Communication link switching method and device and electronic equipment

Abstract

The invention provides a communication link switching method and device and electronic equipment, and relates to the technical field of intelligent power grid communication. The method comprises the following steps: acquiring a task demand type and a first real-time state parameter of a current link; the first real-time state parameters are input into a trend prediction model, predicted state parameters are obtained, and the trend prediction model is a model which is pre-constructed and trained based on a historical state parameter set and used for analyzing the link state change trend; determining whether the current link meets a preset degradation condition or not based on the task demand type and the predicted state parameter; if the current link meets the preset degradation condition, pre-connection operation is executed on a target candidate link, and the target candidate link is determined from a preset candidate link set based on the task demand type; and after the pre-connection operation is completed, switching the current communication task from the current link to the target candidate link. According to the invention, the problem of communication interruption caused by link failure can be avoided, and the stability and real-time performance of intelligent power grid communication are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of smart grid communication, and in particular to a communication link switching method and device and electronic equipment. BACKGROUND

[0002] With the rapid development of smart grids, the dependence on communication systems is increasing. Smart grids need to realize real-time monitoring of power grid operation state, remote control and fast fault response, which puts high requirements on the stability, reliability and real-time performance of communication links. In particular, in scenarios such as single-pilot inspection, emergency rescue, field surveying and mapping, and energy facility inspection, the working environment is complex and variable, often facing strong electromagnetic interference, terrain obstruction, and network coverage fragmentation, etc., resulting in large fluctuations in communication link quality, which seriously affects the operation efficiency and safety of smart grids.

[0003] Currently, in the existing multi-mode communication system of smart grids, terminal devices usually support multiple communication modes such as satellite communication, cellular network, shortwave communication, etc., and adopt a passive mechanism of link failure triggered switching for link switching. When the signal quality of the primary link decreases to a preset threshold, the system automatically switches to the backup link to ensure the continuity of communication.

[0004] However, the existing method has the problem of frequent communication interruption caused by link failure switching mechanism when dealing with complex and variable communication environment of smart grids, which is difficult to ensure the stability and real-time performance of communication. SUMMARY

[0005] The communication link switching method, device and electronic equipment provided by the embodiments of the present application solve the problem that the existing method has frequent communication interruption caused by link failure switching mechanism when dealing with complex and variable communication environment of smart grids, which is difficult to ensure the stability and real-time performance of communication.

[0006] In a first aspect, the embodiments of the present application provide a communication link switching method, comprising:

[0007] obtaining a task demand type and a first real-time state parameter of a current link, the task demand type being used to indicate the performance requirement of the current communication task on the communication link;

[0008] inputting the first real-time state parameter into a trend prediction model to obtain a predicted state parameter, the trend prediction model being a model for analyzing the trend of link state change, which is pre-constructed and trained based on a historical state parameter set;

[0009] determining whether the current link meets a preset degradation condition based on the task demand type and the predicted state parameter;

[0010] If the current link meets the preset degradation condition, a pre-connection operation is performed on a target candidate link, the target candidate link being determined from the preset candidate link set based on the task demand type.

[0011] After the pre-connection operation is completed, the current communication task is switched from the current link to the target candidate link.

[0012] In a possible implementation, the trend prediction model comprises a linear prediction sub-model and a nonlinear correction sub-model; the linear prediction sub-model is configured to perform linear trend analysis on the first real-time state parameter to obtain an initial prediction result; and the nonlinear correction sub-model is configured to perform nonlinear disturbance correction on the initial prediction result to obtain the predicted state parameter.

[0013] In a possible implementation, the linear prediction sub-model is an autoregressive integrated moving average model, and the nonlinear correction sub-model is a random forest model.

[0014] In a possible implementation, whether the current link meets the preset degradation condition is determined based on the task demand type and the predicted state parameter, comprising: determining a performance threshold corresponding to the task demand type; and comparing the predicted state parameter with the performance threshold to determine whether the current link meets the preset degradation condition.

[0015] In a possible implementation, before the pre-connection operation is performed on the target candidate link, the method further comprises: obtaining a second real-time state parameter of each candidate link in the preset candidate link set; determining a comprehensive score of each candidate link based on the task demand type and the second real-time state parameter of each candidate link; and determining the target candidate link from the preset candidate link set based on all the comprehensive scores.

[0016] In a possible implementation, the comprehensive score of each candidate link is determined based on the task demand type and the second real-time state parameter of each candidate link, comprising: obtaining a dynamic weight set according to the task demand type, the dynamic weight set comprising weights corresponding to a plurality of state parameters; determining a normalization strategy for each candidate link based on the task demand type; processing the second real-time state parameter of the candidate link based on the normalization strategy to obtain a normalized parameter set; and determining the comprehensive score of the candidate link based on the normalized parameter set and the dynamic weight set.

[0017] In a possible implementation, the target candidate link is determined from the preset candidate link set based on all the comprehensive scores, comprising: for each comprehensive score, determining whether the comprehensive score is greater than or equal to a preset switching score, the preset switching score being determined based on the comprehensive score of the current link and a preset switching gain threshold; and if the comprehensive score is greater than or equal to the preset switching score, determining the candidate link corresponding to the comprehensive score in the preset candidate link set as the target candidate link.

[0018] In one possible implementation, the method further includes: determining the comprehensive score of the current link based on the task requirement type and the first real-time state parameter; classifying each candidate link based on the comprehensive score of each candidate link and the comprehensive score of the current link, and determining the power consumption monitoring strategy for each candidate link, wherein the power consumption monitoring strategy includes the monitoring wake-up cycle and / or the monitoring power consumption level.

[0019] In one possible implementation, based on the comprehensive score of each candidate link and the comprehensive score of the current link, each candidate link is classified and a power monitoring strategy for each candidate link is determined. This includes: determining a high-score ratio threshold and a low-score ratio threshold based on the comprehensive score of the current link; for each candidate link, if the comprehensive score of the candidate link is greater than or equal to the high-score ratio threshold, the candidate link is determined as a high-score candidate link, wherein the power monitoring strategy for the high-score candidate link is a first power monitoring strategy, and the monitoring wake-up period corresponding to the first power monitoring strategy is a first period; if the comprehensive score of the candidate link is less than the low-score ratio threshold, the candidate link is determined as a low-score candidate link, wherein the power monitoring strategy for the low-score candidate link is a second power monitoring strategy, and the monitoring wake-up period corresponding to the second power monitoring strategy is a second period, wherein the second period is greater than the first period.

[0020] In one possible implementation, if the current link meets the preset degradation conditions, a pre-connection operation is performed on the candidate target link, including: if the current link meets the preset degradation conditions, a pre-connection request is initiated to the target candidate link; and link-layer connection establishment, security authentication, and routing table pre-configuration operations are performed with the target candidate link.

[0021] In one possible implementation, switching the current communication task from the current link to the target candidate link includes: determining a switching strategy based on the task requirement type, the switching strategy including a multi-path TCP migration strategy or a packet mirroring transmission strategy; according to the determined switching strategy, forwarding the data stream of the current communication task to the target candidate link and interrupting data transmission with the current link.

[0022] Secondly, embodiments of this application provide a communication link switching device, comprising:

[0023] The acquisition module is used to acquire the task requirement type and the first real-time status parameter of the current link. The task requirement type is used to indicate the performance requirements of the current communication task on the communication link.

[0024] The prediction module is used to input the first real-time state parameters into the trend prediction model to obtain the predicted state parameters within the prediction time window. The trend prediction model is a model pre-built and trained based on the historical state parameter set to analyze the trend of link state changes. The prediction time window is determined based on the task requirement type.

[0025] The judgment module is used to determine whether the current link meets the preset degradation conditions based on the task requirement type and predicted status parameters.

[0026] The connection module is used to perform a pre-connection operation on the target candidate link if the current link meets the preset degradation conditions. The target candidate link is determined from the preset candidate link set based on the task requirement type.

[0027] The switching module is used to switch the current communication task from the current link to the target candidate link after the pre-connection operation is completed.

[0028] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0029] The memory stores instructions that the computer executes;

[0030] The processor executes computer execution instructions stored in memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0031] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0032] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed, implements the first aspect and / or various possible implementations of the first aspect.

[0033] The communication link switching method, apparatus, and electronic device provided in this application obtain the task requirement type and introduce a trend prediction model trained based on historical state parameter sets. This model analyzes the link state change trend to obtain predicted state parameters. Then, it combines the task requirement type and predicted state parameters to determine whether the current link meets preset degradation conditions. If so, a pre-connection operation is performed on the target candidate link determined from a preset candidate link set based on the task requirements, finally completing the link switching. This method utilizes the trend prediction model to predict potential link degradation and performs pre-connection operations in advance, avoiding communication interruptions caused by switching only after the link has completely failed, thus improving the stability of smart grid communication. Furthermore, the task requirement type is fully considered when determining whether the current link meets the preset degradation conditions and when selecting the target candidate link, ensuring that the performance requirements of the communication task can be met after link switching, making the link switching decision more aligned with actual business scenarios. In addition, the pre-connection operation reduces the connection establishment time during link switching, allowing communication tasks to switch to the target candidate link more quickly, further reducing the possibility of communication interruption and improving the real-time performance and stability of communication. Attached Figure Description

[0034] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0035] Figure 1 A flowchart illustrating the communication link switching method provided in this application;

[0036] Figure 2 A flowchart illustrating the weight determination method provided in this application;

[0037] Figure 3 A flowchart illustrating the selection process for the encryption strategy provided in this application;

[0038] Figure 4 A schematic diagram illustrating the specific process of the communication link switching method provided in this application;

[0039] Figure 5 A schematic diagram of the communication link switching device provided in this application;

[0040] Figure 6 A schematic diagram of the structure of the electronic device provided in this application.

