Power grid data transmission optimization method for mountainous complex environment

Abstract

The invention provides a mountainous area complex environment-oriented power grid data transmission optimization method, which comprises the following steps of: acquiring state data of each terminal node of a mountainous area power grid, and generating a multi-dimensional state vector based on the state data; inputting the time sequence data corresponding to the multi-dimensional state vector into a prediction model, and outputting channel change trend data corresponding to each terminal node in a future time period by the prediction model; inputting the multi-dimensional state vector and the channel change trend data into an edge strategy optimization model, and outputting a transmission strategy corresponding to each terminal node by the edge strategy optimization model; and transmitting the power grid data of each terminal node to the corresponding communication target based on the transmission strategy. According to the power grid data transmission optimization method for the mountainous area complex environment, the problems of poor communication stability, insufficient channel change prediction accuracy and high data transmission energy consumption during power grid data transmission in the mountainous area complex environment can be solved.

Description

Technical Field

[0001] This invention relates to the field of power communication technology, and in particular to a method for optimizing power grid data transmission in complex mountainous environments. Background Technology

[0002] With the rapid development of new power systems and smart grids, the intelligent and refined monitoring and management of power grids increasingly rely on stable and reliable communication systems. In mountainous environments, there are numerous transmission lines, towers, and electrical equipment in substations. The operational status data (such as conductor temperature, video monitoring, etc.) and fault alarm information generated by these devices are crucial to ensuring the safe and stable operation of the power grid.

[0003] Most existing power grid data transmission methods are optimized for scenarios with abundant communication resources, employing traditional optimization methods such as fixed link selection, simple transmit power allocation, or single communication resource scheduling. This approach fails to adapt to communication blind spots and multipath effects caused by undulating mountainous terrain. Reliance on traditional ground base stations or wired deployments results in insufficient coverage and high construction and maintenance costs. Furthermore, it lacks the ability to accurately predict the coupled impact of sudden weather events and micro-topography in mountainous areas, failing to anticipate dynamic channel changes to support reliable link switching. It also fails to address the power constraints of remote monitoring nodes, using continuous high-power connections or frequent probes, leading to energy waste. These methods are unsuitable for the resource-constrained and complex environments of mountainous power grids, resulting in poor communication stability, insufficient accuracy in channel change prediction, and high data transmission energy consumption.

[0004] Existing power grid data transmission methods suffer from poor communication stability, insufficient accuracy in predicting channel changes, and high energy consumption when used in complex mountainous environments. No effective solutions have yet been proposed. Summary of the Invention

[0005] The present invention provides an optimization method for power grid data transmission in complex mountainous environments, which at least solves the problems of poor communication stability, insufficient accuracy of channel change prediction, and high data transmission energy consumption when power grid data transmission is carried out in complex mountainous environments.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] This invention provides a method for optimizing power grid data transmission in complex mountainous environments, comprising the following steps: collecting state data from various terminal nodes of the power grid in the mountainous area, and generating a multi-dimensional state vector based on the state data; wherein the state data includes environmental state data, service state data, and energy consumption state data; inputting the time-series data corresponding to the multi-dimensional state vector into a prediction model, and having the prediction model output channel change trend data corresponding to each terminal node in future time periods; wherein the prediction model is obtained by training a deep learning time-series prediction model using historical channel quality data, historical weather data, and terrain data; inputting the multi-dimensional state vector and the channel change trend data into an edge strategy optimization model, and having the edge strategy optimization model output a transmission strategy corresponding to each terminal node; wherein the edge strategy optimization model is obtained by reinforcement learning training based on the channel change trend data and historical state data, with the goal of maximizing transmission success rewards, avoiding timeout penalties, and reducing energy consumption; the transmission strategy includes transmission links, transmit power ranges, and data transmission methods; and transmitting power grid data from each terminal node to its corresponding communication target based on the transmission strategy.

[0008] Preferably, collecting status data of each terminal node in a mountainous power grid includes the following steps: collecting environmental status data of each terminal node in the mountainous power grid; wherein, the environmental status data includes weather data, terrain data, and channel quality data; the terrain data includes digital elevation model data retrieved from an edge server and real-time location data; collecting service status data of each terminal node in the mountainous power grid; wherein, the service status data includes service status data of each terminal device connected to the terminal node, and the service priority of the service status data; collecting energy consumption data of each terminal node in the mountainous power grid; wherein, the energy consumption data is calculated based on the transmit power data of each terminal node and the operating conditions of each terminal device.

[0009] Preferably, generating a multidimensional state vector based on the state data includes the following steps: calculating the terrain occlusion degree between each terminal node and each communication target based on the digital elevation model data and the positioning data; and generating a multidimensional state vector based on the state data and the terrain occlusion degree.

[0010] Preferably, before inputting the time-series data corresponding to the multidimensional state vector into the prediction model, the method includes the following steps: collecting historical channel quality data, historical weather data, and terrain data, and integrating them into time-series data of a unified dimension as a training dataset; wherein, the historical channel quality data includes any one or more of the channel signal-to-noise ratio, packet loss rate, and latency data; using the training dataset, training a deep learning time-series prediction model based on a self-attention mechanism; when the performance of the deep learning time-series prediction model converges, the training is completed, and the final prediction model is obtained.

[0011] Preferably, before inputting the multidimensional state vector and the channel change trend data into the edge policy optimization model, the method includes the following steps: collecting historical state data and corresponding historical decision data; in a simulation environment constructed based on the historical state data, training the edge policy model deployed on the edge server with the goal of maximizing the reward function, in combination with the historical decision data; wherein the reward function is constructed with successful data transmission as the reward and transmission timeout and energy consumption as the penalty; when the performance of the edge policy model converges, the training is completed, and a trained edge policy model is obtained; the trained edge policy model is continuously interacted with the simulation environment to optimize the network parameters of the edge policy model, and when the comprehensive reward value corresponding to the reward function is stable and meets a preset threshold, the final edge policy optimization model is obtained.

[0012] Preferably, in a simulation environment built based on the historical state data, the edge policy model deployed on the edge server is trained by combining the historical decision data with the objective of maximizing the reward function. This includes the following steps: constructing a reward function with successful data transmission as the basic reward, successful high-priority data transmission as an additional reward, and transmission timeout and energy consumption as penalties; wherein the transmission latency of the high-priority data is lower than a set threshold; constructing a simulation environment on the edge server based on the historical state data; and initializing a policy network and a value network in the simulation environment to train the edge policy model deployed on the edge server with the objective of maximizing the reward function.

[0013] Preferably, transmitting the power grid data of each terminal node to the corresponding communication target based on the transmission strategy includes the following steps: adjusting the transmission power, corresponding communication resources, and data transmission method of each terminal node based on the transmission strategy; wherein, the communication resources include: sky communication resources, ground communication resources, and space communication resources; the ground communication resources have the highest scheduling priority, followed by the sky communication resources, and lastly the space communication resources; the data transmission method includes: immediate transmission, compressed transmission, and local buffering; transmitting the power grid data of each terminal node to the corresponding communication target based on the transmission power, the transmission link corresponding to the communication resources, and the data transmission method; wherein, the communication target includes: base stations, UAV relay platforms, and BeiDou satellites.

[0014] Preferably, transmitting the power grid data of each terminal node to the corresponding communication target based on the transmission power, the transmission link corresponding to the communication resource, and the data transmission method includes the following steps: processing the power grid data of each terminal node based on the data transmission method to obtain transmission strategy matching data; wherein, the data processing method includes any one or more of lossless compression processing, multi-source same-source data aggregation processing, and caching processing; encrypting the transmission strategy matching data to obtain corresponding encrypted data; and transmitting the encrypted data to the corresponding communication target based on the transmission power, the transmission link corresponding to the communication resource, and the data transmission method.

[0015] Preferably, transmitting the encrypted data to the corresponding communication target includes the following steps: when the terminal node is in a signal blind zone, uploading the encrypted data to the terminal node's cache module; obtaining cached data corresponding to the encrypted data from the cache module via a mobile relay station; wherein the mobile relay station and the terminal node communicate via a wireless network or Bluetooth; and transmitting the cached data to the corresponding communication target.

[0016] Another aspect of the present invention provides an electronic device, including: a processor, and a memory storing a program, the program including instructions that, when executed by the processor, cause the processor to perform the aforementioned method.

[0017] In another aspect, the present invention provides a non-transitory machine-readable medium storing computer instructions for causing the computer to perform the method according to the foregoing.

[0018] In another aspect, the present invention provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the aforementioned method.

