Power network-communication network transient and dynamic response collaborative analysis method based on data driving-mechanism fusion

Abstract

The invention relates to the technical field of power grid systems, in particular to a power grid-communication network transient dynamic response collaborative analysis method based on data driving-mechanism fusion, which comprises the following steps: firstly, collecting power transient dynamic, source network load storage response and communication network state data, and carrying out space-time synchronization and exception processing preprocessing; extracting power failure, source network load storage response and communication response characteristics, and constructing a coupling index and time sequence incidence matrix; fusing the physical mechanism model and the data driving model to establish a dynamic coupling model; and finally, the model is applied to scenes such as new energy station grid connection and transformer substation faults, and an optimization scheme is output. The power and communication network transient and dynamic response collaborative analysis is realized, the fault early warning accuracy and the resource optimization efficiency are improved, and support is provided for safe and stable operation of a power grid.

Description

Technical Field

[0001] This invention relates to the field of power grid system technology, specifically to a collaborative analysis method for transient dynamic response of power grid and communication network based on data-driven and mechanism-integrated approaches. Background Technology

[0002] With the rapid development of smart grids and the energy internet, the deep integration of power systems and communication networks has become a core feature of modern power grid construction. The power grid, as the physical carrier of energy transmission, and the communication network, as the nerve center for control commands and information exchange, together constitute the critical infrastructure supporting the operation of society. However, while this coupling enhances system intelligence, it also brings new technological challenges.

[0003] Traditional power system analysis and communication network design often employ independent modeling methods, lacking a deep understanding of cross-network coupling effects. Power system transient processes (such as faults and switching operations) can affect communication performance through power quality and equipment status; conversely, communication network delays and packet loss can also hinder the timely execution of power control commands. This bidirectional coupling is particularly pronounced during transient processes.

[0004] The current mainstream analytical methods mainly suffer from the following technical bottlenecks: (1) Limitations of isolated modeling: Power system analysis is usually based on physical mechanism models (such as differential-algebraic equations), but it is difficult to quantify the impact of communication delay on control performance; communication network performance evaluation focuses on indicators such as bandwidth and packet loss rate, and lacks characterization of power transient disturbance sensitivity; cross-network coupling relationship is often approximated by simplifying assumptions (such as fixed delay), which cannot reflect the actual dynamic characteristics. (2) Insufficient data utilization: There are differences in the spatiotemporal scale between power SCADA, PMU and other data and communication network management data, and there is a lack of effective fusion mechanism; traditional methods are not deep enough in mining the massive heterogeneous data generated in the transient process; and the complementary advantages of data-driven and mechanism models have not been established. (3) Insufficient collaborative control capabilities: It is difficult to achieve collaborative optimization of power restoration and communication routing in fault scenarios; new scenarios such as new energy grid connection lack dynamic resource allocation strategies across networks; existing early warning mechanisms have low sensitivity to identify the chain risk of "power disturbance-communication bottleneck".

[0005] With the high proportion of renewable energy integration and the widespread adoption of power electronic equipment, power system transient processes are exhibiting faster time scales and stronger nonlinear characteristics. Simultaneously, the application of new communication technologies such as 5G and TSN brings new opportunities and challenges to power grid control. Against this backdrop, there is an urgent need to overcome the limitations of traditional analytical methods and establish a comprehensive analytical approach capable of revealing the inherent coupling laws between the power grid and communication network. Summary of the Invention

[0006] The purpose of this invention is to provide a data-driven, mechanism-integrated collaborative analysis method for transient dynamic responses of power grids and communication networks, in order to solve the problems of the power system analysis methods proposed in the background art, which are difficult to effectively characterize the interaction between power transient processes and communication dynamic responses, and have problems such as isolated modeling, insufficient data utilization, and insufficient collaborative control capabilities.

