Edge computing-based electrical automation monitoring data optimization method and system

Abstract

The application discloses an electrical automation monitoring data optimization method and system based on edge computing, and belongs to the field of data optimization. The application synchronously acquires electrical, positioning and communication state data of equipment through an edge computing node, generates a link quality index based on a geometric precision attenuation factor and a communication state, and then constructs a joint feature sequence by correlating a spatial position. Subsequently, a risk prediction model based on an attention mechanism long short-term memory network is used to predict the link risk in a future first time period according to the sequence. When the predicted risk meets the condition, the first electrical monitoring data with high priority is added with a to-be-reconstructed label and locally cached. In a second time period after communication recovery, the cached data and the risk information are sent to a central cloud platform, the cloud platform reconstructs historical data and supplementary data according to the risk information, and finally generates complete target electrical monitoring data, which can improve the completeness and effectiveness of electrical automation monitoring data.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of data optimization, and particularly relates to an electrical automation monitoring data optimization method and system based on edge computing. BACKGROUND

[0002] Electrical automation monitoring data optimization is a key technical support for ensuring efficient and reliable operation of modern industrial systems. By systematically processing the massive real-time monitoring data generated by electrical automation equipment, the quality, availability and transmission efficiency of the data can be improved, and a high-quality data foundation can be provided for intelligent operation and maintenance.

[0003] In the prior art, the electrical automation monitoring data optimization method usually performs preliminary aggregation and compression on the massive raw monitoring data collected at the edge node to reduce the occupation of network bandwidth and improve data processing efficiency. Some electrical automation monitoring data optimization methods detect the communication link state, and when the signal strength or signal-to-noise ratio decreases, the data transmission quality is dynamically improved by adjusting the data transmission rate or switching to a backup communication channel.

[0004] However, in the real-time monitoring scene of mobile industrial equipment, when the equipment moves in complex environments such as mining areas, large factories or tunnels, the quality of the communication link does not decrease smoothly, but often exhibits frequent and sharp instantaneous interruptions. Neither data compression nor transmission strategy adjustment in the prior art can effectively cope with sudden signal loss, which may result in the loss of critical electrical monitoring data during transmission. Therefore, the data optimization method in the prior art cannot guarantee the integrity and effectiveness of the electrical automation monitoring data. SUMMARY

[0005] The application provides an electrical automation monitoring data optimization method, system, device and computer storage medium based on edge computing, which can improve the integrity and effectiveness of electrical automation monitoring data.

[0006] In a first aspect, the application provides an electrical automation monitoring data optimization method based on edge computing, applied to an edge computing node, which comprises:

[0007] Synchronously acquiring electrical monitoring data of a mobile industrial equipment, satellite positioning data and communication state data, the satellite positioning data comprising a geometric dilution of precision factor;

[0008] Based on the communication state data and the geometric dilution of precision factor, a link quality index is generated by weighted fusion, and a joint feature sequence is constructed according to the link quality index and the equipment position data in the satellite positioning data;

[0009] Based on the joint feature sequence, a trained risk prediction model is used to predict first link risk information corresponding to the joint feature sequence in the first time period, and the risk prediction model is trained based on a long short-term memory network using attention mechanism using historical joint feature sequences;

[0010] In a case where the first link risk information meets a preset risk condition, a first electrical monitoring data of high priority associated with the first link risk information is added with a to-be-reconstructed label and locally cached;

[0011] In the second time period, the first electrical monitoring data, the to-be-reconstructed label included in the first electrical monitoring data, and the first link risk information are sent to the central cloud platform, so that the central cloud platform reconstructs the received historical electrical monitoring data and the first electrical monitoring data using the first link risk information to generate target electrical monitoring data when the to-be-reconstructed label is received.

[0012] In an implementable embodiment, the communication state data includes base station position data; the method further comprises:

[0013] Synchronously acquiring environmental point cloud data of the mobile industrial device;

[0014] Taking the device position data as a first end point, taking the base station position data as a second end point, and obtaining a first signal propagation path by determining a spatial straight line segment connecting the first end point and the second end point;

[0015] Calculating a first Fresnel zone of the first signal propagation path, and obtaining a first spatial occlusion factor based on the first spatial occlusion volume ratio and the environmental point cloud data by spatial intersection calculation.

[0016] Based on the communication state data and the geometric dilution of precision factor, a link quality index is generated by weighted fusion, including:

[0017] Based on the communication state data, the geometric dilution of precision factor, and the first spatial occlusion factor, a link quality index is generated by weighted fusion.

[0018] In an implementable embodiment, the satellite positioning data includes satellite position data; the method further comprises:

[0019] Taking the satellite position data as a third end point, obtaining a second signal propagation path by determining a spatial straight line segment connecting the first end point and the third end point; calculating a first Fresnel zone of the second signal propagation path, and obtaining a second spatial occlusion factor based on the first spatial occlusion volume ratio and the environmental point cloud data by spatial intersection calculation.

[0020] According to the link quality index and the device position data in the satellite positioning data, a joint feature sequence is constructed, including:

[0021] According to the link quality index, the second spatial shielding factor and the device position data in the satellite positioning data, a joint feature sequence is constructed.

[0022] In an implementable embodiment, the method further comprises:

[0023] Synchronously acquiring the link quality index of at least one adjacent mobile industrial device in a preset area centered on the mobile industrial device, and calculating a link quality index mean of the link quality index of the at least one adjacent mobile industrial device;

[0024] According to the link quality index, the second spatial shielding factor and the device position data in the satellite positioning data, a joint feature sequence is constructed, including:

[0025] The joint feature sequence is constructed by pairing the link quality index, the link quality index mean, the second spatial shielding factor and the device position data under the same timestamp to form data tuples, and arranging the data tuples in chronological order. In an implementable embodiment, the historical electrical monitoring data includes second electrical monitoring data of low priority in the electrical monitoring data associated with the first link risk information received by the central cloud platform within the first time period in the case that the first link risk information meets the preset risk condition.

[0026] In an implementable embodiment, the second time period is after the first time period;

[0027] In the second time period, the first electrical monitoring data, the to-be-reconstructed label included in the first electrical monitoring data, and the first link risk information are sent to the central cloud platform, including:

[0028] In the case that the second link risk information corresponding to the second time period does not meet the preset risk condition, in the second time period, the first electrical monitoring data, the to-be-reconstructed label included in the first electrical monitoring data, and the first link risk information are sent to the central cloud platform.

[0029] In an implementable embodiment, the historical electrical monitoring data includes non-continuous electrical monitoring data;

[0030] The received historical electrical monitoring data and the first electrical monitoring data are reconstructed by using the first link risk information to generate target electrical monitoring data, including:

[0031] According to the risk level of the first link risk information, a target filling parameter is determined in a preset risk level and filling parameter mapping relationship;

[0032] The historical electrical monitoring data is differentially filled by using the target filling parameter to obtain filled historical electrical monitoring data.

[0033] The filled historical electrical monitoring data and the first electrical monitoring data are spliced according to time points to generate target electrical monitoring data.

[0034] In a second aspect, the present application provides an electrical automation monitoring data optimization system based on edge computing, applied to an edge computing node, the system comprising:

[0035] An acquisition module is configured to synchronously acquire electrical monitoring data, satellite positioning data and communication state data of a mobile industrial device, wherein the satellite positioning data comprises a geometric dilution of precision factor;

[0036] A construction module is configured to generate a link quality index by weighted fusion based on the communication state data and the geometric dilution of precision factor, and construct a joint feature sequence according to the link quality index and device position data in the satellite positioning data;

[0037] A prediction module is configured to predict first link risk information corresponding to the joint feature sequence in a first time period based on the joint feature sequence by using a trained risk prediction model, wherein the risk prediction model is trained by using a long short-term memory network based on an attention mechanism based on historical joint feature sequences;

[0038] A cache module is configured to add a reconstruction pending label to first electrical monitoring data of high priority in electrical monitoring data associated with the first link risk information and perform local caching in a case where the first link risk information meets a preset risk condition;

[0039] A sending module is configured to send the first electrical monitoring data, the reconstruction pending label included in the first electrical monitoring data, and the first link risk information to a central cloud platform in a second time period, so that the central cloud platform reconstructs received historical electrical monitoring data and the first electrical monitoring data by using the first link risk information to generate target electrical monitoring data when the reconstruction pending label is received.