[0019] Compared with the prior art, the above-described technical solution of the present invention has the following advantages:

[0020] This invention provides a method for optimizing power grid data transmission in complex mountainous environments. By collecting multi-dimensional state vectors of environment, services, and energy consumption, and combining them with a deep learning time-series prediction model trained on historical channel quality, weather, and terrain data, the method accurately outputs the channel change trends of each terminal node in the mountainous power grid. This effectively solves the problems of insufficient prediction of the coupling effects of mountainous environments and low channel prediction accuracy in related technologies. Secondly, based on channel change trends and multi-dimensional state data, a reinforcement learning edge strategy optimization model, aimed at maximizing transmission success, avoiding timeout penalties, and reducing energy consumption, dynamically outputs suitable transmission links, transmit power ranges, and data transmission methods. This overcomes the limitations of fixed links and simple transmit power allocation, effectively adapting to communication blind spots and multipath effects caused by mountainous terrain, and significantly improving communication stability. Simultaneously, by inputting energy consumption state data into the edge strategy optimization model, combined with an optimization strategy oriented towards reducing energy consumption, the method avoids energy waste caused by continuous high-power connections and frequent probes, adapts to the power supply constraints of remote nodes, reduces data transmission energy consumption, and ultimately achieves reliable and efficient transmission of power grid data in complex mountainous environments. Attached Figure Description

[0021] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention, and those skilled in the art can obtain other embodiments based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart illustrating an embodiment of the power grid data transmission optimization method for complex mountainous environments, as described in this invention.

[0023] Figure 2 This is a dynamic closed-loop execution flowchart of a power grid data transmission optimization method for complex mountainous environments, as described in an embodiment of the present invention.

[0024] Figure 3 This is a schematic diagram of the structure of a power grid data transmission optimization system for complex mountainous environments, as described in an embodiment of the present invention.

[0025] Figure 4 This is a schematic diagram of the structure of the electronic device created by this invention. Detailed Implementation

[0026] Embodiments of the present invention will now be described in more detail with reference to the accompanying drawings. While some embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the present invention. It should be understood that the drawings and embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.

[0027] To address the problems of poor communication stability, insufficient accuracy in predicting channel changes, and high energy consumption in power grid data transmission in complex mountainous environments, this invention provides an optimization method, device, medium, and product for power grid data transmission in complex mountainous environments.

[0028] Among them, such as Figure 1 As shown, the power grid data transmission optimization method for complex mountainous environments provided by the embodiments of the present invention includes the following steps S1 to S4.

[0029] Step S1: Collect status data of each terminal node of the power grid in the mountainous area, and generate a multi-dimensional status vector based on the status data; wherein, the status data includes environmental status data, business status data, and energy consumption status data.

[0030] Step S2: Input the time series data corresponding to the multidimensional state vector into the prediction model, and the prediction model outputs the channel change trend data corresponding to each terminal node in the future time period; wherein, the prediction model is obtained by training a deep learning time series prediction model using historical channel quality data, historical weather data and terrain data.

[0031] Step S3: Input the multidimensional state vector and channel change trend data into the edge policy optimization model, and the edge policy optimization model outputs the transmission policy corresponding to each terminal node. The edge policy optimization model is obtained by reinforcement learning based on channel change trend data and historical state data, with the goal of maximizing the transmission success reward, avoiding timeout penalties and reducing energy consumption. The transmission policy includes the transmission link, transmit power range and data transmission method.

[0032] Step S4: Based on the transmission strategy, transmit the power grid data of each terminal node to the corresponding communication target.

[0033] Furthermore, mountain power grids refer to power network systems deployed in mountainous, hilly, and other terrain-complex areas, responsible for power generation, transmission, distribution, and consumer-side management in mountainous regions. Due to the unique environmental factors of mountainous areas, such as undulating terrain, variable weather conditions (e.g., strong winds, heavy rain, dense fog), and dense vegetation cover, their data transmission links are susceptible to channel attenuation and interference, necessitating differentiated data transmission strategies adapted to these complex environments.

[0034] Terminal nodes are end devices in mountain power grids that have data acquisition, transmission and status sensing functions. They mainly include: power distribution monitoring terminals, environmental sensing terminals, power consumption acquisition terminals, communication relay terminals and energy storage / microgrid terminals.

[0035] Environmental status data is a set of parameters that reflect the natural environmental characteristics around the terminal nodes of the power grid in mountainous areas and have a direct or indirect impact on the data transmission channel. It can include any one or more of weather data, terrain data, and channel quality data.

[0036] Service status data is a set of parameters characterizing the operational status of data transmission services at terminal nodes. It directly reflects the service quality of the current communication link and can include service status data of various terminal devices connected to the terminal node. Service status data may include: device online / offline status, CPU (Central Processing Unit) utilization, memory usage, data cache queue length, hardware temperature, power supply voltage stability, interface operating status, fault alarm status, etc., and can be obtained from the power grid dispatching system.

[0037] Energy consumption status data is a set of parameters characterizing the energy consumption of the terminal node during its own operation and data transmission. It can be collected by the transmit power sensor on the terminal node to collect real-time operating transmit power and transmission transmit power; or by the node's energy consumption monitoring module to record energy consumption per unit time, cumulative energy consumption, and remaining battery power; or by statistically analyzing energy consumption changes under different business loads based on the device operation log.

[0038] Multidimensional state vectors are generated by selecting core indicators from the collected state data that have a critical impact on the channel change trend and transmission strategy optimization of mountain power grids. Then, preprocessing operations such as cleaning, denoising, and normalization are performed on various indicators to eliminate differences in dimensions. Finally, the preprocessed indicator data are systematically integrated according to the dimensions of environment, service, and energy consumption to form a multidimensional state vector exclusive to each terminal node.

[0039] By integrating state data to obtain multi-dimensional state vectors, scattered and heterogeneous multi-source state data can be transformed into a unified structured vector form. This comprehensively represents the environmental characteristics of the terminal node, the quality of communication service operation, and the energy consumption level. It also adapts to the input requirements of deep learning time series prediction models and reinforcement learning edge strategy optimization models, significantly reducing the processing cost of unstructured data for the model. This provides accurate and efficient basic data support for subsequent channel change trend prediction and transmission strategy optimization.

[0040] The S1 step provided in the embodiment of the present invention collects three types of state data of mountain power grid terminal nodes, namely environment, service and energy consumption, and generates multi-dimensional state vectors. This provides comprehensive and accurate structured data support for subsequent channel change trend prediction and transmission strategy optimization, thereby alleviating the core problems of poor communication stability, insufficient prediction accuracy and high energy consumption of data transmission in mountain power grids from the source.

[0041] The time-series data corresponding to the multidimensional state vector is generated by arranging the multidimensional state vectors of terminal nodes at different acquisition times in chronological order according to timestamps.

[0042] Historical channel quality data refers to the set of quality indicators such as signal-to-noise ratio, packet loss rate, bit error rate, and transmission delay of the past communication links of terminal nodes. It is obtained in the same way as the channel quality data in the environmental status data. It can be obtained from the communication modules deployed on the terminal nodes or retrieved from the historical storage data of the edge gateway.

[0043] Historical weather data refers to the collection of meteorological indicators such as temperature, humidity, and wind speed around the terminal node in the past. It is obtained in the same way as weather data in environmental status data, and can come from sensor data collection and meteorological platform data.

[0044] Terrain data refers to the set of terrain feature parameters such as elevation and slope of the area where the terminal node is located. The acquisition method is the same as the terrain data in the environmental status data. It can come from the Beidou positioning system and the Digital Elevation Model (DEM).

[0045] A Digital Elevation Model (DEM) is a geospatial data model that records the elevation information of discrete points on the Earth's surface in the form of a numerical matrix. It is constructed by collecting accurate elevation values ​​of topographic feature points (such as mountain peaks, valleys, and slopes) and using interpolation algorithms, and can digitally simulate the undulating shape of the Earth's surface.

[0046] Channel change trend data is obtained through prediction models, which shows the changing patterns of indicators such as signal-to-noise ratio, packet loss rate, bit error rate, and transmission delay of terminal node channels in the future time period.

[0047] The prediction model is developed by preprocessing historical channel quality data, historical weather data, and terrain data and then inputting them into a deep learning time series model. The model is then iteratively trained using historical channel quality data as a supervision label, enabling it to capture the spatiotemporal correlation features between the data and obtain a prediction model that meets the requirements for predicting channel change trends.

[0048] Among them, deep learning temporal models include Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), Transformer temporal models, etc.

[0049] Step S2 provided in the embodiment of the present invention solves the problem of insufficient accuracy in channel change prediction in power grid data transmission in mountainous areas by inputting multidimensional state vector time series data into the trained prediction model and outputting channel change trend data, and provides accurate trend basis for subsequent transmission strategy optimization.

[0050] Historical status data refers to status data collected during a historical period, and its collection method is the same as that of status data.

[0051] The edge policy optimization model is deployed on an edge server, which acts as a reinforcement learning agent. When training the edge policy optimization model, the state space is first constructed using channel change trend data and historical state data, and the action space is constructed using link selection, transmit power range, and data transmission method. A reward function is designed that includes positive rewards for successful transmission, negative penalties for timeout transmission, and gradient rewards for energy consumption reduction.

[0052] The agent continuously interacts iteratively with the data transmission environment of the mountain power grid, outputs actions and obtains reward values ​​from the environment, and updates the model parameters in reverse until the comprehensive reward value corresponding to the actions output by the model converges to the optimal value, thus completing the training.

