisition parameters, obtains the current available communication bandwidth, edge node computing power and storage space resource status, downgrades and adjusts the acquisition parameters according to the priority of each acquisition intention, generates the acquisition strategy for each sensor or device, and sends the acquisition strategy to the edge acquisition gateway. The S3 edge gateway drives physical sensors or smart devices to collect data according to the received acquisition strategy. The process includes: dynamically adjusting the sampling frequency according to the data change rate, temporarily increasing the sampling frequency when the data change rate exceeds a preset threshold, and restoring it to the basic cycle after the data stabilizes; uploading data only when the data change exceeds a preset dead zone or meets preset reporting conditions; and dynamically activating edge preprocessing operators according to the instructions in the acquisition strategy, uploading only the calculation results or feature values.

[0007] The collection intent includes the application type, the data variables of interest, the data quality requirements, and the time range.

[0008] The data requirement tuple includes a set of variables, a minimum sampling frequency, a maximum allowable delay, spatial granularity, and priority.

[0009] In step S1, the collection intent is submitted using a structured JSON template, and the resource dependency knowledge graph stores the mapping relationship between application type, physical quantity, sensor type, communication protocol, and data attributes.

[0010] Specifically, the collision detection in step S2 includes: Compare the sampling frequency and accuracy requirements of different acquisition intentions for the same data variable, and take the highest value as the basic acquisition parameter for that variable; if there are differences in the reporting conditions or preprocessing operators for the same variable for different intentions, then configure according to the intention with the highest priority.

[0011] Specifically, step S3, dynamically adjusting the sampling frequency, includes: The edge gateway calculates the data change rate in real time. When the change rate exceeds the first threshold, the sampling frequency is increased to the preset maximum frequency. When the change rate is lower than the second threshold and the duration exceeds the set time window, the sampling frequency is gradually restored to the basic cycle frequency.

[0012] In step S3, the preset reporting conditions specifically include: Data values ​​exceeding the set upper or lower limits, data change rate exceeding the set threshold, data prediction residuals exceeding the set range, or abnormal feature values ​​output by the edge preprocessing operator exceeding the set threshold.

[0013] Secondly, the present invention also provides a new energy data acquisition device for multiple application scenarios, including an intent parsing module, a strategy generation module and an adaptive acquisition execution module, wherein the intent parsing module, the strategy generation module and the adaptive acquisition execution module are connected in sequence. The intent parsing module is used to receive the collection intent submitted by the upper layer application and convert the collection intent into a data requirement tuple based on the pre-built resource dependency knowledge graph. The strategy generation module is used to perform conflict detection on the data requirement tuples of each collection intention when multiple collection intentions exist at the same time, take the highest frequency and highest precision requirement of the same variable as the basic collection parameters, obtain the current available communication bandwidth, edge node computing power and storage space resource status, downgrade the collection parameters according to the priority of each collection intention, generate the collection strategy for each sensor or device, and send the collection strategy to the edge collection gateway. The adaptive acquisition and execution module is deployed on the edge gateway and is used to drive physical sensors or smart devices to acquire data according to the received acquisition strategy. The execution process includes: dynamically adjusting the sampling frequency according to the data change rate; temporarily increasing the sampling frequency when the data change rate exceeds a preset threshold; and restoring to the basic cycle after the data stabilizes; uploading data only when the data change exceeds a preset dead zone or meets preset reporting conditions; and dynamically activating edge preprocessing operators according to the instructions in the acquisition strategy, and uploading only the calculation results or feature values.

[0014] This invention provides a new energy data acquisition device and method for multiple application scenarios. Through intent parsing and multi-intent merging, it automatically selects the highest demand for the same variable as the basic acquisition parameter, avoiding redundant configuration. Combined with resource constraint solving, it prioritizes and degrades the acquisition quality of less important tasks when bandwidth, computing power, or storage is limited. Compared to methods that use a fixed highest frequency acquisition method, this reduces communication traffic and storage space usage, significantly lowering operating costs. Upper-layer applications only need to submit structured acquisition intents (including application type, variables of interest, quality requirements, etc.), which are automatically converted into data requirement tuples through a pre-built knowledge graph, generating corresponding acquisition strategies. When adding a new application scenario, only its intent template needs to be registered; there is no need to modify the underlying acquisition driver or reconfigure sensors, significantly shortening the new service launch cycle. This invention can dynamically adjust acquisition strategies and rationally allocate resources based on the differentiated needs of multiple application scenarios when collecting new energy data, solving the problems of serious resource waste, complex configuration, and difficulty in guaranteeing real-time performance in traditional methods. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0016] Figure 1 This is a flowchart of a new energy data acquisition method for multiple application scenarios according to the present invention.