[0007] To achieve the above objectives, the present invention provides the following technical solution: The collaborative analysis method for transient dynamic response of power grid and communication network based on data-driven and mechanism-integrated approaches includes the following steps: S1 Multi-Source Temporary Dynamic Data Acquisition and Preprocessing: Acquires temporary dynamic power data, source-grid-load-storage response data, and communication network status data. Time alignment is achieved using IEEE1588 or BeiDou clock synchronization to ensure that the timestamp error between power and communication data is <1ms. Establishes spatial mapping relationships of equipment based on the GIS coordinates of a single communication and power map. Impulse noise is removed by sliding window filtering, and power data during communication link outages is completed using interpolation. S2 Temporary Dynamic Feature Joint Extraction and Fusion: Extract power system fault features, source-grid-load-storage response features, and communication network response features, and construct coupling index and time series correlation matrix. The coupling index is defined as communication impact factor = power control command delay / maximum allowable delay. The time series correlation matrix is ​​an m×n matrix, where rows represent power events, columns represent communication responses, and elements are event co-occurrence probabilities. The temporal correlation matrix is ​​an m×n matrix, whose elements are... Indicates power events Communication response The co-occurrence probability is calculated using the following formula: ; in For the event With response The number of co-occurrences within a time window is used to quantify the causal strength between events using conditional probability. ; when When the value is greater than 0.8, an electrical event is determined. Communication response It has a strong triggering effect; S3 Dynamic Coupling Modeling: Establishing a coupling model based on physical mechanisms and a data-driven machine learning model. The expression for the physical mechanism coupling model is as follows: ; in Due to communication delay, The power state vector includes voltage and power. This represents the communication state at the previous moment; the data-driven machine learning model includes a bidirectional LSTM network and a graph neural network (GNN). S4 Collaborative Analysis and Application: The constructed dynamic coupling model is applied to the collaborative scenario of power grid and communication network, and the analysis results and optimization schemes are output.

[0008] Preferably, in step S1, the power transient data includes voltage / current sags, frequency fluctuations, and power angle changes, with a sampling frequency of 100Hz-1kHz, and the sampling devices are PMU and fault recorder; the source-grid-load-storage response data includes energy storage charging and discharging power and load regulation, with a sampling frequency of 10Hz-100Hz, and the sampling devices are energy storage BMS and load controller; the communication network status data includes link delay, bandwidth utilization, and packet loss rate, with a sampling frequency of 1Hz-10Hz, and the sampling devices are network management system (NMS) and OTN.

[0009] Preferably, the power transient data includes voltage / current sags, frequency fluctuations, and power angle changes, with a sampling frequency of 100Hz-1kHz, and the sampling devices are PMU and fault recorders; the communication network status data includes link delay, bandwidth utilization, and packet loss rate, with a sampling frequency of 1Hz-10Hz, and the sampling devices are network management systems (NMS) and OTN.

[0010] Preferably, in step S1, the sliding window filtering adopts Kalman filtering, and the equipment spatial mapping relationship includes the correspondence between substations and communication base stations, and between transmission lines and optical cable segments.

[0011] Preferably, in step S2, the power system fault characteristics include voltage dip depth and rate of change of frequency (ROCOF), wherein the voltage dip depth is less than 60% of the rated voltage; the source-grid-load-storage response characteristics include energy storage response time and load regulation rate, wherein the energy storage response time is <50ms; the communication network response characteristics include delay mutation and bandwidth squeezing, wherein the delay mutation is the communication link delay increasing from 20ms to 100ms at the time of the fault, and the bandwidth squeezing is the communication link utilization rate exceeding 80%.

[0012] Preferably, in step S3, the inputs of the bidirectional LSTM network are a power transient sequence and a communication state sequence. The power transient sequence is the voltage waveform 100ms before the fault, and the bidirectional LSTM network is used to predict the delay changes of the communication link after the fault.

