An edge-computing-based real-time data acquisition method and system for a new energy station

Abstract

The application provides a new energy station real-time data acquisition method and system based on edge computing, which determines the offset vectors of each electric pole by acquiring the first historical marking position and actual position of the first electric pole; performs clustering processing according to the first historical marking position to obtain each first clustering center and a corresponding first cluster group; respectively evaluates the offset vectors of the corresponding electric poles in each first cluster group to obtain corresponding comprehensive offset vectors; then clusters the second historical position of a second electric pole needing rectification with the first historical marking position to obtain a second cluster group, and matches the first cluster group and the second cluster group to determine the first cluster group in which the second electric pole to be maintained subsequently is located; and rectifies the second historical position according to the comprehensive offset vector corresponding to the first cluster group in which the second electric pole is located, so that other maintenance personnel can accurately find the position of the electric pole when performing maintenance on the electric pole subsequently.

Description

Technical Field

[0001] This invention belongs to the field of data acquisition technology, and specifically relates to a real-time data acquisition method and system for new energy power stations based on edge computing. Background Technology

[0002] New energy power plants refer to facilities that generate electricity using renewable energy sources (such as solar, wind, and hydropower). Currently, data collection for new energy power plants typically involves setting up several master nodes with edge computing capabilities and several child nodes attached to the master nodes within the power plant area. These data collection points are responsible for collecting data within their respective areas based on preset fixed collection parameters and transmitting the collected data to the master nodes. When the operating status of equipment within the area becomes abnormal or a certain edge master node fails, without manual intervention and repair, unreasonable data collection or data loss can easily occur. Summary of the Invention

[0003] Based on this, the present invention provides a real-time data acquisition method and system for new energy power stations based on edge computing, which aims to allow adjacent edge master nodes to take over when a certain edge master node fails, thereby avoiding data loss.

[0004] A first aspect of this invention provides a real-time data acquisition method for new energy power stations based on edge computing, applied to a wind farm scenario. The wind farm scenario includes an edge master node and child nodes communicatively connected to the edge master node. The edge master node is used to collect monitoring data from the child nodes in real time, and adjacent edge master nodes are communicatively connected. The method includes:

[0005] Based on the arrangement of wind turbines within the wind farm, the area is divided into several regions, each region including one edge master node and several child nodes;

[0006] The current edge master node is controlled to send heartbeat packets to neighboring edge master nodes at preset intervals. The heartbeat packets include the ID of the current edge master node, the coverage area, and the area status.

[0007] If a neighboring edge master node does not receive a heartbeat packet from the current edge master node within a consecutive preset heartbeat period, a new edge master node is selected from the neighboring edge master nodes based on the load rate of the neighboring edge master nodes and the distance to the abnormal edge master node, and the child nodes within the coverage area of ​​the current edge master node are taken over. The load rate is related to the sampling frequency.

[0008] Furthermore, the step of selecting a new edge master node for region takeover from neighboring edge master nodes based on the load rate of neighboring edge master nodes and the distance to the abnormal edge master node includes:

[0009] Acquire device data and environmental correlation data within the coverage area of ​​each adjacent edge master node. The device data includes at least the real-time wind speed, power, vibration acceleration, and gearbox temperature of the fan. The environmental correlation data includes at least the average wind speed, turbulence intensity, and upstream wake influence coefficient within the coverage area.

[0010] Based on the device data and the environmental correlation data, the abnormal risks and cross-regional correlations of the coverage areas of each neighboring edge master node are assessed, and the predicted sampling frequency of each neighboring edge master node at a preset time is determined based on the assessment results.

[0011] Obtain the current sampling frequency of each neighboring edge master node, and calculate the load rate of each neighboring edge master node based on the current sampling frequency and the predicted sampling frequency;

[0012] Based on the distance between each neighboring edge master node and the abnormal edge master node, as well as the load rate of each neighboring edge master node, a score is determined by a weighted summation method, and the neighboring edge master node with the highest score is determined as the new edge master node.

[0013] Furthermore, the step of assessing the anomaly risk and cross-regional correlation of the coverage area of ​​each neighboring edge master node based on the device data and the environmental correlation data, and determining the predicted sampling frequency of each neighboring edge master node at a preset time based on the assessment results, includes:

[0014] Based on meteorological data, determine the wind direction, and based on the wind direction, determine the upstream and downstream regions of the current adjacent edge master nodes, and obtain the average wind speed and turbulence intensity of the upstream region;

[0015] The device data and environmental correlation data within the coverage area of ​​the current neighboring edge master node, as well as the average wind speed and turbulence intensity in the upstream area of ​​the current neighboring edge master node, are input into the trained regional anomaly risk prediction model, which outputs the anomaly risk probability at a preset time. The regional anomaly risk prediction model is built based on an LSTM neural network.