[0041] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0042] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0043] In existing technologies, a passive mechanism of link failure-triggered switching is used for link switching. When the signal quality of the primary link degrades to a preset threshold, the system automatically switches to the backup link to ensure communication continuity. However, this passive mechanism requires waiting for the current link to completely fail before initiating the switching process when the link suddenly deteriorates (such as satellite signal blockage or cellular base station switching). This results in a long communication interruption time, making it difficult to meet the strict latency requirements of real-time tasks (such as voice commands and emergency decision-making). Furthermore, to ensure timely switching, non-primary links need to maintain full-power monitoring for extended periods, leading to a significant increase in terminal device power consumption and making it difficult to meet the endurance requirements of long-duration tasks such as field operations and individual soldier inspections. In addition, existing link scheduling strategies often use a fixed priority system (such as default satellite > cellular > shortwave), which cannot be dynamically adjusted according to the differentiated needs of different smart grid tasks (such as real-time monitoring, remote control, and data transmission). This causes high-priority tasks to fail due to link mismatch (such as file transfer interruptions due to insufficient bandwidth or loss of emergency commands due to insufficient anti-interference capabilities), resulting in poor communication quality. It is evident that existing methods suffer from poor communication stability and real-time performance when dealing with the complex and ever-changing communication environment of smart grids. In particular, the frequent communication interruptions caused by link failure switching mechanisms result in switching lag and energy waste, further impacting communication stability and real-time performance.

[0044] To address the aforementioned issues, this application provides a communication link switching method, apparatus, and electronic device. By introducing a trend prediction model, the first real-time state parameter of the current link is input into the model. The model is trained based on a historical state parameter set to analyze the link state change trend and obtain predicted state parameters. This allows for early prediction of potential link degradation, rather than waiting until the link is severely degraded before taking action, providing a forward-looking basis for link switching decisions and effectively avoiding communication interruptions caused by delayed switching. Simultaneously, the task requirement type indicates the performance requirements of the current communication task on the communication link. When determining whether the current link meets the preset degradation conditions and selecting target candidate links, the task requirement type is fully considered, making the link switching decision more aligned with actual business scenarios. It allows for the selection of the most suitable link based on the characteristics of different tasks, ensuring communication quality while optimizing the utilization of link resources. Furthermore, after determining that the current link meets the preset degradation conditions, a pre-connection operation is performed on the target candidate links. The target candidate links are determined from a preset candidate link set based on the task requirement type. The pre-connection operation establishes a connection with the target candidate link in advance, reducing the connection establishment time during actual link switching. This makes the link switching process faster and smoother, further reducing the possibility of communication interruption, improving the real-time performance and stability of communication, and effectively solving the problems of communication interruption and delay caused by connection establishment time in the existing link switching mechanism.

[0045] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will be described below with reference to the accompanying drawings.

[0046] The execution subject of the communication link switching method provided in this application embodiment can be a computing device such as a server or server cluster. The server can be a mobile phone, computer, tablet, or other device. This application embodiment does not impose any particular restrictions on the implementation method of the execution subject, as long as the execution subject can obtain the task requirement type and the first real-time state parameter of the current link. The task requirement type is used to indicate the performance requirements of the current communication task on the communication link. The first real-time state parameter is input into a trend prediction model to obtain predicted state parameters. The trend prediction model is a model pre-built and trained based on a historical state parameter set to analyze the trend of link state changes. Based on the task requirement type and the predicted state parameters, it is determined whether the current link meets the preset degradation conditions. If the current link meets the preset degradation conditions, a pre-connection operation is performed on the target candidate link. The target candidate link is determined from the preset candidate link set based on the task requirement type. After the pre-connection operation is completed, the current communication task is switched from the current link to the target candidate link.

[0047] Figure 1 This is a flowchart illustrating the communication link switching method provided in this application. The executing entity of this method can be a system server or other server storing the communication link switching method; this embodiment does not impose any particular limitations. Figure 1 As shown, the method may include:

[0048] S101. Obtain the task requirement type and the first real-time status parameter of the current link. The task requirement type is used to indicate the performance requirements of the current communication task on the communication link.

[0049] The first real-time status parameter can refer to parameters reflecting the real-time operating status of the current communication link. These parameters can include signal strength, signal-to-noise ratio, round-trip time, etc., and are used to intuitively reflect the current working status of the link. The current link can refer to multi-mode links such as satellite communication, cellular network, and shortwave communication.

[0050] In some embodiments, real-time status parameters of each communication link can be collected periodically, such as signal strength (RSSI), signal-to-noise ratio (SNR), packet loss rate (PLR), round-trip time (RTT), bandwidth utilization, remaining terminal power, and link anti-interference level (quantified into 1-10 levels, such as satellite anti-blocking level 8-10 and cellular encryption level 5-7).

[0051] Optionally, after collecting real-time status parameters, outlier filtering is performed on the parameters, and storable time-series data is formed through a sliding window. Further, the 3σ criterion can be used to filter out outliers (such as RSSI jumps exceeding ±30dB caused by sudden impulse interference), and the time-series data can be stored through a sliding window (the window size is adaptively adjusted according to the link dynamics: 5 sampling periods for satellite links (high stability) and 10 sampling periods for shortwave links (high dynamics)) to provide high-quality input for the prediction model.

[0052] Task requirements can be categorized into specific performance requirements for communication links under different business tasks within the smart grid. Examples include emergency commands for power monitoring, file transfers, and voice calls. It's important to note that different communication tasks have significantly different performance requirements for the link. For instance, some tasks (like voice calls) may prioritize real-time performance, where even minor delays can affect call quality; others (like file transfers) prioritize data transmission accuracy or bandwidth. Therefore, it's necessary to determine the differentiated requirements based on the specific task.

[0053] To accurately determine the type of task requirement, a multi-dimensional information identification approach can be adopted. On one hand, task identifiers can be obtained through application-layer APIs. For example, during the operation of a communication system, when a voice call task is initiated, the application-layer API will return a "voice call" task identifier; if there is a video transmission task, it will return a "video transmission" identifier, thus clarifying the specific requirements of different tasks for link performance.

[0054] On the other hand, parsing the DSCP tags of data packets is also an effective method. DSCP tags can assign different priorities and service types to different data streams. For example, data packets marked EF (ExpeditedForwarding) usually correspond to real-time voice services. After obtaining such tags, the system can know the low latency requirements of the task. On the other hand, data packets marked AF4 (AssuredForwarding4) may correspond to high-priority file transfer tasks, indicating their requirements for link bandwidth and transmission reliability.

[0055] Furthermore, user-preset priorities can also help determine the type of task requirements. Users can categorize tasks into levels 0-5 based on their importance, with level 5 being the highest priority. For example, an emergency command transmission task might be set to level 5, meaning that this task has extremely high requirements for link reliability and real-time performance.

[0056] S102. Input the first real-time state parameter into the trend prediction model to obtain the predicted state parameter. The trend prediction model is a model pre-built and trained based on the historical state parameter set to analyze the trend of link state changes.

[0057] In this step, the trend prediction model is a pre-built and trained model based on a set of historical state parameters. Through learning and analyzing a large number of historical state parameters, the model can grasp the changing patterns of the link state with time, environment, and other factors. Predicted state parameters can be used to predict the link's state over a future period. Furthermore, predicted state parameters can refer to state parameters within a preset time window. The preset time window is adaptively adjusted according to the task latency requirements corresponding to the task type. For example, if the latency requirement for a voice task is no more than 200ms, the preset time window can be 200ms; if the latency requirement for a file transfer task is no more than 500ms, the preset time window can be 500ms.

[0058] For example, trend prediction models can employ time series analysis models in machine learning, such as the Autoregressive Integral Moving Average (ARIMA) model. Historical state parameter sets are used as training data to train the model, enabling it to accurately predict future changes in link states. For instance, parameters such as signal strength and bit error rate at different times of day over the past month can be collected as historical data to train the ARIMA model. Then, currently collected parameters are input into the model to obtain predicted state parameters for the next few hours. Neural network models, such as recurrent neural networks or long short-term memory networks, can also be used. These models are better able to handle data with time-series characteristics and capture long-term dependencies in link states.

[0059] A multi-model approach can also be employed to combine the advantages of various models, thereby improving the accuracy and stability of predictions. For example, combining the ARIMA model with the LSTM model leverages the ARIMA model's ability to capture linear relationships and the LSTM model's ability to handle nonlinear relationships, allowing for a more comprehensive analysis of changes in link state parameters and thus obtaining more reliable predicted state parameters. Another example is combining the ARIMA model with a decision tree model (such as a random forest). The ARIMA model captures linear trends, more accurately predicting the linear changes in link state over future time periods. After the ARIMA model provides a prediction based on linear trends, the random forest model can correct the prediction based on the nonlinear characteristics reflected in the current real-time state parameters, thereby improving the overall prediction accuracy.