[0013] As a preferred option, the collaborative scenario in step S4 is grid connection of new energy power plants. The specific analysis process includes: collecting transient current of wind turbine converters, energy storage charging and discharging power, and bandwidth data of power plant communication base stations; using the GARCH model to analyze the time-varying correlation between power fluctuations and communication delays; establishing a dynamic allocation optimization model for new energy output fluctuations and communication bandwidth, and determining that the minimum guaranteed bandwidth for wind turbine control commands is 20Mbps.

[0014] As a preferred option, the collaborative scenario in step S4 is a substation fault. The specific analysis process includes: inputting data on the sudden drop in bus voltage of a 220kV substation and data on the interruption of the main optical cable from the substation to the dispatch center; outputting the optimal backup route and bandwidth allocation scheme through a dynamic coupling model. The optimal backup route is an optical cable transfer through an adjacent substation with a delay compensation of 50ms; the bandwidth allocation scheme prioritizes the transmission of fault waveform data at 10Mbps and suspends non-real-time data.

[0015] Preferably, when the communication influence factor of the coupling index is greater than 1, a collaborative early warning mechanism is triggered. The collaborative early warning mechanism includes sending early warning signals to the power dispatch center and the communication network management center to indicate potential control failure risks.

[0016] As a preferred option, in step S3, the coupling model based on physical mechanisms can quantify the linear relationship between power status and communication indicators. When the voltage of a substation drops by 10%, it can be calculated that the processing delay of the station's communication switch due to power fluctuations increases by 20%.

[0017] Compared with the prior art, the beneficial effects of the present invention are: This invention presents a data-driven, mechanism-integrated power grid-communication network transient response collaborative analysis method. By combining physical mechanism models with data-driven algorithms, it achieves deep collaboration between power system transient processes and communication network responses. Its core advantages are described in detail below from multiple dimensions.

[0018] 1. High degree of integration and model accuracy The core advantage of this method lies in its organic integration of data-driven technology and physical mechanism models, overcoming the limitations of traditional single models in complex scenarios. Through multi-source transient dynamic data acquisition and spatiotemporal synchronous preprocessing, the accuracy and consistency of the data are ensured. This fusion approach not only improves the model's ability to reveal the coupling patterns between power disturbances and communication bottlenecks, but also achieves high-precision predictions through dynamic coupling modeling, significantly enhancing the reliability of transient response analysis.

[0019] 2. Real-time collaboration and rapid response capabilities This method possesses millisecond-level real-time collaborative analysis capabilities. By defining coupling indicators such as communication impact factors, it achieves a dynamic correlation between power disturbances and communication performance. For example, in substation fault scenarios, the model can identify voltage drops and fiber optic cable interruptions within 0.5 seconds and automatically optimize communication routing and bandwidth allocation. This rapid response mechanism effectively avoids system instability caused by control command delays, thus improving grid resilience.

[0020] 3. Multi-scenario adaptability and optimized efficiency This invention is applicable to various typical scenarios, including grid connection of renewable energy power plants and substation faults, demonstrating strong generalization capabilities. In the renewable energy power plant scenario, the time-varying correlation between power fluctuations and communication delays is analyzed using a GARCH model to dynamically allocate bandwidth. In the substation fault scenario, graph neural networks are used to learn fault propagation patterns to achieve resource optimization. This scenario-adaptive design reduces manual intervention and improves operation and maintenance efficiency.

[0021] 4. Data-driven intelligent decision support Based on bidirectional LSTM and GNN machine learning algorithms, transient features can be learned from historical data to generate data-driven decision-making recommendations. For example, the co-occurrence probability of power events and communication responses can be quantified through time-series correlation matrices, providing a basis for preventative control. Furthermore, the model output is visualized, enabling maintenance personnel to quickly formulate recovery strategies and reducing the risk of human error.

[0022] 5. Improved system reliability and economy Through collaborative analysis, this invention effectively avoids power control failures caused by communication bottlenecks and reduces the probability of fault propagation. In the example, the communication routing optimization and dynamic bandwidth allocation scheme avoids unnecessary data transmission, saves communication resources, and ensures the continuity of critical services. Long-term application can reduce system downtime losses and improve the economic efficiency of power grid operation.