[0040] In a third aspect, the present application provides an electronic device, comprising a processor and a memory storing computer program instructions; the processor reads and executes the computer program instructions to implement the electrical automation monitoring data optimization method based on edge computing in any one of the embodiments of the first aspect.

[0041] In a fourth aspect, the present application provides a computer readable storage medium, the computer readable storage medium storing computer program instructions, the computer program instructions being executed by a processor to implement the electrical automation monitoring data optimization method based on edge computing in any one of the embodiments of the first aspect.

[0042] This application presents a method, system, device, and computer storage medium for optimizing electrical automation monitoring data based on edge computing. By acquiring electrical monitoring data, satellite positioning data, and communication status data of mobile industrial equipment at edge computing nodes, and combining this with equipment location and geometric accuracy attenuation factors, a joint feature sequence correlated with spatiotemporal and channel quality is constructed. This allows for the early prediction of communication interruption risks in complex environments using a risk prediction model. Upon risk prediction, high-priority critical electrical monitoring data is proactively cached locally and tagged with "to be reconstructed." After communication is restored, this data is securely transmitted to the central cloud platform, which then reconstructs the data using the risk information and the received data. Therefore, this application effectively avoids the irreversible loss of critical data caused by sudden signal interruptions in complex scenarios, ensuring the integrity and effectiveness of monitoring data acquired in the cloud.

[0043] Furthermore, this application introduces environmental point cloud data and calculates a spatial obstruction factor characterizing the degree of physical obstruction of satellite positioning signals based on the real-time positional relationship between mobile industrial equipment and satellites. This obstruction factor is used as a key dimension in constructing a joint feature sequence, enabling the risk prediction model to learn not only the correlation between historical communication quality and equipment location, but also the deep causal relationship between direct obstruction of satellite positioning signals by the specific physical environment and future communication interruptions. This provides a physical basis for predicting impending communication risks, thereby improving prediction accuracy. Therefore, it further ensures the integrity of critical electrical monitoring data when mobile industrial equipment traverses complex obstructed environments. Attached Figure Description

[0044] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 This is a flowchart illustrating an embodiment of an electrical automation monitoring data optimization method based on edge computing provided in this application.

[0046] Figure 2 This is a flowchart illustrating a method for generating target electrical monitoring data according to an embodiment of this application;

[0047] Figure 3 This is a schematic diagram of the structure of an electrical automation monitoring data optimization system based on edge computing provided in one embodiment of this application;

[0048] Figure 4 This is a schematic diagram of the hardware structure of an electronic device provided in one embodiment of this application. DETAILED DESCRIPTION

[0049] The features and exemplary embodiments of the various aspects of the present application will be described in detail below with reference to the drawings. The following detailed description is merely intended to explain the present application, and is not intended to limit the present application. The present application can be implemented without some of the specific details. The following description of the embodiments is merely provided to give a better understanding of the present application by showing examples of the present application.

[0050] It should be noted that the terms such as first and second, etc., are merely intended to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Also, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that processes, methods, articles or devices that include a series of elements not only include those elements, but also include other elements not explicitly listed, or inherent to such processes, methods, articles or devices. Without more limitations, the elements defined by the statement "comprise" do not exclude the presence of additional identical elements in the processes, methods, articles or devices that include the elements.

[0051] However, in the real-time monitoring scene of mobile industrial equipment, when the equipment moves in a complex environment such as a mine, a large factory or a tunnel, the quality of the communication link does not decrease smoothly, but often shows frequent and sharp instantaneous interruption. Neither data compression nor transmission strategy adjustment in the prior art can effectively cope with the sudden signal loss, which may cause the loss of key electrical monitoring data during transmission. Therefore, the data optimization method in the prior art cannot guarantee the integrity and effectiveness of the electrical automation monitoring data in the real-time monitoring scene of mobile industrial equipment.

[0052] To solve the problems in the prior art, the embodiments of the present application provide an electrical automation monitoring data optimization method, system, device and computer storage medium based on edge computing. First, the electrical automation monitoring data optimization method based on edge computing provided by the embodiments of the present application will be introduced.

[0053] Figure 1 The flowchart of the electrical automation monitoring data optimization method based on edge computing provided by an embodiment of the present application is shown. The method is applied to an edge computing node, such as Figure 1 As shown in the figure, the method includes steps S110 to S150.

[0054] S110: synchronously acquire electrical monitoring data, satellite positioning data and communication status data of the mobile industrial device, the satellite positioning data including a geometric dilution of precision factor.

[0055] In the real-time monitoring scenario of the mobile industrial device, the mobile industrial device can be an unmanned mining truck autonomously driving in a mining area or an automated guided vehicle running in a large automated warehouse. The electrical monitoring data refers to a set of parameters of the operating state of the electrical system of the mobile industrial device, including motor drive current, battery pack voltage, power module temperature, and controller state, etc., which provides a basis for device state evaluation and fault diagnosis. The satellite positioning data refers to information related to the spatial position of the device obtained through a Global Navigation Satellite System (GNSS), mainly including device position data, a geometric dilution of precision factor, and satellite position data; the device position data can be the latitude and longitude coordinates, altitude, and speed of the device, the geometric dilution of precision factor (GDOP) is a factor that reflects the influence of the geometric distribution of the constellation on the determination precision of the three-dimensional position, clock error, etc. of the user, and the satellite position data can be the real-time spatial position information of the satellite currently used for positioning solution, i.e. the coordinate value of the accurate position of the satellite at a certain time in a certain coordinate system, for example, the three-dimensional coordinates in the Earth-Centered, Earth-Fixed (ECEF) coordinate system. The communication status data refers to the parameters of the quality of the wireless data link between the edge computing node and the remote center cloud platform, including Received Signal Strength Indication (RSSI), Signal-to-Noise Ratio (SNR), and the location data of the communication base station providing the service.

[0056] The edge computing node deployed on the mobile industrial device communicates through a local area network to acquire real-time electrical monitoring data, such as motor drive current and battery pack voltage. At the same time, the node acquires satellite positioning data from a multi-mode GNSS receiver module, including device position data, a geometric dilution of precision factor, and satellite position information of each satellite currently used for positioning solution, such as the real-time three-dimensional coordinates of the satellite currently used for positioning solution in the ECEF coordinate system, wherein the geometric dilution of precision factor can be a GDOP value. In addition, the node also acquires current communication status data from a wireless communication module, which includes real-time RSSI and SNR values, and the location data of the currently serving base station.

[0057] Exemplarily, take a self-driving mining truck operating in a large open-pit mine as an example, the truck is driving along the planned path from the open dispatch yard to a vertical mining face. At a certain moment, the edge computing node on board obtains from the vehicle's control unit that the real-time working current of the drive motor is 350 amperes, and the voltage of the battery pack is 600 volts. At the same time, its GNSS receiver module is providing the edge computing node with the precise latitude, longitude and altitude coordinates of the truck, as well as a geometric dilution of precision GDOP value that has increased from 1.5 to 2.5 due to the partial sky being blocked by the mining face rock wall, and the three-dimensional coordinates of the five satellites currently participating in positioning solution in the ECEF coordinate system. In addition, its 5G communication module also reports to the edge computing node that the signal RSSI value connected to the mine edge base station has dropped to -95 dBm, the SNR value is 8 dB, and the precise location coordinates of the base station are provided.

[0058] S120: Based on the communication state data and the geometric dilution of precision, a link quality index is generated by weighted fusion, and a joint feature sequence is constructed according to the link quality index and the device location data in the satellite positioning data.

[0059] The link quality index refers to a numerical value representing the degree of data transmission reliability of the communication link between the edge computing node and the central cloud platform. The joint feature sequence refers to a data set arranged in chronological order, in which each data point binds a link quality index with a specific device spatial position.