[0060] Optionally, before inputting the first real-time state parameters into the trend prediction model, the first real-time state parameters can be filtered to select key indicators (such as SNR and RTT) that have a significant impact on the current link communication. Then, only the key indicator parameters are input into the trend prediction model as the first real-time state parameters for short-term prediction. On the one hand, this can significantly reduce the amount of data, reduce the computational complexity of the model, improve prediction efficiency, and enable the model to output results faster to assist link decision-making. On the other hand, focusing on key indicators allows the model to more accurately capture the changing trends of factors that have a significant impact on link communication, eliminate irrelevant or secondary data interference, thereby enhancing prediction accuracy and providing a more reliable basis for whether to switch links.

[0061] S103. Based on the task requirement type and predicted state parameters, determine whether the current link meets the preset degradation conditions.

[0062] In this step, the preset degradation conditions can be judgment criteria pre-set according to the task requirement type and the normal working range of the link. When the predicted state parameters of the link exceed the standard range, the current link is considered to meet the preset degradation conditions.

[0063] Furthermore, the link performance requirements corresponding to the task requirement type are compared and analyzed with the predicted state parameters to determine whether the predicted state parameters exceed the preset reasonable range.

[0064] In some examples, a set of degradation thresholds can be set for different task requirement types. For instance, for communication tasks with extremely high real-time requirements, lower latency and bit error rate thresholds can be set as degradation conditions. When the predicted latency parameter exceeds the threshold or the bit error rate exceeds the corresponding threshold, the current link is determined to meet the preset degradation conditions.

[0065] Alternatively, a comprehensive evaluation method can be used. Different weights are assigned to each predicted state parameter based on the task requirement type, and a comprehensive degradation index is calculated. When this index exceeds a preset threshold, the current link is determined to meet the preset degradation conditions. For example, for tasks requiring both high bandwidth and low latency, bandwidth and latency parameters are assigned certain weights, and a comprehensive index is obtained through weighted calculation. This index is then compared with a preset threshold to determine whether the link has degraded.

[0066] S104. If the current link meets the preset degradation conditions, then perform a pre-connection operation on the target candidate link. The target candidate link is determined from the preset candidate link set based on the task requirement type.

[0067] Here, the target candidate link refers to a link selected from a preset candidate link set based on task requirements, which can meet the performance requirements of the current communication task. Pre-connection operation refers to establishing a connection with the target candidate link in advance and performing necessary initialization and configuration work before officially switching communication tasks.

[0068] In some embodiments, once it is determined that the current link meets the preset degradation conditions, a suitable target candidate link is selected from the preset candidate link set according to the task requirement type, and then a pre-connection operation is initiated to establish a connection with the target candidate link and perform relevant settings.

[0069] For example, a preset set of candidate links is stored and categorized according to link type (e.g., satellite communication, cellular network, shortwave communication, etc.). Once the task requirement type is determined, the link with the best performance or that meets preset conditions is selected from the corresponding link type category as the target candidate link. Then, a pre-connection command is sent through the link's control interface to perform pre-connection operations such as parameter configuration. For instance, for communication tasks in remote areas, the task requirement type might be more suitable for satellite communication links; therefore, a link with better signal quality is selected from the satellite communication link category as the target candidate link for pre-connection. Further, assuming the current terminal device uses a cellular network link, when performing power facility inspection tasks in mountainous areas, the cellular network signal begins to fluctuate but still maintains a certain communication connection; therefore, it is currently used as the current link. When the cellular network signal quality continues to deteriorate and cannot meet the smart grid's communication needs for real-time monitoring of operational status and voice calls (i.e., the current link meets preset degradation conditions), satellite communication links, shortwave communication links, microwave communication links, etc., can be used as one of the target candidate links.

[0070] Optionally, the preset candidate link set can be dynamically updated based on the real-time availability and performance metrics of the links. When selecting target candidate links, the real-time status of the links and the type of task requirements are taken into account. For example, in areas with good cellular network signal, when the task requirement type demands high bandwidth, links with sufficient bandwidth are selected from the cellular network candidate links as target candidate links, and pre-connection operations are performed.

[0071] S105. After the pre-connection operation is completed, switch the current communication task from the current link to the target candidate link.

[0072] Furthermore, after completing the pre-connection operation with the target candidate link, the current communication task is migrated from the current link to the target candidate link to ensure communication continuity. In some examples, a smooth handover approach can be adopted. During the handover process, a connection with the same communication parameters and protocol as the current link is first established on the target candidate link, and then data traffic is gradually transferred from the current link to the target candidate link. For example, in IP network communication, by changing the routing table information, the transmission path of data packets is switched from the current link to the target candidate link, achieving a smooth migration of communication tasks.

[0073] For communication tasks with extremely high real-time requirements, a rapid switching approach can be adopted. After the pre-connection operation is completed, communication on the current link is immediately interrupted, and the task is switched to the target candidate link. For example, in voice communication, to reduce call interruption time, once the pre-connection is completed, the voice data stream is quickly switched to the target candidate link to ensure call continuity.

[0074] The communication link switching method provided in this application, by acquiring the task requirement type as a foundation, enables link switching decisions to accurately adapt to actual business scenarios, avoiding unreasonable resource occupation caused by blind switching. Simultaneously, addressing the communication interruption problem caused by the lag in traditional passive switching mechanisms, a trend prediction model is introduced. This model is trained using historical state parameter sets to analyze link state change trends, thereby anticipating link degradation tendencies and gaining time for timely intervention. Based on the task requirement type and predicted state parameters, it determines whether the current link meets preset degradation conditions, making the judgment more comprehensive and scientific. Once the conditions are met, a pre-connection operation is performed on the target candidate link determined from the preset candidate link set based on task requirements, establishing a connection in advance to reduce connection establishment time during switching. This method, by predicting link degradation in advance and pre-connecting, effectively solves the problem of frequent communication interruptions under existing link failure switching mechanisms, improving the stability, real-time performance, and resource utilization efficiency of smart grid communication in complex environments.

[0075] Based on the above embodiments, the trend prediction model includes a linear prediction sub-model and a nonlinear correction sub-model; wherein, the linear prediction sub-model is used to perform linear trend analysis on the first real-time state parameter to obtain the initial prediction result; the nonlinear correction sub-model is used to perform nonlinear disturbance correction on the initial prediction result to obtain the predicted state parameter.

[0076] In this embodiment, the linear prediction sub-model can be a mathematical model that predicts based on linear relationships. It assumes that parameter changes follow a linear pattern and makes preliminary predictions about future parameter values ​​by fitting the linear relationship of parameters in historical data. For example, in link status parameters, if a parameter changes roughly in a linear trend over time, the linear prediction sub-model can calculate the slope and intercept of this line based on existing data points, thereby predicting future parameter values.

[0077] The nonlinear correction sub-model is used to correct the initial prediction results of the linear prediction sub-model. Since the changes in actual link state parameters are often not perfectly linear and are affected by various nonlinear factors, such as environmental interference and equipment aging, the nonlinear correction sub-model can identify and correct these nonlinear disturbances, making the prediction results closer to reality.

[0078] The initial prediction result is the predicted value obtained by performing a linear trend analysis on the first real-time state parameters using the linear prediction sub-model. This result is only a preliminary prediction based on the linear assumption and does not consider the influence of nonlinear factors. Nonlinear disturbances refer to the deviations from the linear prediction results caused by nonlinear factors during the change of link state parameters. For example, sudden electromagnetic interference may cause nonlinear fluctuations in the signal strength of the link; such fluctuations are nonlinear disturbances.

[0079] In some embodiments, the linear prediction sub-model can employ a simple linear regression model, a multinomial linear regression model, or a time series analysis model from machine learning. For example, fitting a linear relationship between link parameters and time using the least squares method, the equation of this linear relationship constitutes the linear prediction sub-model, such as predicting the linear decay trend of the signal-to-noise ratio over time. Another example is using the ARIMA model as a linear prediction sub-model.

[0080] Nonlinear correction sub-models can employ machine learning models. The model is trained using extensive historical data, including link state parameters under different nonlinear disturbances. When initial predictions are input, the model can correct them based on the learned nonlinear relationships. For example, random forest models correct for nonlinear disturbances (such as sudden electromagnetic interference).

[0081] By combining linear prediction sub-models and nonlinear correction sub-models, the prediction accuracy of link state parameters is improved, which helps to more accurately determine whether the link meets the preset degradation conditions, thereby improving the timeliness and accuracy of communication link switching and further enhancing the stability and reliability of smart grid communication.

[0082] Based on the above embodiments, the linear prediction sub-model is an autoregressive integral moving average model, and the nonlinear correction sub-model is a random forest model.

[0083] Furthermore, the Autoregressive Integral Moving Average (ARIMA) model is a time series forecasting model suitable for predicting link state parameters with trends and seasonality. Autoregression means the model uses its past values ​​to predict future values. For example, it uses the values ​​of link state parameters from previous times to predict the current value, based on the correlation between the parameters themselves over time. Integration is used to make non-stationary time series data stationary. In link state parameters, if there is a significant trend or seasonal variation over time, these non-stationary factors can be eliminated through differencing (a form of integration). For example, link signal strength may exhibit periodic changes within a day; integration can transform it into a stationary sequence for easier subsequent analysis. Moving average considers the relationship between the predicted value and past prediction errors. It uses a linear combination of past random errors to correct the predicted value, improving prediction accuracy. In link state forecasting, past prediction errors may contain information not fully captured by the model; moving average can compensate for this.

[0084] Random forest is an ensemble learning model composed of multiple decision trees. Each decision tree acts as an expert, independently judging the initial prediction (in the case of link state prediction). During prediction, the random forest model synthesizes the judgments of all decision trees and provides a final corrected result by averaging (for regression problems such as link state parameter prediction). For example, in link bit error rate prediction, multiple decision trees correct the initial predicted bit error rate, and then the random forest model averages these corrections to obtain the final predicted bit error rate.