[0053] During training, the edge strategy optimization model has fully learned the mapping relationship between historical state data, channel change trend data, terminal node transmission requirements, and channel conditions. It can match the optimal action combination based on the real-time input node state and channel trend.

[0054] The transmission strategies output by the edge strategy optimization model include link selection (e.g., sky transmission link, ground transmission link, space transmission link), transmission power selection (e.g., adaptive low transmission power mode, stepped transmission power control mode, emergency high transmission power mode), and transmission method (e.g., direct transmission, compressed transmission, and local caching).

[0055] The embodiments of this invention optimize the transmission link output by the edge strategy model, which can match the optimal link type according to the channel change trend. For example, in valley areas with severe terrain obstruction, a sky link using UAV relay or tethered balloon relay is selected; in conventional distribution areas, a ground link using power line carrier or LTE 230M is selected; and in remote mountainous areas without ground / sky coverage, a space link using BeiDou short message or low-orbit satellite is selected. This effectively avoids interference factors such as mountainous terrain obstruction, signal attenuation, and link coverage blind spots, thereby solving the problem of poor communication stability of power grid data transmission in mountainous areas.

[0056] By determining the transmission power, the transmission power output can be dynamically adjusted based on channel quality. When the channel conditions are good, an adaptive low transmission power is used to reduce energy consumption. When the channel weakens, a stepped transmission power regulation is activated to ensure signal strength. In extreme scenarios, an emergency high transmission power is activated to maintain link connectivity, avoiding energy waste caused by blindly using high transmission power, thereby solving the problem of high energy consumption in data transmission.

[0057] By selecting the transmission method, it is possible to flexibly adapt to data priority and channel trends. High-priority data is transmitted directly to ensure timeliness, while large-volume non-urgent data is transmitted in compressed form to reduce bandwidth consumption. When the channel is poor, local buffering is used to wait for the channel to improve before transmission, effectively avoiding the risk of transmission timeout, thus solving the problem of transmission timeout caused by channel prediction deviation.

[0058] High-priority business data refers to business data generated by terminal nodes that has a strong immediacy and critical impact on the safe and stable operation of the power grid, emergency response to faults, and real-time dispatch decisions. This type of data needs to ensure timely response, i.e., low transmission latency requirements, and can include: power grid fault alarm data, equipment abnormal status data, real-time safety monitoring data, and emergency dispatch command data.

[0059] The embodiment of this invention provides step S3, which outputs a transmission strategy adapted to each terminal node through an edge strategy optimization model. This transforms the accurate channel trend prediction results output in step S2 into a feasible transmission decision, directly solving the core problems of poor data transmission stability, high energy consumption, and easy timeout in mountainous power grids. At the same time, the edge strategy optimization model is deployed on an edge server, enabling rapid local generation of transmission strategies, reducing cloud interaction latency, and adapting to scenarios where terminal nodes are scattered and communication link bandwidth is limited in mountainous power grids. Furthermore, the dynamic optimization characteristics of the strategy can be adjusted in real time according to changes in channel and node status, ensuring the reliability and economy of data transmission in complex environments.

[0060] A communication target refers to the receiving entity of data transmission at the terminal nodes of a mountain power grid. It possesses data reception, storage, parsing, forwarding, or relay functions and serves as a crucial data interaction carrier between the terminal nodes and the power grid management system. Communication targets can include: base stations, UAV relay platforms, BeiDou satellites, etc.

[0061] Based on the transmission strategy output by the edge strategy optimization model, the process of transmitting power grid data from terminal nodes to corresponding communication targets is as follows: when a ground link is selected, the data is transmitted to the nearest dedicated power base station or edge computing node; when a sky link is selected, the data is received and forwarded to the ground base station by a drone relay platform; when a space link is selected, the data is directly transmitted to the Beidou satellite and forwarded to the provincial power grid communication hub; and batches of non-urgent data within the distribution area are transmitted to the local server of the power grid distribution area.

[0062] The transmission power is dynamically adjusted to address the link transmission loss of different communication targets. For short-distance transmission, an adaptive low transmission power mode is used, while for long-distance transmission, a tiered transmission power control or an emergency high transmission power mode is enabled. The transmission method is selected based on data priority. High-priority data is transmitted directly to the base station or BeiDou satellite in real time. Large-volume non-urgent data is compressed to reduce the processing pressure on the communication target. When the channel is poor, the data is first buffered locally and transmitted after the channel improves. Finally, the communication target completes the data integrity verification and realizes the data flow to the power grid management system.

[0063] Step S4 provided in the embodiments of the present invention implements the optimized transmission strategy, relying on the coordinated coverage of communication targets such as base stations, UAV relay platforms, and Beidou satellites, to achieve accurate matching and transmission of terminal node data with corresponding communication targets. This not only fills the communication coverage blind spots in mountainous areas and solves the problem of poor data transmission stability, but also reduces energy consumption through differentiated transmission power control and transmission mode adaptation, while ensuring the timeliness of high-priority data and improving the reliability and economy of power grid data transmission in mountainous areas.

[0064] Therefore, the above-mentioned power grid data transmission optimization method for complex mountainous environments provided by the embodiments of the present invention, by collecting multi-dimensional state vectors of environment, service, and energy consumption, and combining them with a deep learning time series prediction model trained with historical channel quality, weather, and terrain data, accurately outputs the channel change trend of each terminal node of the power grid in mountainous areas, effectively solving the problems of insufficient prediction of the coupling impact of mountainous environment and low channel prediction accuracy in related technologies.

[0065] Secondly, based on channel change trends and multi-dimensional state data, a reinforcement learning edge strategy optimization model is used to maximize transmission success, avoid timeout penalties, and reduce energy consumption. This model dynamically outputs an adapted transmission link, transmit power range, and data transmission method, breaking through the limitations of fixed links and simple transmit power allocation. It effectively adapts to communication blind spots and multipath effects caused by mountainous terrain, significantly improving communication stability.

[0066] Meanwhile, by inputting energy consumption status data into the edge strategy optimization model and combining it with an optimization strategy aimed at reducing energy consumption, the energy waste caused by continuous high-power connections and frequent probes is avoided. This adapts to the power supply constraints of remote nodes, reduces data transmission energy consumption, and ultimately achieves reliable and efficient transmission of power grid data in complex mountainous environments.

[0067] Further, step S1 of this embodiment of the invention, which involves collecting status data of each terminal node in a mountainous power grid, includes the following steps: collecting environmental status data of each terminal node in the mountainous power grid; wherein, the environmental status data includes weather data, terrain data, and channel quality data; the terrain data includes digital elevation model data retrieved from an edge server and real-time location data; collecting service status data of each terminal node in the mountainous power grid; wherein, the service status data includes service status data of each terminal device connected to the terminal node, and the service priority of the service status data; collecting energy consumption data of each terminal node in the mountainous power grid; wherein, the energy consumption data is calculated based on the power data of each terminal node and the operating conditions of each terminal device.

[0068] Specifically, the edge server is pre-installed with a terrain database, which stores digital elevation model data, that is, the terrain elevation data in the digital elevation model.

[0069] Edge servers dynamically maintain terrain data by periodically downloading the latest terrain data from cloud servers and updating the terrain database. The advantage of this design is that it eliminates the need for edge servers to store the entire terrain data long-term, effectively reducing the storage pressure on edge nodes while ensuring the timeliness and accuracy of the terrain data. This provides high-precision terrain support for subsequent power transmission link planning and channel attenuation analysis.

[0070] Each terminal node is also equipped with a power grid status acquisition device, a multi-mode access gateway, and an energy consumption rate monitoring module.

[0071] The power grid status acquisition device is an intelligent monitoring terminal with multiple types of power data acquisition and priority marking functions. It can be a power data acquisition device conforming to the IEC 61850 standard. The power grid status acquisition device can collect business status data from various terminal devices connected to its terminal node, including: primary equipment business status data and secondary equipment business status data. The data types can be further subdivided into electrical quantity data, switching quantity data, non-electrical quantity data, fault alarm data, and fault waveform data.

[0072] After collecting service status data, the power grid status acquisition device prioritizes the data according to the IEC 61850 standard, classifying it into critical alarm data, periodic status data, and non-critical data. Critical alarm data (e.g., relay protection data, fault alarm data) is the highest priority data (QoS Class 0), with an extremely short transmission delay threshold of less than or equal to 20ms. Periodic status data (e.g., fault recording data, circuit breaker opening and closing operation feedback signals) is the second highest priority data (QoS Class 1), with a transmission delay threshold of less than or equal to 100ms. Non-critical data (e.g., equipment ledger information, historical operation statistics) is ordinary monitoring data (QoS Class 2), with a transmission delay threshold of less than or equal to 2s or longer. A transmission timeout is considered occurring when the transmission time of any of these service status data exceeds the corresponding transmission delay threshold.

