Method and device for voltage fluctuation suppression in a metro dc power supply system

Abstract

The application provides a voltage fluctuation suppression method and device for a subway direct-current power supply system, comprising: obtaining voltage amplitude change information of the subway direct-current power supply system at different sampling moments, determining a real-time voltage fluctuation sequence and an operating state feature of each power supply node according to the voltage amplitude change information; performing space-time correlation mapping processing on the real-time voltage fluctuation sequence and the operating state feature to generate a fluctuation correlation feature set, wherein the fluctuation correlation feature set comprises a space-time propagation path feature of voltage fluctuation and a coupling influence feature between nodes; calling a pre-trained fluctuation suppression decision model to perform dynamic suppression strategy generation processing on the fluctuation correlation feature set to obtain a voltage regulation parameter sequence for each power supply node; generating a node-coordinated fluctuation suppression execution instruction set according to the voltage regulation parameter sequence and the operating state feature of each power supply node; and triggering voltage regulation operations of each power supply node according to the fluctuation suppression execution instruction set to suppress voltage fluctuation.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of metro power supply, and particularly relates to a voltage fluctuation suppression method and device for a metro direct-current power supply system. BACKGROUND

[0002] In metro operation, the direct-current power supply system is the key infrastructure to ensure the normal operation of trains. However, due to the dynamic changes of loads, the start and stop of equipment, and the characteristics of the power grid itself during metro operation, the metro direct-current power supply system often has voltage fluctuation problems. Voltage fluctuation not only affects the normal operation of train equipment and reduces the service life of equipment, but also can adversely affect the riding experience of passengers, and even cause safety accidents in serious cases.

[0003] At present, the suppression methods for voltage fluctuation of the metro direct-current power supply system mainly have the following deficiencies. On the one hand, most of the existing methods only focus on the real-time value of voltage fluctuation, ignoring the spatio-temporal correlation of voltage fluctuation between different power supply nodes. The metro power supply system is a complex network, and voltage fluctuation will propagate and interact between different nodes. Only relying on the voltage fluctuation information of a single node for suppression, it is difficult to fully and accurately grasp the essential characteristics of voltage fluctuation, and it is impossible to achieve effective suppression. On the other hand, the traditional suppression strategy is often static and cannot be dynamically adjusted according to the real-time operation state of the power supply system. During metro operation, the load changes frequently and complexly, and the static suppression strategy cannot adapt to such dynamic changes, resulting in poor suppression effect and failing to meet the demand for safe and stable operation of the metro. SUMMARY

[0004] In view of the above-mentioned problems, in combination with the first aspect of the present disclosure, a voltage fluctuation suppression method for a metro direct-current power supply system is provided in the embodiments of the present disclosure, comprising:

[0005] The voltage amplitude variation information of the subway direct-current power supply system at different sampling time points is acquired, and according to the voltage amplitude variation information, a real-time voltage fluctuation sequence of the subway direct-current power supply system and an operating state feature of each power supply node in the subway direct-current power supply system are determined; the real-time voltage fluctuation sequence and the operating state feature are subjected to spatio-temporal correlation mapping processing to generate a fluctuation correlation feature set with node correlation, the fluctuation correlation feature set including a spatio-temporal propagation path feature of voltage fluctuation and an inter-node coupling influence feature; a pre-trained fluctuation suppression decision model is called to generate a voltage regulation parameter sequence for each power supply node by performing dynamic suppression strategy generation processing on the fluctuation correlation feature set; a node-coordinated fluctuation suppression execution instruction set is generated according to the voltage regulation parameter sequence and the operating state feature of each power supply node; and the voltage regulation operation of each power supply node is triggered to suppress voltage fluctuation according to the fluctuation suppression execution instruction set.

[0006] In a possible implementation, the operating state feature includes a load change trend and a device operating mode identifier of each power supply node, and the spatio-temporal correlation mapping processing of the real-time voltage fluctuation sequence and the operating state feature to generate a fluctuation correlation feature set with node correlation includes: partitioning the real-time voltage fluctuation sequence according to the power supply nodes to obtain a local voltage fluctuation sub-sequence corresponding to each power supply node; calculating a similarity index between local voltage fluctuation sub-sequences of adjacent power supply nodes according to the local voltage fluctuation sub-sequences corresponding to each power supply node, the similarity index being used to represent a consistency degree of fluctuation form; constructing a fluctuation propagation adjacency matrix between the power supply nodes according to the similarity index and a topological connection relationship of a power supply network in the subway direct-current power supply system; based on the fluctuation propagation adjacency matrix and the real-time voltage fluctuation sequence, analyzing a diffusion speed and an attenuation characteristic of voltage fluctuation from a starting node to surrounding nodes, and generating a spatio-temporal propagation path feature of voltage fluctuation according to the diffusion speed and the attenuation characteristic; calculating a cross-influence coefficient of the load change trend of different power supply nodes, the cross-influence coefficient being used to describe an influence degree of load change of any power supply node on voltage fluctuation of another power supply node; combining the cross-influence coefficient corresponding to the power supply node and a matching degree of the device operating mode identifier to generate an inter-node coupling influence feature; and concatenating the spatio-temporal propagation path feature and the inter-node coupling influence feature in a time dimension to form a fluctuation correlation feature set with node correlation.

[0007] In a possible implementation, the calculation of the similarity index between the local voltage fluctuation sub-sequences of adjacent power supply nodes according to the local voltage fluctuation sub-sequences of the power supply nodes comprises: performing length normalization processing on the local voltage fluctuation sub-sequences of adjacent power supply nodes, and extracting fluctuation extreme point distribution information of each length-normalized local voltage fluctuation sub-sequence; calculating a voltage amplitude difference between extreme points at corresponding positions in the two local voltage fluctuation sub-sequences according to the positions of maximum points in the fluctuation extreme point distribution information and first voltage amplitudes corresponding to the maximum points, the positions of minimum points in the fluctuation extreme point distribution information and second voltage amplitudes corresponding to the minimum points; calculating an amplitude similarity component based on a cumulative sum of the voltage amplitude difference; analyzing fluctuation period characteristics of the two local voltage fluctuation sub-sequences, extracting a target fluctuation frequency component through Fourier transform, and calculating a matching degree of the target fluctuation frequency component as a period similarity component; and performing weighted summation on the amplitude similarity component and the period similarity component according to a preset weight ratio, to obtain the similarity index between the local voltage fluctuation sub-sequences of adjacent power supply nodes.

[0008] In a possible implementation, the analysis of the diffusion speed and the attenuation characteristic of voltage fluctuation from a starting node to surrounding nodes based on the fluctuation propagation adjacency matrix and the real-time voltage fluctuation sequence, and the generation of the space-time propagation path feature of voltage fluctuation comprise: determining, through a fluctuation starting identification algorithm, a starting node and a corresponding starting time of the voltage fluctuation from the real-time voltage fluctuation sequence; expanding the search of adjacent power supply nodes layer by layer based on a multi-order neighbor node search model constructed by the starting node as a center and the fluctuation propagation adjacency matrix; recording a fluctuation time at which each power supply node first appears voltage fluctuation meeting a preset requirement, and calculating a time difference between each fluctuation time and the starting time as a fluctuation arrival time of the corresponding power supply node; calculating a fluctuation diffusion speed according to the fluctuation arrival time of each power supply node and a physical distance between the power supply nodes, the physical distance being determined based on node coordinates in the power supply network topology graph; extracting a voltage amplitude peak value of each power supply node in the fluctuation propagation process, and taking a ratio of the voltage amplitude peak value to a voltage amplitude peak value of the starting node as an attenuation coefficient of the corresponding power supply node; and arranging the fluctuation arrival time, the fluctuation diffusion speed, and the attenuation coefficient of each power supply node in a spatial distribution order of nodes, to generate a space-time propagation path feature representing a propagation time sequence and intensity change of the voltage fluctuation.

[0009] In a possible implementation, the matching degree of the cross-influence coefficient corresponding to the power supply node and the device operation mode identifier is combined to generate a node-to-node coupling influence feature, including: extracting the device operation mode identifier of each power supply node, and constructing a mode identifier matrix; determining the mode matching degree between any two power supply nodes in the mode identifier matrix according to whether the device operation mode identifiers of the any two power supply nodes are the same; determining the mode influence coefficient of the any two power supply nodes according to the mode matching degree of the any two power supply nodes; performing product operation on the cross-influence coefficient and the mode influence coefficient to obtain a comprehensive influence coefficient; arranging the comprehensive influence coefficient according to the node connection relationship according to the topology structure of the power supply network to form a coupling influence matrix between power supply nodes; deleting elements with the comprehensive influence coefficient less than a preset threshold in the coupling influence matrix between the power supply nodes, and retaining elements with the comprehensive influence coefficient greater than the preset threshold to generate a node-to-node coupling influence feature.

[0010] In a possible implementation, the pre-trained fluctuation suppression decision model is called to perform dynamic suppression strategy generation processing on the fluctuation correlation feature set to obtain a voltage adjustment parameter sequence for each power supply node, including: inputting the fluctuation correlation feature set into a feature encoding layer of the fluctuation suppression decision model to perform high-dimensional feature mapping processing and generate an encoded feature vector with time sequence dependence; performing node importance weight distribution processing on the encoded feature vector by a node attention module of the fluctuation suppression decision model to generate an attention degree coefficient of each power supply node; performing weighted aggregation processing on the encoded feature vector based on the attention degree coefficient to obtain an aggregated feature vector; inputting the aggregated feature vector into a time sequence prediction layer of the fluctuation suppression decision model to perform multi-step voltage fluctuation prediction processing and generate a voltage fluctuation trend prediction sequence for predicting a future time period; determining an adjustment amount dynamic change curve for voltage adjustment of each power supply node according to the voltage fluctuation trend prediction sequence and the node-to-node coupling influence feature in the fluctuation correlation feature set; discretizing the adjustment amount dynamic change curve according to a preset time interval to generate a voltage adjustment parameter sequence for each power supply node.

[0011] In a possible implementation, the node attention module of the fluctuation suppression decision model performs node importance weight allocation processing on the encoded feature vector to generate an attention coefficient of each power supply node, including: analyzing feature components of each power supply node in the encoded feature vector, extracting a fluctuation sensitivity index of each feature component, the fluctuation sensitivity index being positively correlated with a change rate of the voltage fluctuation amplitude; constructing an initial attention vector according to the fluctuation sensitivity index, each element in the initial attention vector corresponding to an initial attention value of one of the power supply nodes; calculating a deviation degree of a load change trend in the operating state feature of each power supply node from an average load change trend of the subway direct-current power supply system to generate a load deviation coefficient; introducing the load deviation coefficient into an attention calculation function, dynamically adjusting the initial attention vector, and combining a cross-influence coefficient in the inter-node coupling influence feature to perform neighborhood propagation processing on the adjusted attention vector; performing normalization processing on the attention vector after the neighborhood propagation processing to generate the attention coefficient of each power supply node.