[0017] Figure 2 This is a structural schematic diagram of a new energy data acquisition device for multiple application scenarios according to the present invention.

[0018] 101 - Intent parsing module, 102 - Strategy generation module, 103 - Adaptive acquisition and execution module. Detailed Implementation

[0019] The embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, but should not be construed as limiting the present invention.

[0020] Firstly, please refer to Figure 1 This invention provides a new energy data acquisition method for multiple application scenarios, including: S1 receives the collection intent submitted by the upper layer application and converts the collection intent into a data requirement tuple based on the pre-built resource dependency knowledge graph. In this step, the data acquisition intent includes the application type, the data variables of interest, data quality requirements, and the time range. The data requirement tuple includes the variable set, minimum sampling frequency, maximum allowable delay, spatial granularity, and priority. The data acquisition intent is submitted using a structured JSON template, and the resource dependency knowledge graph stores the mapping relationships between application type, physical quantity, sensor type, communication protocol, and data attributes.

[0021] In this embodiment of the invention, the upper-layer application submits a data collection intent to the system in the form of a structured JSON template or a natural language statement. The intent includes at least the application type, the data variables of interest, data quality requirements (such as maximum latency, minimum sampling frequency, and accuracy), and time range. After receiving the intent, the system calls a pre-built resource dependency knowledge graph. This knowledge graph stores the mapping relationships between application type, physical quantity, sensor type, communication protocol, and data attributes in the form of a graph database. Through graph reasoning, the natural or semi-natural intent is converted into a standardized data requirement tuple. This tuple includes a set of variables, minimum sampling frequency, maximum allowable latency, spatial granularity, and priority. If the intent is submitted in natural language, key parameters are first extracted through a lightweight natural language processing model and then mapped to the tuple. Each intent is assigned a globally unique identifier and stored in an intent queue for subsequent processing.

[0022] When multiple acquisition intentions exist simultaneously, S2 performs conflict detection on the data requirement tuples of each acquisition intention, takes the highest frequency and highest precision requirement of the same variable as the basic acquisition parameters, obtains the current available communication bandwidth, edge node computing power and storage space resource status, downgrades and adjusts the acquisition parameters according to the priority of each acquisition intention, generates the acquisition strategy for each sensor or device, and sends the acquisition strategy to the edge acquisition gateway. In this step, conflict detection specifically includes: comparing the sampling frequency and accuracy requirements of different acquisition intentions for the same data variable, and taking the highest value as the basic acquisition parameter of the variable; if there are differences in the reporting conditions or preprocessing operators of different intentions for the same variable, then the configuration is performed according to the intention with the highest priority.

[0023] In this embodiment of the invention, when multiple active intents exist in the intent queue, conflict detection is first performed on the data requirement tuples of all intents: the sampling frequency and accuracy requirements of different intents for the same data variable are compared, and the highest value is taken as the basic acquisition parameter of that variable. In the case of conflicting reporting conditions or preprocessing operators, the configuration is performed according to the intent with the highest priority; then, the current available communication bandwidth, edge node computing power, and storage space resource status are obtained, and the intents are sorted from high to low priority. Resource constraints are solved for the acquisition parameters of each intent. If the current remaining resources are insufficient to support the ideal acquisition parameters of a certain intent, a downgrade adjustment is performed. The approach includes extending the sampling period of variables corresponding to low-priority intents, disabling their specified edge preprocessing operators, switching real-time reporting to batch cached reporting, or replacing high-precision sensor data with low-precision alternative data sources. The degradation process ensures that the parameters of high-priority intents are not affected. After conflict detection and resource constraint solving, a detailed acquisition strategy is generated for each sensor or device. This strategy includes device identifier, register address, basic sampling period, adaptive mode enable flag, reporting trigger condition, edge preprocessing operator list, and priority label. Finally, the acquisition strategy is distributed to the corresponding edge acquisition gateway via MQTT or HTTP / 2 protocol.

[0024] The S3 edge gateway drives physical sensors or smart devices to collect data according to the received acquisition strategy. The process includes: dynamically adjusting the sampling frequency according to the data change rate, temporarily increasing the sampling frequency when the data change rate exceeds a preset threshold, and restoring it to the basic cycle after the data stabilizes; uploading data only when the data change exceeds a preset dead zone or meets preset reporting conditions; and dynamically activating edge preprocessing operators according to the instructions in the acquisition strategy, uploading only the calculation results or feature values.

