A frequency control data transmission system and method for an ultra-high voltage direct current sending end

Abstract

The application discloses a kind of for extra-high voltage DC sending end frequency control data transmission system and method, it is related to industrial internet technology field, acquisition multi-source operating data and execute normalization, construct operating state vector;State space is divided, and the core associated link is identified based on link call frequency weight, and the associated mapping table of physical state and link is established;Statistical history data generates link set probability matrix under the description of whole working condition;Utilize long short-term memory network to capture operating state evolution trend and construct prediction model, realize power grid situation advance perception;According to the prediction result, the state interval to which it belongs is positioned, and the target data transmission link set is generated by searching and integrating candidate link, to solve the transmission delay and resource mismatch problem caused by the disconnection of extra-high voltage DC system data transmission link and working condition.

Description

Technical Field

[0001] This invention relates to the field of industrial internet technology, specifically to a data transmission system and method for frequency control at the ultra-high voltage direct current (UHVDC) transmitter. Background Technology

[0002] In ultra-high voltage direct current (UHVDC) sending-end power grids, frequency control is a crucial link in ensuring the safety and stability of grid operation, and it highly depends on the highly reliable and low-latency transmission of control commands between various execution terminals. With the increasing complexity of power grid structures, building an efficient and robust data interaction mechanism to support dynamic frequency regulation has become an important means to improve the DC system's ability to cope with disturbances and ensure the frequency security of the sending-end power grid.

[0003] Traditional data transmission methods, when dealing with the multiple evolving states of UHVDC systems, lack in-depth correlation analysis between system characteristic changes and transmission link call frequencies. This leads to difficulties in ensuring the smooth operation of core interconnected links under specific operating conditions, affecting the transmission redundancy and response time of control commands in emergency situations. Furthermore, existing technologies struggle to proactively reserve link resources for potential operating condition transitions. Moreover, current statistical modeling methods are inadequate for analyzing conditional probabilistic correlations under multi-source heterogeneous operating parameters, resulting in discrepancies between link selection results and actual frequency control requirements. In extreme cases, data transmission delays may trigger chain reactions. Therefore, a new data transmission method for frequency control at the UHVDC sending end is needed. Summary of the Invention

[0004] The purpose of this invention is to provide a data transmission system and method for frequency control at the ultra-high voltage direct current (UHVDC) sending end, in order to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for transmitting frequency control data at the sending end of an ultra-high voltage direct current (UHVDC) transmission line, the method comprising: Step 1: By deploying sensor arrays at key nodes at the sending end, multi-source data reflecting the power grid operation status are acquired in real time. Linear normalization is performed to eliminate the influence of different dimensions on subsequent calculations, and the data are spliced ​​into a multi-dimensional operation status vector. The operation status vector is then stored in the historical database. Step 2: Perform a partitioning operation on the multidimensional parameter space composed of the running state vector to form several state intervals representing specific operating conditions. Calculate the call frequency weight of each data transmission link under different operating state intervals. Use the weight mutation identification function to filter out the core associated links that are highly correlated with specific operating conditions and construct an association mapping table between physical state and link number. Step 3: Generate a link set probability matrix: Statistically analyze the activation frequency of each data transmission link in each state interval of the historical database, analyze the statistical correlation between the state interval and the activation state of the data transmission link in the historical records, and generate a link set probability matrix describing the link calling pattern under all working conditions by calculating the calling condition probability of each data transmission link in the corresponding state interval. Step 4: Extract the operating state vector sequence within the historical sampling period, use a long short-term memory network to capture the nonlinear trend characteristics of the operating state vector evolution over time, construct an operating state change trend prediction model, and realize the prediction of the power grid operating status in the next period. Step 5: Predict the operating state vector at the next moment using the state change trend model, locate the state interval according to the prediction result, retrieve the corresponding associated data transmission link from the link set probability matrix, query the association mapping table, integrate the candidate links that meet the preset conditions into the target data transmission link set, and feed it back to the network management system to perform pre-arranged link resource scheduling.