[0012] In a possible implementation, the determination of the dynamic change curve of the adjustment amount of each power supply node for voltage regulation according to the voltage fluctuation trend prediction sequence and the inter-node coupling influence feature in the fluctuation correlation feature set includes: determining a fluctuation interval of voltage regulation according to a time point at which a voltage prediction value in the voltage fluctuation trend prediction sequence is greater than a preset voltage stability threshold; calculating a deviation amount of the voltage prediction value at each time point in each fluctuation interval from the preset voltage stability threshold to obtain a basic adjustment amount corresponding to each time point; determining a set of associated nodes that have a significant influence on voltage fluctuation of each supply node according to the inter-node coupling influence feature in the fluctuation correlation feature set; extracting an adjustment amount in a voltage regulation parameter sequence of the set of associated nodes of each supply node, calculating a coupling adjustment contribution value of the corresponding supply node; superimposing the basic adjustment amount at the time point and the corresponding coupling adjustment contribution value to obtain a total adjustment amount of the corresponding supply node; and determining a dynamic change curve of the adjustment amount of each power supply node for voltage regulation based on the total adjustment amount through a curve fitting algorithm.

[0013] In a possible implementation, the generating, according to the voltage adjustment parameter sequence and the operation state feature of each power supply node, of a set of node-coordinated fluctuation suppression execution instructions includes: analyzing the voltage adjustment parameter sequence of each power supply node, extracting the voltage adjustment amount and the adjustment duration parameter at each time point; determining the adjustment authority level of each power supply node according to the device operation mode identifier in the operation state feature of each power supply node, the adjustment authority level being used to limit the maximum range of the voltage adjustment amount; performing rationality checking processing on the voltage adjustment amount based on the adjustment authority level, and truncating and correcting the adjustment amount that exceeds the authority range; analyzing the coordination between the voltage adjustment amounts of different power supply nodes at the same time point, calculating a consistency coefficient of the adjustment direction, the consistency coefficient being used to judge the coordination degree of the adjustment operations between the power supply nodes; when the consistency coefficient is lower than a preset threshold, calling a coordination optimization algorithm to generate a node-coordinated adjustment scheme for adjusting the voltage adjustment amount of a target power supply node, the node-coordinated adjustment scheme making the consistency coefficient between the target power supply nodes greater than the preset threshold, wherein the target power supply node includes at least one of the power supply nodes whose consistency coefficient is lower than the preset threshold; and performing association and encapsulation processing on the corrected and adjusted voltage adjustment amount, the adjustment duration parameter, and the corresponding power supply node identifier, to generate a set of node-coordinated fluctuation suppression execution instructions.

[0014] In combination with the second aspect of the application, an embodiment of the application provides a voltage fluctuation suppression device of a metro DC power supply system, the device comprising: a memory having a computer program stored thereon; and a processor configured to execute the computer program stored in the memory to implement the voltage fluctuation suppression method of the metro DC power supply system according to any one of the first aspect.

[0015] The aforementioned technical solution comprehensively grasps the dynamic information of voltage fluctuations and the real-time operation of the power supply system by acquiring the real-time voltage fluctuation sequence and the corresponding operating status characteristics of the metro DC power supply system. It performs spatiotemporal correlation mapping on the real-time voltage fluctuation sequence and operating status characteristics to generate a fluctuation correlation feature set with node correlation. This deeply reveals the spatiotemporal propagation path of voltage fluctuations between different nodes and the coupling influence characteristics between nodes, overcoming the limitations of traditional methods that only focus on voltage fluctuations of a single node, and grasping the laws of voltage fluctuations from a global perspective. A pre-trained fluctuation suppression decision model is invoked to dynamically generate suppression strategies for the fluctuation correlation feature set, resulting in a voltage adjustment parameter sequence for each power supply node. This enables dynamic adjustment of the suppression strategy, adapting to the dynamic changes in the load of the metro power supply system and improving the targeting and effectiveness of suppression. Based on the voltage adjustment parameter sequence and the operating status characteristics of each power supply node, a set of node-coordinated fluctuation suppression execution instructions is generated and sent to the corresponding power supply control module to trigger voltage adjustment operations. Through node coordination, comprehensive and effective suppression of voltage fluctuations is achieved, ensuring the stable operation of the metro DC power supply system and improving the safety and reliability of train operation.

[0016] Other features and advantages of this disclosure will be described in detail in the following detailed description section. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate the present disclosure and form part of the specification. They are used together with the following detailed description to explain the present disclosure, but do not constitute a limitation thereof. In the drawings:

[0018] Figure 1 This is a schematic diagram of the execution flow of the voltage fluctuation suppression method for a subway DC power supply system provided in an embodiment of the present invention.

[0019] Figure 2 This is one implementation provided by an embodiment of the present invention. Figure 1 A flowchart illustrating step S12.

[0020] Figure 3 This is one implementation provided by an embodiment of the present invention. Figure 1 A flowchart of step S13.

[0021] Figure 4 This is one implementation provided by an embodiment of the present invention. Figure 1 A flowchart illustrating step S14.

[0022] Figure 5 This is a schematic diagram of exemplary hardware and software components of the voltage fluctuation suppression device for a subway DC power supply system provided in an embodiment of the present invention. Detailed Implementation

[0023] 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.

[0024] The specific embodiments of this disclosure will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit this disclosure.

[0025] This invention provides a method for suppressing voltage fluctuations in a subway DC power supply system. (See also...) Figure 1 As shown, the method includes:

[0026] In step S11, the voltage amplitude change information of the subway DC power supply system at different sampling times is obtained, and based on the voltage amplitude change information, the real-time voltage fluctuation sequence of the subway DC power supply system and the operating status characteristics of each power supply node in the subway DC power supply system are determined.

[0027] Among these, voltage amplitude variation information refers to the deviation data between the instantaneous voltage value and the reference value (such as the rated voltage) collected in real time by voltage sensors in the subway DC power supply system, used to reflect the dynamic characteristics of voltage changes over time. The real-time voltage fluctuation sequence is a discrete-time series formed by arranging the voltage amplitude variation information at the sampling time in chronological order, used to quantify the amplitude, frequency, and duration of voltage fluctuations. The operating status characteristics of the power supply node can include parameters such as node current, load power, temperature, switch status, and harmonic content, used to describe the real-time operating conditions of the power supply node in the power supply network.

[0028] In this embodiment, a distributed voltage sensor network can be used to acquire the voltage amplitude of the DC bus and each power supply node in the subway DC power supply system at a high-frequency sampling rate (e.g., millisecond level), while simultaneously collecting auxiliary parameters such as current and power. The sampled voltage amplitude change information is then filtered (e.g., using moving average filtering) to eliminate noise. The voltage deviation (ΔV = V_sample - V_rated) is then calculated and arranged in chronological order to generate a real-time voltage time series.

[0029] Furthermore, based on the collected harmonic content, current and power data, combined with the node topology location, the load rate, power factor and other characteristics are calculated, and abnormal states (such as overload and short circuit) are identified through threshold judgment or machine learning models (such as isolated forest), thus obtaining the operating status characteristics of each power supply node.

[0030] In step S12, the real-time voltage fluctuation sequence and the operating state features are spatio-temporally associated and mapped to generate a fluctuation associated feature set with node association, which includes a time-space propagation path feature of voltage fluctuation and a node-to-node coupling influence feature.

[0031] The spatio-temporal association mapping is used to combine the time sequence of voltage fluctuation with the topology structure (spatial dimension) of the power supply network to analyze the propagation path and coupling relationship of the fluctuation among nodes. The time-space propagation path feature in the fluctuation associated feature set is used to describe the propagation direction, delay and attenuation law of the fluctuation from the source node to other nodes. The node-to-node coupling influence feature is used to quantify the mutual influence degree (such as correlation coefficient, causal relationship strength) of voltage fluctuation among nodes. The fluctuation associated feature set is a multi-dimensional feature set that fuses the time-space propagation path feature and the node-to-node coupling influence feature, and is used to represent the node association.

[0032] In the embodiments of the present disclosure, a graph model of the power supply network is constructed, and the nodes represent power supply devices (such as traction substations and contact network segments), and the edges represent physical connection relationships. Based on the graph model of the power supply network, the time difference of voltage fluctuation of different power supply nodes can be calculated by the cross-correlation function, so as to determine the propagation direction of the voltage fluctuation. Then, based on the propagation direction, the voltage fluctuation of each power supply node in the real-time voltage fluctuation sequence is fitted to obtain an amplitude attenuation curve (such as an exponential decay model) of the voltage fluctuation with the propagation distance, so that the time-space propagation path feature of the voltage fluctuation can be generated according to the time difference, the propagation direction and the attenuation curve. Then, it is determined whether the fluctuation of node A has the prediction ability for node B to determine the causal relationship. Then, based on the partial least squares regression (PLSR) or the neural network, a nonlinear mapping relationship of the voltage fluctuation among nodes is established to generate the node-to-node coupling influence feature.

[0033] In step S13, a pre-trained fluctuation suppression decision model is called to generate a dynamic suppression strategy for the fluctuation associated feature set to obtain a voltage adjustment parameter sequence for each power supply node.

[0034] The pre-trained fluctuation suppression decision model is a deep learning model (such as an LSTM-Attention network) trained based on historical data, the input is the fluctuation associated feature, and the output is the voltage adjustment parameter (such as adjustment amplitude and duration) of each power supply node. The dynamic suppression strategy is to dynamically adjust the adjustment parameter according to the real-time fluctuation feature to realize adaptive suppression.

[0035] In the embodiments of the present disclosure, historical voltage fluctuation events and corresponding suppression measures (such as adjusting transformer taps and switching capacitors) are collected, and the optimal adjustment parameters are labeled. An LSTM network is used to capture time dependence, combined with an attention mechanism (Attention) to focus on key node features, pre-train the model, and obtain a pre-trained fluctuation suppression decision model.

[0036] In the embodiments of the present disclosure, the training steps of the fluctuation suppression decision model are as follows: step S210: collect historical voltage fluctuation data of the subway DC power supply system, corresponding power supply node operating state features and historical fluctuation suppression strategy data, and construct a model training data set.