[0073] A multimodal access gateway is an intelligent communication terminal integrating multiple communication standards and positioning functions. It can be equipped with industrial-grade gateway devices that support multiple protocols. The multimodal access gateway integrates a 4G / 5G communication module, a LoRa (Long Range Radio) communication module, and a BeiDou short message communication module. It can collect link channel quality parameters connected to each terminal node, including signal-to-noise ratio, packet loss rate, and transmission latency. The multimodal access gateway also has a pre-installed BeiDou positioning module, which can collect positioning data from each terminal node in real time, including latitude, longitude, and altitude information. The positioning data can be linked with digital elevation model (DEM) data to achieve accurate terrain location matching for terminal nodes.

[0074] The energy consumption rate monitoring module is a hardware unit for monitoring energy consumption at end nodes, comprising a power meter and a microcontroller. The power meter is preferably the INA219 power meter, capable of collecting power data from each end node. Based on the collected power data and the operating conditions of each end device (such as load rate, operating mode, start / stop status, etc.), the microcontroller calculates the energy consumption data of the end node using a power integration algorithm. It can also output an energy consumption rate curve, providing accurate data support for edge node energy consumption optimization and power supply strategy adjustment.

[0075] Environmental micro-meteorological sensing units are also installed within the deployment area of ​​each terminal node. These units are sensing devices capable of collecting multiple meteorological elements, and industrial-grade micro-meteorological monitoring terminals can be selected. Through these units, weather data from the deployment area of ​​each terminal node can be collected in real time. Furthermore, via a multi-modal access gateway, they can connect to external meteorological service interfaces to periodically obtain short-term (2-3 hour) weather forecast data corresponding to the geographical location of each terminal node.

[0076] This invention, through its embodiments, utilizes a terrain database and a BeiDou positioning module to separately schedule digital elevation model data and collect positioning data to obtain terrain data. This achieves lightweight storage and precise matching of terrain data, providing crucial terrain support for predicting channel attenuation in mountainous power grids and optimizing transmission line routes, thus reducing the impact of complex terrain on data transmission. Through an environmental micro-meteorological sensing unit, weather data from the deployment areas of each terminal node can be collected and fused with forecasts, enabling early prediction of the impact of extreme weather on power grid equipment and communication links, thereby improving the anti-interference capability of mountainous power grids. By collecting channel quality parameters of each link through a multi-modal access gateway and periodically acquiring external weather forecast data, a basis for dynamic selection of transmission links is provided, addressing the problem of insufficient coverage by a single communication standard in mountainous areas and improving the reliability of data transmission.

[0077] By collecting business status data from various terminal devices through the power grid status collector and marking business priorities, differentiated data transmission control is achieved, ensuring ultra-low latency transmission of critical alarm data, avoiding the risk of power grid fault escalation, and improving the transmission efficiency of non-critical data.

[0078] The energy consumption rate monitoring module can collect energy consumption data, enabling refined management of energy consumption at terminal nodes, providing data support for the formulation of energy-saving strategies for edge nodes, and reducing the operation and maintenance costs of power grids in mountainous areas.

[0079] Therefore, the state data acquisition process of the present invention realizes comprehensive acquisition and collaborative management of multi-dimensional data such as terrain, equipment, channel, energy consumption, and environment, providing complete and accurate input data for subsequent edge strategy optimization models, and ensuring the real-time performance, reliability, and economy of power grid data transmission in mountainous areas.

[0080] Furthermore, in step S1 of the present invention, generating a multidimensional state vector based on state data includes the following steps: calculating the terrain occlusion degree between each terminal node and each communication target based on digital elevation model data and positioning data; and generating a multidimensional state vector based on state data and terrain occlusion degree.

[0081] Specifically, terrain occlusion refers to the degree to which the terrain between the terminal node and the communication target obstructs the wireless communication link, quantitatively representing the impact of terrain undulation on signal transmission. The calculation first obtains the latitude, longitude, and altitude coordinates of the terminal node and the communication target based on positioning data. Then, it uses digital elevation model (DEM) data to extract the terrain elevation profile corresponding to the line connecting the two points. By determining whether the highest point in the profile is higher than the skyline of the line connecting the two points, and combining the elevation difference with the distance ratio, a quantitative value of the occlusion degree is calculated (e.g., a range of 0-1, where 0 represents no occlusion and 1 represents complete occlusion).

[0082] The preferred method for generating a multidimensional state vector is to first standardize the state data (equipment service status, channel quality, energy consumption, meteorological data, etc.) and the quantified values ​​of terrain occlusion, then extract the core features of each data (such as key alarm data identifiers, average signal-to-noise ratio, energy consumption rate, occlusion level, etc.), and integrate them into a vector containing multidimensional information such as terrain, equipment, channel, and environment in a preset dimensional order, thus forming a multidimensional state vector that represents the communication and operating conditions of the terminal node.

[0083] The steps provided in the embodiments of this invention integrate terrain occlusion into the multi-dimensional state vector, which makes up for the shortcomings of traditional state vectors that do not consider the influence of terrain. This allows the vector to more comprehensively reflect the actual operation and communication environment of terminal nodes under complex mountainous terrain, providing more accurate input data for subsequent edge strategy optimization models. This helps the model formulate transmission strategies adapted to terrain conditions and improves the pertinence and reliability of power grid data transmission in mountainous areas.

[0084] Furthermore, prior to step S2 in this embodiment of the invention, the method further includes the following steps: collecting historical channel quality data, historical weather data, and terrain data, and integrating them into time series data of a unified dimension as a training dataset; wherein, the historical channel quality data includes any one or more of the channel's signal-to-noise ratio, packet loss rate, and latency data; using the training dataset, training a deep learning time series prediction model based on a self-attention mechanism; when the performance of the deep learning time series prediction model converges, the training is completed, and the final prediction model is obtained.

[0085] Specifically, the deep learning time series prediction model based on the self-attention mechanism is an intelligent prediction model that can capture the long-distance dependencies and feature associations of multi-dimensional time series data. Its core self-attention mechanism can accurately focus on key data segments that have a significant impact on channel quality by calculating the attention weights of features at different times and in different dimensions in the data sequence (such as periods when the signal-to-noise ratio drops sharply during rainstorms or periods when the packet loss rate is abnormal in areas with sudden changes in terrain occlusion).

[0086] The preferred deep learning time series prediction model based on the self-attention mechanism is the spatiotemporal Transformer model. This model adds a terrain spatial feature encoding branch to the Transformer encoder structure, which can simultaneously analyze the weather-channel evolution law in the time dimension and the influence mechanism of terrain-channel in the spatial dimension, and adapt to the scenario characteristics of mountain power grid channel status being constrained by both weather and terrain.

[0087] During model training, the collected historical channel quality data, historical weather data, and terrain data are first preprocessed, including data normalization to eliminate dimensional differences, missing value interpolation to ensure data integrity, and timestamp alignment to construct a time-series dataset with unified dimensions. Subsequently, the preprocessed dataset is divided into training, validation, and test sets according to a preset ratio. The training set data drives the spatiotemporal Transformer model to learn the feature correlation patterns. The channel quality parameters (signal-to-noise ratio, packet loss rate, and latency) for future time periods are used as the prediction targets. The mean squared error (MSE) is used as the loss function, and the model parameters are iteratively optimized through backpropagation. At the same time, the model's generalization ability is monitored in real time based on the validation set. When the validation set loss value tends to stabilize for several consecutive rounds without significant decrease, the model performance is considered to have converged, the training is completed, and the final prediction model is obtained.

[0088] The spatiotemporal Transformer prediction model trained by the above steps provided in the embodiments of this invention can deeply explore the potential coupling patterns between historical channel quality and weather and terrain, accurately predict the channel change trend of mountain power grid in future periods, and provide a forward-looking channel state input basis for subsequent edge strategy optimization models. This allows the output transmission strategy to adapt to the dynamic fluctuations of the channel in advance, effectively avoid transmission timeouts and data packet loss caused by sudden deterioration of the channel in mountainous areas, and significantly improve the reliability and adaptability of data transmission in mountainous power grids.

[0089] Furthermore, prior to step S3 in this embodiment of the invention, the method preferably includes the following steps: collecting historical state data and corresponding historical decision data; training an edge policy model deployed on an edge server in a simulation environment constructed based on the historical state data, combining the historical decision data with the goal of maximizing the reward function; wherein the reward function is constructed with successful data transmission as the reward and transmission timeout and energy consumption as the penalty; when the performance of the edge policy model converges, the training is completed, and a trained edge policy model is obtained; the trained edge policy model is continuously interacted with the simulation environment to optimize the network parameters of the edge policy model, and when the comprehensive reward value corresponding to the reward function is stable and meets the preset threshold, the final edge policy optimization model is obtained.

[0090] Historical state data is input into the state space of the edge strategy optimization model. This historical state data may include: historical real-time channel quality (signal-to-noise ratio, packet loss rate, and latency) for each available link, historical terminal node energy consumption, future channel quality (signal-to-noise ratio, packet loss rate, and latency) output by the Transformer, terrain occlusion, future short-term weather forecast data (temperature, precipitation probability, and wind speed), service priority of the data to be transmitted, and the current time.