[0060] Firstly, the communication state data and the geometric dilution of precision are normalized, mapping the numerical values of all parameters to a unified interval, and assigning a preset weight coefficient to the communication state data and the geometric dilution of precision. The size of the weight coefficient reflects the contribution of the parameter to the evaluation of the final link quality, and a link quality index is obtained by weighted summation. Subsequently, the edge computing node pairs the generated link quality index with the device location data at the same timestamp to form a data tuple. The edge computing node can continuously generate such joint feature sequences and store them in the queue of the joint feature sequence in chronological order.

[0061] Exemplarily, the edge computing node on the unmanned mining truck obtains real-time data of RSSI value of -95 dBm, SNR value of 8 dB, and GDOP value of 2.5. The node calculates a link quality index, for example, a value of 0.3, by weighting and summing the three values according to preset weight coefficients. The node pairs the link quality index 0.3 with the accurate longitude and latitude coordinates of the mining truck currently close to the mining surface to form a new data tuple. The newly generated data tuple is stored in a queue, and a plurality of historical data tuples arranged in time sequence in the queue collectively constitute a joint feature sequence. For example, the joint feature sequence can be [(timestamp T-2, {longitude A, latitude B}, link quality index: 0.9), (timestamp T-1, {longitude C, latitude D}, link quality index: 0.8), (timestamp T, {longitude E, latitude F}, link quality index: 0.3)].

[0062] S130: predicting first link risk information corresponding to the joint feature sequence in a first time period based on the joint feature sequence and using a trained risk prediction model, the risk prediction model being trained based on a long short-term memory network with attention mechanism and using historical joint feature sequences.

[0063] The risk prediction model refers to a pre-trained and deployed computing model on the edge computing node, which is used to infer the future communication link state. The link risk information is the output result of the risk prediction model, which is a prediction of a series of continuous states of the communication link in a future time period, which can be a set of risk levels sorted by time, such as good, poor, very poor, and interruption risk. The first time period refers to a short time window in the future, for example, 5 seconds in the future, when the risk prediction model makes a prediction. The long short-term memory network with attention mechanism refers to a specific structure of deep learning network, which is trained and learned offline based on a large number of historical joint feature sequences, and finally obtains a risk prediction model with prediction ability.

[0064] First, the edge computing node will input the constructed and maintained joint feature sequence into a locally running risk prediction model as a time series. The model includes an input layer, an attention layer, a long short-term memory (LSTM) layer, and an output layer. The input layer is responsible for receiving joint feature sequence data; the attention layer weights each time step in the input sequence, assigning higher weights to historical data points that are more indicative of future risk prediction; then, the weighted sequence is fed into the LSTM layer, which captures complex long-term temporal dependencies and change patterns in the sequence data through its internal gating unit structure; finally, the output of the LSTM layer is processed by a fully connected output layer to calculate and output a first link risk information for the first time period, i.e., the risk level that the communication link may experience at each time in the future few seconds.

[0065] Exemplarily, the edge computing node on the unmanned mining truck will provide a joint feature sequence containing multiple data tuples in recent time to its locally running risk prediction model as an input. The joint feature sequence records the whole process of the link quality index gradually decreasing from 0.9 in the open field to 0.3 near the mining face as the vehicle travels. When processing this sequence, the attention layer in the model identifies that the segment of rapid decline in the index is the key to predicting future risk and assigns higher analysis weights to these data points. Then, the weighted data is fed into the LSTM layer for time series pattern analysis. Finally, the output layer of the model calculates and generates a first link risk information for the next 5 seconds, which can be a specific risk level sequence, exemplarily represented as: [Time: T+1 second, Risk Level: Good], [Time: T+2 second, Risk Level: Poor], [Time: T+3 second, Risk Level: Very Poor], [Time: T+4 second, Risk Level: Interruption Risk], [Time: T+5 second, Risk Level: Interruption Risk].

[0066] As another implementation of the present application, before step S130: predicting the first link risk information corresponding to the joint feature sequence in the first time period based on the joint feature sequence using the trained risk prediction model, the method further includes a process of obtaining the trained risk prediction model. The process specifically includes:

[0067] First, a training sample set containing a plurality of historical training samples is obtained, wherein each training sample includes a historical joint feature sequence collected in a past time period and a real link risk information label recording a subsequent real communication link state corresponding to the historical joint feature sequence. Subsequently, for each training sample, the following steps are performed: inputting the historical joint feature sequence in the training sample into an initialized risk prediction model based on an attention mechanism long short-term memory network structure to obtain a predicted link risk information label; calculating a prediction loss function value of the current model according to the real link risk information label corresponding to the training sample and the predicted link risk information label output by the model; determining whether the prediction loss function value meets a preset training stop condition, for example, whether the loss value is less than a preset convergence threshold or whether the training iteration number has reached an upper limit. In the case where the prediction loss function value does not meet the training stop condition, the model parameters of each layer in the risk prediction model are adjusted by a back propagation algorithm, such as the weights of the attention layer and the internal parameters of the long short-term memory network layer, to obtain an updated risk prediction model. Then, the step of inputting the historical joint feature sequence into the updated risk prediction model to obtain a new predicted link risk information label is returned, and the iteration process is repeated until the prediction loss function value of the model meets the training stop condition, at which time a trained risk prediction model is obtained.

[0068] S140: In the case where the first link risk information meets the preset risk condition, adding a to-be-reconstructed label to the first electrical monitoring data of high priority in the electrical monitoring data associated with the first link risk information and performing local caching.

[0069] The preset risk condition refers to a trigger standard for judging whether the link risk reaches a need for intervention measure, which can be a threshold corresponding to the risk level in the link risk information, for example, when the predicted risk level is extremely poor, that is, the preset risk condition is met. The first electrical monitoring data refers to the part of data that is predefined as being crucial to the safe operation and state evaluation of the equipment in all electrical monitoring data, and these data usually have higher priority, such as the core working current of the motor, the key voltage of the battery pack and the state instruction of the safety controller. The to-be-reconstructed label is an identifier attached to the cached data, which indicates to the central cloud platform that this data needs subsequent data reconstruction processing.

[0070] Firstly, the edge computing node compares the first link risk information with a preset risk condition. When the comparison result shows that the risk level of a certain future time point in the first link risk information meets the preset risk condition, for the electrical monitoring data collected at the high-risk time point, according to the preset priority division rule, the first electrical monitoring data is extracted, and the remaining low-priority data is processed by reducing the transmission frequency. Subsequently, a to-be-reconstructed label is added to the first electrical monitoring data, and the data with the label is stored in the local cache area of the node, waiting for the subsequent transmission opportunity. At the same time, for the remaining low-priority data, the node dynamically adjusts the processing strategy based on the current predicted risk level for transmission. The node adjusts the processing strategy for low-priority data based on the current predicted risk level, which can be to reduce the data collection and transmission frequency, or to aggregate and compress the data. The higher the risk level, the greater the degree of frequency reduction or compression.

[0071] Illustratively, the edge computing node on the unmanned mining truck receives the predicted link risk information for the next 5 seconds, in which the risk level from the 3rd second is extremely poor. The preset risk condition is set to trigger when the risk level reaches extremely poor. From the next second, the edge computing node executes the caching strategy: when the electrical monitoring data is collected at the predicted 3rd second, according to the priority rule, the first electrical monitoring data such as motor current and battery voltage is extracted and a to-be-reconstructed label is added to it and written to the local cache. At the same time, for low-priority data such as power module temperature that changes slowly, the edge computing node reduces the sampling frequency of the originally high-frequency low-priority data stream, for example, the originally collected power module temperature data every second is now collected and attempted to be transmitted every ten seconds; or aggregate and calculate the low-priority data within a period of time, for example, calculate an average value of the temperature data within 10 seconds, and then only attempt to send the one aggregated average value.

[0072] S150: The first electrical monitoring data, the to-be-reconstructed label included in the first electrical monitoring data, and the first link risk information are transmitted to the central cloud platform within a second time period, so that the central cloud platform reconstructs the received historical electrical monitoring data and the first electrical monitoring data using the first link risk information when the to-be-reconstructed label is received, to generate target electrical monitoring data.