[0006] Furthermore, the process of collecting operational characteristic parameters of the UHVDC system in step 1 includes acquiring static and dynamic indicators. Static indicators include the per-unit voltage of the AC bus at the sending end of the UHVDC system and the real-time opening and closing status of circuit breakers at each voltage level of the DC system. Dynamic indicators include DC line power fluctuations, sending-end grid frequency deviations, and the rate of change of power over time. These parameters are acquired using a synchronous phasor measurement device at a preset sampling frequency to ensure that the transient characteristics of the UHVDC system during high-power disturbances can be captured.

[0007] Furthermore, the generation process of the operating state vector involves performing deviation standardization on each piece of raw data. For the raw numerical sequence of each collected operating feature parameter, the maximum and minimum values ​​within the observation window are calculated. The ratio of the difference between the original value and the minimum value to the difference between the maximum and minimum values ​​yields the standardized mapping value of the original value for that feature. All standardized mapping values ​​are concatenated to form the operating state vector representing the system's operating profile at the current moment.

[0008] Furthermore, the parameter space partitioning method in step 2 employs vector partitioning logic based on cluster analysis. The set of operating state vectors in the historical database is divided into several vector clusters, each defined as an independent state interval representing a typical power grid operating condition. A preset sampling period is set; when the operating state vectors for several consecutive sampling periods all fall into the same vector cluster, the system is determined to have entered that operating condition.

[0009] Furthermore, for a specific state interval, the activation frequency of each data transmission link in each sampling period is statistically analyzed, and the activation frequency is calculated by the ratio of the activation frequency to the sampling period length. For example, for the i-th data transmission link, its activation frequency in the current sampling period and its activation frequency in the previous sampling period are obtained, and the weight value of this link in the current period is calculated. The weight value is obtained through a weight calculation formula, the parameters of which include the change in activation frequency, the sum of activation frequencies, and the old weight value. The old weight value is the weight value of the i-th data transmission link in the previous monitoring interval; if there is no historical record, a preset initial value is used.

[0010] Furthermore, the process of constructing the association mapping table involves core link identification based on weight gradients. For each vector cluster, its corresponding historical records are traversed, and the average weight of each data transmission link within the vector cluster is calculated. All involved transmission links are sorted in descending order according to their average weights, and the difference between the average weights of adjacent links is calculated. When the difference between the calculated average weight of the previous link and the average weight of the next link is greater than a preset multiple of the global average weight, a weight abrupt change is determined, and all high-weight data transmission links before that position are associated with that state interval.

[0011] Furthermore, in step 3, for a state interval labeled Zm, the total number of activations of each data transmission link is obtained from the historical operation records. The activation probability of the j-th data transmission link in this state is calculated. All state intervals are mapped and aggregated with the conditional probabilities of all data transmission links to form a link set probability matrix where rows represent state intervals and columns represent transmission links.

[0012] Furthermore, step 4 establishes the operational state change trend model using a deep learning architecture based on a long short-term memory network. The input of the operational state change trend model receives the operational state vector sequence from the previous several time steps, and the output of the operational state change trend model generates the predicted value of the operational state vector for the next time step.

[0013] Furthermore, the Long Short-Term Memory (LSTM) network employs multiple gating units to filter and memorize time-series features. The LTM network's internal mechanisms, including the forget gate, input gate, temporary cell state, current cell state, output gate, and hidden layer output, are all calculated using a nonlinear transformation function containing a predefined weight matrix and bias term. Finally, a predicted vector is output through a fully connected layer, where the weight matrix and bias term are obtained through backpropagation training on a historical sample set.

[0014] Furthermore, in step 5, the predicted next-moment running state vector is projected onto the parameter space, and its state interval is determined using a nearest neighbor matching algorithm. The probability row vector corresponding to this state interval is extracted from the link set probability matrix, and links with activation probabilities greater than a preset probability threshold are selected. Simultaneously, the core associated links corresponding to this state interval are retrieved from the association mapping table. The two results are then combined to generate the target data transmission link set.

[0015] The complete operational logic of this invention is as follows: During the operation of a certain ultra-high voltage direct current (UHVDC) sending-end power grid, various operational characteristic parameters, including the per-unit value of AC bus voltage, the power fluctuation value of DC lines, and the frequency deviation of the sending-end power grid, are monitored in real time. After deviation standardization processing, and combined with the preset extreme values ​​of each parameter, the standardized values ​​of each parameter are obtained. The concatenated current-time operational state vector is then input into a pre-trained long short-term memory (LSTM) network model.