[0091] Historical decision data is input into the action space of the edge policy optimization model. Historical decision data may include: link selection (selecting a single link or redundant dual links from 4G / 5G, LoRa, and BeiDou satellites), power control (adjusting the transmit power within a dynamic range), and data transmission method (immediate transmission, compressed transmission, and local caching).

[0092] The reward function is the core evaluation metric in reinforcement learning frameworks that guides marginal policy models to learn optimal decisions. The reward function is obtained by weighting and combining reward and penalty terms: the reward term is set as a positive value positively correlated with the data transmission success rate; that is, when data transmission is successful and the latency threshold is met, the model receives a positive reward corresponding to its weight. The penalty term consists of two parts: a negative value positively correlated with the number of transmission timeouts and a negative value positively correlated with energy consumption; that is, when transmission timeouts or energy consumption exceed the limit occur, the model will have its corresponding reward score deducted.

[0093] The edge policy optimization model in this invention is trained using the Proximal Policy Optimization (PPO) algorithm. Based on the advantage function calculated from the reward function, a truncated agent objective function is constructed by combining the probability ratio of the old and new policies, and this objective function serves as the optimization target for the policy network. This objective function limits the policy update magnitude through a pruning mechanism, ensuring that the model iterates towards higher transmission success rates, lower timeout rates, and lower energy consumption, while avoiding training instability caused by policy mutations.

[0094] The core process of building a simulation environment based on historical state data is as follows: First, feature extraction and scenario classification are carried out on the historical data to divide typical operating conditions such as terrain complexity, meteorological conditions, and business load; then, virtual terminal nodes and communication links are built on the power grid simulation platform to map the feature parameters of the historical data into the operating parameters of the simulation environment; finally, channel fluctuation patterns and equipment energy consumption models are implanted to reproduce the real operating state of the mountain power grid and provide a high-fidelity verification scenario for edge strategy model training.

[0095] In the constructed simulation environment, historical state data is input into the edge policy model. After the model outputs transmission policy decisions, the reward value is calculated and the advantage function is derived by combining the execution effect feedback from the simulation environment (whether the transmission is successful, whether there is a timeout, and whether the energy consumption exceeds the limit). Then, based on the truncated agent objective function of the PPO algorithm, the model parameters are iteratively updated through backpropagation. The pruning mechanism avoids policy mutations and continuously optimizes the decision logic of transmission link, transmission power, and transmission mode until the decision effect of the model in various typical scenarios is stable, the performance is judged to have converged, and the basic training is completed.

[0096] When the trained edge policy model interacts continuously with the simulation environment, the network parameters are optimized by utilizing the transmission results and reward values ​​fed back by the environment in real time, thereby enhancing the adaptive decision-making ability for unknown scenarios. During the process, the fluctuation of the comprehensive reward value and core indicators (transmission success rate, latency compliance rate, and energy consumption control rate) are continuously monitored. When the comprehensive reward value fluctuates below the preset threshold for several consecutive rounds and the core indicators meet the requirements, the model is judged to have reached the optimal performance, and the final edge policy optimization model is obtained.

[0097] The steps provided in the embodiments of this invention construct a high-fidelity simulation environment based on real historical data and guide model training and iterative optimization using a reward function with differentiated weights. This enables the final edge strategy optimization model to deeply adapt to the complex and variable terrain and weather conditions of mountain power grids, learn the optimal decision logic that balances transmission reliability, low latency, and low energy consumption, effectively avoid transmission timeouts or energy waste caused by strategy formulation deviating from actual operating conditions, and provide high-performance model support for subsequent output of optimal transmission strategies based on real-time status data.

[0098] Furthermore, in the embodiments of this invention, in a simulation environment constructed based on historical state data, and in conjunction with historical decision data, the edge policy model deployed on the edge server is trained with the goal of maximizing the reward function. Preferably, this includes the following steps: constructing a reward function with successful data transmission as the basic reward, successful high-priority data transmission as an additional reward, and transmission timeout and energy consumption as penalties; wherein the transmission latency of high-priority data is lower than a set threshold; constructing a simulation environment on the edge server based on historical state data; and initializing the policy network and value network in the simulation environment, and training the edge policy model deployed on the edge server with the goal of maximizing the reward function.

[0099] High-priority service data may include only the highest-priority service data marked above, or it may include both the highest-priority service data and the second-highest-priority service data. The threshold here can be set according to the transmission delay threshold corresponding to the highest-priority service data and the second-highest-priority service data.

[0100] The preferred expression for the reward function is:

[0101] .

[0102] in, For the reward function, For successful transmission of instructions, This is a timeout indication. This is an estimated energy consumption value. An additional reward for the successful transmission of high-priority data. There are four adjustable weighting coefficients.

[0103] In this embodiment of the invention, a reward for successful high-priority data transmission is added to the reward item of the reward function. This can guide the edge strategy model to prioritize the low-latency transmission needs of high-priority data during the decision-making process, deeply binding the model's decision-making logic with the priority marking rules of the IEC 61850 standard. This significantly improves the transmission success rate and latency compliance rate of high-priority data, avoids the core risk of critical data transmission timeout from the reward mechanism level, and balances the overall data transmission efficiency and energy consumption control objectives.

[0104] In a high-fidelity simulation environment built based on historical state data, the parameters of the policy network and value network of the edge policy model are first initialized. The policy network is responsible for outputting the optimal combination of transmission policies for terminal nodes (transmission link selection, transmit power range configuration, and data transmission method determination), while the value network is responsible for evaluating the expected cumulative reward corresponding to the policy in the current state. Subsequently, the preprocessed historical state data is input into the model, and the policy network, combined with the historical decision data, outputs an initial transmission policy. The simulation environment executes this policy and provides real-time feedback on the decision results (including data transmission success indication). Timeout Indicator Energy consumption estimates High-priority data successful transmission indication ).

[0105] The reward function value is calculated based on the feedback decision results, and the advantage function is calculated using the state value output by the value network. Based on this, the policy network (aiming to maximize cumulative reward) and the value network (aiming to reduce value prediction error) are updated respectively, and the adjustable weight coefficients are dynamically adjusted. Optimize the impact weights of different reward / penalty items (e.g., increase) (To enhance the reward ratio of high-priority data transmission), iteratively train the model until the decision-making effect of the model in various typical scenarios tends to be stable and the performance converges.

[0106] The steps provided in the embodiments of this invention construct an accurate evaluation function with differentiated rewards for high-priority data and adopt a dual-network architecture of policy network and value network to conduct model training in a high-fidelity simulation environment. This enables the trained edge policy model to balance the protection of high-priority data transmission with the optimization of overall transmission performance. It not only meets the low-latency transmission requirements of key alarm and fault recording data in mountain power grids, but also effectively controls the energy consumption of terminal nodes, significantly improving the pertinence and adaptability of model decision-making. This provides high-performance and highly reliable model support for the rapid output of the optimal transmission strategy in real-time conditions.

[0107] Furthermore, step S4 of the present invention preferably includes the following steps: adjusting the power, corresponding communication resources, and data transmission methods of each terminal node based on the transmission strategy; wherein, the communication resources include: sky communication resources, ground communication resources, and space communication resources; ground communication resources have the highest scheduling priority, followed by sky communication resources, and lastly space communication resources; the data transmission methods include: immediate transmission, compressed transmission, and local caching; based on the transmission link and data transmission method corresponding to the power and communication resources, transmitting the power grid data of each terminal node to the corresponding communication target; wherein, the communication targets include: base stations, UAV relay platforms, and BeiDou satellites.

[0108] Specifically, terrestrial communication resources, including 4G / 5G and LoRa networks, are the lowest cost and can be prioritized. Airborne communication resources, including high-altitude UAV relay platforms, utilize OLSRv2 (Optimized Link State Routing Protocol Version 2) to build a shortest-path mesh network as a supplement to regional coverage and can be activated on demand. Space communication resources, based on BeiDou satellite short messages, are the most expensive and can serve as a final communication guarantee.

[0109] Immediate transmission refers to the method of transmitting data collected by the terminal node directly to the target node through the selected communication link without additional processing such as compression or buffering. This corresponds to the transmission requirements of high-priority data, such as critical alarm data of QoS Class 0 (relay protection action commands, fault trip signals) and second-highest priority data of QoS Class 1 (fault waveform data).

[0110] Sending after compression refers to the method of first performing lossless or light lossy compression on the collected data to reduce the data volume before transmission. This method is suitable for medium-priority transmission needs with large data volumes, such as waveform data of fault recordings and batch periodic monitoring data.

[0111] Local caching refers to temporarily storing collected data in the local storage module of the terminal node, and then uploading it in batches when the communication link quality is good and the network load is low. This is suitable for the transmission needs of low-priority data, such as non-critical data in QoS Class 2 (equipment ledger, historical operation statistics).