[0073] The second time period refers to a subsequent time window in which the communication link quality is restored to a stable and suitable state for data transmission after the end of the first time period. The central cloud platform refers to a remote data center responsible for receiving and aggregating data from all edge nodes and completing data reconstruction. The historical electrical monitoring data refers to a set of electrical monitoring data that has been successfully received and stored on the central cloud platform side, including non-continuous low-priority electrical monitoring data received at a reduced transmission frequency during the first time period. The target electrical monitoring data refers to a complete data stream of electrical automation detection data of all priorities obtained by reconstruction.

[0074] First, the transmission opportunity of entering the second time period is determined by the edge computing node. After the first time period, the current communication link quality is continuously evaluated, and second link risk information is generated. When the second link risk information shows that the current link no longer meets the preset risk condition, i.e., the link has been restored to be stable, the first electrical monitoring data with a reconstruction tag is read from the local cache and is packaged and sent to the central cloud platform together with the first link risk information that caused the cache. Subsequently, the central cloud platform identifies the reconstruction tag after receiving the data package and starts the reconstruction process. The central cloud platform uses the received first link risk information to determine a filling parameter for filling the sparse historical electrical monitoring data, such as a target filling parameter corresponding to the risk level. The historical electrical monitoring data is filled by difference using the target filling parameter to obtain filled historical electrical monitoring data, and the filled historical electrical monitoring data and the first electrical monitoring data are spliced according to time points to generate target electrical monitoring data.

[0075] In this embodiment, the electrical monitoring data, satellite positioning data, and communication state data of the mobile industrial device are obtained at the edge computing node, and the device position and geometric dilution of precision factor are combined to construct a joint feature sequence associated with space-time and channel quality, so that the risk prediction model can be used to predict the communication interruption risk that is about to occur in a complex environment in advance; when the risk is predicted, the high-priority key electrical monitoring data is actively cached and added with a reconstruction tag, and is safely sent to the central cloud platform after the communication is restored, and then the cloud platform reconstructs the data using the risk information and the received data. Therefore, the application effectively avoids the irreversible loss of key data caused by sudden signal interruption of the mobile industrial device in a complex scenario, and the completeness and effectiveness of the monitoring data obtained in the cloud.

[0076] In an implementable embodiment, the communication state data includes base station position data; the method further includes:

[0077] Synchronously acquiring environmental point cloud data of a mobile industrial device. The environmental point cloud data refers to a mass of point sets of a three-dimensional form of a physical environment around the mobile industrial device acquired by a sensor. Each point contains accurate coordinate information in a three-dimensional space, which can be X, Y, and Z coordinate values. The environmental point cloud data can digitally reproduce the geometric shapes of topography, buildings, vegetation, and other obstacles around the device.

[0078] An edge computing node deployed on the mobile industrial device acquires environmental point cloud data from one or more three-dimensional laser radar sensors or depth cameras carried by the vehicle. These sensors continuously emit laser beams or structured light outward and calculate three-dimensional coordinates on the surfaces of surrounding objects by measuring time differences or phase differences of returned signals. The edge computing node collects and preliminarily processes the point cloud data output by these sensors in real time to generate a point cloud data frame with a time stamp that can reflect the physical environment around the device at the current time.

[0079] The device position data is taken as a first end point, the base station position data is taken as a second end point, and a first signal propagation path is obtained by determining a spatial straight line segment connecting the first end point and the second end point. The first signal propagation path refers to an ideal straight line connecting the position of a base station antenna providing communication services and the position of a receiving antenna on the mobile industrial device in a three-dimensional space, representing the shortest and most direct geometric path of radio signals from the transmitting end to the receiving end without any obstacles.

[0080] The base station position data and the device position data are both three-dimensional space coordinates and are converted to the same coordinate system. The edge computing node determines a spatial straight line segment between the base station position coordinate as the second end point and the device position coordinate as the first end point by spatial geometric calculation, and the straight line segment is defined as the first signal propagation path.

[0081] A first Fresnel zone of the first signal propagation path is calculated, a spatial occlusion volume ratio is calculated by spatial intersection based on the first Fresnel zone of the first signal propagation path and the environmental point cloud data, and a first spatial occlusion factor is obtained.

[0082] In electromagnetic wave propagation theory, the Fresnel zone refers to the set of spatial regions that satisfy a specific phase difference condition along the propagation path of an electromagnetic wave between two points. According to Huygens' Fresnel principle, each wavefront can be decomposed into countless secondary wavelet sources. These wavelets coherently superimpose during propagation in space, and the resulting wavefront is composed of the vector sum of contributions from all wavelets. The first Fresnel zone is a rugby ball-shaped three-dimensional spatial region surrounding the propagation path of the first signal. The sum of the distances from any point on the boundary of this region to the transmitter and receiver, which is the straight-line distance between the two points, exceeds half a wavelength. This region defines the energy range that contributes the most to the free-space propagation of the signal; any obstacle intruding into this region can have a significant negative impact on the signal strength and phase. The spatial obstruction factor is a quantitative indicator used to accurately measure the severity of physical obstruction of the first signal propagation path.

[0083] (1)

[0084] Among them, R n λ is the lateral radius of the nth Fresnel zone at a specific point on the signal propagation path, where n is the order of the Fresnel zone; λ is the wavelength of the communication signal, which is determined by the frequency of the signal, for example, the wavelength of a 2.4 GHz Wi-Fi signal is approximately 0.125 meters; d is the distance from one end of the signal path to the specific point where the radius is calculated; d is the distance from the other end of the signal path to the same specific point.

[0085] The edge computing node first constructs a three-dimensional mathematical model of the first Fresnel zone (i.e., when n=1) corresponding to the determined first signal propagation path and the wavelength of the communication signal used, using the Fresnel zone geometric formula as shown in formula (1). Specifically, the construction process involves the node sampling equidistantly or adaptively along the propagation path from the starting point to the ending point, generating a series of discrete specific points. Subsequently, for each specific point, the lateral radius of the first Fresnel zone at that point is calculated using formula (1) based on its distances d and d' to the two ends of the path, i.e., the base station and the mobile industrial equipment. For example, at the midpoint of a propagation path with a total length of 1000 meters, the calculated maximum radius for a signal with a wavelength of 0.125 meters is approximately 7.9 meters. By connecting and rotating all these calculated radius points, an oval-shaped ellipsoid with the first signal propagation path as its central axis can be constructed in memory. This ellipsoid is the three-dimensional mathematical model of the first Fresnel zone, providing a precise geometric boundary for subsequent spatial intersection calculations.

[0086] After the three-dimensional mathematical model of the first Fresnel zone is constructed, the edge computing node then performs a spatial intersection calculation to obtain the first spatial occlusion factor. This process first traverses each three-dimensional coordinate point in the environmental point cloud data in the same coordinate system as the three-dimensional model of the Fresnel zone, and determines whether each point falls within the spatial volume defined by the first Fresnel zone one by one through a point-in-ellipsoid determination algorithm, thereby screening out all point cloud subsets that constitute physical occlusions. In order to calculate the spatial occlusion volume ratio, the node first calculates the total volume of the Fresnel zone according to its geometric parameters. At the same time, the node uses a voxelization method to divide the space occupied by the first Fresnel zone into a three-dimensional grid, for example, into a plurality of 1 cubic meter size cubic units, i.e. voxels, and then estimates the total volume occupied by the occlusion object by counting the number of voxels that contain at least one occlusion point cloud subset point. Finally, the node divides the estimated occlusion volume by the total volume of the first Fresnel zone, and the quotient obtained is determined as the first spatial occlusion factor. For example, if the occlusion volume is estimated to be 400 cubic meters, and the total volume of the Fresnel zone is 1000 cubic meters, then the first spatial occlusion factor obtained is 0.4.

[0087] In step S120, the link quality index is generated by weighted fusion based on the communication state data and the geometric dilution of precision factor, including: generating the link quality index by weighted fusion based on the communication state data, the geometric dilution of precision factor and the first spatial occlusion factor.