[0085] In some examples, firstly, the time series data of the link state parameters are tested for stationarity. If the data is non-stationary, it is transformed into a stationary series through differencing. For example, the unit root test is used to determine whether the data is stationary; if not, first-order or higher-order differencing is performed. Then, the model parameters are determined, including the autoregressive order (p), integral order (d), and moving average order (q). These parameters can be determined using the autocorrelation function (ACF) and partial autocorrelation function (PACF). Finally, the ARIMA model with determined parameters is used to perform linear trend analysis on the first real-time state parameters to obtain initial prediction results. For the random forest model, a large amount of historical link state data is collected, including the first real-time state parameters, initial prediction results, and corresponding actual state parameters. This data is used to train the random forest model. During training, the random forest randomly selects subsets of data and features to construct multiple decision trees. When the initial prediction results are input, each decision tree in the trained random forest model undergoes nonlinear correction, and then the results of all decision trees are combined to output the final predicted state parameters.

[0086] The ARIMA model can effectively capture the linear trend of link state parameters; the random forest model can accurately correct nonlinear disturbances in linear prediction. Combining the two makes the predicted state parameters more accurate, which can further optimize communication link switching decisions and improve the stability and communication quality of smart grid communication in complex environments.

[0087] In practical smart grid communication, different service tasks have significantly different requirements for link performance. The lack of a mechanism to determine link degradation based on task requirements and predicted parameters may lead to untimely or inaccurate link switching, affecting the normal operation of the system. Therefore, based on the above embodiments, the method described in S103 for determining whether the current link meets preset degradation conditions based on task requirement type and predicted state parameters may include: determining a performance threshold corresponding to the task requirement type; comparing the predicted state parameters with the performance threshold to determine whether the current link meets the preset degradation conditions.

[0088] In this embodiment, the performance threshold can be a critical value of the link performance parameters preset according to different task requirements. Different tasks have different requirements for link performance, and the performance threshold is a standard used to measure whether the link performance meets the task requirements. Furthermore, the predicted state parameters output by the trend prediction model are numerically compared with the corresponding performance threshold, and the comparison result is used to determine whether the current link meets the preset degradation conditions.

[0089] Optionally, a performance threshold can be preset for state parameters that significantly impact communication tasks. For example, in a smart grid power equipment condition monitoring task, the task requirement is real-time acquisition of equipment operating data, and the corresponding performance threshold is set to a link transmission round-trip delay of no more than 100ms. After obtaining the predicted round-trip delay parameter of the link over a future period using a trend prediction model, it is compared with the 100ms performance threshold. If the predicted round-trip delay exceeds 100ms at some future point, the current link is determined to meet the preset degradation condition.

[0090] Alternatively, a multi-state parameter comprehensive comparison method can be chosen. For some complex communication tasks, such as remote video monitoring and control of substations, the task requirements combine real-time performance and data accuracy. In this case, multiple performance thresholds are set, such as a round-trip time (RTD) of less than or equal to 50ms and a signal-to-noise ratio (SNR) of greater than or equal to 8dB. The RTD and SNR in the predicted state parameters are compared with their corresponding performance thresholds. When one or more parameters exceed the threshold, the current link is determined to meet the preset degradation condition. For example, if the predicted RTD will be greater than 50ms in the future, even if the SNR is within the threshold range, the link is determined to meet the degradation condition.

[0091] For example, performance thresholds may include signal-to-noise ratio (SNR) thresholds and round-trip time (RTT) thresholds. Preset degradation conditions can be set as follows: when SNR < the task-specific threshold, or RTT > the task's maximum tolerable latency, the degradation standard is reached. The task-specific thresholds can differ depending on the task requirement type. For instance, if a voice task requires SNR ≥ 10dB and an emergency command requires SNR ≥ 8dB, then the task-specific threshold for the voice task is 10dB, and the threshold for the emergency command is 8dB. During a voice call, when SNR < 10dB, the degradation standard is considered to have been reached.

[0092] Furthermore, the performance requirements of tasks on the power grid link may vary depending on the operating phase of the smart grid. For example, during peak electricity consumption periods, the real-time requirements for power dispatch command transmission are higher. In this case, performance thresholds can be dynamically adjusted based on the grid's operating status. Assuming the normal RTT for dispatch command transmission is 80ms, it can be adjusted to 50ms during peak electricity consumption periods. The predicted state parameters are compared with the dynamically adjusted performance thresholds to determine whether the link meets the degradation conditions.

[0093] By clearly defining the specific methods for determining whether the current link meets the preset degradation conditions, it is possible to accurately determine whether the link status has deteriorated based on the characteristics and requirements of different tasks, providing a clear and reasonable basis for link switching, thereby improving the accuracy and effectiveness of communication link switching and ensuring the stable operation of smart grid communication.

[0094] In actual communication link switching processes, if the target candidate link is not selected properly, the switched link may fail to meet the requirements of the service task, affecting the stability of smart grid communication. Therefore, this embodiment provides a target candidate link selection method. Based on the above embodiment, before performing the pre-connection operation on the target candidate link as described in S104, the method may further include: obtaining the second real-time status parameter of each candidate link in a preset candidate link set; determining the comprehensive score of each candidate link based on the task requirement type and the second real-time status parameter of each candidate link; and determining the target candidate link from the preset candidate link set based on all the comprehensive scores.

[0095] In this embodiment, the second real-time state parameter is similar to the first real-time state parameter described above; it is a parameter used to describe the current state of the candidate link. It may include indicators such as signal strength, signal-to-noise ratio, and round-trip time, which are obtained through real-time monitoring to reflect the current performance status of the candidate link.

[0096] The comprehensive score is an evaluation score given to each candidate link through a certain calculation method. This score takes into account both the link's suitability for the current task and its own performance status, and is used to measure the merits of the candidate link as a target link.

[0097] The comprehensive score can be calculated using a weighted scoring method, which involves setting different weights for real-time status parameters based on different task requirements. These weights can be set according to the real-time and accuracy requirements of different tasks. Then, the comprehensive score for each candidate link is calculated by weighting.

[0098] Fuzzy comprehensive evaluation can also be used to calculate the comprehensive score. For example, in the communication of a distributed energy management system in a smart grid, the task requirement is to reliably transmit energy production and consumption data. A pre-set candidate link set includes various wireless communication links. Since link state parameters have a certain degree of fuzziness, such as descriptions of signal strength as "strong" or "weak," a fuzzy comprehensive evaluation method is adopted. The evaluation factor set (i.e., the second real-time state parameter) and the evaluation citation set (e.g., excellent, good, average, poor) are determined. By establishing a fuzzy relation matrix and weight vector, the fuzzy comprehensive evaluation result for each candidate link is calculated and then converted into a comprehensive score.

[0099] In some examples, for each candidate link, after obtaining the second real-time status parameter, the weight allocation of each status parameter of the link is determined according to the requirement type of the current business task. For example, for tasks with high real-time requirements, the weight of link latency will be set higher; for tasks with large data volumes, the weight of bandwidth may be greater. Then, combined with the second real-time status parameter of each candidate link, a comprehensive score for each candidate link is calculated according to the set scoring rules. Finally, the comprehensive scores of all candidate links are compared, and the candidate link with the highest comprehensive score or that meets the preset scoring threshold is selected as the target candidate link.

[0100] By acquiring the second real-time status parameters of candidate links and calculating a comprehensive score, the ability of each candidate link to meet task requirements can be comprehensively and objectively evaluated. This allows for more accurate determination of target candidate links from the preset candidate link set, improving the rationality and effectiveness of link switching and ensuring the stable operation of the smart grid communication system under different task scenarios.

[0101] Based on the above embodiments, the method for determining the comprehensive score of each candidate link based on the task requirement type and the second real-time state parameter of each candidate link may include: obtaining a dynamic weight set according to the task requirement type, the dynamic weight set including weights corresponding to multiple state parameters; determining a normalization strategy for each candidate link based on the task requirement type; processing the second real-time state parameter of the candidate link based on the normalization strategy to obtain a normalized parameter set; and determining the comprehensive score of the candidate link based on the normalized parameter set and the dynamic weight set.

[0102] In this embodiment, the dynamic weight set can be a set of weight values ​​dynamically generated according to different task requirements. Each weight corresponds to a link state parameter, such as bandwidth, latency, and security. Different task requirements place different degrees of emphasis on link state parameters, and the dynamic weight set can flexibly adjust the weights of each parameter according to task requirements to more accurately evaluate candidate links. For example, for tasks with extremely high real-time requirements, the weight of the latency parameter will be set higher.

[0103] The normalization strategy is designed to eliminate the impact of differences in units, orders of magnitude, etc., between different second real-time state parameters on the overall score. Since the value ranges and units of various parameters may differ, the normalization strategy processes the second real-time state parameters to bring them into a uniform numerical range, facilitating subsequent calculation of the overall score.

[0104] Optionally, the normalization strategy may include different normalization methods for positive and negative parameters. Further, a min-max normalization strategy (mapped to [0,1]) can be used for each parameter, where positive parameters (such as SNR, anti-interference level) take the original value, and negative parameters (such as PLR, RTT) take 1 minus the original value.