[0037] In order to train a decision model that can accurately generate fluctuation suppression strategies, a large amount of historical data can be collected. The historical voltage fluctuation data includes voltage amplitude change sequences of each power supply node under different time periods and different working conditions; the power supply node operating state features include historical load change trends and device operating mode identifiers; and the historical fluctuation suppression strategy data includes adjustment parameter sequences and actual suppression effect data taken for historical voltage fluctuations.

[0038] The collected data is preprocessed, including data cleaning, missing value filling, and outlier processing, to ensure the quality and consistency of the data. Then, the preprocessed data is divided into training set, validation set and test set according to a certain proportion, and a complete model training data set is constructed.

[0039] Step S220: The historical voltage fluctuation data and power supply node operating state features in the model training data set are processed in the same way as step S120, to generate a historical fluctuation correlation feature set.

[0040] The historical voltage fluctuation data and power supply node operating state features in the model training data set are processed in the same way as step S120, to generate a historical fluctuation correlation feature set.

[0041] Through this processing, the historical data can be converted into feature data consistent with the format of the fluctuation correlation feature set in actual application, providing suitable input for model training.

[0042] Step S230: Construct the network structure of the fluctuation suppression decision model, which includes a feature encoding layer, a node attention module, a time series prediction layer, and a strategy generation layer.

[0043] The network structure of the fluctuation suppression decision model is built using a deep learning framework. The feature encoding layer uses a combination of convolutional neural networks and recurrent neural networks to map high-dimensional features and capture temporal dependencies from the input set of historical fluctuation-related features. The node attention module uses a self-attention mechanism to assign importance weights to the encoded feature vectors. The time series prediction layer uses a long short-term memory network to predict future voltage fluctuation trends based on aggregated feature vectors. The strategy generation layer uses a fully connected neural network to generate voltage regulation parameter sequences based on predicted fluctuation trends and inter-node coupling characteristics.

[0044] Neurons connect between layers, forming a complete network structure. The initial values of the network parameters are set by random initialization.

[0045] Step S240: Train the fluctuation suppression decision model using the set of historical fluctuation-related features as input and the voltage regulation parameter sequence in the corresponding historical fluctuation suppression strategy data as expected output.

[0046] The set of historical fluctuation-related features is input into the constructed fluctuation suppression decision model. The model calculates the predicted voltage regulation parameter sequence through forward propagation. The predicted results are compared with the expected output (i.e., the actual voltage regulation parameter sequence) in the historical fluctuation suppression strategy data to calculate the loss value between them.

[0047] The loss function uses the mean square error function to measure the difference between the predicted value and the expected output. Through the backpropagation algorithm, the parameters of each layer of the model are adjusted based on the loss value to reduce the prediction error.

[0048] During training, the training set is used to iteratively train the model. After a certain number of iterations, the validation set is used to evaluate the performance of the model. If the loss value of the validation set no longer decreases or shows an upward trend, training is stopped to avoid overfitting of the model.

[0049] Step S250: Use the test set to evaluate the performance of the trained fluctuation suppression decision model. When the prediction accuracy of the model reaches the preset standard, the model training is considered complete.

[0050] After training, the test set is used to evaluate the performance of the fluctuation suppression decision model. Evaluation indicators include the similarity between the predicted voltage regulation parameter sequence and the actual sequence, the fluctuation suppression effect after voltage regulation based on the predicted sequence, etc.

[0051] When the prediction accuracy of the model reaches the preset standard, such as the similarity between the predicted sequence and the actual sequence is above the preset threshold, and the fluctuation suppression effect based on the predicted sequence meets the system requirements, it is determined that the model training is completed. The trained model can be deployed in the actual metro DC power supply system to generate real-time fluctuation suppression strategies.

[0052] Step S260: Periodically update and optimize the trained fluctuation suppression decision model, adjust the model parameters according to new operation data and suppression effect feedback to adapt to changes in system operation state.

[0053] The operation state of the metro DC power supply system will change over time, such as equipment aging, load mode change, etc., so the fluctuation suppression decision model can be periodically updated and optimized.

[0054] Periodically collect new operation data and fluctuation suppression effect feedback, add these data to the model training data set, retrain the model according to the method of steps S220 to S240, and adjust the parameters of the model. In this way, the model can continuously adapt to changes in the system operation state and maintain good fluctuation suppression strategy generation capability.

[0055] During the model updating and optimization process, incremental training can be used to avoid completely retraining the model to improve update efficiency. At the same time, the performance of the updated model will be evaluated to ensure that its performance is not lower than that of the model before updating.

[0056] Further, the generated spatiotemporal propagation path features and inter-node coupling influence features are input into the pre-trained fluctuation suppression decision model, and the adjustment parameter sequence (such as ΔV_i(t), t=1,2,…,T) of each power supply node is output. Through reinforcement learning (such as PPO algorithm), the model parameters are optimized online to adapt to new fluctuation patterns. Thus, end-to-end mapping from fluctuation features to suppression strategies can be realized, and decision efficiency can be improved.

[0057] In step S14, a set of node-coordinated fluctuation suppression execution instructions is generated according to the voltage adjustment parameter sequence and the operation state features of each power supply node.

[0058] Among them, node coordination is to realize global voltage stability through coordinated control of multiple power supply nodes, avoiding new fluctuations caused by local adjustment. The fluctuation suppression execution instruction set includes adjustment target values, execution times, priority levels and other instruction parameters of each power supply node.

[0059] In the embodiments of the present disclosure, power balance constraints are used to ensure that the total power demand after adjustment matches the power supply capacity, and device amplitude constraints are used to limit the adjustment amplitude (such as ΔV i(t) ≤ 5% V_rated) to avoid equipment overload, AV i (t) is the adjustment amplitude of the ith power supply node, and V_rated is a preset amplitude threshold.

[0060] Further, a distributed optimization algorithm (such as ADMM) is used to decompose the global problem into node-level sub-problems, which are independently solved by each node and then coordinated through communication to reach an agreement. The priority of the instruction can be dynamically allocated according to the node criticality (such as the priority of the traction substation being higher than that of the ordinary load node). Thus, the executability and global optimality of the adjustment instruction are ensured.

[0061] In one possible implementation, the target voltage adjustment amount AV i (t) of each node within a time window t ∈ [1, T] is determined. i The operating state characteristics of the power supply node can include the real-time load rate λ i , power factor cosφ , device health status (such as temperature, aging degree), and the like.

[0062] Preprocessing operation: normalization processing, mapping AV i (t), λ i and other characteristics to the interval [0, 1] to eliminate dimensional differences.

[0063] State label mapping, converting the device health status into a weight coefficient (such as weight = 1 when the health status is "good" and weight = 0.5 when the health status is "fault warning"), which is used for subsequent constraint reinforcement.

[0064] Further, the global adjustment cost is minimized while satisfying all constraint conditions:

[0065]

[0066] where C i (ΔV i (t)) is the adjustment cost (such as device loss, operation times) of the power supply node i, and ρ is a penalty coefficient.

[0067] where ω i is the priority weight coefficient of the power supply node i, which is used to quantify the relative importance of different nodes in global optimization, and the influence of different nodes on system stability is different. For example, the voltage fluctuation of the traction substation may directly affect the safety of train operation, while the fluctuation of the ordinary section load has less influence. ω i By giving higher weight to the key node, it is ensured that the adjustment instruction is executed in priority. For example, the weight can dynamically change with the real-time state (such as load rate λ i ) of the node. For example, when the load rate of node i exceeds the threshold, ω i can be increased in real time to strengthen its contribution to global optimization.

[0068] where Penalty(constraints) is a scalar function quantifying the degree of violation of constraints in the optimization process. By converting hard constraints (such as device clipping, power balance) into penalty terms, it ensures that the optimization solution satisfies physical feasibility. For example, if the node voltage adjustment ΔV i (t) exceeds the device clipping (such as ±5%), the penalty term significantly increases the objective function value, forcing the optimizer to adjust the solution. Soft constraint handling: For non-critical constraints (such as adjustment rate limits), a certain degree of violation can be allowed by adjusting the penalty coefficient ρ to balance optimization efficiency and constraint strictness.

[0069] Further, the global optimization is split into N node-level sub-problems, each solving a local objective function (such as minimizing its own adjustment cost) independently. The coupling information between nodes (such as power balance) is exchanged through Lagrange multipliers, iteratively updated until convergence, for example, the convergence condition can be that the global objective function value changes less than a threshold ϵ (such as ϵ = 10 −4 ) between adjacent two iterations, ensuring coordination consistency.

[0070] In step S15, according to the fluctuation suppression execution instruction set, the voltage adjustment operation of each power supply node is triggered to suppress the voltage fluctuation.

[0071] where the voltage adjustment operation can be adjusting the node voltage through power electronic devices (such as DC / DC converters, SVG) or mechanical devices (such as on-load voltage regulating switches).

[0072] In the embodiments of the present disclosure, communication with the adjustment device is performed through, for example, IEC 61850 protocol or Modbus, and instructions (such as setting the SVG reactive power output reference value) are issued. The adjusted node voltage is monitored in real time, and if the deviation exceeds the threshold (such as ±2%V_rated), secondary adjustment is triggered. PID controllers or model predictive control (MPC) can also be used to dynamically adjust the adjustment amount to eliminate steady-state errors.

[0073] The aforementioned technical solution comprehensively grasps the dynamic information of voltage fluctuations and the real-time operation of the power supply system by acquiring the real-time voltage fluctuation sequence and the corresponding operating status characteristics of the metro DC power supply system. It performs spatiotemporal correlation mapping on the real-time voltage fluctuation sequence and operating status characteristics to generate a fluctuation correlation feature set with node correlation. This deeply reveals the spatiotemporal propagation path of voltage fluctuations between different nodes and the coupling influence characteristics between nodes, overcoming the limitations of traditional methods that only focus on voltage fluctuations of a single node, and grasping the laws of voltage fluctuations from a global perspective. A pre-trained fluctuation suppression decision model is invoked to dynamically generate suppression strategies for the fluctuation correlation feature set, resulting in a voltage adjustment parameter sequence for each power supply node. This enables dynamic adjustment of the suppression strategy, adapting to the dynamic changes in the load of the metro power supply system and improving the targeting and effectiveness of suppression. Based on the voltage adjustment parameter sequence and the operating status characteristics of each power supply node, a set of node-coordinated fluctuation suppression execution instructions is generated and sent to the corresponding power supply control module to trigger voltage adjustment operations. Through node coordination, comprehensive and effective suppression of voltage fluctuations is achieved, ensuring the stable operation of the metro DC power supply system and improving the safety and reliability of train operation.