[0016] Furthermore, the frequency deviation evolution trend at the next moment is predicted by the long short-term memory network model, and the predicted operating state vector is output. The corresponding predicted value of the operating state vector indicates that the system will enter a specific frequency adjustment condition and is located to the corresponding state interval number.

[0017] Furthermore, within a given state interval, the link set probability matrix is ​​queried to find that the activation probability of the corresponding data transmission link meets a preset condition. Simultaneously, the association mapping table is queried to obtain the core associated links corresponding to that state interval. Through union processing, the target data transmission link set is obtained. Upon receiving feedback, the network management system immediately allocates additional bandwidth redundancy to the target link set and elevates the transmission priority of control commands to a preset level.

[0018] A data transmission system for frequency control at the ultra-high voltage direct current (UHVDC) sending end, comprising: a data acquisition module, a feature processing module, a historical database storage module, a correlation analysis module, a prediction algorithm module, and a link decision module; The data acquisition module is used to obtain the operating characteristic parameters of the ultra-high voltage direct current system; The feature processing module is used to normalize the running feature parameters and serialize them to construct the running state vector; The historical database storage module is used to store the historical logs of the running state vector and the associated data transmission links, forming a time-series dataset for training and mining; The association analysis module is used to analyze the link call patterns through clustering algorithms and weight mutation identification logic, and generate an association mapping table between the running status and frequent links; The prediction algorithm module is used to output the next predicted operating state vector based on the real-time operating condition sequence; The link decision module is used to perform mapping operations between the predicted running state vector and the link set probability matrix to generate the target data transmission link set in real time.

[0019] Compared with the prior art, the beneficial effects of the present invention are: 1. By constructing an operational state vector and partitioning the parameter space, the traditional method overcomes the disconnect between link configuration and operational conditions. Utilizing a weighted mutation identification function and an association mapping table, the core data transmission links that play a crucial role under specific power grid disturbance conditions are identified, significantly improving identification accuracy. Furthermore, by combining a link set probability matrix, dynamic link matching based on statistical laws is achieved, ensuring the uniqueness and accuracy of control command transmission targets and effectively avoiding the link resource mismatch problem in traditional data transmission methods.

[0020] 2. A learning model of operational status change trends is constructed to predict the next state of the system, transforming passive triggering into proactive reservation. Before a substantial deterioration in the UHVDC system's state, the retrieval and feedback of the target data transmission link set are completed, reserving sufficient bandwidth adjustment time for the network management system. This proactive scheduling mechanism ensures the success rate of frequency control command transmission under conditions such as UHV high-power shortages, reducing the risk of grid instability caused by data transmission delays. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the structure of a frequency control data transmission system for ultra-high voltage direct current transmitting end according to the present invention; Figure 2 This is a flowchart illustrating a method for transmitting frequency control data at the ultra-high voltage direct current transmitting end according to the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Example: Figures 1-2 As shown, the present invention provides a technical solution: a method for transmitting data for frequency control at the ultra-high voltage direct current transmitting end.

[0024] Step 1: Real-time acquisition of operational characteristic parameters of the UHVDC system, construction of an operational state vector, and storage of this vector in the historical operational database. The operational characteristic parameter system encompasses multi-dimensional indicators reflecting the system's static stability and dynamic transient characteristics, specifically including the per-unit value of AC bus voltage and the DC line power fluctuation value ΔP. dc Frequency deviation Δf of the sending-end power grid g The power change rate is defined as follows: DC line power fluctuation is the difference in DC line power per unit sampling time; sending-end grid frequency deviation is the deviation between the sending-end grid frequency and the reference frequency; and power change rate is the derivative of the power-time function.

[0025] In the process of constructing the running state vector, the collected raw data is first normalized. Due to the significant differences in the dimensions of different physical quantities, to eliminate the influence of dimensions on subsequent feature extraction, linear normalization is used to map the raw data to the real number interval [0,1]. The normalization formula is: x' = (x - xmin) / (xmax - xmin), where x' represents the normalized value, x represents the original value, and xmax and xmin are the maximum and minimum values ​​of the physical quantity in the historical records, respectively. The normalized values ​​are then concatenated to form the running state vector, and the timestamp of acquiring the running state vector is recorded and stored in the historical running database.