[0023] This method, centered on data-driven and mechanism-based integration, achieves deep collaboration between the transient responses of power grids and communication networks through innovative modeling frameworks and algorithm design. Its advantages lie in high precision, real-time performance, multi-scenario adaptability, intelligent decision-making, and system economy, providing key technical support for building a resilient smart grid. This method can be further extended to multi-energy system collaboration, promoting the development of the energy internet. Attached Figure Description

[0024] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are explained in detail together with the embodiments of the invention, but do not constitute a limitation thereof.

[0025] Figure 1 This is a schematic diagram of the process of the present invention; Figure 2 This is a monitoring table of key parameters for grid connection of new energy power plants according to an embodiment of the present invention; Figure 3 This is a comparison table of key data before and after a substation fault according to an embodiment of the present invention. Detailed Implementation

[0026] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0027] I. Key Modeling Steps and Methods of the Data-Driven-Mechanism Fusion-Based Collaborative Analysis Method for Temporary Dynamic Response of Power Grid and Communication Network in this Invention (1) Acquisition and preprocessing of multi-source transient dynamic data 1) Data types and sources Data categories Specific indicators sampling frequency Typical equipment Temporary power data Voltage / current sags, frequency fluctuations, power angle changes 100Hz 1kHz PMU, fault recorder Source-grid-load-storage response Energy storage charging and discharging power, load regulation 10Hz 100Hz Energy storage BMS, load controller Communication network status Link latency, bandwidth utilization, packet loss rate 1Hz 10Hz Network Management System (NMS), OTN 2) Spatiotemporal synchronization preprocessing Time alignment: IEEE1588 or BeiDou clock synchronization is adopted to ensure that the timestamp error of power and communication data is <1ms; Spatial mapping: Based on the coordinates of a single GIS map of communication and power, establish the correspondence between equipment (such as communication base stations in substations and optical cable sections of transmission lines). Abnormal data processing: Impulse noise is removed by sliding window filtering (such as Kalman filtering), and power data during communication link outages is completed by interpolation.

[0028] (2) Joint extraction and fusion of transient dynamic features 1) Temporal characteristics of power systems Fault characteristics: voltage drop depth (e.g., below 60% of rated voltage), rate of change of frequency (ROCOF); Source-grid-load-storage response characteristics: energy storage response time (<50ms), load regulation rate (such as the power change rate of air conditioning clusters).

[0029] 2) Communication network response characteristics Delay mutation: The communication link delay jumps at the moment of failure (e.g., from 20ms to 100ms). Bandwidth squeeze: Burst transmission of power control commands causes communication link utilization to exceed a threshold (e.g., >80%).

[0030] 3) Feature Fusion Construction Coupling index: Define communication impact factor = power control command delay / maximum allowable delay. When the factor > 1, a coordinated early warning is triggered. Temporal correlation matrix: Construct an m×n matrix where rows represent power events (such as fault types), columns represent communication responses (such as link switching times), and elements are the co-occurrence probabilities of events.

[0031] (3) Dynamic coupling modeling method 1) Coupling model based on physical mechanism Power communication mapping model: ; in: For communication metrics (such as delay) at time t. It is the power state vector (voltage, power). This represents the communication status of the previous time period.

[0032] Case: When the voltage of a substation drops by 10%, the processing delay of the station's communication switch increases by 20% due to power fluctuations. The linear relationship between voltage and delay can be quantified through mechanism modeling.

[0033] 2) Data-driven machine learning models Bidirectional LSTM network: Input power transient sequence (such as voltage waveform 100ms before a fault) and communication state sequence to predict the delay change of communication link after a fault; Graph Neural Networks (GNNs): Using power equipment communication nodes as graph nodes and physical connections and data interactions as edges, they learn the propagation patterns of faults in power communication networks.