[0074] In one possible implementation, the operating status characteristics include the load change trend of each power supply node and the equipment operating mode identifier, wherein the load change trend is used to characterize the load power change pattern of the power supply node in the time dimension, typically expressed as a time series (e.g., P). i (t) or a trend function (such as linear, periodic). Equipment operating mode identifiers are discrete labels used to characterize the current operating state of power supply equipment (such as "normal mode", "derating mode", "fault mode"), and are used to quantify the differences in the equipment's response characteristics to voltage fluctuations.

[0075] See Figure 2 As shown, in step S12, the spatiotemporal correlation mapping processing of the real-time voltage fluctuation sequence and the operating state features to generate a set of fluctuation correlation features with node correlation includes:

[0076] In step S121, the real-time voltage fluctuation sequence is divided into partitions according to the power supply nodes to obtain local voltage fluctuation sub-sequences corresponding to each power supply node.

[0077] Among them, the local voltage fluctuation subsequence is a time subsequence that contains only the voltage fluctuation data of a single node after the global voltage fluctuation sequence is divided according to the power supply node.

[0078] In this embodiment of the disclosure, the real-time voltage fluctuation sequence V(t) is aligned to the sampling time of each power supply node according to the timestamp to ensure time synchronization.

[0079] The node partitioning can be to divide V(t) into N subsequences according to the power supply network topology, each subsequence containing only voltage fluctuation data of node i. A sliding window process is performed, for example, a sliding window (such as a window length Tw=10 seconds) can be applied to each subsequence to extract local fluctuation segments and reduce non-stationary interference.

[0080] In this embodiment, since the real-time voltage fluctuation sequence is a set of voltage amplitudes of the entire power supply system at different sampling times, in order to analyze the voltage fluctuation of each power supply node itself, the power supply nodes can be partitioned. According to the number and distribution range of the power supply nodes, the voltage amplitude data belonging to the same power supply node in the real-time voltage fluctuation sequence is extracted to form a local voltage fluctuation subsequence corresponding to the node.

[0081] For example, assuming that the power supply system has multiple power supply nodes, which are numbered as node A, node B, node C, etc. From the real-time voltage fluctuation sequence, the voltage amplitudes at the sampling times marked as node A are extracted and arranged in chronological order to obtain the local voltage fluctuation subsequence of node A; similarly, the local voltage fluctuation subsequences of node B, node C, etc. can be obtained. Each local voltage fluctuation subsequence completely records the change of the voltage amplitude of the corresponding node with time within the monitoring period.

[0082] In step S122, according to the local voltage fluctuation subsequences corresponding to each of the power supply nodes, a similarity index between the local voltage fluctuation subsequences of adjacent power supply nodes is calculated, and the similarity index is used to represent the consistency degree of the fluctuation form.

[0083] The similarity index is an index (such as a dynamic time warping distance or a Pearson correlation coefficient) for quantifying the consistency degree of the form of two local voltage fluctuation subsequences.

[0084] In the embodiments of the present disclosure, the local subsequences V i (t) and V j (t) of adjacent nodes (such as nodes i and j) are calculated by using a dynamic time warping (DTW) algorithm to calculate a similarity index S ij :

[0085]

[0086] DTW is a dynamic time warping distance, and the smaller the value is, the more similar the form is, len(V i ) represents the length of the voltage time sequence V i (t) of the power supply node i, and len(V j ) represents the length of the voltage time sequence V j (t) of the power supply node j.

[0087] Further, the similarity index S is mapped to the interval [0, 1], where 1 represents complete consistency and 0 represents complete irrelevance. The similarity index reflects the consistency of fluctuation patterns, and a high similarity indicates that there may be a direct or indirect fluctuation propagation path between nodes. ij ij ij

[0088] In step S123, a fluctuation propagation adjacency matrix between the power supply nodes is constructed according to the similarity index and the topological connection relationship of the power supply network in the subway DC power supply system.

[0089] The fluctuation propagation adjacency matrix is a matrix used to describe the voltage fluctuation propagation relationship between power supply nodes, and the element value represents the strength or probability of fluctuation transmission.

[0090] In the embodiments of the present disclosure, in combination with the power supply network topology (such as the traction substation-catenary-train node chain), the propagation relationship is calculated only for the physically connected node pairs. Then, an N x N matrix A is constructed, where the element A ij in the matrix A is: If nodes i and j are physically connected, A ij is the product of the similarity index corresponding to nodes i and j and the reciprocal of the electrical distance from node i to j.

[0091] The fluctuation propagation adjacency matrix can quantify the propagation strength of fluctuations between nodes, and the closer the electrical distance and the more similar the pattern, the higher the propagation strength of the node pair.

[0092] In step S124, based on the fluctuation propagation adjacency matrix and the real-time voltage fluctuation sequence, the diffusion speed and the attenuation characteristic of voltage fluctuation from the starting node to the surrounding nodes are analyzed, and the space-time propagation path feature of the voltage fluctuation is generated according to the diffusion speed and the attenuation characteristic.

[0093] The diffusion speed is used to describe the time delay required for voltage fluctuation to propagate from the starting node to the adjacent node. The attenuation characteristic is used to describe the law (such as exponential attenuation, linear attenuation) that the amplitude of the voltage fluctuation decreases with the increase of the distance during the propagation process.

[0094] In the embodiments of the present disclosure, for each node i, the voltage fluctuation starting time t i,start is identified by a mutation detection algorithm, and then the time delay Δt ij of the fluctuation propagation to the adjacent node j is calculated as Δt j,start = t i,start − t ij .

[0095] Further, the diffusion speed v ij = d ij / Δt ijis the electrical distance between node i and j. The attenuation characteristic can be a fitted wave amplitude decay curve A(d) = A0 x e −αd where A0 is the initial node amplitude and a is the attenuation coefficient. a is estimated by least square method to quantify the speed of attenuation.

[0096] The space-time propagation path feature is obtained by path feature coding. For example, the diffusion speed and the attenuation coefficient can be spliced into a feature vector pi = [vi1, …, viN, a i] to describe the wave propagation characteristics of node i. The space-time propagation path feature is used to describe the diffusion law and energy attenuation mechanism of the wave in space.

[0097] In step S125, the cross-influence coefficient of the load change trend of different power supply nodes is calculated, and the cross-influence coefficient is used to describe the influence degree of the load change of any power supply node on the voltage fluctuation of another power supply node.

[0098] where the cross-influence coefficient is a scalar value used to quantify the influence degree of the load change of one node on the voltage fluctuation of another node.

[0099] In the embodiments of the present disclosure, the load change trend is an important factor affecting the voltage fluctuation. The load changes of different power supply nodes can influence each other and further affect the voltage fluctuation of each other. Therefore, the cross-influence coefficient of the load change trend of different nodes can be calculated.

[0100] First, the load change trend is extracted from the operating state feature of each power supply node, and the load change trend can be represented by a load change curve in a period of time. Then, for any two nodes, the relationship between the load change trend of one node and the voltage fluctuation of another node is analyzed.

[0101] Through statistical analysis, when the load of one node changes, the voltage fluctuation amplitude change of another node is determined, and then the cross-influence coefficient is calculated. If the load change of one node has a greater influence on the voltage fluctuation of another node, the cross-influence coefficient is larger; otherwise, the cross-influence coefficient is smaller.

[0102] In step S126, the cross-influence coefficient corresponding to the power supply node and the matching degree of the equipment operation mode identifier are combined to generate the inter-node coupling influence feature.

[0103] The cross-influence coefficient reflects the influence degree of the load change on the voltage fluctuation, and the matching degree of the equipment operation mode identifier reflects the similarity of different nodes in the equipment operation characteristics. Combining the two can more comprehensively describe the coupling influence relationship between nodes to generate the inter-node coupling influence feature.

[0104] In step S127, the space-time propagation path feature and the inter-node coupling influence feature are spliced in the time dimension to form a wave fluctuation correlation feature set with node correlation.

[0105] The space-time propagation path feature describes the propagation of voltage fluctuation in time and space, and the inter-node coupling influence feature describes the mutual influence relationship between nodes. Splicing the two features in the time dimension can integrate them into a unified feature set, i.e., a wave fluctuation correlation feature set.

[0106] In the embodiments of the present disclosure, splicing in the time dimension means that the space-time propagation path feature and the inter-node coupling influence feature in the same time segment are combined together, so that the feature set can not only reflect the voltage fluctuation propagation state at different time points, but also reflect the inter-node coupling influence at the corresponding time point, thereby completely presenting the voltage fluctuation feature with node correlation.

[0107] In a possible implementation, in step S122, the similarity index between the local voltage fluctuation sub-sequences of adjacent power supply nodes is calculated according to the local voltage fluctuation sub-sequences of each power supply node, including:

[0108] In step S1221, the local voltage fluctuation sub-sequences of adjacent power supply nodes are subjected to length normalization processing, and the fluctuation extreme point distribution information of each length-normalized local voltage fluctuation sub-sequence is extracted.

[0109] Since the local voltage fluctuation sub-sequences of different adjacent power supply nodes may differ in sampling duration or sampling point number, in order to ensure the accuracy and comparability of subsequent calculation, the sub-sequences can be subjected to length normalization processing first. By interpolation or truncation, the sub-sequences of different lengths are adjusted to the same length.

[0110] After length normalization processing, the fluctuation extreme point distribution information of each sub-sequence is extracted. The extreme points include maximum points and minimum points. The maximum point refers to a point where the voltage amplitude reaches the highest in a certain time period, and the minimum point refers to a point where the voltage amplitude reaches the lowest. For each local voltage fluctuation sub-sequence, the change of voltage amplitude is analyzed segment by segment, and the positions (i.e., the corresponding sampling time) of all maximum points and minimum points and the voltage amplitudes corresponding to the positions are determined. These information collectively constitute the fluctuation extreme point distribution information.

[0111] In step S1222, according to the positions of the maximum points in the fluctuation extreme point distribution information and the first voltage amplitude corresponding to the maximum points, the positions of the minimum points in the fluctuation extreme point distribution information and the second voltage amplitude corresponding to the minimum points, the voltage amplitude difference of the extreme points at the corresponding positions in the two local voltage fluctuation sub-sequences is calculated.

[0112] In the embodiments of the present disclosure, for the local voltage fluctuation sub-sequences of two adjacent power supply nodes, after the extreme point distribution information is extracted, the extreme point positions corresponding in time are found. The corresponding positions here refer to the extreme points at the same or similar positions on the normalized time axis.

[0113] In step S1223, the amplitude similarity component is calculated based on the cumulative sum of the voltage amplitude difference.