[0026] Step 2: By analyzing the historical records in the historical operation database, calculate the call frequency weight of each data transmission link under different operating states, and identify the frequently used links associated with specific operating states.

[0027] In practice, the logic for identifying frequently used links is as follows: Communication logs under running state vectors with the same or similarity greater than a threshold are extracted from historical data. Using a clustering algorithm, historical sample points with an Euclidean distance less than a preset distance threshold to the current running state vector are selected from the historical database. For these historical sample points, the activation frequency of each transmission link within the monitoring interval's time window is statistically analyzed.

[0028] In a certain operating state, the weight of the i-th transmission link is denoted as WI, and the calculation formula is as follows: In the formula, f1 represents the current monitoring interval, for example, with a duration of T. The activation frequency of the i-th transmission link is calculated as the ratio of the activation frequency N1 of the monitoring interval to the duration T, i.e., f1 = N1 / T; f0 represents the activation frequency of the i-th transmission link in the previous monitoring interval, Δf = f1 - f0; K is the protection coefficient, which takes a positive integer value to prevent... The coefficient term has a negative value; W oldThis represents the initial weight value of the i-th transmission link in the previous monitoring interval. This weight calculation model introduces a gain on the rate of change of link activity through a differential frequency term, enabling it to keenly capture sudden increases in link call demands.

[0029] Based on a given operating state, all historical records are traversed, and each transmission link is sorted in descending order according to its weight. Further, the first-order difference of the weight sequence is calculated to find weight mutation points, and all transmission links before these mutation points are selected. The current operating state vector is then associated with the selected set of transmission links, establishing an association mapping table with the operating state vector as the index and the data transmission link number as the element.

[0030] As a preferred embodiment of the present invention, the specific process of establishing the association between the running state vector and the data transmission link is as follows: Set the partition radius of the running state space, and divide all historical state vectors into different state clusters. For each state cluster, count the set of active links corresponding to all historical moments falling within that cluster, and calculate the average weight of each link within that cluster, where the average weight of the d-th data transmission link is denoted as... The average weight of the (d+1)th data transmission link is denoted as Define a weight mutation detection function. When G(d) exceeds a times the average weight of all data transmission links, preferably a∈(1,3), d is determined to be a weight mutation point, and the data transmission links from sequence 1 to d are stored in the associated link set corresponding to the state cluster.

[0031] Step 3: Establish a probability matrix for the link set by mapping the value ranges of different state vectors to the probability of each link being invoked.

[0032] Extract the sequence of operational state vectors arranged chronologically from the historical operational database. For a given state interval, labeled Zm, obtain the total number of activations of each data transmission link from the historical operational records. Specifically, let Cm,j be the activation count of the j-th data transmission link, and Cmt be the total activation count of all data transmission links in state interval Zm. Calculate the ratio of the activation count Cm,j to the total activation count Cmt to obtain the conditional probability P(Lj, Zm) of the j-th data transmission link Lj in that state. Map all state intervals to the conditional probabilities of all data transmission links to form a structured matrix. The rows of this matrix represent different state intervals, the columns represent all candidate transmission links in the system, and the elements in the matrix are the corresponding conditional probability values. This link set probability matrix not only records the link's activation frequency but also reflects the inherent coupling strength between the link and specific physical conditions.

[0033] Step S4: Based on the historical operation database, construct a prediction model for the operation status of the UHVDC system.

[0034] By arranging the running state vectors in historical records in chronological order, capturing the changing patterns of the running state vectors over time, and constructing a running state prediction model based on a long short-term memory network.

[0035] Obtain the sequence of running state vectors at several time points with a step size of h before time t, denoted as V. t-h , ..., V t-1 V t The sequence is memorized and learned through a Long Short-Term Memory (LSTM) network, and a prediction vector is output. ; These include: Forgotten Gate: ; Input Gate: ; Temporary cell state: ; Current cell state: ; Output gate: ; Hidden output layer: ; Fully connected output prediction vector: ; Among them W f W i W C W o and W v Let b be the weight matrix in the model. f b i b C b o and b v This refers to the bias term in the model.

[0036] Step 5: Based on the changing trend of the current operating status, predict the next operating status of the UHVDC system using the UHVDC system operating status prediction model, and determine the target data transmission link set corresponding to the next operating status by combining the data transmission links under the predicted operating status.