[0016] Based on the abnormal risk probability and regional association weight at a preset time, the cross-regional association index is calculated within the regional set consisting of the coverage area, upstream area, and downstream area of ​​the current neighboring edge master node. The cross-regional association index is used to judge whether the abnormal risk is caused by the regional linkage effect.

[0017] Determine whether the probability of abnormal risk and the cross-regional correlation index at the preset time are both greater than the corresponding thresholds;

[0018] If so, the predicted sampling frequency of the current neighboring edge master node at the preset time is determined based on the abnormal risk probability and cross-regional correlation index at the preset time.

[0019] Furthermore, the step of determining the predicted sampling frequency of the current neighboring edge master node at a preset time based on the abnormal risk probability and cross-regional correlation index at a preset time includes:

[0020] Obtain the predicted sampling frequency and the actual sampling frequency of the current neighboring edge master node at a preset time, and optimize the calculation coefficient of the region association weight based on the predicted sampling frequency and the actual sampling frequency.

[0021] Furthermore, the step of obtaining the predicted sampling frequency and the actual sampling frequency of the current neighboring edge master node at a preset time, and optimizing the calculation coefficients of the region association weight based on the predicted sampling frequency and the actual sampling frequency includes:

[0022] Several predicted sampling frequencies and actual sampling frequencies at the preset time are obtained respectively. The average deviation of the neighboring edge master nodes is calculated, and the calculation coefficient of the regional association weight is optimized in the short term based on the average deviation.

[0023] A federated learning model based on federated gradient boosting trees is adopted. Based on the average deviation, regional distance and historical calculation coefficients of regional association weights, the target calculation coefficients of regional association weights are output for long-term calibration.

[0024] Furthermore, in the step of calculating the cross-regional correlation index within the set of regions consisting of the coverage area, upstream region, and downstream region of the current neighboring edge master node based on the abnormal risk probability and regional correlation weight at a preset time, the calculation formula is as follows:

[0025] ;

[0026] in, This is a cross-regional correlation index, where S is the set of regions. For the associated region R j The probability of abnormal risks This represents the regional association weight.

[0027] Furthermore, the regional anomaly risk prediction model consists of an input layer, a hidden layer, and an output layer. The input layer contains several input nodes, the number of which is the same as the number of input data types, and is used to receive different types of input data.

[0028] The hidden layer consists of five sub-layers. The first hidden sub-layer has 64 nodes and uses the LeakyReLU function to extract features from the input data to find key features related to the output results. The second and third hidden sub-layers each have 128 nodes and use the Sigmoid function to further mine the features of the output data of the first hidden sub-layer and capture the intrinsic relationship between nodes. The fourth and fifth hidden sub-layers each contain 64 nodes and use the ReLU activation function to perform secondary processing on the features of the output data of the third hidden sub-layer.

[0029] The output layer has one output node, and the output layer uses a linear activation function to output the final prediction result.

[0030] A second aspect of this invention provides a real-time data acquisition system for new energy power stations based on edge computing, used to implement the real-time data acquisition method for new energy power stations based on edge computing described in the first aspect, the system comprising:

[0031] The partitioning module is used to divide the wind farm into several regions based on the arrangement of wind turbines within the wind farm. Each region includes one edge master node and several child nodes.

[0032] The control module is used to control the current edge master node to send heartbeat packets to neighboring edge master nodes at preset intervals. The heartbeat packets include the ID of the current edge master node, the coverage area, and the area status.

[0033] The selection module is used to select a new edge master node from the neighboring edge master nodes based on the load rate of the neighboring edge master nodes and the distance to the abnormal edge master node when the neighboring edge master nodes do not receive the heartbeat packet sent by the current edge master node within a continuous preset heartbeat period, and take over each of the child nodes in the coverage area of ​​the current edge master node. The load rate is related to the sampling frequency.

[0034] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the real-time data acquisition method for new energy power stations based on edge computing provided in the first aspect.

[0035] A fourth aspect of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the real-time data acquisition method for new energy power stations based on edge computing provided in the first aspect.