[0025] In this step, dynamically adjusting the sampling frequency specifically includes: the edge gateway calculates the data change rate in real time; when the change rate exceeds a first threshold, the sampling frequency is increased to a preset maximum frequency; when the change rate is below a second threshold and the duration exceeds a set time window, the sampling frequency is gradually restored to the basic cycle frequency. The preset reporting conditions specifically include: data values ​​exceeding a set upper or lower limit, data change rate exceeding a set threshold, data prediction residuals exceeding a set range, or abnormal feature values ​​output by the edge preprocessing operator exceeding a set threshold.

[0026] In this embodiment of the invention, after receiving the acquisition strategy, the edge gateway parses it and stores it in a local acquisition task table, with each task running independently. The gateway driver layer periodically polls physical sensors or subscribes to data pushes from smart devices based on the basic sampling period in the task table. During the acquisition process, it calculates the data change rate in real time. When the change rate exceeds a preset first threshold, it automatically and temporarily increases the sampling frequency to the highest frequency specified in the strategy. When the change rate is lower than a second threshold and the duration exceeds a set time window, it gradually restores the sampling frequency to the basic period. For reporting control, the gateway only reports when the data value change exceeds a preset dead zone (such as absolute value change or percentage change) or meets preset reporting conditions (such as data value exceeding limits, change rate exceeding limits, data prediction residual exceeding range, or abnormal feature values ​​output by the edge preprocessing operator exceeding a set threshold). Data is encapsulated and uploaded to the cloud; otherwise, it is cached in a local circular buffer. Simultaneously, the gateway dynamically loads and executes edge preprocessing operators according to the acquisition strategy. These operators include moving average filtering, dead-zone filtering, rate of change detection, fast Fourier transform, envelope analysis, and a lightweight anomaly detection model. The gateway only uploads the calculation results or extracted feature values ​​after operator processing; the original data is only retained locally for a set duration before being discarded. All uploaded data packets include a timestamp, device ID, and strategy version number. After receiving the data, the cloud stores it in a time-series database by variable name for various applications to subscribe to. If an acquisition task finds that edge gateway resources (CPU, memory) exceed a safety threshold during execution, the execution of the lowest priority edge operator is automatically paused, retaining only basic acquisition and reporting functions. The operator is reactivated once resources are restored.

[0027] This invention provides a new energy data acquisition method for multiple application scenarios. Through intent parsing and multi-intent merging, it automatically selects the highest demand for the same variable as the basic acquisition parameter, avoiding redundant configuration. Combined with resource constraint solving, it prioritizes and degrades the acquisition quality of lower-importance tasks when bandwidth, computing power, or storage is limited. Compared to methods that use a fixed highest frequency acquisition method, this reduces communication traffic and storage space usage, significantly lowering operating costs. Upper-layer applications only need to submit structured acquisition intents (including application type, variables of interest, quality requirements, etc.), which are automatically converted into data requirement tuples through a pre-built knowledge graph, generating corresponding acquisition strategies. When adding a new application scenario, only its intent template needs to be registered; there is no need to modify the underlying acquisition driver or reconfigure sensors, significantly shortening the new service launch cycle. This invention can dynamically adjust acquisition strategies and rationally allocate resources based on the differentiated needs of multiple application scenarios when collecting new energy data, solving the problems of severe resource waste, complex configuration, and difficulty in guaranteeing real-time performance in traditional methods.

[0028] Secondly, please refer to Figure 2The present invention also provides a new energy data acquisition device for multiple application scenarios, including an intent parsing module 101, a strategy generation module 102 and an adaptive acquisition execution module 103, wherein the intent parsing module 101, the strategy generation module 102 and the adaptive acquisition execution module 103 are connected in sequence. The intent parsing module 101 is used to receive the collection intent submitted by the upper layer application and convert the collection intent into a data requirement tuple based on the pre-built resource dependency knowledge graph. The strategy generation module 102 is used to perform conflict detection on the data requirement tuples of each acquisition intention when multiple acquisition intentions exist at the same time, take the highest frequency and highest precision requirement of the same variable as the basic acquisition parameters, obtain the current available communication bandwidth, edge node computing power and storage space resource status, downgrade the acquisition parameters according to the priority of each acquisition intention, generate the acquisition strategy for each sensor or device, and send the acquisition strategy to the edge acquisition gateway. The adaptive acquisition execution module 103 is deployed on the edge gateway and is used to drive physical sensors or smart devices to acquire data according to the received acquisition strategy. The execution process includes: dynamically adjusting the sampling frequency according to the data change rate; temporarily increasing the sampling frequency when the data change rate exceeds a preset threshold; and restoring it to the basic cycle after the data stabilizes; uploading data only when the data change exceeds a preset dead zone or meets preset reporting conditions; and dynamically activating the edge preprocessing operator according to the instructions in the acquisition strategy, and uploading only the calculation results or feature values.