[0105] For example, in a remote meter reading system for a smart grid, the task requirement is to collect user electricity consumption data periodically and accurately. Link parameters include transmission rate, bit error rate, and network availability. Based on the task's requirements for data accuracy and stability, the bit error rate weight in the dynamic weight set is set to 0.5, the transmission rate weight to 0.3, and the network availability weight to 0.2. For the normalization strategy, a range normalization method is used. For example, for the transmission rate parameter, the transmission rate values ​​of all candidate links are collected, the maximum and minimum values ​​are found, and the transmission rate value of each candidate link is mapped to the [0,1] interval using a formula. The bit error rate, network availability, and other parameters are normalized using the same method to obtain a normalized parameter set. Finally, the normalized parameters are multiplied by the corresponding dynamic weights and summed to obtain the comprehensive score of the candidate link.

[0106] In power dispatch automation systems, the task requirement is to transmit dispatch instructions in real-time and reliably. Link parameters include network latency, packet loss rate, and signal strength. Considering the extremely high real-time requirements, the dynamic weight of network latency is non-linearly assigned. The weight increases rapidly when latency is low, and slows down as latency increases. Assuming the final weight for network latency is 0.6, the weight for packet loss rate is 0.25, and the weight for signal strength is 0.15, the normalization strategy uses standard deviation normalization. Taking network latency as an example, the mean and standard deviation of the network latency for all candidate links are calculated. The network latency value of each candidate link is then transformed into a value related to the mean and standard deviation, making it conform to a standard normal distribution, thus achieving normalization. After normalizing each parameter, a normalized parameter set is obtained, which is then combined with the dynamic weight set to calculate the comprehensive score.

[0107] In one example, the task requirements include: voice calls, file transfers, and emergency commands; the real-time status parameters include: latency, bandwidth, stability (packet loss resistance), and security (interference resistance). In this case, a task-weight mapping table (as shown in Table 1) can be pre-defined to determine the dynamic weight set for different task requirement types.

[0108] Table 1 Task-Weight Mapping Table

[0109]

[0110] For example, Figure 2 A flowchart illustrating the weight determination method provided in this application is shown below. Figure 2 As shown, the process first identifies the task type to determine the task requirement type. Based on Table 1, it determines whether the task requirement type is a voice call. If so, the latency weight is increased to 40%. If not, it determines whether the task requirement type is a file transfer. If so, the bandwidth weight is increased to 40%. If not, the security weight is increased to 50%. This allows for the adjustment of the weights of each parameter. Based on the adjusted weights, the link score is calculated.

[0111] As can be seen, the dynamic weights output based on the task-weight mapping table can be dynamically updated according to the task stage. For example, the video backhaul task assigns different weights to different link parameters at different stages. For instance, the latency weight is 40% in the "link establishment stage" (first 5 seconds), and the bandwidth weight increases to 40% in the "transmission stage".

[0112] Optionally, the conditions for adjusting or changing the weight of each parameter may include: switching the task requirement type (such as switching from voice to file transfer), sudden changes in link status (a certain indicator deviates from the sliding window average by more than 20%), and manual user intervention (adjusting the priority via terminal buttons).

[0113] In one example, the overall score can be determined using a dynamic scoring model that satisfies:

[0114] ;

[0115] in, The comprehensive score for the i-th candidate link; The weight of the k-th parameter (satisfying) ); It is the normalized value of the k-th parameter of the i-th candidate link.

[0116] The dynamic weight set can accurately match the importance of each parameter to the task requirements, and the normalization strategy eliminates the differences between parameters. This allows the comprehensive score to reflect the performance of candidate links in meeting task requirements in a comprehensive and objective manner. It provides a strong basis for accurately selecting appropriate target candidate links from the preset candidate link set, improves the rationality of link selection and the accuracy and effectiveness of communication link switching, avoids link mismatch problems caused by fixed priority strategies, and thus improves task completion efficiency.

[0117] Based on the above embodiments, the method for determining a target candidate link from a preset candidate link set based on all comprehensive scores may include: for each comprehensive score, determining whether the comprehensive score is greater than or equal to a preset handover score, wherein the preset handover score is determined based on the comprehensive score of the current link and a preset handover gain threshold; if the comprehensive score is greater than or equal to the preset handover score, then the candidate link in the preset candidate link set corresponding to the comprehensive score is determined as the target candidate link.

[0118] In this embodiment, the preset switching score is used to measure whether the candidate links in the preset candidate link set have sufficient advantages to replace the current link, and serves as the basis for judging link switching.

[0119] The preset handover gain threshold can be a pre-set value used to adjust the relationship between the preset handover score and the current link score. By setting different preset handover gain thresholds, the conditions for link handover can be flexibly controlled to meet the link performance requirements of different business scenarios.

[0120] For example, the preset handover score = the score of the current link + the preset handover gain threshold. The preset handover gain threshold can be a percentage of the score of the current link, such as 5% of the score of the current link. The score of the current link can be determined in the same way as the comprehensive score of the candidate links.

[0121] Optionally, when there are multiple target candidate links, a pre-connection priority can be set. Based on the target link scores in descending order, the links with the highest scores and stable predictions (e.g., index fluctuation ≤ 5%) are given priority for pre-connection.

[0122] By determining the preset handover score based on the current link comprehensive score and the preset handover gain threshold, and determining the target candidate link based on the comparison between the comprehensive score and the preset handover score, the candidate link with significant performance advantages can be selected more scientifically and reasonably.

[0123] In practical applications, due to the link switching mechanism, candidate links are continuously monitored. However, prolonged full-power monitoring of non-primary links leads to a surge in power consumption, wasting system resources and reducing system battery life. Therefore, to address this issue, based on the above embodiments, the method may further include: determining the comprehensive score of the current link based on the task requirement type and a first real-time state parameter; classifying each candidate link based on the comprehensive score of each candidate link and the comprehensive score of the current link, and determining the power consumption monitoring strategy for each candidate link, wherein the power consumption monitoring strategy includes the monitoring wake-up cycle and / or the monitoring power consumption level.

[0124] In this embodiment, the first real-time state parameter refers to various performance indicators during the real-time operation of the current link. The power consumption monitoring strategy can be a strategy formulated to monitor and manage the power consumption of candidate links. The monitoring wake-up period refers to the wake-up time interval for power consumption monitoring of candidate links; the monitoring power consumption level is a different level divided according to the power consumption of the link, used for more granular management of power consumption monitoring.

[0125] In some embodiments, different weights are assigned to the first real-time state parameter of the current link according to different task requirements, and a comprehensive score for the current link is obtained through weighted calculation or other methods. The comprehensive scores of candidate links are compared with the comprehensive score of the current link, and the candidate links are classified according to certain rules. For example, candidate links with a significantly higher comprehensive score than the current link are classified as Level 1, those with a certain advantage are classified as Level 2, and those with a lower comprehensive score than the current link are classified as Level 3, and so on. Based on the classification results, corresponding power consumption monitoring strategies are determined for candidate links of different levels. Level 1 candidate links may require more frequent monitoring (shorter monitoring wake-up cycle) and a higher monitoring power consumption level; while lower-level candidate links can have a longer monitoring wake-up cycle and a relatively lower monitoring power consumption level.

[0126] By comparing the performance of candidate links with that of the current link, power consumption monitoring resources can be allocated reasonably, improving the system's resource utilization efficiency. At the same time, the timing of link switching can be better grasped, ensuring the stable and efficient operation of communication links in the smart grid system.

[0127] Based on the above embodiments, the method for classifying each candidate link and determining the power monitoring strategy for each candidate link based on the comprehensive score of each candidate link and the comprehensive score of the current link may include: determining a high-score ratio threshold and a low-score ratio threshold based on the comprehensive score of the current link; for each candidate link, if the comprehensive score of the candidate link is greater than or equal to the high-score ratio threshold, then the candidate link is determined to be a high-score candidate link, wherein the power monitoring strategy of the high-score candidate link is a first power monitoring strategy, and the monitoring wake-up period corresponding to the first power monitoring strategy is a first period; if the comprehensive score of the candidate link is less than the low-score ratio threshold, then the candidate link is determined to be a low-score candidate link, wherein the power monitoring strategy of the low-score candidate link is a second power monitoring strategy, and the monitoring wake-up period corresponding to the second power monitoring strategy is a second period, wherein the second period is greater than the first period.

[0128] In this embodiment, the high score ratio threshold and the low score ratio threshold are two scoring thresholds determined based on the overall score of the current link, used to classify candidate link levels. The high score ratio threshold is typically set to a value higher than the overall score of the current link by a certain percentage, used to filter out candidate links with significant performance advantages; the low score ratio threshold is set to a value lower than the overall score of the current link by a certain percentage, used to filter out candidate links with relatively poor performance. For example, the high score ratio threshold is 80% of the overall score of the current link, and the low score ratio threshold is 50% of the overall score of the current link.

[0129] The first and second power monitoring strategies are power monitoring schemes designed for candidate links of different levels. The first power monitoring strategy is suitable for high-scoring candidate links, characterized by a shorter monitoring wake-up cycle to monitor power changes more promptly; the second power monitoring strategy is suitable for low-scoring candidate links, with a longer monitoring wake-up cycle to reduce unnecessary monitoring resource consumption, thus achieving differentiated management of power monitoring frequency for candidate links of different levels.

[0130] In one example, the first power consumption monitoring strategy for high-scoring candidate links (Score ≥ 80% of the current link's overall score) is: low-power monitoring (wake-up every 2 seconds, sampling period 500ms, power consumption reduced to 30% of full power). The second power consumption monitoring strategy for low-scoring candidate links (Score < 50% of the current link's overall score) is: ultra-low frequency monitoring (wake-up every 10 seconds, sampling period 100ms, power consumption reduced to 10% of full power).