[0036] This invention provides a real-time data acquisition method and system for new energy power stations based on edge computing. The system divides the wind turbine layout within the wind farm into several regions, each region including one edge master node and several child nodes. The current edge master node is controlled to send heartbeat packets to neighboring edge master nodes at preset intervals. The heartbeat packets include the ID of the current edge master node, its coverage area, and the region status. If a neighboring edge master node does not receive a heartbeat packet from the current edge master node within a consecutive preset heartbeat period, a new edge master node is selected from the neighboring edge master nodes based on their load rate and distance from the faulty edge master node. This new edge master node then takes over the child nodes within its coverage area. The load rate is related to the sampling frequency. Specifically, when an edge master node fails, all neighboring edge master nodes are comprehensively evaluated. The neighboring edge master node with the best evaluation result automatically takes over the child nodes in its coverage area to ensure uninterrupted data acquisition. Attached Figure Description

[0037] Figure 1 This is a flowchart illustrating the implementation of a real-time data acquisition method for new energy power stations based on edge computing, as provided in Embodiment 1 of the present invention.

[0038] Figure 2 This is a structural block diagram of a real-time data acquisition system for new energy power stations based on edge computing, provided in Embodiment 2 of the present invention.

[0039] Figure 3 This is a structural block diagram of an electronic device provided in Embodiment 3 of the present invention. Detailed Implementation

[0040] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Several embodiments of the invention are illustrated in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0041] It should be noted that when a component is said to be "fixed to" another component, it can be directly on the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0042] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0043] Example 1

[0044] According to an embodiment of the present invention, a method for real-time data acquisition of new energy power stations based on edge computing is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0045] This embodiment provides a real-time data acquisition method for new energy power stations based on edge computing. This method can be used in electronic devices, such as computers, and is applied to wind farm scenarios. The wind farm scenario includes an edge master node and child nodes communicatively connected to the edge master node. The edge master node collects monitoring data from the child nodes in real time, and adjacent edge master nodes are communicatively connected. The edge master node can be an industrial-grade edge computing gateway, while the child nodes use dedicated sensors based on different monitoring parameters; for example, an ultrasonic anemometer. (See also...) Figure 1 , Figure 1 The flowchart of a real-time data acquisition method for new energy power stations based on edge computing provided in Embodiment 1 of the present invention is shown, specifically including steps S01 to S03.

[0046] Step S01: Based on the arrangement of wind turbines within the wind farm, the area is divided into several regions, each region including one edge master node and several child nodes.

[0047] Specifically, based on equipment density and communication distance, wind farms are divided into areas automatically or manually. For example, each wind farm is divided into areas of 3-5 wind turbines, and signal boosters are set at the boundaries of the areas to ensure the communication distance between the sub-nodes and the edge master nodes.

[0048] In this embodiment of the invention, the edge master node collects monitoring data from the child nodes in real time and calculates the fluctuation coefficient of the device's operating indicators using a sliding window algorithm. When the fluctuation coefficient exceeds a preset threshold (e.g., power fluctuation exceeds 5%), it is determined to be an abnormal area, and the sampling frequency is immediately increased; if the fluctuation coefficient is below the threshold for 30 consecutive minutes, the sampling frequency is automatically reduced.

[0049] Step S02: Control the current edge master node to send heartbeat packets to neighboring edge master nodes at preset intervals. The heartbeat packets include the ID of the current edge master node, the coverage area, and the area status.

[0050] Each edge master node has a unique ID. In addition, it mainly sends the status of child nodes within the coverage area of ​​the current edge master node, as well as whether the area status is normal or abnormal, to nearby edge master nodes.

[0051] Step S03: When a neighboring edge master node does not receive a heartbeat packet from the current edge master node within a consecutive preset heartbeat period, a new edge master node is selected from the neighboring edge master nodes based on the load rate of the neighboring edge master nodes and the distance to the abnormal edge master node, and the child nodes within the coverage area of ​​the current edge master node are taken over. The load rate is related to the sampling frequency.

[0052] It should be noted that if a neighboring edge master node does not receive a heartbeat packet from the current edge master node within a consecutive preset heartbeat cycle, it indicates that the current edge master node may be malfunctioning. Therefore, device data and environmental correlation data within the coverage area of ​​each neighboring edge master node are obtained. The device data includes at least the real-time wind speed, power, vibration acceleration, and gearbox temperature of the fan. The environmental correlation data includes at least the average wind speed, turbulence intensity, and upstream wake influence coefficient within the coverage area.