[0088] First, the communication state data, the geometric dilution of precision factor and the first spatial occlusion factor need to be normalized, mapping the numerical values of all parameters to a unified interval, for example, 0 to 1, where a value of 1 represents the best state and a value of 0 represents the worst state. For parameters such as received signal strength RSSI and signal-to-noise ratio SNR, which are better the higher the value, linear mapping can be used for normalization; while for parameters such as geometric dilution of precision factor GDOP and first spatial occlusion factor, which are better the lower the value, reverse processing is needed during normalization to ensure that the final value also conforms to the unified standard that higher values represent better states. After all input parameters are normalized, a preset weight coefficient is assigned to the communication state data, the geometric dilution of precision factor and the first spatial occlusion factor, where the first spatial occlusion factor, as a parameter that can directly reflect the risk of physical occlusion, can be assigned a higher weight coefficient. Subsequently, a comprehensive link quality index is obtained by weighted summation, and the calculation process can be as shown in formula (2):

[0089] (2)

[0090] wherein, LQI is a link quality index; Wr, Ws, Wg and Ws are preset weight coefficients of RSSI, SNR, GDOP and spatial occlusion factor respectively; Vr, Vs, Vg and Vs represent communication state data, GDOP and spatial occlusion factor respectively.

[0091] Exemplarily, the three-dimensional laser radar of the unmanned mining truck synchronously acquires the environmental point cloud data of the rock wall of the front mining face. The edge computing node determines a first signal propagation path pointing to the base station by using the known base station position and the accurate device position data of the truck itself, and constructs a three-dimensional mathematical model of the first Fresnel zone around the path based on the wavelength of the communication signal. By performing spatial intersection calculation, the node determines that a part of the point set in the collected rock wall point cloud data falls into the interior of the Fresnel zone, and further estimates that this part of the occlusion occupies a certain proportion, for example, 40%, of the total volume of the Fresnel zone, so as to directly calculate the first spatial occlusion factor as 0.4.

[0092] After that, the four key data are normalized to uniformly map their values to the interval of 0 to 1. For example, the RSSI value of -95dBm is normalized to 0.25, the SNR value of 8dB is normalized to 0.27, the GDOP value of 2.5 is inversely normalized to 0.7, and the first spatial occlusion factor with a value of 0.4 is also inversely normalized to 0.6. After all the parameters are normalized, preset weight coefficients are assigned to these normalized values, for example, the weight of the first spatial occlusion factor is 0.4, the weight of RSSI is 0.3, the weight of SNR is 0.2, and the weight of GDOP is 0.1. Finally, the link quality index is calculated as 0.44 by multiplying these normalized values by the corresponding weights and summing them up.

[0093] In an implementable embodiment, the satellite positioning data comprises satellite position data;

[0094] The method further comprises:

[0095] The satellite position data is taken as a third end point, and the second signal propagation path is obtained by determining a spatial straight line segment connecting the first end point and the third end point. The second signal propagation path refers to a straight line connecting a certain satellite currently used for positioning solution and the GNSS receiving antenna on the mobile industrial device in the three-dimensional space. Since the device simultaneously receives signals of multiple satellites, there will be multiple second signal propagation paths, each of which represents the shortest geometric propagation route of a satellite signal in the case of no occlusion.

[0096] The edge computing node utilizes the device location data and the satellite location data of each satellite. For each satellite participating in the positioning solution, the node takes the satellite location coordinates as the starting point, i.e., the third endpoint, and the device location coordinates as the end point, i.e., the first endpoint, to determine a unique spatial straight line segment through spatial geometric calculation, which is defined as the second signal propagation path corresponding to the satellite. This process is repeated for all satellites participating in the solution to obtain a set of second signal propagation paths.

[0097] The first Fresnel zone of the second signal propagation path is calculated, and based on the first Fresnel zone of the second signal propagation path and the environmental point cloud data, the spatial occlusion volume ratio is calculated through spatial intersection calculation to obtain the second spatial occlusion factor.

[0098] The first Fresnel zone of the second signal propagation path refers to the rugby-shaped three-dimensional spatial region around each second signal propagation path. The second spatial occlusion factor is a quantitative index for accurately measuring the severity of the comprehensive occlusion of the satellite signal propagation paths participating in the positioning solution by the physical environment. The factor can be the average or weighted average of the occlusion factors of all second signal propagation paths, and its value directly reflects the overall smoothness and reliability of the satellite positioning signal.

[0099] The edge computing node repeats the similar operation for each second signal propagation path as when calculating the first spatial occlusion factor. The node first calculates the first Fresnel zone three-dimensional model corresponding to each second signal propagation path based on the wavelength of the GNSS signal. Then, the node performs a spatial intersection calculation with the environmental point cloud data for each first Fresnel zone to obtain the volume ratio of the satellite signal being physically occluded. After calculating the occlusion ratios of all satellite signal paths, the node aggregates these individual occlusion ratios into a comprehensive value representing the overall occlusion of the current satellite positioning signal by, for example, taking the arithmetic mean, and determines this value as the second spatial occlusion factor.

[0100] In step S120, a joint feature sequence is constructed according to the link quality index and the device location data in the satellite positioning data, including: constructing a joint feature sequence according to the link quality index, the second spatial occlusion factor, and the device location data in the satellite positioning data.

[0101] The edge computing node pairs the link quality index, the calculated second spatial occlusion factor, and the device location data under the same timestamp to form a data tuple containing multi-dimensional information. Such data tuples can be continuously generated and stored in a queue in chronological order, and the queue constitutes a joint feature sequence containing multi-source physical occlusion information and link quality information.

[0102] Exemplarily, the edge computing node first acquires the device location data of the miner card itself, for example, the coordinates are , and acquires the satellite location data of each of the five satellites currently participating in positioning, for example, the coordinates of satellite 1 are , the coordinates of satellite 2 are , and so on. By performing coordinate conversion on the coordinates of each satellite and the coordinates of the miner card and then connecting the two points, five second signal propagation paths from the satellites to the miner card are determined. Subsequently, the Fresnel zone analysis and spatial intersection calculation are performed on each path. Taking satellite 1 as an example, the node calculates the corresponding first Fresnel zone three-dimensional model based on the signal wavelength, and determines that there are 1500 points in the front mining surface environment point cloud data falling into the Fresnel zone by the spatial intersection algorithm, and estimates that the shielding volume ratio is 0.5. Similarly, the node calculates the shielding ratios of satellite 2, satellite 3, satellite 4 and satellite 5 as 0.6, 0.4, 0 and 0 respectively. By calculating the arithmetic mean of the five independent shielding ratios (0.5, 0.6, 0.4, 0, 0), a second spatial shielding factor that can comprehensively reflect the overall shielding of the satellite signal is finally obtained, and the value is 0.3. Finally, when constructing the joint feature sequence, a new data tuple is generated at the current time T. This data tuple contains the current latitude and longitude coordinates of the miner card, the link quality index with a value of 0.44, and the second spatial shielding factor with a value of 0.3.

[0103] The embodiment introduces environment point cloud data and calculates a spatial shielding factor representing the degree of physical shielding of the satellite positioning signal based on the real-time position relationship between the mobile industrial device and the satellite. The shielding factor is used as a key dimension for constructing a joint feature sequence, so that the risk prediction model can not only learn the correlation between historical communication quality and device location, but also learn the deep causal relationship between the direct shielding of the specific physical environment to the satellite positioning signal and the future communication interruption, and can have physical reality basis for predicting the impending communication risk, thereby improving the accuracy of the prediction. Therefore, the integrity of the key electrical monitoring data of the mobile industrial device when passing through a complex shielding environment is further ensured.

[0104] In an implementable embodiment, the method further comprises:

[0105] synchronously acquiring a link quality index of at least one adjacent mobile industrial device in a preset area centered on the mobile industrial device, and calculating a link quality index mean of the link quality index of the at least one adjacent mobile industrial device.

[0106] The neighboring mobile industrial device refers to other working devices of the same or similar type as the device and currently located within a certain spatial range around the device. The preset area is a circular space defined by a preset radius with the current location of the mobile industrial device as the center. The mean of the link quality index is a statistical indicator representing the quality of the communication environment in the local area where the device is located.