[0034] II. Typical Application Scenarios and Modeling Examples Scenario: Communication and power coordination during grid connection of new energy power plants

[0035] Modeling steps: 1) Collect transient current of wind turbine converter, energy storage charging and discharging power, and bandwidth data of station communication base station; 2) Use the GARCH model to analyze the time-varying correlation between power fluctuations and communication delays; 3) Establish an optimization model for dynamic allocation of communication bandwidth for fluctuations in new energy output, and determine the minimum guaranteed bandwidth (e.g., 20Mbps) for wind turbine control commands.

[0036] (4) Monitoring table of key parameters for grid connection of new energy power plants, as shown in the table below:

[0037] (5) such as Figure 2The data showcases the monitoring of key parameters during grid connection of new energy power plants, including transient current of wind turbine converters, energy storage charging and discharging power, bandwidth utilization of the power plant's communication base station, and communication latency over time. The graph shows that all parameters reach their peak values ​​between 30 and 40 seconds, then gradually decrease to a stable state. Example: Communication route optimization during substation faults

[0038] 1) Model input: Power data: The bus voltage of the 220kV substation suddenly dropped (from 1.0pu to 0.6pu within 0.5s). Communication data: The main optical cable from the substation to the dispatch center is interrupted (delay increased from 30ms to ∞).

[0039] 2) Model output: Optimal backup route: Optical cable relay via adjacent substation B, with delay compensation to 50ms; Bandwidth allocation scheme: Prioritize the transmission of fault recording data (10Mbps), and suspend non-real-time data (such as equipment status inspection).

[0040] (3) A comparison table of key data before and after the substation fault is shown below:

[0041] (4) such as Figure 3 The presentation compares key data points before and after a substation fault, including bus voltage, communication delay, data transmission rate, and bandwidth utilization. The charts clearly show that communication delay becomes infinitely large during a fault, but is effectively compensated for after routing optimization, ensuring the transmission of fault recording data and demonstrating the effectiveness of the collaborative analysis method.

[0042] The proposed data-driven, mechanism-integrated approach for collaborative analysis of transient responses in power grids and communication networks has significant advantages in the field of smart grid and communication network integration. This method, by combining physical mechanism models with data-driven algorithms, achieves deep collaboration between power system transient processes and communication network responses. Its core advantages are elaborated in detail below from multiple dimensions.

[0043] 1. High degree of integration and model accuracy The core advantage of this method lies in its organic integration of data-driven technology and physical mechanism models, overcoming the limitations of traditional single models in complex scenarios. Through multi-source transient dynamic data acquisition (such as voltage drop depth and communication delay abrupt changes) and spatiotemporal synchronous preprocessing (time alignment error <1ms), the accuracy and consistency of the data are ensured. This fusion approach not only improves the model's ability to reveal the coupling law between power disturbances and communication bottlenecks, but also achieves high-precision prediction through dynamic coupling modeling (such as bidirectional LSTM and graph neural networks), significantly enhancing the reliability of transient response analysis.

[0044] 2. Real-time collaboration and rapid response capabilities This method possesses millisecond-level real-time collaborative analysis capabilities. By defining coupling indicators such as communication impact factors (e.g., triggering an early warning when the factor > 1), it achieves a dynamic correlation between power disturbances and communication performance. For example, in substation fault scenarios, the model can identify voltage drops and fiber optic cable interruptions within 0.5 seconds and automatically optimize communication routing (e.g., delay compensation to 50ms) and bandwidth allocation (prioritizing the transmission of fault waveform data). This rapid response mechanism effectively avoids system instability caused by control command delays, enhancing grid resilience.

[0045] 3. Multi-scenario adaptability and optimized efficiency This invention is applicable to various typical scenarios, including grid connection of renewable energy power plants and substation faults, demonstrating strong generalization capabilities. In the renewable energy power plant scenario, the time-varying correlation between power fluctuations and communication delays is analyzed using a GARCH model to dynamically allocate bandwidth (e.g., ensuring a minimum bandwidth of 20Mbps for wind turbine control commands). In the substation fault scenario, graph neural networks are used to learn fault propagation patterns to achieve resource optimization. This scenario-adaptive design reduces manual intervention and improves operation and maintenance efficiency.