[0053] Based on the device data and the environmental correlation data, the abnormal risks and cross-regional correlations of the coverage areas of each neighboring edge master node are assessed, and based on the assessment results, the predicted sampling frequency of each neighboring edge master node at a preset time is determined. Specifically, based on meteorological data, the wind direction is determined, and based on the wind direction, the upstream and downstream areas of the current neighboring edge master node are determined, and the average wind speed and turbulence intensity of the upstream area are obtained.

[0054] The device data and environmental correlation data within the coverage area of ​​the current neighboring edge master node, as well as the average wind speed and turbulence intensity in the upstream area of ​​the current neighboring edge master node, are input into the trained regional anomaly risk prediction model, which outputs the anomaly risk probability at a preset time. The regional anomaly risk prediction model is built based on an LSTM neural network and consists of an input layer, a hidden layer, and an output layer. The input layer contains several input nodes, the number of which is the same as the number of input data types, and is used to receive different types of input data.

[0055] The hidden layer consists of five sub-layers. The first hidden sub-layer has 64 nodes and uses the LeakyReLU function to extract features from the input data to find key features related to the output results. The second and third hidden sub-layers each have 128 nodes and use the Sigmoid function to further mine the features of the output data of the first hidden sub-layer and capture the intrinsic relationship between nodes. The fourth and fifth hidden sub-layers each contain 64 nodes and use the ReLU activation function to perform secondary processing on the features of the output data of the third hidden sub-layer.

[0056] The output layer has one output node, and the output layer uses a linear activation function to output the final prediction result;

[0057] Based on the abnormal risk probability and regional association weight at a preset time, the cross-regional association index within the regional set consisting of the coverage area, upstream area, and downstream area of ​​the current neighboring edge master node is calculated. The calculation formula is as follows:

[0058] ;

[0059] in, This is a cross-regional correlation index, where S is the set of regions. For the associated region R j The probability of abnormal risks For the current region R r With associated region R j The regional correlation weight, the cross-regional correlation index is used to judge whether the abnormal risk is caused by regional linkage effect. In addition, the regional correlation weight W r,j = e^(-λ×d r,j In this context, λ represents the computational coefficient to be optimized, and a default value is usually given first.

[0060] Furthermore, the predicted sampling frequency and the actual sampling frequency of the current neighboring edge master nodes at a preset time are obtained. Based on the predicted sampling frequency and the actual sampling frequency, the calculation coefficients of the regional association weight are optimized. Specifically, several predicted sampling frequencies and actual sampling frequencies at the preset time are obtained respectively, and the average deviation of the neighboring edge master nodes is calculated. The calculation formula is as follows:

[0061] ;

[0062] M is the sample size. For the associated region R j The absolute value of the difference between the predicted sampling frequency and the actual sampling frequency. The average deviation;

[0063] Based on the average deviation, the calculation coefficients of the regional association weights are optimized in the short term. Specifically, for the current region R... r according to Real-time optimization of λ, the calculation formula is:

[0064] ;

[0065] These are the initial coefficients. The reference sampling frequency is given by the following constraints: To ensure λ is not less than 0.3 × λ0 (to avoid the coefficients being too small and causing the weights to lose their physical meaning), it should be noted that during this process, each calculation... If λ is updated synchronously, it will be applied to the next association weight calculation.

[0066] A federated learning model based on federated gradient boosting trees is adopted. Based on the average deviation, regional distance, and historical calculation coefficients of regional association weights, target calculation coefficients for regional association weights are output for long-term calibration. In this embodiment, the training process of the federated learning model is as follows: each master node acts as a client of the federated learning, trains the model subtree based on local data, and only uploads the model parameters (without disclosing the original data) to the cloud aggregation server; the server aggregates parameters every 3 days, generates a global model, and distributes it to each master node. Further, the master node, based on the suggested values ​​output by the global model (i.e., the calculation coefficients of the long-term calibrated regional association weights), combines them with the calculation coefficients of the short-term optimized regional association weights, and obtains the final target calculation coefficients for the regional association weights using a weighted summation method. It is understood that the adjustment of the calculation coefficients of the short-term optimized regional association weights takes precedence over the adjustment of the calculation coefficients of the long-term calibrated regional association weights, ensuring that real-time deviations can quickly affect weight calculations. The update of the calculation coefficients of the long-term calibrated regional association weights serves as a basic configuration, providing more reasonable initial coefficients for short-term optimization. Simultaneously, the coefficients generated during short-term optimization... The λ data serves as training samples for the long-term calibration model, enabling the global model to continuously learn the dynamic changes of the wind farm (such as changes in airflow characteristics caused by seasonal changes); the target calculation coefficients output by the long-term calibration provide initial parameters for newly deployed master nodes or newly divided areas, shortening the convergence time of short-term optimization.