[0114] In the embodiments of the present disclosure, the voltage amplitude difference of the extreme points at the corresponding positions is calculated, that is, the voltage amplitude of the extreme points of one sub-sequence is subtracted from the voltage amplitude of the extreme points at the corresponding positions of the other sub-sequence to obtain a series of difference values. Then the difference values are cumulatively summed, and the amplitude similarity component is calculated according to the size of the cumulative sum. The smaller the cumulative sum is, the closer the voltage amplitudes of the extreme points of the two sub-sequences are, and the larger the amplitude similarity component is. Conversely, the larger the cumulative sum is, the smaller the amplitude similarity component is.

[0115] In step S1224, the fluctuation period characteristics of the two local voltage fluctuation sub-sequences are analyzed, the target fluctuation frequency component is extracted through Fourier transform, and the matching degree of the target fluctuation frequency component is calculated as the period similarity component.

[0116] The target fluctuation frequency component can be a main fluctuation frequency component. By analyzing the fluctuation period of the local voltage fluctuation sub-sequence, the repetition rule of the voltage fluctuation can be understood. The Fourier transform is performed on the local voltage fluctuation sub-sequences of the two adjacent power supply nodes respectively, the time domain signal is converted into the frequency domain signal, and thus the main fluctuation frequency components of the two sub-sequences are extracted.

[0117] In the embodiments of the present disclosure, the main fluctuation frequency component refers to the frequency value with relatively high energy in the frequency domain. Then the matching degree of the main fluctuation frequency components of the two sub-sequences is calculated, that is, whether the main frequency components of the two sub-sequences are the same or similar is compared. The higher the matching degree is, the more similar the fluctuation period characteristics of the two sub-sequences are, and the larger the period similarity component is. Conversely, the smaller the period similarity component is.

[0118] In step S1225, the amplitude similarity component and the period similarity component are weighted and summed according to a preset weight ratio, and the similarity index between the local voltage fluctuation sub-sequences of the adjacent power supply nodes is obtained.

[0119] In the embodiments of the present disclosure, after obtaining the amplitude similarity component and the period similarity component, they are weighted and summed according to a preset weight ratio. The preset weight ratio is determined according to the influence degree of the amplitude variation and the period variation on the consistency of the fluctuation form in the actual power supply system. For example, if the amplitude variation has a greater influence on the consistency of the fluctuation form in the subway DC power supply system, a higher weight is set for the amplitude similarity component; otherwise, a higher weight is set for the period similarity component.

[0120] The result obtained after the weighted sum is the similarity index between the local voltage fluctuation sub-sequences of the adjacent power supply nodes. In order to facilitate subsequent calculation and analysis, the similarity index is normalized to adjust its value range to a preset feature interval, such as normalizing it to between 0 and 1. The normalization processing can eliminate the dimensional differences between different similarity indexes, so that they have comparability.

[0121] In a possible implementation, in step S124, the diffusion speed and the attenuation characteristic of the voltage fluctuation from the starting node to the surrounding nodes are analyzed based on the fluctuation propagation adjacency matrix and the real-time voltage fluctuation sequence, and a space-time propagation path feature of the voltage fluctuation is generated according to the diffusion speed and the attenuation characteristic, including:

[0122] In step S1241, the starting node and the corresponding starting time of the voltage fluctuation are determined from the real-time voltage fluctuation sequence by a fluctuation starting identification algorithm.

[0123] In the embodiments of the present disclosure, the fluctuation starting identification algorithm is an algorithm specially used for detecting the starting position and time of the voltage fluctuation. The real-time voltage fluctuation sequence is analyzed node by node and time by time, and when the voltage amplitude of a node deviates from the normal stable range and this deviation lasts for a certain time, the node is determined as the starting node of the voltage fluctuation, and the starting time corresponding to the deviation is the starting time.

[0124] For example, in the real-time voltage fluctuation sequence, when the voltage amplitude of node A suddenly exceeds the normal range at a certain time and continues to deviate subsequently, node A is determined as the starting node, and the time of the sudden exceeding is the starting time.

[0125] In step S1242, based on the multi-order neighbor node search model constructed by the fluctuation propagation adjacency matrix with the starting node as the center, the adjacent power supply nodes are searched layer by layer.

[0126] In the embodiments of the present disclosure, after the starting node is determined, a multi-order neighbor node search model is constructed using the wave propagation adjacency matrix with the node as the center. The first-order neighbor node refers to the node directly connected to the starting node, i.e., the node with a non-zero element value in the row or column corresponding to the starting node in the wave propagation adjacency matrix; the second-order neighbor node refers to the node directly connected to the first-order neighbor node but not directly connected to the starting node, and so on, the adjacent nodes are expanded layer by layer. The range of all nodes to which the voltage fluctuation may propagate can be determined.

[0127] In step S1243, the fluctuation time when each of the power supply nodes first appears the voltage fluctuation meeting the preset requirement is recorded, and the time difference between each of the fluctuation times and the starting time is calculated as the fluctuation arrival time of the corresponding power supply node.

[0128] In the embodiments of the present disclosure, after the range of nodes that may be affected by the fluctuation is determined, the local voltage fluctuation subsequence of these nodes is monitored, and the time when each node first appears the obvious voltage fluctuation is recorded. The obvious voltage fluctuation refers to the voltage amplitude deviating from the normal range and reaching a certain fluctuation amplitude.

[0129] Then, the fluctuation arrival time of each node, i.e., the time difference between the time when the node first appears the obvious voltage fluctuation and the starting time of the starting node, is calculated. The fluctuation arrival time reflects the time required for the voltage fluctuation to propagate from the starting node to the node.

[0130] In step S1244, the fluctuation diffusion speed is calculated according to the fluctuation arrival time of each of the power supply nodes and the physical distance between the power supply nodes, which is determined based on the node coordinates in the power supply network topology graph.

[0131] The fluctuation diffusion speed reflects the speed of the voltage fluctuation in space, and the fluctuation diffusion speeds of different nodes may be different due to factors such as the connection mode between nodes and line impedance.

[0132] In the embodiments of the present disclosure, the physical distance between nodes can be calculated according to the coordinates of each node in the power supply network topology graph, and the coordinate information is a fixed parameter determined during the design of the power supply network. For each node, the physical distance between the node and the starting node is divided by the fluctuation arrival time to obtain the diffusion speed of the fluctuation from the starting node to the node.

[0133] In step S1245, the voltage amplitude peak value of each of the power supply nodes in the fluctuation propagation process is extracted, and the ratio of the voltage amplitude peak value to the voltage amplitude peak value of the starting node is taken as the attenuation coefficient of the corresponding power supply node.

[0134] In the embodiments of the present disclosure, in the process of voltage fluctuation propagating to each node, the voltage amplitude of the node will have a peak value, i.e., the maximum voltage amplitude in the fluctuation process. The voltage amplitude peak value of each node is extracted and compared with the voltage amplitude peak value of the starting node.

[0135] The ratio of the voltage amplitude peak value of each node to the voltage amplitude peak value of the starting node is calculated, which is the attenuation coefficient. The attenuation coefficient reflects the intensity attenuation of the voltage fluctuation in the propagation process. The smaller the attenuation coefficient, the more obvious the fluctuation intensity attenuation.

[0136] In step S1246, the fluctuation arrival time, the fluctuation diffusion speed and the attenuation coefficient of each power supply node are arranged in the order of spatial distribution of the nodes to generate a space-time propagation path feature representing the propagation timing and intensity change of the voltage fluctuation.

[0137] In the embodiments of the present disclosure, the fluctuation arrival time, the diffusion speed and the attenuation coefficient of each node are arranged in the order of spatial distribution of the nodes to form a multi-dimensional feature set, which is the space-time propagation path feature of the voltage fluctuation.

[0138] For example, the fluctuation arrival time, the diffusion speed and the attenuation coefficient of each node are listed in the order of the distance between the node and the starting node from near to far, which can clearly show the propagation order of the voltage fluctuation in time and the intensity change in space, and completely describe the propagation path feature of the voltage fluctuation.

[0139] In a possible implementation, in step S126, the matching degree of the cross-influence coefficient corresponding to the power supply node and the device operation mode identifier is combined to generate a node-coupling influence feature, including:

[0140] In step S1261, the device operation mode identifier of each power supply node is extracted to construct a mode identifier matrix.

[0141] In the embodiments of the present disclosure, the device operation mode identifier is extracted from the operation state feature of each power supply node, and the device operation mode identifier of each node is represented by a specific symbol or code, representing the device operation type and parameter setting of the node.

[0142] These device operation mode identifiers are arranged in the order of the node number to construct a mode identifier matrix. The rows and columns of the matrix correspond to the power supply nodes, and the element values in the matrix are the device operation mode identifiers of the corresponding two nodes.

[0143] In step S1262, it is determined whether the device operation mode identifiers of any two power supply nodes in the mode identification matrix are the same according to the mode identifiers of the any two power supply nodes.

[0144] In the embodiments of the present disclosure, for elements corresponding to any two node pairs in the mode identification matrix, it is compared whether the device operation mode identifiers of the elements are the same. If the device operation mode identifiers of the two nodes are the same, the matching degree of the two nodes is high; if the device operation mode identifiers of the two nodes are different, the matching degree of the two nodes is low.

[0145] The high and low of the matching degree reflects the similarity of the two nodes in the device operation characteristics. The same device operation mode means that the two nodes may have more similarities in the power supply capability, the adjustment mode and the like.

[0146] In step S1263, the mode influence coefficient of the any two power supply nodes is determined according to the mode matching degree of the any two power supply nodes.

[0147] In the embodiments of the present disclosure, in order to facilitate subsequent calculation and analysis, the matching degree is converted into a specific mode influence coefficient value. It is set that the mode influence coefficient value corresponding to the high matching degree is larger, and the mode influence coefficient value corresponding to the low matching degree is smaller.

[0148] For example, the mode influence coefficient corresponding to the high matching degree can be set as a larger numerical range, and the mode influence coefficient corresponding to the low matching degree can be set as a smaller numerical range. The specific numerical range is determined according to the actual situation, and the influence degree of different matching degrees can be effectively distinguished.

[0149] In step S1264, the cross-influence coefficient and the mode influence coefficient are multiplied to obtain a comprehensive influence coefficient.

[0150] In the embodiments of the present disclosure, the cross-influence coefficient reflects the influence of the load change on the voltage fluctuation, and the mode influence coefficient reflects the degree of the influence of the device operation mode similarity on the nodes. The multiplication of the two coefficients can obtain a comprehensive influence coefficient which comprehensively considers the two factors.

[0151] The larger the comprehensive influence coefficient is, the more significant the coupling influence between the two nodes is; on the contrary, the smaller the comprehensive influence coefficient is, the weaker the coupling influence between the two nodes is.