[0131] Optionally, the system's power parameters can be introduced, and the power consumption monitoring strategy can be dynamically adjusted in conjunction with preset power thresholds. For example, if the current power of the system device is ≥ 30% of its full power, the power consumption monitoring strategy will not be adjusted; if the current power of the system device is 15% ≤ 30% and the current power of the system device is < 30%, only the top two candidate links with the highest comprehensive scores will be monitored, and the current link transmission rate will be reduced by 50% (e.g., from 2Mbps to 1Mbps); if the current power of the system device is < 15%, only the current main link communication will be maintained, and all candidate link monitoring will be turned off.

[0132] By setting power consumption monitoring strategies with different monitoring cycles, candidate links can be managed hierarchically and more scientifically, optimizing the allocation of power consumption monitoring resources and improving the efficiency and accuracy of power consumption monitoring for candidate links. Furthermore, the monitoring frequency of non-primary links can be dynamically adjusted based on the comprehensive link score and power status, reducing system terminal power consumption while ensuring link availability.

[0133] Based on the above embodiments, the method described in S104 for performing a pre-connection operation on a candidate target link if the current link meets the preset degradation conditions may include: if the current link meets the preset degradation conditions, initiating a pre-connection request to the target candidate link; and performing link-layer link establishment, security authentication, and routing table pre-configuration operations with the target candidate link.

[0134] In this embodiment, a pre-connection request can be a signal or instruction sent to a target candidate link to establish a connection relationship in advance. The purpose is to establish a certain degree of association with the target candidate link in advance before the current link deteriorates and affects the communication quality, in order to prepare for a possible formal connection later.

[0135] Link-layer connection establishment refers to the process in a communication network where the link layer is responsible for data transmission between adjacent nodes. Link-layer connection establishment involves establishing a link-layer connection between the two endpoints of a target candidate link, determining the physical channel and related parameters for data transmission, such as transmission rate and frame format, to ensure normal data transmission at the link layer.

[0136] Security authentication is used to ensure the security of communication on a target candidate link by verifying the identities of devices at both ends of the link to prevent unauthorized access and malicious attacks. Common security authentication methods include password verification and digital certificate verification. Only devices that pass security authentication can transmit data on the link.

[0137] A routing table is a crucial data structure in a network used to guide packet forwarding. Routing table pre-configuration refers to configuring routing table information in advance during the connection establishment process with a target candidate link, determining the forwarding path of data packets on that link, so that data transmission can be performed quickly and accurately when officially switching to that link.

[0138] Furthermore, when the current link is detected to meet preset degradation conditions, indicating that the current main link may degrade and affect communication quality within a certain period of time (such as within a preset time window), the system sends a pre-connection request to the target candidate link, informing it of the need to prepare for connection. Upon receiving the pre-connection request, the target candidate link initiates link-layer connection establishment, creating a physical transmission channel and determining relevant parameters. After link-layer connection establishment is completed, security authentication is performed to ensure the legitimacy of both communicating parties and the security of data transmission. After successful security authentication, routing table pre-configuration is performed to plan forwarding paths for potential subsequent data transmissions.

[0139] Optionally, to ensure the security of data transmitted in the link and improve communication security, a dynamic encryption strategy can be introduced. Furthermore, an encryption algorithm can be selected for the pre-connection operation based on the security level associated with the task requirement type; specifically, a first encryption algorithm is selected for high-security-level tasks, and a second encryption algorithm is selected for ordinary-security-level tasks, with the first encryption algorithm having a higher computational complexity than the second encryption algorithm.

[0140] In communication tasks within systems such as smart grids, different tasks have varying degrees of importance and security requirements. Based on the nature of the task and the importance of the data involved, tasks are categorized into different security levels, such as high security and ordinary security. High-security tasks may involve control commands for critical equipment or sensitive user information; ordinary security tasks may involve general monitoring data transmission. Encryption algorithms are techniques used to encode data to prevent unauthorized access and tampering during transmission and storage. Different encryption algorithms have different encryption strengths, computational complexities, and applicable scenarios.

[0141] Furthermore, in one example, the corresponding task sensitivity is determined based on the task requirement type, and then the corresponding encryption algorithm is determined based on the task sensitivity (e.g., level 0-5, with level 5 being the highest) and the link latency. For example, Figure 3 A flowchart illustrating the encryption strategy selection process provided in this application is shown below. Figure 3 As shown, when the task sensitivity is low (e.g., less than level 4), the RC4 low-power encryption algorithm is selected; when the task sensitivity is high (e.g., greater than or equal to level 4) and the link is a low-latency link (e.g., when emergency command + satellite link, the link RTT is less than 100ms), the ChaCha20 encryption algorithm is used to meet the requirements of low encryption latency; when the task sensitivity is high and the link is a high-latency link (e.g., when emergency command + shortwave link, the link RTT is greater than or equal to 100ms), the AES-256 / SM4 encryption algorithm is used to ensure sufficient encryption strength.

[0142] Optionally, a temporary session key (TSK) can be generated based on the pre-shared root key (PSK). During link switching, only the validity of the temporary session key needs to be verified, without the need for a complete key exchange process. This allows for the reasonable selection of encryption methods based on different tasks and link characteristics while ensuring data security, reducing key negotiation time, improving link switching efficiency, and meeting the diverse communication security needs of smart grid systems.

[0143] By pre-connecting, target candidate links are prepared for connection in advance, enabling them to be quickly put into use during link switching, thereby improving the switching efficiency and stability of the communication system under link degradation.

[0144] Based on the above embodiments, the method for switching the current communication task from the current link to the target candidate link described in S105 may include: determining a switching strategy based on the task requirement type, the switching strategy including a multi-path TCP migration strategy or a packet mirroring transmission strategy; forwarding the data stream of the current communication task to the target candidate link according to the determined switching strategy, and interrupting data transmission with the current link.

[0145] In this embodiment, Multipath TCP (MPTCP) is a TCP extension protocol that allows data to be transmitted simultaneously on multiple network paths. The Multipath TCP migration strategy refers to utilizing the characteristics of MPTCP during link switching to gradually migrate the data stream of the current communication task to the target candidate link, achieving a smooth data transition. Simultaneously, it can fully utilize the bandwidth resources of multiple links, improving the efficiency and reliability of data transmission.

[0146] The packet mirroring transmission strategy involves simultaneously copying the data packets on the current link and transmitting them to the target candidate link. During link switching, the target candidate link can receive the same data as the current link in real time. When the switching conditions are met, it can quickly and seamlessly take over the communication tasks, reducing data loss and communication interruption time.

[0147] Data stream forwarding refers to the process of guiding the data generated by the current communication task from the current link to the target candidate link, ensuring that the data can be transmitted according to the new link.

[0148] Furthermore, if the task requirement type indicates that the current communication task is a real-time task (such as voice or emergency commands), a multi-path TCP migration strategy is used for primary substream switching; if the task requirement type indicates that the current communication task is a non-real-time task (such as file transfer), a packet mirroring transmission strategy is used for switching. For example, real-time tasks use MPTCP, and during switching, a 0-30ms interruption is achieved through rapid substream migration; non-real-time tasks use packet mirroring transmission (the primary and secondary links simultaneously send the first N packets, where N = window size / packet length), and primary link transmission stops after confirming the target link is available, with an interruption ≤50ms.

[0149] Optionally, the pre-connection process can be completed using the current communication gap (such as the packet sending interval) without consuming the main link bandwidth.

[0150] By selecting appropriate switching methods based on the characteristics of different tasks, the efficiency and reliability of link switching can be improved, data loss and communication interruption time can be reduced, and the smooth operation of communication tasks in the smart grid can be guaranteed.

[0151] The communication link switching method provided in this application is applicable to the intelligent scheduling of communication links for multi-mode communication terminals such as satellite, cellular, and shortwave in complex and dynamic scenarios. For example, typical scenarios include: during individual soldier patrols, real-time transmission of voice commands, video feeds, and emergency instructions is required, facing terrain obstruction, signal blockage, and interference from movement paths. During field surveying, terminal equipment operates for extended periods in remote areas (such as mountains, oceans, and deserts), requiring a balance between communication continuity and terminal battery life.

[0152] Taking the communication link switching method applied to the multi-mode communication system of smart grid as an example, the multi-mode communication system of smart grid may include a state perception module, a task parsing module, a processor, a power management module, a multi-mode communication module, and an external network. Figure 4 A detailed flowchart illustrating the communication link switching method provided in this application is shown below. Figure 4 As shown, in this embodiment... Figure 1 Based on the embodiments, the communication link switching method is described in detail, which includes:

[0153] (1) Communication status awareness and preprocessing.

[0154] The status parameters of the multimode communication link are collected by the status awareness module to achieve real-time parameter acquisition.

[0155] Furthermore, the status awareness module can periodically collect real-time parameters of each communication link: signal strength (RSSI), signal-to-noise ratio (SNR), packet loss rate (PLR), round-trip time (RTT), bandwidth utilization, remaining terminal power, and link anti-interference level (quantified into 1-10 levels, such as satellite anti-blocking level 8-10, cellular encryption level 5-7).

[0156] The system employs the 3σ criterion to filter out outliers (such as RSSI jumps exceeding ±30dB caused by sudden pulse interference) and stores time series data through a sliding window (the window size is adaptively adjusted according to the link dynamics: 5 sampling periods for satellite links (high stability) and 10 sampling periods for shortwave links (high dynamics)) to provide high-quality input for the prediction model.