[0037] Specifically, once the corresponding state interval is identified, a dual retrieval operation is performed, including querying the link set probability matrix and querying the association mapping table. First, the probability row vector corresponding to the state interval is extracted from the link set probability matrix, and all links with an activation probability greater than a preset probability threshold are selected. These links represent communication resources that are statistically highly likely to be invoked. Simultaneously, the association mapping table is queried to obtain the core associated links corresponding to the state interval. These links represent the critical paths with physical determinism under this operating condition. The results of these two retrievals are then combined to generate the target data transmission link set.

[0038] Furthermore, a complete embodiment is provided to demonstrate the operational logic of the present invention: Within the current monitoring window, raw data of the operating characteristic parameters of the UHVDC system at a certain moment have been acquired in real time. Specific parameters include: AC bus voltage at the sending end is 515 kV, per unit value is 1.03, current power fluctuation of the DC line is 20 MW, frequency deviation of the sending end grid is -0.05 Hz, and power change rate is 15 MW per second.

[0039] As required in step 1, deviation standardization is performed on the original data, and the current running state vector Q formed by splicing the standardized values ​​is stored in the historical database.

[0040] Proceed to the calculation phase of steps 2 and 3. The current operating state vector is identified as being in state interval Z5. At this point, three main data transmission links L1, L2, and L3 are detected within the system. For link L1, the activation frequency f0 in the previous sampling period was 10 times / second, and the activation frequency f1 in the current sampling period is 12 times / second, then the frequency increment Δf = 2. If the old weight value W of this link... old =1, the protection coefficient K value is 1, and the weight calculation formula is: WI=(1+(2×2) / (10+12))×1=1.182.

[0041] A mapping table was constructed by traversing historical data. In state Z5, the core associated links were calculated and determined to be {L1, L2}. Simultaneously, the link set probability matrix was queried. In the row corresponding to Z5, the probabilities of each link being invoked were P(L1, Z5) = 0.85, P(L2, Z5) = 0.7, and P(L3, Z5) = 0.35, with a probability threshold set at 0.5.

[0042] In step 4, the Long Short-Term Memory (LSTM) network model predicts the operating state for the next time step. The model receives a sequence of state vectors from the past 20 time steps. After nonlinear transformation by the internal control unit, the predicted operating state vector is obtained. The corresponding predicted state vector indicates that the system will enter step Z8.

[0043] In step 5, the system performs a retrieval based on the predicted state interval Z8. In the association mapping table, the core association links corresponding to Z8 are {L1, L4}. In the link set probability matrix, the set of links with an activation probability greater than 0.5 in the row corresponding to Z8 is {L1, L2}. Taking the union of these two sets yields the target data transmission link set {L1, L2, L4}.

[0044] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for transmitting data for frequency control at the transmitting end of an ultra-high voltage direct current (UHVDC) transmission line, characterized in that: the method include: The operating characteristic parameters of the UHVDC system are collected, linear normalization is performed and the data are concatenated into an operating state vector, which is then stored in the historical database. The parameter space formed by the running state vector is divided into several state intervals. The call frequency weight of each data transmission link in different running state intervals is calculated. The core associated links are screened using the weight mutation identification function, and an association mapping table between physical state and link number is constructed. The activation frequency of each data transmission link within each state interval is statistically analyzed. The statistical correlation between each state interval and the activation state of each data transmission link in the historical record is statistically analyzed. The conditional probability of each data transmission link being invoked in the corresponding state interval is calculated, and a link set probability matrix is ​​generated. By using the sequence of operating state vectors from historical sampling periods, the trend characteristics of the operating state vectors changing over time are obtained, and a model of the trend of operating state changes is established. The operating state change trend model is used to predict the operating state vector of the next state; based on the state interval to which the predicted operating state vector belongs, the probability matrix of the link set is retrieved and the associated data transmission links are extracted. The association mapping table is queried, and the data transmission links that meet the conditions are recorded in the target data transmission link set and fed back to the network management system. Methods for obtaining the target data transmission link set include: Obtain the prediction vector output by the state change trend model, match the prediction vector in the parameter space to obtain the state interval corresponding to the prediction vector; retrieve the probability vector corresponding to the state interval from the link set probability matrix, obtain the corresponding data transmission link according to the probability vector, retrieve the data transmission link associated with the state interval from the association mapping table, and combine the two parts of data transmission links into the target data transmission link set after performing a union operation.