[0107] The edge computing node on the mobile industrial device periodically broadcasts its ID and location through a vehicle-to-vehicle (V2V) short-range communication method and simultaneously listens to information broadcast from other devices. When the node receives information from other devices, it first determines whether the device is located within a preset area centered on itself, such as a circular area with a radius of 20 meters, based on the received location information. For all neighboring mobile industrial devices located within the area, extract their link quality index calculated at the same time from their respective broadcast information. After collecting the link quality index of all neighboring devices, an arithmetic mean is calculated to obtain a mean of the link quality index representing the overall communication condition of the current local environment.

[0108] According to the link quality index, the second spatial obstruction factor, and the device location data in the satellite positioning data, a joint feature sequence is constructed, including: pairing the link quality index, the mean of the link quality index, the second spatial obstruction factor, and the device location data at the same timestamp to form a data tuple, and arranging the data tuples in chronological order to construct the joint feature sequence.

[0109] The edge computing node pairs the link quality index, the mean of the link quality index, the second spatial obstruction factor, and the device location data at the same timestamp as a data tuple, and stores it in a queue in chronological order to construct the joint feature sequence.

[0110] Exemplarily, after the edge computing node of the mining truck A calculates its link quality index as 0.44 and the second spatial obstruction factor as 0.3, the edge computing node of the mining truck A finds, through V2V short-range communication, that there is another unmanned mining truck B traveling in the same direction within its preset area of 20 meters. The mining truck A extracts from the broadcast information of the mining truck B that the mining truck B, due to its closer position to the mining face, has a link quality index of 0.25 calculated at the same time. The edge computing node of the mining truck A then calculates the mean link quality index as 0.345. Finally, in constructing the joint feature sequence, a data tuple is generated at the current time T, which contains the current latitude and longitude coordinates of the mining truck A, the link quality index of the mining truck A as 0.44, the second spatial obstruction factor as 0.3, and the mean link quality index as 0.345. For example, the joint feature sequence can be: [timestamp T, {latitude X, longitude Y}, self LQI: 0.44, mean LQI: 0.345, obstruction factor: 0.3].

[0111] In an implementable embodiment, the historical electrical monitoring data includes second electrical monitoring data of low priority in the electrical monitoring data associated with the first link risk information received by the central cloud platform within the first time period, in the case that the first link risk information satisfies the preset risk condition.

[0112] The historical electrical monitoring data particularly includes those second electrical monitoring data of lower priority which are still attempted to be transmitted through the unstable link when the edge computing node locally caches the first electrical monitoring data of high priority due to the predicted communication risk within the first time period. Since the edge node actively adopts the transmission strategy of frequency reduction or aggregation to process these low-priority data in the high-risk period, for example, changing the originally high-frequency collection to sparse sampling transmission, and the unstable link itself also causes random packet loss, ultimately resulting in that the historical electrical monitoring data successfully received by the central cloud platform within the time period is actually a sparse and possibly aggregated low-priority data stream of a group of time discontinuous data points.

[0113] Since these second electrical monitoring data, such as slowly changing power module temperature, are synchronized in time with the cached first electrical monitoring data, the central cloud platform can fill in these sparse second electrical monitoring data and reconstruct the complete electrical monitoring data with the subsequently received first electrical monitoring data.

[0114] In an implementable embodiment, the second time period is after the first time period; and the sending, in step S150, of the first electrical monitoring data, the to-be-reconstructed label included in the first electrical monitoring data, and the first link risk information to the central cloud platform within the second time period comprises:

[0115] In the case that the second link risk information corresponding to the second time period does not satisfy the preset risk condition, the first electrical monitoring data, the first electrical monitoring data including the to-be-reconstructed label, and the first link risk information are sent to the central cloud platform within the second time period.

[0116] The second link risk information refers to the continuous prediction and evaluation of the communication link quality by the edge computing node in the first time period and in the current and a short future time window. The information is generated in the same way as the first link risk information, and is used to determine whether the link has recovered from the high-risk state.

[0117] After completing the local cache operation of the high-priority first electrical monitoring data, the edge computing node does not immediately send the cached data, but enters a continuous link state monitoring and evaluation cycle. Steps S120 and S130 are repeatedly executed to generate real-time second link risk information. The risk level of each prediction time point in the newly generated second link risk information is determined one by one. Only when all risk levels in the second link risk information show that the preset risk condition is not met, for example, the risk levels of all prediction time points are good, the edge computing node determines that the link has completely recovered. For example, at time T+6 seconds, the second link risk information predicted by the model for the next 5 seconds can be [good, good, good, good, good]. Since all levels in this risk information are lower than the preset trigger condition of extreme difference, i.e., the preset risk condition is no longer met, the edge computing node confirms that the communication link has recovered and stabilized. Once the link is confirmed to be recovered, the node immediately reads all the first electrical monitoring data with the to-be-reconstructed label from its local cache area, and packs it with the first link risk information that caused the cache, i.e., the motor current and battery voltage data saved for 3 seconds with the to-be-reconstructed label and the associated first link risk information [poor, poor, extremely poor, interruption risk, interruption risk].

[0118] Figure 2 A flowchart of a method for generating target electrical monitoring data is shown. As shown in Figure 2 the method includes steps S210 to S210.

[0119] In an implementable embodiment, the historical electrical monitoring data includes non-continuous electrical monitoring data, i.e., non-continuous sparse low-priority electrical monitoring data received in the first time period.

[0120] In step S150, the received historical electrical monitoring data and the first electrical monitoring data are reconstructed using the first link risk information to generate target electrical monitoring data, including:

[0121] S210: According to the risk level of the first link risk information, determine the target filling parameter in the preset risk level and filling parameter mapping relationship.

[0122] The filling parameter mapping relationship is a pre-established lookup table, which includes the correspondence between different risk levels in the link risk information and the order of the polynomial interpolation algorithm. The target filling parameter refers to the order of the corresponding polynomial interpolation obtained by querying the mapping relationship according to the specific risk level. The higher the risk level, the sparser the historical electrical monitoring data, i.e. the data points of the low-priority electrical monitoring data, and the higher-order polynomial interpolation needs to be used to fit the trend of change between data points.

[0123] The central cloud platform first extracts the first link risk information from the received data, for example [poor, poor, very poor, interruption risk, interruption risk]. The preset risk condition is set to trigger when the risk level reaches very poor, starting from the beginning of the first link risk information sequence, and determining the first risk level that meets the preset risk condition. Subsequently, in the correspondence table between the risk level and the order of the polynomial interpolation algorithm as shown in Table 1, the corresponding order of the target polynomial interpolation is determined according to the first risk level that meets the preset risk condition. For example, the first risk level that meets the preset risk condition in the first link risk information is very poor, and the order of the corresponding polynomial interpolation algorithm is 2.

[0124] Table 1: Risk level and polynomial interpolation order correspondence table

[0125] ;

[0126] The risk level and polynomial interpolation order correspondence table in Table 1 defines the corresponding rules between the risk level of the link and the order of the polynomial interpolation, i.e. the risk level and filling parameter mapping relationship. As the link risk level increases, the data points received by the central cloud platform in this period will become more and more sparse. Higher-order polynomial functions need to be used to fit and fill the potential trend of change between missing data points. When the risk level is poor, less data is lost and first-order interpolation is used; when the risk level is very poor, second-order interpolation is used as the data loss increases; and when the risk reaches the highest interruption risk, the data may be largely missing, and third-order polynomial is selected for interpolation reconstruction.

[0127] S220: Differentiate and fill the historical electrical monitoring data using the target filling parameter to obtain the filled historical electrical monitoring data.

[0128] The second electrical monitoring data of low priority received by the central cloud platform in the first time period, such as temperature data, has only two time points of T+2s and T+8s in the time period of T to T+10s. The central cloud platform performs interpolation according to the polynomial order specified by the target filling parameter. For example, the target filling parameter specifies the order of 2, that is, a second-order polynomial interpolation, and solves a quadratic polynomial equation that can optimally fit all known sparse second electrical monitoring data points by taking all known sparse second electrical monitoring data points as input. Using this solved equation, the temperature values corresponding to all missing time points between T and T+10s, such as T+1, T+3, T+4, T+5, T+6, T+7, T+9, and T+10s, are calculated. These newly calculated values are inserted into the original sparse historical electrical monitoring data stream to obtain a complete temperature change curve, that is, an approximately continuous historical electrical monitoring data in time.