[0157] (2) Task identification and dynamic mapping of requirements.

[0158] Specifically, the task parsing module can parse and classify various communication task requirements from the smart grid and determine a dynamic weight set. For example, the task parsing module identifies the current task type based on application identifiers, data packet tags, and user-preset priorities, and dynamically outputs weight requirements for latency, bandwidth, stability, and security according to the task type and execution stage.

[0159] Furthermore, the task requirement type can be identified from multiple dimensions: the task parsing module obtains task identifiers (such as "voice call", "video transmission", "emergency command") through the application layer API, parses the DSCP markers of the data packets (such as EF marker for real-time voice, AF4 for high-priority files), and combines them with user-preset priorities (levels 0-5, with level 5 being the highest); after determining the task requirement type, a dynamic weight set can be output based on the task-weight mapping table.

[0160] (3) Link scoring and weight adaptive adjustment, and link quality trend prediction and switching decision.

[0161] The processor, based on information provided by the state awareness module and the task parsing module, enables adaptive adjustment of link scoring and weight, as well as prediction of link quality trends and switching decisions.

[0162] For example, the processor uses a normalized model to calculate the overall score for each link. This overall score is calculated based on dynamic weights and the normalized values ​​of each link's parameters. The sum of the dynamic weights is 1, and they are adaptively adjusted according to the triggering conditions. In the normalized model, the parameters are normalized using the min-max method, with positive indicators taking their original values ​​and negative indicators taking 1 minus their original values.

[0163] For example, the processor uses a lightweight hybrid prediction model to predict the short-term trends of key parameters of the current link, determining whether the predicted values ​​will degrade to below the threshold corresponding to the task within a preset time window. The preset time window is adaptively adjusted based on the task's tolerance for latency. If the prediction indicates deterioration and there is a target link with a higher overall score, parallel pre-connection is performed on the target link according to the score priority.

[0164] Further steps may include:

[0165] ① Normalization processing: Min-max normalization (mapped to [0,1]) is applied to each index (i.e., state parameters), where positive indices (SNR, anti-interference level) take the original value, and negative indices (PLR, RTT) take 1 minus the original value;

[0166] ② Determine the comprehensive score based on the dynamic scoring model;

[0167] Among them, the weight adjustment trigger conditions can be set: task type switching (such as switching from voice to file transfer), sudden change in link status (a certain indicator deviates from the sliding window average by more than 20%), and manual user intervention (adjusting priority via terminal buttons).

[0168] ③ Trend prediction model: A lightweight hybrid model (ARIMA captures linear trends + random forest corrects nonlinear perturbations) is adopted. The model parameters are trained offline using historical data and fine-tuned online (updated every 10 seconds). Short-term predictions are made for key indicators (such as SNR and RTT). The time window for this short-term trend prediction is 200~500ms. The prediction window is adaptively adjusted according to the task latency requirements: 200ms for voice tasks (latency tolerance 200ms) and 500ms for file transfer (latency tolerance 500ms).

[0169] ④ Degradation judgment: If the predicted value meets the conditions of "SNR < task corresponding threshold (e.g., voice requires SNR ≥ 10dB, emergency command requires SNR ≥ 8dB)" or "RTT > task maximum tolerable latency", and there are other links whose predicted scores are ≥ current link scores + handover gain threshold (5%, to avoid frequent handover), then handover preparation is triggered.

[0170] (4) Parallel pre-connection and seamless switching execution, and a preset security-performance balance mechanism.

[0171] The multi-mode communication module supports multiple communication protocols and interfaces, enabling rapid and seamless switching between different communication links based on processor switching instructions. It maintains stable communication connections across various network environments, ensuring real-time data transmission and reliable system operation. For example, the multi-mode communication module hides the pre-connection process within the current communication gap, and after pre-connection is complete, it selects between multi-path TCP or packet mirroring strategies based on task type to achieve seamless switching.

[0172] Furthermore, prioritize the pre-connection: sort the target links in descending order of their scores, and prioritize the links with the highest scores and stable predictions (index fluctuation ≤5%) for pre-connection (including channel negotiation, encrypted handshake, and routing table pre-configuration). The pre-connection process is completed using the current communication gap (such as the packet sending interval) and does not occupy the main link bandwidth.

[0173] Optionally, a seamless switching strategy can be pre-configured: Real-time tasks (voice, emergency commands): Multi-path TCP (MPTCP) is used, and the interruption is 0~30ms during switching by fast migration of sub-streams; Real-time tasks (file transfer): Data packet mirroring transmission is used (the first N data packets are sent simultaneously by the master and slave links, N = window size / packet length), and the main link transmission is stopped after the target link is confirmed to be available, with an interruption of ≤50ms.

[0174] To ensure the security of data transmitted in the link and improve communication security, a suitable encryption algorithm can be adopted based on a preset security-performance balance mechanism (i.e., a dynamic encryption strategy). Furthermore, the security-performance balance mechanism may include:

[0175] ① Adaptive selection of encryption algorithm: Based on the sensitivity of the task (level 0-5, with level 5 being the highest) and the link latency, ChaCha20 (encryption latency <5ms) is used for high-sensitivity + low-latency links (such as emergency commands + satellite links), and AES-256 / SM4 (encryption strength ≥256 bits) is used for high-sensitivity + high-latency links (such as emergency commands + shortwave links).

[0176] ② Fast key negotiation: A temporary session key (TSK) is generated based on the pre-shared root key (PSK). During switching, only the validity of the TSK needs to be verified (time <10ms), avoiding the need for a full key exchange (time >100ms).

[0177] (5) Power consumption optimization and adaptive power adjustment.

[0178] The power management module dynamically adjusts the monitoring power consumption of each link in the multi-mode communication module based on information such as the urgency of the communication task and system load, ensuring that the device's battery life is maximized without affecting communication quality. For example, the power management module can perform tiered low-power monitoring of non-primary links and dynamically adjust the monitoring frequency and primary link transmission rate based on the remaining battery power.

[0179] Furthermore, the hierarchical management of each candidate link can be achieved as follows:

[0180] ① High-scoring candidate link (Score ≥ 80% of the current link): Low-power monitoring (wakes up every 2 seconds, sampling period 500ms, power consumption reduced to 30% of full power);

[0181] ② Low-scoring link (Score < 50% of current link): Ultra-low frequency monitoring (wake up every 10s, sampling period 100ms, power consumption reduced to 10% of full power).

[0182] In addition, the power consumption monitoring strategy can be dynamically adjusted based on the power threshold:

[0183] ① Battery level ≥ 30%: Monitor according to the above strategy;

[0184] ②15%≤Battery<30%: Only the top 2 candidate links in the score will be kept under monitoring, and the main link transmission rate will be reduced by 50% (e.g., from 2Mbps to 1Mbps).

[0185] ③ Battery level <15%: Maintain only the main link communication and disable listening on all candidate links.

[0186] The communication link switching method provided in this application utilizes multi-mode fusion scheduling technology that actively predicts link degradation, dynamically adapts to task requirements, and optimizes terminal power consumption to achieve seamless switching and efficient resource utilization of communication links in complex environments. Furthermore, by predicting link degradation in advance within 200-500ms, combined with parallel pre-connection and differentiated switching strategies, communication interruption time is reduced to meet the real-time task requirements of voice, emergency commands, etc. A dynamic scoring model is used to match the differentiated requirements of tasks for latency, bandwidth, and security, improving task completion efficiency and task adaptability compared to existing fixed-priority strategies. The combination of tiered low-power monitoring and dynamic power adjustment extends the battery life of terminal devices. Adaptive selection of encryption algorithms and fast key negotiation reduce switching encryption latency while ensuring the security of highly sensitive tasks, balancing security and performance requirements. It is compatible with multi-mode links such as satellite, cellular, and shortwave, and through dynamic prediction and parameter adjustment, it can work stably in complex environments such as mountainous areas, oceans, and disaster areas, improving environmental adaptability.

[0187] Figure 5 A schematic diagram of the communication link switching device provided in this application is shown below. Figure 5 As shown, the communication link switching device 50 provided in this embodiment includes:

[0188] The acquisition module 501 is used to acquire the task requirement type and the first real-time status parameter of the current link. The task requirement type is used to indicate the performance requirements of the current communication task on the communication link.

[0189] The prediction module 502 is used to input the first real-time state parameters into the trend prediction model to obtain the predicted state parameters within the prediction time window. The trend prediction model is a model pre-built and trained based on the historical state parameter set to analyze the trend of link state changes. The prediction time window is determined based on the task requirement type.

[0190] The judgment module 503 is used to determine whether the current link meets the preset degradation conditions based on the task requirement type and the predicted state parameters.

[0191] The connection module 504 is used to perform a pre-connection operation on the target candidate link if the current link meets the preset degradation conditions. The target candidate link is determined from the preset candidate link set based on the task requirement type.

[0192] The switching module 505 is used to switch the current communication task from the current link to the target candidate link after the pre-connection operation is completed.

[0193] In one possible implementation, the judgment module 503 can also be used to: determine the performance threshold corresponding to the task requirement type; compare the predicted state parameters with the performance threshold to determine whether the current link meets the preset degradation conditions.

[0194] In one possible implementation, the judgment module 503 can also be used to: obtain the second real-time status parameter of each candidate link in the preset candidate link set; determine the comprehensive score of each candidate link based on the task requirement type and the second real-time status parameter of each candidate link; and determine the target candidate link from the preset candidate link set based on all the comprehensive scores.