[0112] In the transmission strategy output by the edge strategy optimization model, if space communication resources are selected, the corresponding transmission link is the BeiDou short message communication link. The power range needs to match the low-power transmission threshold of the BeiDou terminal (usually the minimum transmission power to meet satellite signal reception, avoiding excessive energy consumption). The corresponding communication target is the BeiDou satellite. The terminal node sends data to the BeiDou satellite through the BeiDou short message module, and the satellite relays the data to the power grid dispatch center or edge server to complete data transmission in extreme scenarios.

[0113] In the transmission strategy output by the edge strategy optimization model, if terrestrial communication resources are selected, the corresponding transmission link is either a 4G / 5G link or a LoRa link. The power range needs to be dynamically adjusted according to the link type and real-time channel quality. For 4G / 5G links, the transmission power is adapted based on the base station signal strength (reducing power to save energy when the signal is strong and appropriately increasing power to ensure link stability when the signal is weak). For LoRa links, based on their low power consumption and wide coverage characteristics, a low-power stable output range is set to match the needs of long-distance, low-data-volume transmission in mountainous areas. The corresponding communication target is a terrestrial base station, suitable for conventional scenarios such as mountainous towns with good base station signal coverage and along power transmission lines. Relying on the advantages of low cost and high bandwidth of terrestrial communication resources, it meets the daily data transmission needs of the vast majority of terminal nodes.

[0114] In the transmission strategy output by the edge strategy optimization model, if sky communication resources are selected, the corresponding transmission link is a drone relay mesh network link built based on the OLSRv2 routing protocol. The model calculates the shortest communication path between the terminal node and the drone, and between drones, using the OLSRv2 protocol to ensure link transmission latency and stability. The power range needs to be adapted to the communication distance between the terminal and the drone relay platform, and is set to a medium power output range to ensure signal penetration while avoiding excessive energy consumption. The corresponding communication target is the drone relay platform, which is suitable for scenarios such as blind spots in ground communication resource coverage and ground link interruptions caused by sudden natural disasters. As a supplementary means of regional communication coverage, the data transmission path in the area can be quickly restored by activating the drone relay as needed.

[0115] The above steps of the embodiments of the present invention, by constructing a three-level hierarchical communication resource scheduling system of ground, sky, and space and configuring priorities, combined with differentiated data transmission methods such as immediate transmission, compressed transmission, and local caching, and synchronously and dynamically adjusting the transmission power of terminal nodes, achieve precise adaptation of communication resources, transmission methods and power grid data priorities. It not only relies on ground resources to ensure low-cost and efficient transmission in conventional scenarios, but also uses sky relays to supplement blind spot coverage and space satellites to build a solid emergency communication baseline, significantly improving the reliability, timeliness and economy of power grid data transmission in mountainous areas.

[0116] Furthermore, the method of transmitting power grid data from each terminal node to the corresponding communication target based on the transmission link and data transmission method corresponding to power and communication resources in the embodiments of the present invention includes the following steps: processing the power grid data of each terminal node based on the data transmission method to obtain transmission strategy matching data; wherein, the data processing method includes any one or more of lossless compression processing, multi-source same-source data aggregation processing, and caching processing; encrypting the transmission strategy matching data to obtain corresponding encrypted data; and transmitting the encrypted data to the corresponding communication target based on the transmission link and data transmission method corresponding to power and communication resources.

[0117] Lossless compression processing can be achieved using lossless compression algorithms such as LZ77 (Lempel-Ziv 1977), DEFLATE, or LZMA (Lempel-Ziv-Markov chain-Algorithm). These algorithms can reduce data volume by eliminating internal redundancy (such as duplicate data segments and spatial correlations) without losing any original data information. They are suitable for critical power grid data with extremely high requirements for data integrity (such as relay protection commands and fault waveforms).

[0118] Multi-source homogeneous data aggregation processing refers to a processing method that integrates, deduplicates, and extracts features from power grid data of the same type collected from multiple terminal nodes. It is suitable for batch periodic monitoring data (such as temperature and voltage monitoring data from multiple nodes). The specific implementation process is as follows: first, the homogeneous data collected from multiple terminals are classified and collected by type; then, invalid and duplicate data are removed; finally, statistical feature values ​​or key trend node data are extracted through aggregation calculation to reduce transmission redundancy.

[0119] Caching refers to the process of temporarily storing data to be transmitted in the local storage module of the terminal node, and then uploading it in batches after the preset trigger conditions are met. It is suitable for non-critical data of QoS Class 2.

[0120] Encryption refers to the process of converting plaintext into ciphertext through cryptographic algorithms and key management mechanisms, ensuring data security throughout the entire process. Specifically, encryption can employ algorithms such as SM2 (SM2 Elliptic Curve Public Key Cryptography Algorithm) or SM4 (SM4 Block Cipher Algorithm). Using the SM2 algorithm enables two-way authentication, digital signatures, and key negotiation; using the SM4 algorithm provides end-to-end authentication and encryption for business data, thereby ensuring data security.

[0121] The steps provided in the embodiments of this invention, through a differentiated data preprocessing strategy of lossless compression, multi-source data aggregation, and caching, effectively reduce the data transmission volume, lower link bandwidth usage and terminal node energy consumption, while ensuring the integrity of critical data and the transmission efficiency of non-critical data; combined with encryption processing, it achieves identity authentication, anti-tampering, and anti-theft of data transmission, solves the data security risks in the complex communication environment of mountain power grids, and significantly improves the efficiency, security, and reliability of power grid data transmission.

[0122] Furthermore, in the embodiments of the present invention, transmitting encrypted data to the corresponding communication target preferably includes the following steps: when the terminal node is in a signal blind zone, uploading the encrypted data to the terminal node's cache module; obtaining the cached data corresponding to the encrypted data from the cache module through a mobile relay station; wherein the mobile relay station and the terminal node communicate via a wireless network or Bluetooth; and transmitting the cached data to the corresponding communication target.

[0123] When a terminal node is in a signal blind zone of a ground base station and a signal blind zone where the UAV relay platform has not completed its aerial coverage, the encrypted power grid data is first stored in the terminal node's local cache module according to priority and marked with a cache timestamp. After the mobile relay station (such as the relay equipment carried by the inspection vehicle or the portable relay terminal) enters the communication coverage area of ​​the terminal node according to the preset inspection route, the relay station and the terminal node automatically establish a wireless network or Bluetooth communication connection. The terminal node retrieves the corresponding encrypted data from the cache module and sends it to the mobile relay station. After receiving the data, the mobile relay station first verifies the data integrity and encryption signature. After confirming that there are no errors, it forwards the cached data to the corresponding communication target (base station, edge server or power grid dispatch center).

[0124] The steps provided in the embodiments of this invention effectively solve the problem of encrypted data retention at terminal nodes in mountainous power grid signal blind spots by combining local caching with the cache-relay transmission mode of mobile relay stations. The short-range communication characteristics of wireless networks or Bluetooth ensure the stability of the data relay process, while retaining the encrypted data state until it is transmitted to the target node, avoiding security risks in the relay process, and further improving the full coverage and security of data transmission in mountainous power grids.

[0125] like Figure 2 As shown in the embodiment of the present invention, the power grid data transmission optimization method for complex mountainous environments first collects multi-dimensional status data of each terminal node of the power grid in the mountainous area. This data is then input into an intelligent decision-making module composed of a prediction model and an edge strategy optimization model. The prediction model outputs the channel change trend, assisting the edge strategy optimization model in generating a transmission strategy that includes communication resource scheduling, transmission power adjustment, and data transmission mode selection. Subsequently, the strategy is executed to complete data transmission through a specified link, and the transmission results (success rate, timeout rate, energy consumption, etc.) are fed back to the edge strategy optimization model for online learning and optimization, iteratively updating its network parameters. Finally, it is determined whether the data transmission is complete. If not, the process returns to the data acquisition stage to regenerate the strategy until the transmission is successful, at which point the process ends. The entire process forms a self-optimizing closed loop of "perception-decision-execution-feedback," continuously improving the reliability and adaptability of power grid data transmission in complex mountainous scenarios.

[0126] like Figure 3 As shown, this invention also provides a power grid data transmission optimization method for complex mountainous environments, applicable to the power grid data transmission optimization method for complex mountainous environments provided in the above embodiments. The system includes: a perception layer, a decision layer, a resource layer, an execution layer, a data security layer, and a transmission protection layer.

[0127] Specifically, the perception layer is the "data input source" of the entire system, responsible for collecting and preprocessing multi-dimensional state data of terminal nodes, including: multimodal intelligent terminals, data collection devices, and DEM terrain database.

[0128] Among them, the multimodal intelligent terminal corresponds to the multimodal access gateway, power grid status collector, environmental micro-meteorological sensor unit, and energy consumption rate monitoring module in the above embodiments. It integrates multiple communication standards and acquisition functions to collect data such as environmental status (weather, terrain, channel quality), service status (equipment operation, data priority), and energy consumption status (power, energy consumption rate).