[0067] Determine whether the probability of abnormal risk and the cross-regional correlation index at the preset time are both greater than the corresponding thresholds;

[0068] If so, the predicted sampling frequency of the current neighboring edge master node at the preset time is determined based on the abnormal risk probability and cross-regional correlation index at the preset time. Specifically, a mapping relationship between the normal risk probability, cross-regional correlation index, and sampling frequency can be established in advance through expert evaluation. Once the abnormal risk probability and cross-regional correlation index are known, the predicted sampling frequency is output according to the corresponding mapping relationship. It is understandable that in a wind farm, the operating status of a single wind turbine (such as power fluctuations and abnormal vibrations) is often affected by the airflow environment of the surrounding area (such as turbulence and wind speed gradients). The airflow environment has cross-regional correlation (such as the wake of an upstream wind turbine affecting a downstream wind turbine). Therefore, cross-regional correlation verification is used to reduce the false alarm rate of single-region anomaly prediction.

[0069] Obtain the current sampling frequency of each neighboring edge master node, and calculate the load rate of each neighboring edge master node based on the current sampling frequency and the predicted sampling frequency. The formula for calculating the load rate can be expressed as:

[0070] L = α×(L1×β) + (1-α)×(L2×β);

[0071] Understandably, the sampling frequency indirectly affects the load pressure on the master node by influencing the data upload volume of child nodes and the data processing volume of the master node. L represents the total load rate, L1 represents the data processing load, L2 represents the communication load, α represents the weighting coefficient, and β represents the sampling frequency correction coefficient, reflecting the resource utilization multiple of the current sampling frequency relative to the base frequency. The data processing load is related to the current CPU utilization, current memory utilization, and the proportion of data processing volume. This proportion is further related to the data processing volume per second at the current sampling frequency and the master node's maximum data processing capacity. The communication load is related to the current uplink bandwidth utilization, current downlink bandwidth utilization, and the proportion of data transmission volume. This proportion is also related to the data transmission volume per second at the current sampling frequency and the master node's maximum communication bandwidth.

[0072] Based on the distance between each neighboring edge master node and the abnormal edge master node, as well as the load rate of each neighboring edge master node, a score is determined by a weighted summation method. The neighboring edge master node with the highest score is then identified as the new edge master node. Understandably, the closer the neighboring edge master node is and the lower its load rate, the higher its score.

[0073] In summary, the edge computing-based real-time data acquisition method for new energy power stations described in the above embodiments of the present invention divides the wind turbine layout within the wind farm into several regions, each region including one edge master node and several child nodes. The method controls the current edge master node to send heartbeat packets to neighboring edge master nodes at preset intervals. The heartbeat packets include the ID of the current edge master node, its coverage area, and the region status. If a neighboring edge master node does not receive a heartbeat packet from the current edge master node within a consecutive preset heartbeat period, a new edge master node is selected from the neighboring edge master nodes based on their load rate and distance from the faulty edge master node. This new edge master node then takes over the child nodes within its coverage area. The load rate is related to the sampling frequency. Specifically, when an edge master node fails, a comprehensive evaluation of all neighboring edge master nodes is performed. The neighboring edge master node with the best evaluation result automatically takes over the child nodes within its coverage area to ensure uninterrupted data acquisition.

[0074] Example 2

[0075] Please see Figure 2 , Figure 2 This is a structural block diagram of a real-time data acquisition system for new energy power stations based on edge computing, provided in Embodiment 2 of the present invention. This edge computing-based real-time data acquisition system 200 is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the apparatus described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0076] Specifically, the edge computing-based real-time data acquisition system 200 for new energy power stations includes: a partitioning module 21, a control module 22, and a selection module 23, wherein:

[0077] The partitioning module 21 is used to divide the wind turbines in the wind farm into several regions according to their arrangement. Each region includes one edge master node and several child nodes.