2. The data transmission method for frequency control at the UHVDC transmitting end according to claim 1, characterized in that: Methods for obtaining the running state vector include: The operation characteristic parameters of the UHVDC system are collected. The operation characteristic parameters include static and dynamic indicators of the electrical quantities of the UHVDC system. The static indicators include the per-unit value of AC bus voltage and the opening and closing status of circuit breakers. The dynamic indicators include the DC line power fluctuation value, the frequency deviation of the sending-end grid, and the power change rate. Raw data of various operational characteristic parameters are collected, and deviation standardization is performed on each raw data to obtain standardized mapping values ​​of each operational characteristic parameter. The standardized mapping values ​​are then concatenated to obtain the operational state vector.

3. The data transmission method for frequency control at the UHVDC transmitting end according to claim 2, characterized in that: The methods for calculating call frequency weights include: The running state vector in the parameter space is divided into several vector clusters, each vector cluster corresponds to a state interval, and a fixed sampling period is set. When several consecutive running state vectors fall into the same state interval, the activation frequency of each data transmission link in each sampling period is counted. The activation frequency of each data transmission link in each sampling period is calculated by the ratio of the activation frequency to the sampling period length. For a given state interval, obtain the activation frequency f1 of the i-th data transmission link in a certain sampling period and the activation frequency f0 in the previous sampling period of the same sampling period, and calculate the weight value WI of the i-th data transmission link in a certain sampling period. Where K is the protection coefficient, which takes a positive integer value, Δf = f1 - f0, W old This represents the weight value or initial weight value of the i-th transmission link in the previous monitoring interval of a certain monitoring interval.

4. The data transmission method for frequency control at the UHVDC transmitting end according to claim 1, characterized in that: Methods for constructing association mapping tables include: For each vector cluster, iterate through all its historical records, calculate the average weight of each data transmission link within the vector cluster, sort the data transmission links in descending order of average weight, and calculate the difference between adjacent average weights for each transmission link. When the difference between the current average weight and the next average weight is greater than a times the global average weight, where a is a multiple and a∈(1,3), associate the current average weight and all data transmission links before the current average weight with the state intervals corresponding to the vector cluster, and collect all the association relationships between the state intervals and data transmission links to form an association mapping table.

5. A method for transmitting data for frequency control at the UHVDC transmitting end according to claim 1, characterized in that: Methods for constructing a link set probability matrix include: For a certain state interval Zm, the activation count of each data transmission link is obtained from the historical operation record, where the activation count of the j-th data transmission link is Cm,j, and the total activation count of the data transmission links in a certain state interval Zm is Cmt. The activation probability P(Lj, Zm) of the ith data transmission link Lj is: P(Lj, Zm) = Cm,j / Cmt. The correspondence between all state intervals and the conditional probabilities of all data transmission links is collected to form a link set probability matrix.

6. A method for transmitting data for frequency control at the UHVDC transmitting end according to claim 1, characterized in that: Methods for establishing operational status change trend models include: The system retrieves operational state vectors from historical databases, arranges them in chronological order, captures the patterns of operational state vector changes over time, and constructs an operational state change trend model to predict the next operational state vector.

7. A data transmission system for frequency control at the transmitting end of an ultra-high voltage direct current (UHVDC) transmission line, used to execute the data transmission method for frequency control at the transmitting end of an UHVDC transmission line as described in any one of claims 1-6, characterized in that: The system includes: a data acquisition module, a feature processing module, a historical database storage module, a correlation analysis module, a prediction algorithm module, and a link decision module; The data acquisition module is used to acquire the operating characteristic parameters of the UHVDC system; the feature processing module is used to normalize the operating characteristic parameters and serialize them to construct an operating state vector; the historical database storage module is used to store the operating state vector and the historical logs of the associated data transmission links, forming a time-series dataset for training and mining; the association analysis module is used to analyze the link call patterns through clustering algorithms and weight mutation identification logic, and generate an association mapping table between operating states and frequent links; the prediction algorithm module is used to output the next predicted operating state vector based on the real-time operating condition sequence; the link decision module is used to perform mapping operations between the predicted operating state vector and the link set probability matrix to generate the target data transmission link set in real time.

Images