[0046] 4. Data-driven intelligent decision support Based on machine learning algorithms such as bidirectional LSTM and GNN, transient features can be learned from historical data to generate data-driven decision-making recommendations. For example, the co-occurrence probability of power events and communication responses can be quantified through time-series correlation matrices, providing a basis for preventative control. Furthermore, the model outputs are visualized (such as a collaborative efficiency index), supporting maintenance personnel in quickly developing recovery strategies and reducing the risk of human error.

[0047] 5. Improved system reliability and economy Through collaborative analysis, this invention effectively avoids power control failures caused by communication bottlenecks and reduces the probability of fault propagation. In the example, the communication routing optimization and dynamic bandwidth allocation scheme avoids unnecessary data transmission, saves communication resources, and ensures the continuity of critical services (such as fault recording). Long-term application can reduce system downtime losses and improve the economic efficiency of power grid operation.

[0048] This method, centered on data-driven and mechanism-based integration, achieves deep collaboration between the transient responses of power grids and communication networks through innovative modeling frameworks and algorithm design. Its advantages lie in high precision, real-time performance, multi-scenario adaptability, intelligent decision-making, and system economy, providing key technical support for building a resilient smart grid. This method can be further extended to multi-energy system collaboration, promoting the development of the energy internet.

[0049] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A collaborative analysis method for transient dynamic response of power grid and communication network based on data-driven and mechanism-integrated approach, specifically including the following steps: S1 Multi-source transient dynamic data acquisition and preprocessing: Acquires power transient dynamic data, source-grid-load-storage response data and communication network status data, and uses IEEE1588 or Beidou clock synchronization to achieve time alignment, ensuring that the timestamp error of power and communication data is <1ms. Establish spatial mapping relationships for equipment based on a single GIS coordinate map of communication and power systems; Impulse noise is removed by sliding window filtering, and power data during communication link outages is completed by interpolation. S2 Temporary Dynamic Feature Joint Extraction and Fusion: Extract power system fault features, source-grid-load-storage response features, and communication network response features, and construct coupling index and time series correlation matrix. The coupling index is defined as communication impact factor = power control command delay / maximum allowable delay. The time series correlation matrix is ​​an m×n matrix, where rows represent power events, columns represent communication responses, and elements are event co-occurrence probabilities. The temporal correlation matrix is ​​an m×n matrix, whose elements are... Indicates power events With communication response The co-occurrence probability is calculated using the following formula: ； in For the event With response The number of co-occurrences within a time window is used to quantify the causal strength between events using conditional probability. ； when When the value is greater than 0.8, an electrical event is determined. Communication response It has a strong triggering effect; S3 Dynamic Coupling Modeling: Establishing a coupling model based on physical mechanisms and a data-driven machine learning model. The expression for the physical mechanism coupling model is as follows: ； in Due to communication delay, The power state vector includes voltage and power. This represents the communication state at the previous moment; the data-driven machine learning model includes a bidirectional LSTM network and a graph neural network (GNN). S4 Collaborative Analysis and Application: The constructed dynamic coupling model is applied to the collaborative scenario of power grid and communication network, and the analysis results and optimization schemes are output.

2. The collaborative analysis method for transient dynamic response of power grid and communication network based on data-driven and mechanism-integrated fusion as described in claim 1, characterized in that, In step S1, the power transient data includes voltage / current sags, frequency fluctuations, and power angle changes, with a sampling frequency of 100Hz-1kHz and sampling devices including a PMU and a fault recorder; the source-grid-load-storage response data includes energy storage charging and discharging power and load regulation, with a sampling frequency of 10Hz-100Hz and sampling devices including an energy storage BMS and a load controller; the communication network status data includes link delay, bandwidth utilization, and packet loss rate, with a sampling frequency of 1Hz-10Hz and sampling devices including a network management system (NMS) and an OTN.