[0078] Control module 22 is used to control the current edge master node to send heartbeat packets to neighboring edge master nodes at preset intervals. The heartbeat packets include the ID of the current edge master node, the coverage area, and the area status.

[0079] Selection module 23 is used to select a new edge master node from the neighboring edge master nodes based on the load rate of the neighboring edge master nodes and the distance to the abnormal edge master node when the neighboring edge master node does not receive a heartbeat packet sent by the current edge master node within a continuous preset heartbeat period, and to take over each of the child nodes in the coverage area of ​​the current edge master node, wherein the load rate is related to the sampling frequency.

[0080] Furthermore, in some optional embodiments of the present invention, the selection module 23 includes:

[0081] The acquisition unit is used to acquire device data and environmental correlation data within the coverage area of ​​each adjacent edge master node. The device data includes at least the real-time wind speed, power, vibration acceleration and gearbox temperature of the fan. The environmental correlation data includes at least the average wind speed, turbulence intensity and upstream wake influence coefficient within the coverage area.

[0082] The evaluation unit is used to evaluate the abnormal risks and cross-regional correlations of the coverage areas of each neighboring edge master node based on the device data and the environmental correlation data, and to determine the predicted sampling frequency of each neighboring edge master node at a preset time based on the evaluation results.

[0083] The calculation unit is used to obtain the current sampling frequency of each neighboring edge master node, and calculate the load rate of each neighboring edge master node based on the current sampling frequency and the predicted sampling frequency.

[0084] The determination unit is used to determine the score based on the distance between each neighboring edge master node and the abnormal edge master node and the load rate of each neighboring edge master node, using a weighted summation method, and to determine the neighboring edge master node with the highest score as the new edge master node.

[0085] Furthermore, in some optional embodiments of the present invention, the evaluation unit includes:

[0086] The first determining subunit is used to determine the wind direction based on meteorological data, determine the upstream and downstream regions of the current adjacent edge master node based on the wind direction, and obtain the average wind speed and turbulence intensity of the upstream region.

[0087] The input subunit is used to input the device data and environmental correlation data within the coverage area of ​​the current neighboring edge master node, as well as the average wind speed and turbulence intensity of the upstream area of ​​the current neighboring edge master node, into the trained regional anomaly risk prediction model, and output the anomaly risk probability at a preset time. The regional anomaly risk prediction model is built based on an LSTM neural network. The regional anomaly risk prediction model consists of an input layer, a hidden layer, and an output layer. The input layer contains a number of input nodes, the number of which is the same as the number of input data types, and is used to receive different types of input data.

[0088] The hidden layer consists of five sub-layers. The first hidden sub-layer has 64 nodes and uses the LeakyReLU function to extract features from the input data to find key features related to the output results. The second and third hidden sub-layers each have 128 nodes and use the Sigmoid function to further mine the features of the output data of the first hidden sub-layer and capture the intrinsic relationship between nodes. The fourth and fifth hidden sub-layers each contain 64 nodes and use the ReLU activation function to perform secondary processing on the features of the output data of the third hidden sub-layer.

[0089] The output layer has one output node, and the output layer uses a linear activation function to output the final prediction result;

[0090] The calculation subunit is used to calculate the cross-regional correlation index within the set of regions consisting of the coverage area, upstream region, and downstream region of the current neighboring edge master node, based on the anomaly risk probability and regional correlation weight at a preset time. The cross-regional correlation index is used to determine whether the anomaly risk is caused by regional linkage effect, and the calculation formula is as follows:

[0091] ;

[0092] in, This is a cross-regional correlation index, where S is the set of regions. For the associated region R j The probability of abnormal risks For regional association weights;

[0093] The judgment sub-unit is used to determine whether the abnormal risk probability and cross-regional correlation index at a preset time are both greater than the corresponding thresholds;

[0094] The second determining subunit is used to determine the prediction sampling frequency of the current neighboring edge master node at the preset time based on the abnormal risk probability and cross-regional correlation index at the preset time when both the abnormal risk probability and cross-regional correlation index at the preset time are greater than the corresponding threshold.

[0095] Furthermore, in some optional embodiments of the present invention, the evaluation unit further includes:

[0096] The optimization subunit obtains the predicted sampling frequency and the actual sampling frequency of the current neighboring edge master node at a preset time, and optimizes the calculation coefficient of the regional association weight based on the predicted sampling frequency and the actual sampling frequency. Specifically, it obtains several predicted sampling frequencies and actual sampling frequencies at the preset time, calculates the average deviation of the neighboring edge master node, and optimizes the calculation coefficient of the regional association weight in the short term based on the average deviation.