[0029] In this embodiment, the upper-layer application submits a collection intent to the intent parsing module 101 in the form of a structured JSON template or a natural language statement. The intent includes at least the application type, the data variables of interest, data quality requirements (such as maximum latency, minimum sampling frequency, and accuracy), and time range. After receiving the intent, the intent parsing module 101 calls a pre-built resource dependency knowledge graph. This knowledge graph stores the mapping relationships between application type, physical quantity, sensor type, communication protocol, and data attributes in the form of a graph database. Through graph reasoning, the natural or semi-natural intent is converted into a standardized data requirement tuple. This tuple includes a variable set, minimum sampling frequency, maximum allowable latency, spatial granularity, and priority. If the intent is submitted in natural language, the device first extracts key parameters through a lightweight natural language processing model and then maps them to the tuple. Each intent is assigned a globally unique identifier and stored in an intent queue for subsequent processing. The strategy generation module 102 is used to, when there are multiple active intents in the intent queue, first perform conflict detection on the data requirement tuples of all intents: compare the sampling frequency and accuracy requirements of different intents for the same data variable, and take the highest value as the basic acquisition parameter of that variable. In the case of conflict in reporting conditions or preprocessing operators, the configuration is performed according to the intent with the highest priority; then, obtain the current available communication bandwidth, edge node computing power, and storage space resource status, sort each intent from high to low priority, and solve the resource constraint for the acquisition parameters of each intent. If the current remaining resources are insufficient to support the ideal acquisition parameters of a certain intent, a downgrade adjustment is performed. The degradation methods include extending the sampling period of variables corresponding to low-priority intents, disabling their specified edge preprocessing operators, switching real-time reporting to batch cached reporting, or replacing high-precision sensor data with low-precision alternative data sources. The degradation process ensures that the parameters of high-priority intents are not affected. After conflict detection and resource constraint solving, a detailed acquisition strategy is generated for each sensor or device. This strategy includes device identifier, register address, basic sampling period, adaptive mode enable flag, reporting trigger condition, edge preprocessing operator list, and priority label. Finally, the acquisition strategy is distributed to the corresponding edge acquisition gateway via MQTT or HTTP / 2 protocol.After receiving the acquisition strategy, the adaptive acquisition execution module 103 parses it and stores it in the local acquisition task table. Each task runs independently. The gateway driver layer periodically polls physical sensors or subscribes to data pushes from smart devices according to the basic sampling period in the task table. During the acquisition process, it calculates the data change rate in real time. When the change rate exceeds a preset first threshold, it automatically increases the sampling frequency temporarily to the highest frequency specified in the strategy. When the change rate is lower than the second threshold and the duration exceeds the set time window, it gradually restores the sampling frequency to the basic period. For reporting control, the gateway only reports when the data value change exceeds a preset dead zone (such as absolute value change or percentage change) or meets preset reporting conditions (such as data value exceeding limits, change rate exceeding limits, data prediction residual exceeding range, or abnormal feature values ​​output by the edge preprocessing operator exceeding a set threshold). Data is only encapsulated and uploaded to the cloud when it is available; otherwise, it is cached in a local circular buffer. Simultaneously, the gateway dynamically loads and executes edge preprocessing operators according to the instructions in the acquisition strategy. These operators include moving average filtering, dead zone filtering, rate of change detection, fast Fourier transform, envelope analysis, and a lightweight anomaly detection model. The gateway only uploads the calculation results or extracted feature values ​​after operator processing; the original data is only retained locally for a set period before being discarded. All uploaded data packets include a timestamp, device ID, and strategy version number. After receiving them, the cloud stores them in a time-series database by variable name for various applications to subscribe to. If an acquisition task finds that edge gateway resources (CPU, memory) exceed a safety threshold during execution, the execution of the lowest priority edge operator is automatically paused, retaining only basic acquisition and reporting functions. The operator is reactivated once resources are restored.

[0030] This invention provides a new energy data acquisition device for multiple application scenarios. Through intent parsing and multi-intent merging, it automatically selects the highest demand for the same variable as the basic acquisition parameter, avoiding redundant configuration. Combined with resource constraint solving, it prioritizes and degrades the acquisition quality of less important tasks when bandwidth, computing power, or storage is limited. Compared to a fixed highest frequency acquisition method, this reduces communication traffic and storage space usage, significantly lowering system operating costs. This invention can dynamically adjust acquisition strategies and rationally allocate resources based on the differentiated needs of multiple application scenarios when acquiring new energy data.