3. The collaborative analysis method for transient dynamic response of power grid and communication network based on data-driven and mechanism-integrated fusion as described in claim 1, characterized in that, The transient power data includes voltage / current sags, frequency fluctuations, and power angle changes, with a sampling frequency of 100Hz-1kHz and sampling devices including a PMU and a fault recorder; the communication network status data includes link delay, bandwidth utilization, and packet loss rate, with a sampling frequency of 1Hz-10Hz and sampling devices including a network management system (NMS) and an OTN.

4. The collaborative analysis method for transient dynamic response of power grid and communication network based on data-driven and mechanism-integrated fusion as described in claim 1, characterized in that, In step S1, the sliding window filtering adopts Kalman filtering, and the equipment spatial mapping relationship includes the correspondence between substations and communication base stations, and between transmission lines and optical cable segments.

5. The collaborative analysis method for transient dynamic response of power grid and communication network based on data-driven and mechanism-integrated fusion as described in claim 1, characterized in that, In step S2, the power system fault characteristics include voltage sag depth and rate of change of frequency (ROCOF), wherein the voltage sag depth is less than 60% of the rated voltage; the source-grid-load-storage response characteristics include energy storage response time and load regulation rate, wherein the energy storage response time is <50ms; the communication network response characteristics include delay mutation and bandwidth squeezing, wherein the delay mutation is the communication link delay increasing from 20ms to 100ms at the time of the fault, and the bandwidth squeezing is the communication link utilization rate exceeding 80%.

6. The collaborative analysis method for transient dynamic response of power grid and communication network based on data-driven and mechanism-integrated fusion as described in claim 1, characterized in that, In step S3, the inputs of the bidirectional LSTM network are a power transient sequence and a communication state sequence. The power transient sequence is the voltage waveform 100ms before the fault, and the bidirectional LSTM network is used to predict the delay changes of the communication link after the fault.

7. The collaborative analysis method for transient dynamic response of power grid and communication network based on data-driven and mechanism-integrated fusion as described in claim 1, characterized in that, The collaborative scenario in step S4 is the grid connection of new energy power plants. The specific analysis process includes: collecting transient current of wind turbine converters, energy storage charging and discharging power, and bandwidth data of power plant communication base stations; using the GARCH model to analyze the time-varying correlation between power fluctuations and communication delays; establishing a dynamic allocation optimization model for new energy output fluctuations and communication bandwidth, and determining that the minimum guaranteed bandwidth for wind turbine control commands is 20Mbps.

8. The collaborative analysis method for transient dynamic response of power grid and communication network based on data-driven and mechanism-integrated fusion as described in claim 1, characterized in that, The collaborative scenario in step 4 is a substation fault. The specific analysis process includes: inputting data on the sudden drop in bus voltage of a 220kV substation and data on the interruption of the main optical cable from the substation to the dispatch center; outputting the optimal backup route and bandwidth allocation scheme through a dynamic coupling model. The optimal backup route is an optical cable transfer via the adjacent substation B, with a delay compensation of 50ms; the bandwidth allocation scheme prioritizes the transmission of fault waveform data at 10Mbps and suspends non-real-time data.

9. The collaborative analysis method for transient dynamic response of power grid and communication network based on data-driven and mechanism-integrated fusion as described in claim 1, characterized in that, In step S2, when the coupling index communication influence factor > 1, a collaborative early warning mechanism is triggered. The collaborative early warning mechanism includes sending early warning signals to the power dispatch center and the communication network management center to indicate potential control failure risks.

10. The collaborative analysis method for transient dynamic response of power grid and communication network based on data-driven and mechanism-integrated fusion as described in claim 1, characterized in that, In step S3, the coupling model based on physical mechanism can quantify the linear relationship between power status and communication indicators. When the voltage of a substation drops by 10%, it can be calculated that the processing delay of the station's communication switch due to power fluctuations increases by 20%.

Images