[0129] The data collection device, corresponding to the various data acquisition and preprocessing processes in the above embodiments, is responsible for integrating scattered multi-source data into a unified multi-dimensional state vector, providing structured input for subsequent models.

[0130] The DEM terrain database, corresponding to the terrain database pre-installed on the edge server in the above embodiments, stores digital elevation model data and is used to calculate the terrain occlusion between terminal nodes and communication targets, providing high-precision terrain support for channel prediction and strategy optimization.

[0131] The decision-making layer is the "central brain" of the system, responsible for generating the optimal transmission strategy adapted to the mountainous environment, including: the attention channel prediction module and the PPO near-end strategy optimization and reinforcement model.

[0132] The attention channel prediction module, corresponding to the spatiotemporal Transformer prediction model in the aforementioned embodiments, uses historical channel quality, weather, and terrain data as training inputs. It captures spatiotemporal correlation features through a self-attention mechanism and outputs channel change trend data (signal-to-noise ratio, packet loss rate, and other change patterns) for future periods, addressing the problem of insufficient channel prediction accuracy in mountainous areas. The PPO near-end policy optimization and reinforcement model, corresponding to the edge policy optimization model in the above embodiments, is deployed on an edge server and includes a state space, action space, and reward function.

[0133] The resource layer is the system's "communication resource scheduling center," responsible for dynamically allocating communication resources at the air, ground, and space levels. The resource layer includes air, ground, and space communication resources and a resource scheduler.

[0134] Among them, the ground-air-space communication resources correspond to the three-level communication resource system in the above embodiments, namely, ground communication resources, sky communication resources, and space communication resources. The resource scheduler corresponds to the communication resource scheduling logic in the above embodiments. Based on the output of the edge strategy optimization model, it dynamically matches communication resources for different scenarios (such as using ground for normal scenarios, sky for blind spots, and space for extreme scenarios) to achieve efficient utilization of resources.

[0135] The execution layer is the system's "strategy executor," responsible for implementing the transmission strategies generated by the decision layer. The execution layer includes: a protocol conversion and fast switching unit and a data adaptive processing unit.

[0136] The protocol conversion and fast switching unit corresponds to the multi-protocol compatibility and link switching capability of the multi-modal access gateway in the above embodiments. It supports protocol conversion of different communication links (such as 4G / 5G, LoRa, BeiDou) and fast link switching (such as switching to sky / space link when the ground link is interrupted) to ensure communication continuity.

[0137] The adaptive data processing unit corresponds to the data preprocessing and caching mechanisms in the above embodiments, including lossless compression, multi-source data aggregation, and caching processing. It adapts the data according to the transmission method to reduce transmission volume and energy consumption. It also supports the "caching-transfer" mode to solve the problem of data retention in signal blind spots.

[0138] The data security layer is a security module that runs through the entire process. Corresponding to the encryption processing mechanism mentioned above, it is responsible for encrypting the data matched by the transmission strategy. It uses algorithms such as SM2 / SM4 to achieve identity authentication, anti-tampering and anti-theft, and solves the data security risks in the complex communication environment of mountainous areas.

[0139] The transmission assurance layer is the "stability guarantee" for the transmission process, corresponding to the link and power regulation mechanism mentioned in the text. Relying on the coordinated coverage of air, ground, and space communication resources, the transmission power is dynamically adjusted (e.g., low power saving when the channel is good, step-by-step power increase when the channel is attenuated, and emergency high power in extreme scenarios). Through the OLSRv2 routing protocol of UAV relay and the emergency guarantee of Beidou short messages, it fills the communication coverage blind spots in mountainous areas and solves the problem of poor data transmission stability.

[0140] The system provided in the embodiments of this invention is applied to the above methods. By collecting data at the perception layer, generating strategies at the decision layer, scheduling resources at the resource layer, transmitting data at the execution layer, and providing full protection at the data security layer and transmission guarantee layer, it achieves a self-optimizing closed loop and continuously improves the reliability, timeliness, and economy of power grid data transmission in complex mountainous scenarios.

[0141] The following describes the method and system provided by the embodiments of the present invention through specific scenario simulations.

[0142] When a monitoring node detects an abnormal signal of a sudden increase in transmission line current at a certain moment, lasting for 8ms, and fault characteristic analysis confirms a flashover fault in the insulator (critical data), the system responds immediately, completing communication link status determination and service priority adjustment within 150ms of the fault occurrence. After determining the fault information, the system will undergo the following steps:

[0143] First, the system upgrades the service priority from QoS Class 2 to QoS Class 0, with data transmission latency within 20ms and reliability requirements exceeding 99.99%. Meanwhile, the multimodal access gateway in the perception layer detects that while the 4G link is initially usable, its signal-to-noise ratio (SNR) has dropped to 5dB, the real-time packet loss rate has reached 25%, and latency fluctuations are severe, indicating rapid deterioration. The LoRa link has an SNR of 4dB, a stable packet loss rate of 5%, and latency around 280ms, demonstrating good stability. The BeiDou satellite link has an SNR of 10dB, positioning accuracy of less than 5m, and multiple satellites can provide services simultaneously, indicating a good link status.

[0144] II. Based on the Transformer architecture, the prediction model accurately predicts the trajectory of the 4G link being completely interrupted within the next 3 minutes by analyzing historical data from the past 60 minutes: after 30 seconds, the signal-to-noise ratio will drop to 3.2dB and the packet loss rate will rise to 45%; after 1 minute, the signal-to-noise ratio will drop to 1.5dB and the packet loss rate will reach 68%; after 3 minutes, the signal-to-noise ratio will drop to -2.1dB and the packet loss rate will exceed 90%, basically losing communication capability.

[0145] III. The edge strategy optimization model, upon receiving predictive information and combining it with current data, completes a decision within 80 minutes. The decision-making process considers multiple factors, including the rapid deterioration of 4G signals, the highest priority of faulty data, and the urgency of the situation. The strategy network output shows that the probability of selecting BeiDou satellite as the primary transmission link is 0.92, the probability of using LoRa as a backup link is 0.65, while the probability of continuing to use the 4G link is only 0.01. Regarding power control, the probability of selecting the maximum transmit power +20dBm is as high as 0.95. In terms of data processing strategy, the probability of immediate transmission without compression reaches 0.98.

[0146] IV. The final system adopts a dual-link structure with BeiDou satellite primary transmission and LoRa redundancy backup to immediately transmit uncompressed fault data at maximum power. BeiDou satellite transmission completes the transmission of fault data within 760 ms and receives confirmation from the ground station 40 ms later, with a total energy consumption of 80 mJ. LoRa redundant transmission takes 1500 ms, with a total energy consumption of 150 mJ, forming a data backup. This dual-link structure achieves highly reliable data transmission.

[0147] V. The total energy consumption of this emergency transmission was 230mJ, and the reward function, through multi-dimensional evaluation, gave a high score of +1.48. Specifically, successful transmission earned a base reward of +1.0, the absence of timeout avoided a penalty of -0.8, the energy consumption item generated a negative evaluation of -0.12, and successfully transmitting the highest priority data received an additional reward of +0.6. This result effectively enhanced the system's "reliability over energy consumption" strategy in emergency situations, and the relevant state-action-reward experience data was stored in the training database for subsequent model optimization.

[0148] Therefore, the method and system provided by the embodiments of the present invention can achieve the following technical effects:

[0149] 1. The edge strategy optimization model does not require a pre-set precise mountain channel model. It can learn the optimal strategy autonomously through real-time interaction with the environment, effectively cope with the complex and ever-changing dynamic environment in mountainous areas, and solve the limitations of traditional optimization methods in mountainous applications.

[0150] 2. By designing a reward function, the multi-objective optimization problem, including reliability, latency, and energy consumption, is internalized into the learning objectives of the intelligent agent. It can autonomously find the best balance point among competing objectives, avoiding the result bias caused by manually setting weights.

[0151] 3. By taking power control and data transmission strategies as core decision-making actions, the system can proactively determine the severity of the environment and the value of the data, and intelligently select energy-saving strategies, rather than blindly trying high-power transmission and increasing equipment energy consumption.

[0152] Embodiments of the present invention also provide a non-transitory machine-readable medium storing a computer program, wherein the computer program, when executed by a computer's processor, is used to cause the computer to perform a method according to an embodiment of the present invention.

[0153] Embodiments of the present invention also provide a computer program product, including a computer program, wherein the computer program, when executed by a computer's processor, is used to cause the computer to perform the method of an embodiment of the present invention.

[0154] An embodiment of the present invention also provides an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor. The memory stores a computer program executable by the at least one processor, which, when executed by the at least one processor, causes the electronic device to perform the method of the embodiment of the present invention.