[0152] In step S1265, according to the topology structure of the power supply network, the comprehensive influence coefficient is arranged according to the node connection relationship to form a coupling influence matrix between power supply nodes.

[0153] In the embodiments of the present disclosure, the topology structure of the power supply network indicates the connection relationship between nodes, and the calculated comprehensive influence coefficients are arranged according to the corresponding node pairs to form a coupling influence matrix between nodes according to the connection relationship.

[0154] The rows and columns of the matrix correspond to each power supply node respectively, and the element values in the matrix are the comprehensive influence coefficients between the corresponding two nodes. Through the matrix, it can be directly seen which nodes have strong coupling influence.

[0155] In step S1266, elements in the coupling influence matrix between the power supply nodes whose comprehensive influence coefficients are less than a preset threshold are deleted, and elements whose comprehensive influence coefficients are greater than the preset threshold are retained to generate a coupling influence feature between nodes.

[0156] In the embodiments of the present disclosure, the coupling influence matrix may contain a large number of elements and information. In order to simplify subsequent processing, feature compression processing can be performed. The feature compression processing can use principal component analysis and other methods to extract the main influence relationship information in the matrix and ignore the secondary and less influential information.

[0157] After the feature compression processing, the obtained result is the coupling influence feature between nodes, which is a refined representation of the coupling influence relationship between nodes, can reduce the data amount while retaining the key information, and is convenient for subsequent model processing and analysis.

[0158] In a possible implementation manner, referring to FIG. 13, Figure 3 As shown in FIG. 13, in step S13, the pre-trained fluctuation suppression decision model is called to perform dynamic suppression strategy generation processing on the fluctuation correlation feature set to obtain a voltage adjustment parameter sequence for each power supply node, including:

[0159] In step S131, the fluctuation correlation feature set is input into the feature encoding layer of the fluctuation suppression decision model to perform high-dimensional feature mapping processing and generate an encoded feature vector with time sequence dependence.

[0160] In the embodiments of the present disclosure, the feature encoding layer of the fluctuation suppression decision model is responsible for processing the input fluctuation correlation feature set and converting it into a high-dimensional feature form more suitable for subsequent processing of the model. The fluctuation correlation feature set contains information of multiple dimensions and multiple time points, and the feature encoding layer integrates and maps these information through a series of nonlinear transformations and feature extraction operations.

[0161] During the processing, the feature encoding layer focuses on the time sequence relationship between the features, associates the feature information at different time points, and generates an encoded feature vector with time sequence dependence. Such an encoded feature vector can more effectively express the key information in the fluctuation correlation feature set. For example, the feature encoding layer can extract feature patterns in a local time range through convolution operation, and then capture the dependence relationship between different time segments through recurrent connection, thereby generating a high-dimensional encoded feature vector containing time sequence information.

[0162] In step S132, the node importance weight allocation processing is performed on the encoded feature vector by the node attention module of the fluctuation suppression decision model, and the attention coefficients of each power supply node are generated.

[0163] The node attention module determines the importance of different nodes in the current voltage fluctuation suppression decision according to the feature information of each power supply node in the encoded feature vector, and then allocates the corresponding attention coefficients to them. Nodes with high importance will be allocated higher attention coefficients. In step S133, the encoded feature vector is weighted and aggregated based on the attention coefficients to obtain an aggregated feature vector.

[0164] In the embodiments of the present disclosure, the generated attention coefficients are used for weighted aggregation processing of the encoded feature vector. Specifically, the feature components of each power supply node in the encoded feature vector are multiplied by the corresponding attention coefficients, and then the weighted feature components are spliced to form an aggregated feature vector.

[0165] Through this processing, the feature components of nodes with high attention coefficients will occupy a more important position in the aggregated feature vector, and their information will be strengthened, while the influence of the feature components of nodes with low attention coefficients will be relatively weakened. The aggregated feature vector can focus on the information of key nodes.

[0166] In step S134, the aggregated feature vector is input into the time sequence prediction layer of the fluctuation suppression decision model, and multi-step voltage fluctuation prediction processing is performed to generate a voltage fluctuation trend prediction sequence for predicting the future period.

[0167] The main function of the time sequence prediction layer is to predict the voltage fluctuation of each power supply node in the future period according to the input aggregated feature vector. Multi-step prediction means not only predicting the voltage fluctuation at a future time point, but also predicting the voltage fluctuation at multiple consecutive time points to form a sequence.

[0168] In the embodiments of the present disclosure, the time sequence prediction layer can adopt a structure such as a recurrent neural network to capture the time sequence dependence contained in the aggregated feature vector by using its processing capability for time sequence data, and then predict the future voltage fluctuation trend.

[0169] In step S135, according to the voltage fluctuation trend prediction sequence and the node-to-node coupling influence feature in the fluctuation correlation feature set, a dynamic change curve of the adjustment amount of each power supply node for voltage adjustment is determined.

[0170] The voltage fluctuation trend prediction sequence indicates the possible changes of the voltage of each node in the future, and in combination with the node-to-node coupling influence feature in the fluctuation correlation feature set, i.e., the mutual influence relationship between the nodes, the dynamic change curve of the adjustment amount of each power supply node for suppressing voltage fluctuation with time can be calculated, i.e., the dynamic change curve of the voltage adjustment amount.

[0171] In step S136, the dynamic change curve of the adjustment amount is discretized according to a preset time interval to generate a voltage adjustment parameter sequence for each power supply node.

[0172] The dynamic change curve of the voltage adjustment amount is continuous, and the actual voltage adjustment operation can be performed according to a certain time interval, so the curve can be discretized. According to the preset time interval, the corresponding adjustment amount values are extracted from the dynamic change curve, and these values are arranged in time sequence to form the voltage adjustment parameter sequence for each power supply node.

[0173] In the embodiments of the present disclosure, the preset time interval can be determined according to the adjustment accuracy and response speed of the power supply system, and the smaller the time interval, the more accurate the adjustment parameter sequence, but at the same time, the data volume and processing complexity are also increased.

[0174] In a possible implementation, in step S132, the node attention module of the fluctuation suppression decision model performs node importance weight allocation processing on the encoded feature vector to generate an attention coefficient of each power supply node, including:

[0175] The feature components of each power supply node in the encoded feature vector are analyzed, and a fluctuation sensitivity index of each feature component is extracted, and the fluctuation sensitivity index is positively correlated with the change rate of the voltage fluctuation amplitude.

[0176] In the embodiments of the present disclosure, the encoded feature vector is first analyzed to separate the feature components of each power supply node contained therein. Each feature component contains various information of the corresponding node in terms of voltage fluctuation. The fluctuation sensitivity index is extracted from the feature components, and the index can reflect the sensitivity of the node to voltage fluctuation.

[0177] The fluctuation sensitivity index is positively correlated with the change rate of the voltage fluctuation amplitude, i.e., when the change rate of the voltage fluctuation amplitude of the node is larger, the value of the fluctuation sensitivity index is also larger, indicating that the node is more sensitive to voltage fluctuation and can be paid more attention to in the suppression decision.

[0178] According to the fluctuation sensitivity indicators, an initial attention degree vector is constructed, each element in the initial attention degree vector corresponding to an initial attention degree value of a power supply node.

[0179] In the embodiments of the present disclosure, after obtaining the fluctuation sensitivity indicators of each power supply node, the indicators are arranged in the order of node numbers as initial attention degree values to construct an initial attention degree vector. Each element in the initial attention degree vector corresponds to a power supply node, and the greater the value of the element, the higher the importance of the node in the initial stage.

[0180] The deviation degree of the load change trend in the running state feature of each power supply node from the average load change trend of the subway DC power supply system is calculated to generate a load deviation coefficient.

[0181] In the embodiments of the present disclosure, the system average load change trend is obtained by calculating the average value of the load change trends of all power supply nodes. For each power supply node, the deviation degree between its load change trend and the system average load change trend is calculated, and the greater the deviation degree, the more obvious the difference between the load change of the node and the load change of the overall system.

[0182] According to the deviation degree, a load deviation coefficient is generated, and the greater the deviation degree, the greater the load deviation coefficient of the node, and the smaller the deviation degree, the smaller the load deviation coefficient of the node.

[0183] The load deviation coefficient is introduced into an attention calculation function, the initial attention degree vector is dynamically adjusted, and the adjusted attention degree vector is processed by neighborhood propagation in combination with the cross-influence coefficient in the coupling influence feature between nodes.

[0184] In the embodiments of the present disclosure, the attention calculation function is a function in the node attention module for adjusting the attention degree value, which dynamically adjusts the initial attention degree vector according to the input load deviation coefficient. For nodes with a larger load deviation coefficient, i.e., nodes with a larger load fluctuation, the attention calculation function increases the attention degree value of the node in the initial attention degree vector to highlight the importance of these nodes in voltage fluctuation suppression. The adjusted attention degree vector can better reflect the importance difference of different nodes under the current load condition.

[0185] The cross-influence coefficient in the coupling influence feature between nodes reflects the mutual influence degree between nodes, and the cross-influence coefficient between nodes with strong coupling relationship is larger. On the basis of the adjusted attention degree vector, neighborhood propagation processing is performed in combination with the cross-influence coefficient, i.e., the attention degree values of nodes with strong coupling relationship influence each other and tend to be similar.

[0186] For example, if node A and node B have a strong coupling relationship, when the attention value of node A is high, the attention value of node B will also be increased through the neighborhood propagation process, and vice versa. In this way, it can be ensured that in the decision-making process, the nodes that have a greater mutual influence are given consistent attention.

[0187] The attention value vector after the neighborhood propagation process is normalized to generate the attention coefficients of each power supply node.

[0188] In the embodiments of the present disclosure, the element values in the attention value vector after the propagation process may have large differences. In order to facilitate subsequent weighted aggregation processing, the element values can be normalized to a preset range, such as between 0 and 1.

[0189] Each element in the normalized attention value vector is the final attention coefficient of the corresponding power supply node, and these coefficients can accurately reflect the importance of each node in the current voltage fluctuation suppression decision.

[0190] In one possible implementation, in step S135, the determination of the adjustment amount dynamic change curve of each power supply node for voltage adjustment according to the voltage fluctuation trend prediction sequence and the coupling influence feature between nodes in the fluctuation correlation feature set comprises:

[0191] According to the time point at which the voltage prediction value in the voltage fluctuation trend prediction sequence is greater than the preset voltage stability threshold, the fluctuation interval for voltage adjustment is determined.