[0195] In one possible implementation, the judgment module 503 can also be used to: obtain a dynamic weight set according to the task requirement type, the dynamic weight set including the weights corresponding to multiple state parameters; for each candidate link, determine a normalization strategy based on the task requirement type; process the second real-time state parameters of the candidate link based on the normalization strategy to obtain a normalized parameter set; and determine the comprehensive score of the candidate link based on the normalized parameter set and the dynamic weight set.

[0196] In one possible implementation, the judgment module 503 can also be used to: for each comprehensive score, determine whether the comprehensive score is greater than or equal to a preset handover score, the preset handover score being determined based on the comprehensive score of the current link and a preset handover gain threshold; if the comprehensive score is greater than or equal to the preset handover score, then the candidate link in the preset candidate link set corresponding to the comprehensive score is determined as the target candidate link.

[0197] In one possible implementation, the judgment module 503 can also be used to: determine the comprehensive score of the current link based on the task requirement type and the first real-time status parameter; classify each candidate link based on the comprehensive score of each candidate link and the comprehensive score of the current link, and determine the power consumption monitoring strategy of each candidate link, wherein the power consumption monitoring strategy includes the monitoring wake-up cycle and / or the monitoring power consumption level.

[0198] In one possible implementation, the judgment module 503 can also be used to: determine a high-score ratio threshold and a low-score ratio threshold based on the comprehensive score of the current link; for each candidate link, if the comprehensive score of the candidate link is greater than or equal to the high-score ratio threshold, then the candidate link is determined to be a high-score candidate link, wherein the power consumption monitoring strategy of the high-score candidate link is a first power consumption monitoring strategy, and the monitoring wake-up period corresponding to the first power consumption monitoring strategy is a first period; if the comprehensive score of the candidate link is less than the low-score ratio threshold, then the candidate link is determined to be a low-score candidate link, wherein the power consumption monitoring strategy of the low-score candidate link is a second power consumption monitoring strategy, and the monitoring wake-up period corresponding to the second power consumption monitoring strategy is a second period, wherein the second period is greater than the first period.

[0199] In one possible implementation, the connection module 504 can also be used to: initiate a pre-connection request to the target candidate link if the current link meets the preset degradation conditions; and perform link-layer link establishment, security authentication, and routing table pre-configuration operations with the target candidate link.

[0200] In one possible implementation, the switching module 505 can also be used to: determine a switching strategy based on the task requirement type, the switching strategy including a multi-path TCP migration strategy or a packet mirroring transmission strategy; according to the determined switching strategy, forward the data stream of the current communication task to the target candidate link and interrupt data transmission with the current link.

[0201] The communication link switching device provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0202] Figure 6 A schematic diagram of the structure of the electronic device provided in this application. Figure 6 As shown, the electronic device 60 provided in this embodiment includes at least one processor 601 and a memory 602. Optionally, the device 60 further includes a communication component 603. The processor 601, memory 602, and communication component 603 are connected via a bus 604.

[0203] In a specific implementation, at least one processor 601 executes computer execution instructions stored in memory 602, causing at least one processor 601 to perform the above-described method.

[0204] The specific implementation process of processor 601 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0205] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0206] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0207] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0208] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0209] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0210] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0211] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0212] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0213] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0214] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0215] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0216] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0217] It should be understood that the terms “comprising” and “having”, and any variations thereof, in the specification, claims, and accompanying drawings of this application are intended to cover but not exclude inclusion. For example, a product or device that includes a series of components is not necessarily limited to those components that are explicitly listed, but may include other components that are not explicitly listed or that are inherent to such product or device.

[0218] As used in this application, the term "module" means any known or subsequently developed hardware, software, firmware, artificial intelligence, fuzzy logic, or combination of hardware and / or software code capable of performing the functions associated with that element.

[0219] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A communication link switching method, characterized in that, include: Obtain the task requirement type and the first real-time status parameter of the current link, wherein the task requirement type is used to indicate the performance requirements of the current communication task on the communication link; The first real-time state parameter is input into the trend prediction model to obtain the predicted state parameter. The trend prediction model is a model pre-built and trained based on the historical state parameter set for analyzing the trend of link state changes. Based on the task requirement type and the predicted state parameters, determine whether the current link meets the preset degradation conditions; If the current link meets the preset degradation conditions, a pre-connection operation is performed on the target candidate link, which is determined from the preset candidate link set based on the task requirement type; After the pre-connection operation is completed, the current communication task is switched from the current link to the target candidate link.

2. The method according to claim 1, characterized in that, The trend prediction model includes a linear prediction sub-model and a nonlinear correction sub-model; wherein, the linear prediction sub-model is used to perform linear trend analysis on the first real-time state parameter to obtain an initial prediction result; the nonlinear correction sub-model is used to perform nonlinear perturbation correction on the initial prediction result to obtain the predicted state parameter.

3. The method according to claim 2, characterized in that, The linear prediction sub-model is an autoregressive integral moving average model, and the nonlinear correction sub-model is a random forest model.

4. The method according to claim 1, characterized in that, The step of determining whether the current link meets the preset degradation conditions based on the task requirement type and the predicted state parameters includes: Determine the performance threshold corresponding to the task requirement type; The predicted state parameters are compared with the performance threshold to determine whether the current link meets the preset degradation condition.

5. The method according to any one of claims 1-4, characterized in that, Before performing the pre-connection operation on the target candidate link, the method further includes: Obtain the second real-time status parameter of each candidate link in the preset candidate link set; Based on the task requirement type and the second real-time status parameter of each candidate link, a comprehensive score for each candidate link is determined. Based on all the comprehensive scores, the target candidate link is determined from the preset candidate link set.

6. The method according to claim 5, characterized in that, The determination of a comprehensive score for each candidate link based on the task requirement type and the second real-time status parameter of each candidate link includes: Based on the task requirement type, a dynamic weight set is obtained, which includes weights corresponding to multiple state parameters. For each candidate link, a normalization strategy is determined based on the task requirement type; Based on the normalization strategy, the second real-time state parameters of the candidate links are processed to obtain a normalized parameter set; The comprehensive score of the candidate link is determined based on the normalized parameter set and the dynamic weight set.

7. The method according to claim 5, characterized in that, The step of determining the target candidate link from the preset candidate link set based on all comprehensive scores includes: For each comprehensive score, it is determined whether the comprehensive score is greater than or equal to a preset handover score, which is determined based on the score of the current link and a preset handover gain threshold. If the overall score is greater than or equal to the preset switching score, then the candidate link in the preset candidate link set that corresponds to the overall score is determined as the target candidate link.

8. The method according to claim 5, characterized in that, The method further includes: Based on the task requirement type and the first real-time status parameter, determine the comprehensive score of the current link; Based on the comprehensive score of each candidate link and the comprehensive score of the current link, each candidate link is classified and a power consumption monitoring strategy for each candidate link is determined. The power consumption monitoring strategy includes the monitoring wake-up cycle and / or the monitoring power consumption level.

9. The method according to claim 8, characterized in that, The process of classifying each candidate link based on its comprehensive score and the comprehensive score of the current link, and determining the power consumption monitoring strategy for each candidate link, includes: Based on the comprehensive score of the current link, determine the high score ratio threshold and the low score ratio threshold; For each candidate link, if the comprehensive score of the candidate link is greater than or equal to the high score ratio threshold, then the candidate link is determined to be a high score candidate link. The power consumption monitoring strategy of the high score candidate link is a first power consumption monitoring strategy, and the monitoring wake-up period corresponding to the first power consumption monitoring strategy is a first period. If the overall score of the candidate link is less than the low score ratio threshold, the candidate link is determined to be a low score candidate link. The power consumption monitoring strategy of the low score candidate link is a second power consumption monitoring strategy, and the monitoring wake-up period corresponding to the second power consumption monitoring strategy is a second period, wherein the second period is longer than the first period.

10. The method according to any one of claims 1-4, characterized in that, If the current link meets the preset degradation condition, then a pre-connection operation is performed on the target candidate link, including: If the current link meets the preset degradation conditions, a pre-connection request is initiated to the target candidate link; Perform link-layer link establishment, security authentication, and routing table pre-configuration operations with the target candidate link.

11. The method according to any one of claims 1-4, characterized in that, The step of switching the current communication task from the current link to the target candidate link includes: Based on the task requirement type, a switching strategy is determined, which includes a multi-path TCP migration strategy or a packet mirroring transmission strategy. According to the determined switching strategy, the data stream of the current communication task is forwarded to the target candidate link, and data transmission with the current link is interrupted.

12. A communication link switching device, characterized in that, include: The acquisition module is used to acquire the task requirement type and the first real-time status parameter of the current link, wherein the task requirement type is used to indicate the performance requirements of the current communication task on the communication link. The prediction module is used to input the first real-time state parameter into the trend prediction model to obtain the predicted state parameter within the prediction time window. The trend prediction model is a model pre-built and trained based on the historical state parameter set for analyzing the trend of link state changes. The prediction time window is determined based on the task requirement type. The judgment module is used to determine whether the current link meets the preset degradation conditions based on the task requirement type and the predicted state parameters. A connection module is used to perform a pre-connection operation on a target candidate link if the current link meets a preset degradation condition. The target candidate link is determined from a preset candidate link set based on the task requirement type. The switching module is used to switch the current communication task from the current link to the target candidate link after the pre-connection operation is completed.

13. An electronic device, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-11.

Images