[0155] refer to Figure 4 The present invention will now describe a structural block diagram of an electronic device that can serve as an embodiment of the present invention, serving as an example of a hardware device applicable to various aspects of the present invention. The electronic device is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0156] like Figure 4As shown, the electronic device includes a computing unit 501, which can perform various appropriate actions and processes based on a computer program stored in ROM (Read-Only Memory) 502 or loaded from storage unit 508 into RAM (Random Access Memory) 503. RAM 503 can also store various programs and data required for the operation of the electronic device. The computing unit 501, ROM 502, and RAM 503 are interconnected via bus 504. An I / O interface (Input / Output Interface) 505 is also connected to bus 504.

[0157] Multiple components in the electronic device are connected to I / O interface 505, including: input unit 506, output unit 507, storage unit 508, and communication unit 509. Input unit 506 can be any type of device capable of inputting information into the electronic device. Input unit 506 can receive input digital or character information and generate key signal inputs related to user settings and / or function control of the electronic device. Output unit 507 can be any type of device capable of presenting information and may include, but is not limited to, a display, speaker, video / audio output terminal, vibrator, and / or printer. Storage unit 508 may include, but is not limited to, disks and optical discs. Communication unit 509 allows the electronic device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, and / or wireless communication transceivers, such as Bluetooth devices, WiFi devices, WiMax devices, cellular communication devices, and / or the like.

[0158] The computing unit 501 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a CPU (Central Processing Unit), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing units, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 501 performs the various methods and processes described above. For example, in some embodiments, the method embodiments of the present invention can be implemented as a computer program tangibly contained in a machine-readable medium, such as storage unit 508. In some embodiments, part or all of the computer program can be loaded and / or installed on an electronic device via ROM 502 and / or communication unit 509. In some embodiments, the computing unit 501 can be configured to perform the methods described above by any other suitable means (e.g., by means of firmware).

[0159] Computer programs for implementing the methods of embodiments of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0160] In the context of embodiments of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable signal medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, or infrared systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0161] It should be noted that the term "comprising" and its variations used in the embodiments of this invention are open-ended, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The modifications of "one" and "a plurality" mentioned in the embodiments of this invention are illustrative and not restrictive, and those skilled in the art should understand that unless explicitly indicated otherwise in the context, they should be understood as "one or more".

[0162] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the embodiments of this invention are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0163] The steps described in the method embodiments provided by the present invention can be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of protection of the present invention is not limited in this respect.

[0164] The term "embodiment" in this specification refers to a specific feature, structure, or characteristic described in connection with an embodiment that may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily imply the same embodiment, nor does it imply independence or alternativeity from other embodiments. The various embodiments in this specification are described in a related manner, with reference to each other for similar or identical parts. In particular, for apparatus, device, and system embodiments, since they are substantially similar to method embodiments, the description is relatively simple, and relevant details are referred to in the description of the method embodiments.

[0165] The above-described embodiments are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of protection. It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims

1. A method for optimizing power grid data transmission in complex mountainous environments, characterized in that, Includes the following steps: Status data of each terminal node of the power grid in the mountainous area is collected, and a multi-dimensional status vector is generated based on the status data; wherein, the status data includes environmental status data, business status data, and energy consumption status data; The time series data corresponding to the multidimensional state vector is input into the prediction model, and the prediction model outputs the channel change trend data corresponding to each terminal node in the future time period; wherein, the prediction model is obtained by training a deep learning time series prediction model using historical channel quality data, historical weather data and terrain data. The multidimensional state vector and the channel change trend data are input into the edge policy optimization model, which outputs the transmission policy corresponding to each terminal node. The edge policy optimization model is obtained by reinforcement learning training based on the channel change trend data and historical state data, with the goal of maximizing transmission success reward, avoiding timeout penalty and reducing energy consumption. The transmission policy includes transmission link, transmit power range and data transmission method. Based on the transmission strategy, the power grid data of each terminal node is transmitted to the corresponding communication target.

2. The power grid data transmission optimization method for complex mountainous environments according to claim 1, characterized in that, Collecting status data from various terminal nodes of the power grid in mountainous areas includes the following steps: Environmental status data of each terminal node of the power grid in the mountainous area is collected; wherein, the environmental status data includes weather data, terrain data and channel quality data; the terrain data includes digital elevation model data retrieved from the edge server and real-time location data; Collect service status data of each terminal node of the mountain power grid; wherein, the service status data includes the service status data of each terminal device connected to the terminal node, and the service priority of the service status data; Energy consumption data of each terminal node of the mountain power grid is collected; wherein, the energy consumption data is calculated based on the transmission power data of each terminal node and the operating conditions of each terminal device.

3. The power grid data transmission optimization method for complex mountainous environments according to claim 2, characterized in that, Generating a multidimensional state vector based on the state data includes the following steps: Based on the digital elevation model data and the positioning data, the terrain occlusion degree between each terminal node and each communication target is calculated. A multidimensional state vector is generated based on the state data and the terrain occlusion degree.

4. The power grid data transmission optimization method for complex mountainous environments according to claim 1, characterized in that, Before inputting the time series data corresponding to the multidimensional state vector into the prediction model, the method includes the following steps: Historical channel quality data, historical weather data, and terrain data are collected and integrated into a unified time series data set as a training dataset; wherein, the historical channel quality data includes any one or more of the following: channel signal-to-noise ratio, packet loss rate, and latency data. Using the training dataset, a deep learning time series prediction model based on a self-attention mechanism is trained. When the performance of the deep learning time series prediction model converges, training is complete, and the final prediction model is obtained.

5. The power grid data transmission optimization method for complex mountainous environments according to claim 1, characterized in that, Before inputting the multidimensional state vector and the channel change trend data into the edge policy optimization model, the method includes the following steps: Collect historical status data and corresponding historical decision data; In a simulation environment built based on the historical state data, and in conjunction with the historical decision data, the edge policy model deployed on the edge server is trained with the goal of maximizing the reward function; wherein the reward function is constructed with successful data transmission as the reward and transmission timeout and energy consumption as the penalty. When the performance of the edge policy model converges, the training is complete, and the trained edge policy model is obtained. The trained edge policy model is continuously interacted with the simulation environment to optimize the network parameters of the edge policy model. When the comprehensive reward value corresponding to the reward function is stable and meets the preset threshold, the final edge policy optimization model is obtained.

6. The power grid data transmission optimization method for complex mountainous environments according to claim 5, characterized in that, In a simulation environment built based on the historical state data, and in conjunction with the historical decision data, the edge policy model deployed on the edge server is trained with the objective of maximizing the reward function, including the following steps: A reward function is constructed with successful data transmission as the basic reward, successful high-priority data transmission as an additional reward, and transmission timeout and energy consumption as penalties; wherein the transmission delay of the high-priority data is lower than a set threshold. In the edge server, a simulation environment is constructed based on the historical state data; In the simulation environment, the policy network and value network are initialized to train the edge policy model deployed on the edge server with the goal of maximizing the reward function.

7. The power grid data transmission optimization method for complex mountainous environments according to claim 1, characterized in that, Based on the aforementioned transmission strategy, transmitting the power grid data of each terminal node to its corresponding communication target includes the following steps: Based on the transmission strategy, the transmission power, corresponding communication resources, and data transmission methods of each terminal node are adjusted; wherein, the communication resources include: sky communication resources, ground communication resources, and space communication resources; the ground communication resources have the highest scheduling priority, followed by the sky communication resources, and lastly the space communication resources; the data transmission methods include: immediate transmission, compressed transmission, and local caching; Based on the transmission power, the transmission link corresponding to the communication resources, and the data transmission method, the power grid data of each terminal node is transmitted to the corresponding communication target; wherein, the communication target includes: base station, UAV relay platform and Beidou satellite.

8. The power grid data transmission optimization method for complex mountainous environments according to claim 7, characterized in that, Based on the transmission power, the transmission link corresponding to the communication resources, and the data transmission method, the power grid data of each terminal node is transmitted to the corresponding communication target, including the following steps: Based on the data transmission method, the power grid data of each terminal node is processed to obtain transmission strategy matching data; wherein, the data processing method includes any one or more of lossless compression processing, multi-source homogeneous data aggregation processing, and caching processing; The data matching the transmission strategy is encrypted to obtain the corresponding encrypted data. Based on the transmission power, the transmission link corresponding to the communication resources, and the data transmission method, the encrypted data is transmitted to the corresponding communication target.

9. The power grid data transmission optimization method for complex mountainous environments according to claim 8, characterized in that, Transmitting the encrypted data to the corresponding communication target includes the following steps: When the terminal node is in a signal blind zone, the encrypted data is uploaded to the terminal node's cache module; The encrypted data is retrieved from the cache module via a mobile relay station; wherein the mobile relay station and the terminal node communicate via a wireless network or Bluetooth. The cached data is then transmitted to the corresponding communication target.

10. An electronic device, comprising: A processor and a memory storing a program, characterized in that the program includes instructions that, when executed by the processor, cause the processor to perform the method according to any one of claims 1 to 9.

11. A non-transitory machine-readable medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method according to any one of claims 1 to 9.

12. A computer program product, comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the method of any one of claims 1 to 9.

Images