[0097] A federated learning model based on federated gradient boosting trees is adopted. Based on the average deviation, regional distance and historical calculation coefficients of regional association weights, the target calculation coefficients of regional association weights are output for long-term calibration.

[0098] Example 3

[0099] In another aspect, the present invention also proposes an electronic device, please refer to [link to relevant documentation]. Figure 3 The image shows an electronic device according to Embodiment 3 of the present invention, including a memory 20, a processor 10, and a computer program 30 stored in the memory and executable on the processor. When the processor 10 executes the computer program 30, it implements the real-time data acquisition method for new energy power stations based on edge computing as described above.

[0100] In some embodiments, the processor 10 may be a central processing unit (CPU), controller, microcontroller, microprocessor or other data processing chip, used to run program code stored in memory 20 or process data, such as executing access restriction programs.

[0101] The memory 20 includes at least one type of readable storage medium, such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 20 can be an internal storage unit of an electronic device, such as the hard disk of the electronic device. In other embodiments, the memory 20 can also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. Furthermore, the memory 20 can include both internal and external storage units of the electronic device. The memory 20 can be used not only to store application software and various types of data of the electronic device, but also to temporarily store data that has been output or will be output.

[0102] It should be pointed out that, Figure 3 The structure shown does not constitute a limitation on the electronic device. In other embodiments, the electronic device may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0103] This invention also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned method for real-time data acquisition at new energy power stations based on edge computing.

[0104] Those skilled in the art will understand that the logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can mean any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0105] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0106] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0107] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0108] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this patent should be determined by the appended claims.

Claims

1. A real-time data acquisition method for new energy power stations based on edge computing, characterized in that, The method is applied to a wind farm scenario, which includes an edge master node and child nodes communicatively connected to the edge master node. The edge master node is used to collect monitoring data from the child nodes in real time, and adjacent edge master nodes are communicatively connected. The method includes: Based on the arrangement of wind turbines within the wind farm, the area is divided into several regions, each region including one edge master node and several child nodes; The current edge master node is controlled to send heartbeat packets to neighboring edge master nodes at preset intervals. The heartbeat packets include the ID of the current edge master node, the coverage area, and the area status. When a neighboring edge master node does not receive a heartbeat packet from the current edge master node within a consecutive preset heartbeat period, the current edge master node is determined to be an abnormal edge master node. Then, based on the load rate of the neighboring edge master nodes and the distance to the abnormal edge master node, a new edge master node is selected from the neighboring edge master nodes, and the child nodes within the coverage area of ​​the current edge master node are taken over. The load rate is related to the sampling frequency. The step of selecting a new edge master node for region takeover based on the load rate of neighboring edge master nodes and the distance to the abnormal edge master node includes: Acquire device data and environmental correlation data within the coverage area of ​​each adjacent edge master node. The device data includes at least the real-time wind speed, power, vibration acceleration, and gearbox temperature of the fan. The environmental correlation data includes at least the average wind speed, turbulence intensity, and upstream wake influence coefficient within the coverage area. Based on the device data and the environmental correlation data, the abnormal risks and cross-regional correlations of the coverage areas of each neighboring edge master node are assessed, and the predicted sampling frequency of each neighboring edge master node at a preset time is determined based on the assessment results. Obtain the current sampling frequency of each neighboring edge master node, and calculate the load rate of each neighboring edge master node based on the current sampling frequency and the predicted sampling frequency; Based on the distance between each neighboring edge master node and the abnormal edge master node, as well as the load rate of each neighboring edge master node, a score is determined by a weighted summation method, and the neighboring edge master node with the highest score is determined as the new edge master node.