[0031] The above-disclosed embodiments are merely one or more preferred embodiments of this application and should not be construed as limiting the scope of this application. Those skilled in the art can understand that all or part of the processes for implementing the above embodiments and equivalent changes made in accordance with the claims of this application still fall within the scope of this application.

Claims

1. A new energy data acquisition method for multiple application scenarios, characterized in that, include: S1 receives the collection intent submitted by the upper layer application and converts the collection intent into a data requirement tuple based on the pre-built resource dependency knowledge graph. When multiple acquisition intentions exist simultaneously, S2 performs conflict detection on the data requirement tuples of each acquisition intention, takes the highest frequency and highest precision requirement of the same variable as the basic acquisition parameters, obtains the current available communication bandwidth, edge node computing power and storage space resource status, downgrades and adjusts the acquisition parameters according to the priority of each acquisition intention, generates the acquisition strategy for each sensor or device, and sends the acquisition strategy to the edge acquisition gateway. The S3 edge gateway drives physical sensors or smart devices to collect data according to the received collection strategy. The process includes: dynamically adjusting the sampling frequency according to the data change rate; temporarily increasing the sampling frequency when the data change rate exceeds a preset threshold; and restoring the sampling frequency to the basic cycle after the data stabilizes; and uploading data only when the data change exceeds a preset dead zone or meets preset reporting conditions. And dynamically enable edge preprocessing operators according to the instructions in the acquisition strategy, and only upload the calculation results or feature values.

2. The new energy data acquisition method for multiple application scenarios as described in claim 1, characterized in that, The collection intent includes the application type, the data variables of interest, data quality requirements, and the time range.

3. The new energy data acquisition method for multiple application scenarios as described in claim 2, characterized in that, The data requirement tuple includes the set of variables, minimum sampling frequency, maximum allowable delay, spatial granularity, and priority.

4. The new energy data acquisition method for multiple application scenarios as described in claim 3, characterized in that, In step S1, the collection intent is submitted using a structured JSON template, and the resource dependency knowledge graph stores the mapping relationship between application type, physical quantity, sensor type, communication protocol, and data attributes.

5. The new energy data acquisition method for multiple application scenarios as described in claim 4, characterized in that, The collision detection in step S2 specifically includes: Compare the sampling frequency and accuracy requirements of different acquisition intentions for the same data variable, and take the highest value as the basic acquisition parameter for that variable; if there are differences in the reporting conditions or preprocessing operators for the same variable for different intentions, then configure according to the intention with the highest priority.

6. The new energy data acquisition method for multiple application scenarios as described in claim 5, characterized in that, In step S3, dynamically adjusting the sampling frequency specifically includes: The edge gateway calculates the data change rate in real time. When the change rate exceeds the first threshold, the sampling frequency is increased to the preset maximum frequency. When the change rate is lower than the second threshold and the duration exceeds the set time window, the sampling frequency is gradually restored to the basic cycle frequency.

7. The new energy data acquisition method for multiple application scenarios as described in claim 6, characterized in that, In step S3, the preset reporting conditions specifically include: Data values ​​exceeding the set upper or lower limits, data change rate exceeding the set threshold, data prediction residuals exceeding the set range, or abnormal feature values ​​output by the edge preprocessing operator exceeding the set threshold.

8. A new energy data acquisition device for multiple application scenarios, employing the new energy data acquisition method for multiple application scenarios as described in any one of claims 1-7, characterized in that, It includes an intent parsing module, a policy generation module, and an adaptive acquisition and execution module, which are connected in sequence. The intent parsing module is used to receive the collection intent submitted by the upper layer application and convert the collection intent into a data requirement tuple based on the pre-built resource dependency knowledge graph. The strategy generation module is used to perform conflict detection on the data requirement tuples of each collection intention when multiple collection intentions exist at the same time, take the highest frequency and highest precision requirement of the same variable as the basic collection parameters, obtain the current available communication bandwidth, edge node computing power and storage space resource status, downgrade the collection parameters according to the priority of each collection intention, generate the collection strategy for each sensor or device, and send the collection strategy to the edge collection gateway. The adaptive acquisition and execution module is deployed on the edge gateway and is used to drive physical sensors or smart devices to acquire data according to the received acquisition strategy. The execution process includes: dynamically adjusting the sampling frequency according to the data change rate; temporarily increasing the sampling frequency when the data change rate exceeds a preset threshold; and restoring to the basic cycle after the data stabilizes; uploading data only when the data change exceeds a preset dead zone or meets preset reporting conditions; and dynamically activating edge preprocessing operators according to the instructions in the acquisition strategy, and uploading only the calculation results or feature values.

Images