[0129] S230: Splicing the filled historical electrical monitoring data and the first electrical monitoring data by time point to generate target electrical monitoring data.

[0130] Traverse each timestamp in the first time period, and at each timestamp, extract the value at the corresponding time point from the filled historical electrical monitoring data and the received first electrical monitoring data, respectively. For example, at T+5s, the power module temperature value at this time point is extracted from the filled historical data as 46.5 degrees Celsius, and the motor current value at this time point is extracted from the first electrical monitoring data as 350 amperes. Then, the two values of different physical meanings are combined into a new data record containing multi-dimensional information, that is, timestamp: T+5s, motor current: 350A, temperature: 46.5℃. By performing this alignment and combination operation on all time points in the entire time period, two independent single-dimensional data streams are finally integrated into a multi-dimensional target electrical monitoring data stream that is complete and continuous in time.

[0131] Through the above two-stage data reconstruction and combination process, the application not only restores the high-priority key electrical monitoring data that is completely missing due to communication interruption, but also intelligently completes the auxiliary information dimension of the data stream using the low-priority data intermittently received under unstable links. This complete solution effectively overcomes the defects of the prior art that cannot guarantee the completeness and effectiveness of the monitoring data when facing sudden signal loss of mobile industrial devices in complex scenarios.

[0132] The data reconstruction and combination process of the embodiment not only restores the low-priority data missing due to the frequency reduction transmission, but also ensures that the high-priority data cached completely can be aligned and combined with a reference data stream that is physically associated and continuous in time, thereby generating high-fidelity target electrical monitoring data that is complete in information dimension and completely continuous in time at the central cloud platform side. The defects of the prior art in which the monitoring data integrity and effectiveness cannot be guaranteed when a mobile industrial device experiences sudden signal loss in a complex scene are effectively overcome.

[0133] Based on the same concept, the embodiment of the present application provides an electrical automation monitoring data optimization system based on edge computing. The following will be described in combination with Figure 3 The electrical automation monitoring data optimization system based on edge computing provided by the embodiment of the present application will be described in detail.

[0134] Figure 3 is a structural block diagram of an electrical automation monitoring data optimization system based on edge computing shown by the embodiment of the present application.

[0135] As Figure 3 shown, the electrical automation monitoring data optimization system based on edge computing is applied to an edge computing node, and the system comprises:

[0136] The acquisition module 310 is configured to synchronously acquire electrical monitoring data of a mobile industrial device, satellite positioning data, and communication state data, and the satellite positioning data comprises a geometric dilution of precision factor.

[0137] The construction module 320 is configured to generate a link quality index by weighted fusion based on the communication state data and the geometric dilution of precision factor, and construct a joint feature sequence according to the link quality index and device position data in the satellite positioning data.

[0138] The prediction module 330 is configured to predict first link risk information corresponding to the joint feature sequence in a first time period based on the joint feature sequence by using a trained risk prediction model, and the risk prediction model is obtained by training a long short-term memory network based on an attention mechanism based on historical joint feature sequences.

[0139] The cache module 340 is configured to add a to-be-reconstructed label to and locally cache first electrical monitoring data of high priority in electrical monitoring data associated with the first link risk information in a case where the first link risk information satisfies a preset risk condition.

[0140] The sending module 350 is configured to send the first electrical monitoring data, the to-be-reconstructed label included in the first electrical monitoring data, and the first link risk information to the central cloud platform within a second time period, so that the central cloud platform reconstructs the received historical electrical monitoring data and the first electrical monitoring data by using the first link risk information when the to-be-reconstructed label is received, and generates target electrical monitoring data.

[0141] In one embodiment, the communication state data includes base station position data; the acquisition module 310 is further configured to synchronously acquire environment point cloud data of the mobile industrial device; the device position data is taken as a first endpoint, the base station position data is taken as a second endpoint, a first signal propagation path is obtained by determining a straight line segment connecting the first endpoint and the second endpoint in space; a first Fresnel zone of the first signal propagation path is calculated, a first spatial occlusion factor is obtained by calculating a spatial occlusion volume ratio through spatial intersection based on the first Fresnel zone of the first signal propagation path and the environment point cloud data; and a link quality index is generated by weighted fusion based on the communication state data, the geometric accuracy attenuation factor, and the first spatial occlusion factor.

[0142] In one embodiment, the satellite positioning data includes satellite position data; the construction module 320 is further configured to take the satellite position data as a third endpoint, obtain a second signal propagation path by determining a straight line segment connecting the first endpoint and the third endpoint in space; calculate a first Fresnel zone of the second signal propagation path, obtain a second spatial occlusion factor by calculating a spatial occlusion volume ratio through spatial intersection based on the first Fresnel zone of the second signal propagation path and the environment point cloud data; and construct a joint feature sequence according to the link quality index, the second spatial occlusion factor, and the device position data in the satellite positioning data.

[0143] In one embodiment, the acquisition module 310 is further configured to synchronously acquire a link quality index of at least one adjacent mobile industrial device in a preset area centered on the mobile industrial device, and calculate a link quality index mean of the link quality index of the at least one adjacent mobile industrial device; form data tuples by pairing the link quality index, the link quality index mean, the second spatial occlusion factor, and the device position data under the same timestamp, and arrange the data tuples in chronological order to construct a joint feature sequence.

[0144] In one embodiment, the historical electrical monitoring data includes second electrical monitoring data of a low priority in electrical monitoring data associated with the first link risk information received by the central cloud platform within the first time period in a case where the first link risk information meets a preset risk condition.

[0145] In an embodiment, the second time period is after the first time period; and the sending module 350 is specifically configured to send the first electrical monitoring data, the to-be-reconstructed label included in the first electrical monitoring data, and the first link risk information to the central cloud platform in the second time period.

[0146] In an embodiment, the historical electrical monitoring data includes discontinuous electrical monitoring data; the sending module 350 is specifically configured to determine a target filling parameter in a preset risk level and filling parameter mapping relationship according to a risk level of the first link risk information; perform difference filling on the historical electrical monitoring data by using the target filling parameter to obtain filled historical electrical monitoring data; and splice the filled historical electrical monitoring data and the first electrical monitoring data according to time points to generate the target electrical monitoring data.

[0147] Figure 3 Each module in the system has the function of implementing each step in the method and can achieve the corresponding technical effects. For brevity, the details are not repeated here. Figures 1 to 2

[0148] Figure 4 Fig. 1 shows a hardware structure schematic diagram of an electronic device provided by an embodiment of the present application.

[0149] The electronic device can include a processor 410 and a memory 420 having computer program instructions stored therein.

[0150] Specifically, the processor 410 described above can include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or can be configured to implement one or more integrated circuits of the embodiments of the present application.

[0151] The memory 420 can include a mass storage for data or instructions. By way of example and not limitation, the memory 420 can include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. Where appropriate, the memory 420 can include removable or non-removable (or fixed) media. Where appropriate, the memory 420 can be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, the memory 420 is a non-volatile solid-state memory.

[0152] ​The memory can include read-only memory (ROM), random access memory (RAM), a magnetic disk storage medium device, an optical storage medium device, a flash memory device, an electrical, optical, or other physical / tangible memory storage device. Thus, generally, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., a memory device) encoded with software that, when executed (e.g., by one or more processors), is operable to perform the operations described with reference to the method according to the first aspect of the present disclosure.

[0153] The processor 410 implements the edge computing-based electrical automation monitoring data optimization method in any of the above embodiments by reading and executing the computer program instructions stored in the memory 420.

[0154] In one example, the electronic device can further include a communication interface 430 and a bus 440. As shown, the processor 410, the memory 420, and the communication interface 430 are connected through the bus 440 and complete communication with each other. Figure 4

[0155] The communication interface 430 is mainly used to realize the communication between the modules, devices, units and / or equipment in the embodiments of the present application.