2. The real-time data acquisition method for new energy power stations based on edge computing according to claim 1, characterized in that, The step of assessing the anomaly risk and cross-regional correlation of the coverage area of ​​each neighboring edge master node based on the device data and the environmental correlation data, and determining the predicted sampling frequency of each neighboring edge master node at a preset time based on the assessment results, includes: Based on meteorological data, determine the wind direction, and based on the wind direction, determine the upstream and downstream regions of the current adjacent edge master nodes, and obtain the average wind speed and turbulence intensity of the upstream region; The device data and environmental correlation data within the coverage area of ​​the current neighboring edge master node, as well as the average wind speed and turbulence intensity in the upstream area of ​​the current neighboring edge master node, are input into the trained regional anomaly risk prediction model, which outputs the anomaly risk probability at a preset time. The regional anomaly risk prediction model is built based on an LSTM neural network. Based on the abnormal risk probability and regional association weight at a preset time, the cross-regional association index is calculated within the regional set consisting of the coverage area, upstream area and downstream area of ​​the current neighboring edge master node. The cross-regional association index is used to judge whether the abnormal risk is caused by the regional linkage effect. Determine whether the probability of abnormal risk and the cross-regional correlation index at the preset time are both greater than the corresponding thresholds; If so, the predicted sampling frequency of the current neighboring edge master node at the preset time is determined based on the abnormal risk probability and cross-regional correlation index at the preset time.

3. The real-time data acquisition method for new energy power stations based on edge computing according to claim 2, characterized in that, The step of determining the predicted sampling frequency of the current neighboring edge master node at a preset time based on the abnormal risk probability and cross-regional correlation index at a preset time includes: Obtain the predicted sampling frequency and the actual sampling frequency of the current neighboring edge master node at a preset time, and optimize the calculation coefficient of the region association weight based on the predicted sampling frequency and the actual sampling frequency.

4. The real-time data acquisition method for new energy power stations based on edge computing according to claim 3, characterized in that, The step of obtaining the predicted sampling frequency and the actual sampling frequency of the current neighboring edge master node at a preset time, and optimizing the calculation coefficients of the region association weight based on the predicted sampling frequency and the actual sampling frequency includes: Several predicted sampling frequencies and actual sampling frequencies at the preset time are obtained respectively. The average deviation of the neighboring edge master nodes is calculated, and the calculation coefficient of the regional association weight is optimized in the short term based on the average deviation. A federated learning model based on federated gradient boosting trees is adopted. Based on the average deviation, regional distance and historical calculation coefficients of regional association weights, the target calculation coefficients of regional association weights are output for long-term calibration.

5. The real-time data acquisition method for new energy power stations based on edge computing according to claim 4, characterized in that, In the step of calculating the cross-regional correlation index within the set of regions consisting of the coverage area, upstream area, and downstream area of ​​the current neighboring edge master node based on the abnormal risk probability and regional correlation weight at a preset time, the calculation formula is as follows: ； in, This is a cross-regional correlation index, where S is the set of regions. For the associated region R j The probability of abnormal risks This represents the regional association weight.

6. The real-time data acquisition method for new energy power stations based on edge computing according to claim 5, characterized in that, The regional anomaly risk prediction model consists of an input layer, a hidden layer, and an output layer. The input layer contains several input nodes, the number of which is the same as the number of input data types, and is used to receive different types of input data. The hidden layer consists of five sub-layers. The first hidden sub-layer has 64 nodes and uses the LeakyReLU function to extract features from the input data to find key features related to the output results. The second and third hidden sub-layers each have 128 nodes and use the Sigmoid function to further mine the features of the output data of the first hidden sub-layer and capture the intrinsic relationship between nodes. The fourth and fifth hidden sub-layers each contain 64 nodes and use the ReLU activation function to perform secondary processing on the features of the output data of the third hidden sub-layer. The output layer has one output node, and the output layer uses a linear activation function to output the final prediction result.

7. A real-time data acquisition system for new energy power stations based on edge computing, characterized in that, The system is used to implement the real-time data acquisition method for new energy power stations based on edge computing as described in any one of claims 1-6, the system comprising: The partitioning module is used to divide the wind farm into several regions based on the wind turbine layout within the wind farm. Each region includes one edge master node and several child nodes. The control module is used to control the current edge master node to send heartbeat packets to neighboring edge master nodes at preset intervals. The heartbeat packets include the ID of the current edge master node, the coverage area, and the area status. The selection module is used to determine that the current edge master node is an abnormal edge master node when a neighboring edge master node does not receive a heartbeat packet sent by the current edge master node within a consecutive preset heartbeat period. Then, based on the load rate of the neighboring edge master nodes and the distance to the abnormal edge master node, a new edge master node is selected from the neighboring edge master nodes, and the child nodes in the coverage area of ​​the current edge master node are taken over. The load rate is related to the sampling frequency.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the real-time data acquisition method for new energy power stations based on edge computing as described in any one of claims 1-6.

9. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the real-time data acquisition method for new energy power stations based on edge computing as described in any one of claims 1-6.

Images