[0192] In the embodiments of the present disclosure, the preset voltage stability threshold is determined according to the safety operation standard of the subway DC power supply system and is a basis for judging whether the voltage is adjusted. The voltage prediction value at each time point in the voltage fluctuation trend prediction sequence is compared with the voltage stability threshold. If the prediction value exceeds the range of the stability threshold, the interval in which the time point is located is determined as the fluctuation interval for adjustment. The fluctuation interval for adjustment may be a continuous time period or multiple discontinuous time periods, depending on the comparison result of the voltage fluctuation trend prediction sequence and the stability threshold.

[0193] The deviation amount of the voltage prediction value at each time point in each fluctuation interval from the preset voltage stability threshold is calculated to obtain the basic adjustment amount corresponding to each time point.

[0194] In the embodiments of the present disclosure, for each determined fluctuation interval for adjustment, the difference between the voltage prediction value at each time point in the interval and the voltage stability threshold is calculated, and the difference is the basic adjustment amount. The basic adjustment amount reflects the adjustment amount of the node itself without considering the influence of other nodes in order to return the voltage to the stability threshold range.

[0195] For example, if the voltage prediction value is higher than the upper limit of the stability threshold in a certain fluctuation interval, the basic adjustment amount is positive, indicating that the voltage needs to be lowered; if the voltage prediction value is lower than the lower limit of the stability threshold, the basic adjustment amount is negative, indicating that the voltage needs to be raised.

[0196] According to the inter-node coupling influence feature in the fluctuation correlation feature set, a set of associated nodes that have a significant influence on the voltage fluctuation of each supply node is determined.

[0197] In the embodiments of the present disclosure, the comprehensive influence coefficient and other information in the inter-node coupling influence feature reflect the influence degree between nodes. By analyzing these features, other nodes that have a significant influence on the voltage fluctuation of the current node can be found, and these nodes collectively constitute the set of associated nodes.

[0198] The determination of significant influence can be realized by setting an influence degree threshold. When the comprehensive influence coefficient between two nodes is greater than the threshold, it is considered that there is a significant influence between them.

[0199] The adjustment amount in the voltage adjustment parameter sequence of the set of associated nodes of each supply node is extracted, and the corresponding coupling adjustment contribution value of the power supply node is calculated.

[0200] In the embodiments of the present disclosure, the voltage adjustment parameter sequence of the set of associated nodes can be generated in the previous processing or calculated synchronously in this processing. The adjustment amount of these associated nodes in the corresponding fluctuation interval is extracted, and then the coupling adjustment contribution value of the current node is calculated according to the cross-influence coefficient and other information in the inter-node coupling influence feature.

[0201] The calculation of the coupling adjustment contribution value can be the sum of the products of the adjustment amount of the associated nodes and the corresponding cross-influence coefficient, which reflects the influence degree of the adjustment operation of the associated nodes on the voltage fluctuation of the current node.

[0202] The basic adjustment amount at the time point is superimposed with the corresponding coupling adjustment contribution value to obtain the total adjustment amount of the power supply node.

[0203] In the embodiments of the present disclosure, the basic adjustment amount is the adjustment amount required by the current node itself, and the coupling adjustment contribution value is the influence of the adjustment of the associated nodes on the current node. By superimposing the two, the total adjustment requirement of the current node to achieve voltage stability is obtained.

[0204] The total adjustment amount comprehensively considers the voltage deviation of itself and the influence of the associated nodes, and can more accurately reflect the actual adjustment amount required by the current node.

[0205] Based on the total adjustment amount, a curve fitting algorithm is used to determine an adjustment amount dynamic change curve for voltage adjustment of each power supply node.

[0206] In the embodiments of the present disclosure, the curve fitting algorithm can fit a smooth curve according to the values of the total adjustment demand at each time point in the fluctuation interval, and the curve is the voltage adjustment amount dynamic change curve of the power supply node. The voltage adjustment amount dynamic change curve can clearly show the continuous change of the adjustment amount with time in the future fluctuation interval.

[0207] In a possible implementation, referring to FIG. 14, in step S14, the generation of the set of fluctuation suppression execution instructions for node coordination according to the sequence of voltage adjustment parameters and the operating state characteristics of each power supply node includes: Figure 4

[0208] In step S141, the sequence of voltage adjustment parameters of each power supply node is analyzed to extract the voltage adjustment amount and the adjustment duration parameter at each time point.

[0209] In the embodiments of the present disclosure, the sequence of voltage adjustment parameters of each power supply node is analyzed to extract the voltage adjustment amount required at each time point and the duration of the adjustment amount, i.e., the adjustment duration parameter.

[0210] These parameters are the basis for generating execution instructions, and clearly indicate when each node adjusts, how much it adjusts, and how long it adjusts.

[0211] In step S142, the adjustment authority level of each power supply node is determined according to the device operating mode identifier in the operating state characteristics of each power supply node, and the adjustment authority level is used to limit the maximum range of the voltage adjustment amount.

[0212] Different device operating modes correspond to different adjustment capabilities and safety limits. According to the device operating mode identifier of each power supply node, the corresponding adjustment authority level can be determined. The adjustment authority level specifies the maximum adjustment range that the node can allow during voltage adjustment, preventing damage to devices or systems due to excessive adjustment.

[0213] In the embodiments of the present disclosure, a node in standby mode may have a lower adjustment authority level and a smaller maximum adjustment range, while a node in normal operating mode and with good device performance may have a higher adjustment authority level and allow a larger range of adjustment.

[0214] In step S143, the voltage adjustment amount is reasonably checked based on the adjustment authority level, and the adjustment amount that exceeds the authority range is truncated and corrected.

[0215] ​In the embodiments of the present disclosure, the voltage adjustment amount of each time point extracted is reasonably checked according to the determined adjustment authority level. If the adjustment amount is within the maximum range allowed by the authority level, the adjustment amount is considered reasonable; if the adjustment amount exceeds the authority range, the adjustment amount can be truncated and corrected to the maximum or minimum value within the authority range.

[0216] Through the checking and correction processing, it is ensured that the adjustment operation of each node is performed within a safe and feasible range.

[0217] In step S144, the coordination between the voltage adjustment amounts of different power supply nodes at the same time point is analyzed, and a consistency coefficient of the adjustment direction is calculated, which is used to judge the coordination degree of the adjustment operation between the power supply nodes.

[0218] At the same time point, the voltage adjustment operations of different power supply nodes can be coordinated with each other to achieve the best fluctuation suppression effect. The consistency coefficient of the adjustment direction is calculated by analyzing the adjustment direction of each node at the same time point, i.e., whether to raise or lower the voltage.

[0219] In the embodiments of the present disclosure, the calculation of the consistency coefficient can be realized by the proportion of the number of nodes with the same adjustment direction to the total number of nodes, and the higher the proportion, the greater the consistency coefficient, indicating that the coordination degree of the adjustment operation between the nodes is higher; otherwise, the smaller the consistency coefficient, the lower the coordination degree.

[0220] In step S145, when the consistency coefficient is lower than a preset threshold, a cooperative optimization algorithm is called to generate a node cooperative adjustment scheme for adjusting the voltage adjustment amount of the target power supply node, the node cooperative adjustment scheme makes the consistency coefficient between the target power supply nodes greater than the preset threshold, wherein the target power supply node includes at least one of the power supply nodes with a consistency coefficient lower than the preset threshold.

[0221] In the embodiments of the present disclosure, when the consistency coefficient is lower than the preset threshold, it indicates that the coordination of the adjustment operation between the nodes is poor, which may affect the fluctuation suppression effect, and even cause mutual interference. At this time, the voltage adjustment amount of the related node can be adjusted synchronously by calling the cooperative optimization algorithm to improve the coordination.

[0222] In step S146, the voltage adjustment amount, the adjustment duration parameter and the corresponding power supply node identifier after correction and adjustment are associated and encapsulated to generate a set of node cooperative fluctuation suppression execution instructions.

[0223] In the embodiments of the present disclosure, the node coordination regulation scheme determines the regulation amount and execution time of each node, and on this basis, the regulation duration parameter of each power supply node is updated according to the size of the regulation amount and the response characteristics of the system. The regulation duration parameter can be matched with the regulation amount and the execution time to ensure that the regulation operation can fully play a role and make the voltage stable within the target range.

[0224] The updated voltage regulation parameter sequence and the regulation duration parameter are encapsulated in a preset format, and these encapsulated information constitutes the core content of the node coordination fluctuation suppression execution instruction set. In the encapsulation process, the parameters are structured to facilitate parsing and execution by the power supply control module.

[0225] In a possible implementation, in step S145, when the consistency coefficient is lower than the preset threshold, a coordination optimization algorithm is called to generate a node coordination regulation scheme for synchronously adjusting the voltage regulation amount of the target power supply node, including:

[0226] The power supply node with the consistency coefficient lower than the preset threshold is added to the set of nodes to be optimized as a power supply node to be regulated.

[0227] In the embodiments of the present disclosure, according to the calculation result of the consistency coefficient, the nodes with poor consistency in the adjustment direction at the same time point are screened out, and the consistency coefficient is lower than the preset threshold, to form the set of nodes to be optimized. The adjustment operation of these nodes needs to be adjusted to improve the coordination.

[0228] An optimization function is constructed to target the regulation complementarity between power supply nodes, and the optimization function is positively correlated with the coordination effect of the voltage regulation amount of the power supply node.

[0229] In the embodiments of the present disclosure, the optimization function aims to improve the complementarity between the regulation amounts of the nodes in the set of nodes to be optimized, that is, by adjusting the regulation amount, the adjustment operations of the nodes can cooperate with each other to form a better coordination effect. The value of the optimization function is positively correlated with the coordination effect of the node regulation amount, and the better the coordination effect, the larger the value of the optimization function.

[0230] For example, the optimization function can include a coordination term between the regulation amounts of the nodes, and when the adjustment directions between the nodes are consistent or can complement each other, the value of the coordination term is larger, thereby increasing the value of the optimization function.

[0231] The constraint conditions of the optimization variable are set, and the constraint conditions include the upper and lower limits of the voltage regulation amount, the regulation rate limit, and the node load capacity limit.

[0232] In the embodiments of the present disclosure, in order to ensure the feasibility and safety of the optimized adjustment amount, a constraint condition of the optimization variable can be set. The upper and lower limits of the voltage adjustment amount are determined by the adjustment authority level of each node; the adjustment rate limit refers to the maximum change amplitude of the adjustment amount per unit time, which prevents the adjustment from being too fast to cause impact on the system; and the node load capacity limit considers the load bearing capacity of the node to ensure that the adjusted voltage does not cause the load to exceed its capacity range.

[0233] These constraint conditions together constitute the feasible region of the optimization problem, and the optimization process must be carried out within the feasible region.