[0156] The bus 440 includes hardware, software or both to couple the components of the online data traffic billing device to each other. By way of example, and not limitation, the bus can include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand™ interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or another suitable bus or combination of two or more of these. Where suitable, the bus 440 can include one or more buses. Although specific buses are described and illustrated, the present application contemplates any suitable bus or interconnect.

[0157] The electronic device can perform the edge computing-based electrical automation monitoring data optimization method in the embodiments of the present application, thereby realizing the edge computing-based electrical automation monitoring data optimization method described in combination Figures 1 to 2 with the above embodiments.

[0158] ​In addition, in combination with the electrical automation monitoring data optimization method based on edge computing in the above embodiments, an embodiment of the present application can provide a computer readable storage medium for implementation. The computer readable storage medium has computer program instructions stored thereon; the computer program instructions are executed by a processor to implement any one of the electrical automation monitoring data optimization methods based on edge computing in the above embodiments.

[0159] It should be noted that the present application is not limited to the specific configurations and processes described above and illustrated in the drawings. For the sake of brevity, detailed descriptions of well-known methods are omitted. In the above embodiments, several specific steps are described and illustrated as examples. However, the method processes of the present application are not limited to the specific steps described and illustrated, and those skilled in the art can make various changes, modifications and additions, or change the order of steps, after understanding the spirit of the present application.

[0160] The functional blocks shown in the structural block diagrams described above can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a functional card, etc. When implemented in software, the elements of the present application are program or code segments used to perform the required tasks. The program or code segments can be stored in a machine readable medium or transmitted through a data signal carried in a carrier wave over a transmission medium or communication link. The "machine readable medium" can include any medium capable of storing or transmitting information. Examples of machine readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segments can be downloaded via a computer network such as the Internet, an intranet, etc.

[0161] It should also be noted that the exemplary embodiments mentioned in the present application describe some methods or systems based on a series of steps or devices. However, the present application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiments, or in an order different from the embodiments, or several steps can be performed simultaneously.

[0162] The computer program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other processing device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other processing device to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0163] The above only describes specific implementation of the present application. For the convenience and brevity of description, the specific working processes of the above-described system, module and unit can refer to the corresponding processes in the foregoing method embodiments, which will not be described herein. It should be understood that the protection scope of the present application is not limited in this way. Any person skilled in the art can easily think of various equivalent modifications or replacements within the technical range disclosed in the present application, and these modifications or replacements should be covered by the protection scope of the present application.

Claims

1. An edge computing-based electrical automation monitoring data optimization method applied to an edge computing node, characterized in that, The method comprises: synchronously acquiring electrical monitoring data, satellite positioning data and communication state data of a mobile industrial device, the satellite positioning data comprising a geometric dilution of precision factor; based on the communication state data and the geometric dilution of precision factor, generating a link quality index through weighted fusion, and pairing the link quality index with device position data in the satellite positioning data at the same timestamp to form a data tuple, arranging a plurality of data tuples in chronological order to construct a joint feature sequence; based on the joint feature sequence, predicting first link risk information corresponding to the joint feature sequence in a first time period by using a trained risk prediction model, the risk prediction model being trained by using a historical joint feature sequence based on an attention mechanism long short-term memory network; in a case where the first link risk information meets a preset risk condition, adding a reconstruction pending label to first electrical monitoring data of high priority in the electrical monitoring data associated with the first link risk information and performing local caching; sending the first electrical monitoring data, the reconstruction pending label included in the first electrical monitoring data, and the first link risk information to a central cloud platform in a second time period, so that the central cloud platform reconstructs received historical electrical monitoring data and the first electrical monitoring data by using the first link risk information to generate target electrical monitoring data when the reconstruction pending label is received, the historical electrical monitoring data including non-continuous electrical monitoring data; the reconstruction of the received historical electrical monitoring data and the first electrical monitoring data by using the first link risk information to generate target electrical monitoring data comprises: determining a target filling parameter in a preset risk level and filling parameter mapping relationship according to a risk level of the first link risk information; differentially filling the historical electrical monitoring data by using the target filling parameter to obtain filled historical electrical monitoring data; splicing the filled historical electrical monitoring data and the first electrical monitoring data by time point to generate the target electrical monitoring data.

2. The method of claim 1, wherein, The communication state data comprises base station position data; the method further comprises: synchronously acquiring environmental point cloud data of the mobile industrial device; taking the device position data as a first end point and the base station position data as a second end point, obtaining a first signal propagation path by determining a spatial straight line segment connecting the first end point and the second end point; calculating a first Fresnel zone of the first signal propagation path, and obtaining a first spatial occlusion factor by spatial intersection calculation based on the first Fresnel zone of the first signal propagation path and the environmental point cloud data; the link quality index is generated by weighted fusion based on the communication state data and the geometric dilution of precision factor, comprising: the link quality index is generated by weighted fusion based on the communication state data, the geometric dilution of precision factor and the first spatial occlusion factor.

3. The method of claim 2, wherein, The satellite positioning data comprises satellite position data; the method further comprises: The satellite position data is taken as a third endpoint, a second signal propagation path is obtained by determining a spatial straight line segment connecting the first endpoint and the third endpoint; A first Fresnel zone of the second signal propagation path is calculated, and a second spatial occlusion factor is calculated by spatial intersection based on the first Fresnel zone of the second signal propagation path and the environmental point cloud data; The method further comprises: The method further comprises:

4. The method of claim 3, wherein, The method further comprises: The method further comprises: The method further comprises: The method further comprises:

5. The method of claim 1, wherein, The historical electrical monitoring data comprises second electrical monitoring data of low priority in the electrical monitoring data associated with the first link risk information received by the central cloud platform in the first time period, in a case where the first link risk information meets the preset risk condition.

6. The method of claim 1, wherein, The second time period is after the first time period; The method further comprises: The method further comprises:

7. An edge computing based electrical automation monitoring data optimization system, applied to an edge computing node, characterized in that, The system comprises: The acquisition module is configured to synchronously acquire electrical monitoring data, satellite positioning data and communication state data of a mobile industrial device, the satellite positioning data comprising a geometric dilution of precision factor; The construction module is configured to generate a link quality index by weighted fusion based on the communication state data and the geometric dilution of precision factor, and pair the link quality index with device position data in the satellite positioning data at the same timestamp to form a data tuple, and arrange a plurality of data tuples in chronological order to construct a joint feature sequence; The prediction module is configured to predict first link risk information corresponding to the joint feature sequence in a first time period based on the joint feature sequence, by using a trained risk prediction model, the risk prediction model being trained by using a historical joint feature sequence on a long short-term memory network based on an attention mechanism. The cache module is configured to add a reconstruction label to first electrical monitoring data of high priority in the electrical monitoring data associated with the first link risk information and perform local caching in a case where the first link risk information meets a preset risk condition. The sending module is configured to send the first electrical monitoring data, the reconstruction label included in the first electrical monitoring data, and the first link risk information to a central cloud platform in a second time period, so that the central cloud platform reconstructs received historical electrical monitoring data and the first electrical monitoring data by using the first link risk information to generate target electrical monitoring data when the reconstruction label is received, the historical electrical monitoring data includes non-continuous electrical monitoring data, and the reconstructing received historical electrical monitoring data and the first electrical monitoring data by using the first link risk information to generate target electrical monitoring data includes: determining a target filling parameter in a preset risk level and filling parameter mapping relationship according to a risk level of the first link risk information; performing difference filling on the historical electrical monitoring data by using the target filling parameter to obtain filled historical electrical monitoring data; and splicing the filled historical electrical monitoring data and the first electrical monitoring data according to time points to generate the target electrical monitoring data.

8. An electronic device, comprising: The device comprises a processor and a memory storing computer program instructions; The processor executes the computer program instructions to implement the electrical automation monitoring data optimization method based on edge computing according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer program instructions, and the computer program instructions are executed by the processor to implement the electrical automation monitoring data optimization method based on edge computing according to any one of claims 1-6.

Images