[0234] The particle swarm optimization algorithm is called to solve the optimization function to obtain the optimal adjustment amount combination of the set of nodes to be optimized.

[0235] The particle swarm optimization algorithm is a kind of optimization algorithm based on swarm intelligence, which simulates the movement of particles in the search space, constantly updates the position and speed of the particles, and finds the optimal solution of the optimization function.

[0236] In the embodiments of the present disclosure, the adjustment amount of the set of nodes to be optimized is taken as the position variable of the particle, and the optimization function is solved under the constraint condition to finally obtain the optimal adjustment amount combination that maximizes the value of the optimization function. The optimal adjustment amount combination can make the adjustment operation of the set of nodes to be optimized have the best synergistic effect.

[0237] The optimal adjustment amount combination is compared with the voltage adjustment amount before the optimization adjustment of each node to be adjusted to obtain the adjustment amplitude of each node to be adjusted.

[0238] In the embodiments of the present disclosure, the optimal adjustment amount combination obtained by the particle swarm optimization algorithm is compared with the original voltage adjustment amount to calculate the change amplitude of the adjustment amount of each node, i.e., the adjustment amplitude. The adjustment amplitude reflects the degree to which the adjustment amount of each node can be changed in order to achieve the optimal synergistic effect.

[0239] According to the adjustment amplitude of the node to be adjusted, the voltage adjustment amount of the target power supply node is synchronously corrected to generate an adjusted voltage adjustment parameter sequence.

[0240] In the embodiments of the present disclosure, according to the calculated adjustment amplitude, the voltage adjustment amount of the related node is synchronously corrected, i.e., the original adjustment amount is changed according to the adjustment amplitude to form a new voltage adjustment parameter sequence. The synchronous correction ensures that the change of the adjustment amount of each node can be coordinated with each other, avoiding the destruction of the overall synergy caused by the individual adjustment of a node. For example, if the adjustment amplitude of node A is to increase a certain adjustment amount and the adjustment amplitude of node B is to decrease a corresponding adjustment amount, the synchronous correction will ensure that the adjustment amount changes of the two nodes are carried out at the same time to achieve the expected synergistic effect.

[0241] The consistency of the voltage adjustment parameter sequence after the synchronization correction is verified, and the consistency coefficient of the adjustment direction is recalculated until the consistency coefficient reaches above the preset threshold.

[0242] In the embodiments of the present disclosure, after the synchronization correction, the voltage adjustment parameter sequence needs to be verified again for consistency. The consistency coefficient of the adjustment direction of each node at the same time point is recalculated, and whether it reaches the preset threshold is checked. If the consistency coefficient is still below the threshold, the process of synchronization correction can be repeated, the optimization function is reconstructed, the constraint condition is set, the particle swarm optimization algorithm is called to solve and the synchronization correction is performed until the consistency coefficient meets the requirements. This process of multiple verification and adjustment can ensure that the final voltage adjustment parameter sequence has good consistency.

[0243] The voltage adjustment parameter sequence verified for consistency is associated with the corresponding timestamp to generate a node coordination adjustment scheme for synchronously adjusting the voltage adjustment amount of the target power supply node.

[0244] In the embodiments of the present disclosure, after the voltage adjustment parameter sequence passes the consistency verification, each adjustment amount is associated with the corresponding timestamp. The timestamp explicitly specifies the specific time point at which each adjustment amount is executed, so that the adjustment scheme has a clear arrangement in the time dimension. Through this association, the final node coordination adjustment scheme is formed, which specifies the adjustment amount of each power supply node at different time points in detail.

[0245] In the embodiments of the present disclosure, a voltage fluctuation suppression device of a subway DC power supply system is provided, which comprises a memory having a computer program stored thereon, and a processor configured to execute the computer program stored in the memory to implement the voltage fluctuation suppression method of the subway DC power supply system according to any one of the preceding embodiments.

[0246] Figure 5 The voltage fluctuation suppression device 100 of the subway DC power supply system shown comprises a processor 1001 and a memory 1003. The processor 1001 and the memory 1003 are connected, for example, through a bus 1002. Optionally, the voltage fluctuation suppression device 100 of the subway DC power supply system can further comprise a communication component 1004, which can be used for data interaction, such as data sending and / or data receiving, between the device 100 and other devices. It should be noted that the communication component 1004 is not limited to one in actual scheduling, and the structure of the voltage fluctuation suppression device 100 of the subway DC power supply system does not constitute a limitation on the embodiments of the present disclosure.

[0247] The processor 1001 can be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic device, transistor logic device, hardware component or any combination thereof. It can implement or execute the various exemplary logical blocks, modules and circuits described in connection with the disclosure. The processor 1001 can also be a combination of computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0248] The bus 1002 can include a path for transmitting information between the above-mentioned components. The bus 1002 can be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The bus 1002 can be divided into an address bus, a data bus, a control bus, etc. For convenience of representation, Figure 5 In the figure, only one thick line is used to represent the bus, but it does not mean that there is only one bus or only one type of bus.

[0249] The memory 1003 can be a ROM (Read Only Memory) or other type of static storage device that can store static information and instructions, a RAM (Random Access Memory) or other type of dynamic storage device that can store information and instructions, an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, an optical disk storage (including a compact disk, a laser disk, an optical disk, a digital versatile disk, a Blu-ray disk, etc.), a magnetic disk storage medium, other magnetic storage device, or any other medium that can be used to carry or store program codes and can be read by a computer, without limitation.

[0250] The memory 1003 is configured to store program codes for implementing the embodiments of the present disclosure, and the processor 1001 is configured to control the execution of the program codes stored in the memory 1003. The processor 1001 is configured to execute the program codes stored in the memory 1003 to implement the steps of the above-mentioned embodiments of the voltage fluctuation suppression method for metro DC power supply system.

[0251] The preferred embodiments of the present disclosure are described in detail above with reference to the drawings, but the present disclosure is not limited to the specific details of the above-described embodiments. Various changes, modifications, replacements and variations can be made to the embodiments within the technical concept of the present disclosure, and all the changes, modifications, replacements and variations shall fall within the protection scope of the present disclosure.

[0252] In addition, it should be noted that each specific technical feature described in the above specific embodiments can be combined in any appropriate manner without contradiction, and shall be considered as disclosed in the present disclosure. In order to avoid unnecessary repetition, various possible combinations are not described again in the present disclosure. The technical scope of the present application shall be determined by the scope of the claims.

Claims

1. A method of suppressing voltage fluctuation in a metro DC power supply system, characterized by, The method comprises: obtaining voltage amplitude variation information of the subway DC power supply system at different sampling times, and determining a real-time voltage fluctuation sequence of the subway DC power supply system and an operating state feature of each power supply node in the subway DC power supply system according to the voltage amplitude variation information; performing spatio-temporal correlation mapping processing on the real-time voltage fluctuation sequence and the operating state feature to generate a fluctuation correlation feature set with node correlation, wherein the fluctuation correlation feature set comprises a spatio-temporal propagation path feature of voltage fluctuation and an inter-node coupling influence feature; calling a pre-trained fluctuation suppression decision model to perform dynamic suppression strategy generation processing on the fluctuation correlation feature set to obtain a voltage regulation parameter sequence for each power supply node; generating a node-coordinated fluctuation suppression execution instruction set according to the voltage regulation parameter sequence and the operating state feature of each power supply node; triggering voltage regulation operations of each power supply node to suppress voltage fluctuation according to the fluctuation suppression execution instruction set; wherein the operating state feature comprises a load change trend and a device operating mode identifier of each power supply node, and the spatio-temporal correlation mapping processing on the real-time voltage fluctuation sequence and the operating state feature to generate a fluctuation correlation feature set with node correlation comprises: partitioning the real-time voltage fluctuation sequence according to the power supply nodes to obtain a local voltage fluctuation sub-sequence corresponding to each power supply node; calculating a similarity index between local voltage fluctuation sub-sequences of adjacent power supply nodes according to the local voltage fluctuation sub-sequences corresponding to each power supply node, wherein the similarity index is used to represent a consistency degree of fluctuation form; constructing a fluctuation propagation adjacency matrix between the power supply nodes according to the similarity index and a topological connection relationship of a power supply network in the subway DC power supply system; based on the fluctuation propagation adjacency matrix and the real-time voltage fluctuation sequence, analyzing a diffusion speed and a decay characteristic of voltage fluctuation from a starting node to surrounding nodes, and generating a spatio-temporal propagation path feature of voltage fluctuation according to the diffusion speed and the decay characteristic; calculating a cross-influence coefficient of the load change trend of different power supply nodes, wherein the cross-influence coefficient is used to describe an influence degree of load change of any power supply node on voltage fluctuation of another power supply node; generating an inter-node coupling influence feature in combination with a matching degree of the cross-influence coefficient and the device operating mode identifier corresponding to the power supply nodes; performing spatio-temporal correlation mapping processing on the real-time voltage fluctuation sequence and the operating state feature to generate a fluctuation correlation feature set with node correlation, comprising:

2. The voltage fluctuation suppression method for a subway DC power supply system according to Claim 1, wherein performing length normalization processing on local voltage fluctuation sub-sequences of adjacent power supply nodes to extract fluctuation extreme point distribution information of each length-normalized local voltage fluctuation sub-sequence; ​ According to the position of the maximum point in the fluctuation extreme point distribution information and the first voltage amplitude corresponding to the maximum point, the position of the minimum point in the fluctuation extreme point distribution information and the second voltage amplitude corresponding to the minimum point, the voltage amplitude difference of the extreme points at the corresponding positions in the two local voltage fluctuation subsequences is calculated; Based on the cumulative sum of the voltage amplitude difference, the amplitude similarity component is calculated; The fluctuation cycle characteristics of the two local voltage fluctuation subsequences are analyzed, the target fluctuation frequency component is extracted through Fourier transform, and the matching degree of the target fluctuation frequency component is calculated as the cycle similarity component; According to the preset weight proportion, the amplitude similarity component and the cycle similarity component are weighted and summed to obtain the similarity index between the local voltage fluctuation subsequences of adjacent power supply nodes.

3. The voltage fluctuation suppression method for a subway DC power supply system according to Claim 1, wherein Based on the fluctuation propagation adjacency matrix and the real-time voltage fluctuation sequence, the diffusion speed and attenuation characteristics of voltage fluctuation from the starting node to the surrounding nodes are analyzed, and the space-time propagation path characteristics of voltage fluctuation are generated according to the diffusion speed and the attenuation characteristics, including: The starting node and the corresponding starting time of the voltage fluctuation are determined from the real-time voltage fluctuation sequence through a fluctuation starting identification algorithm; Based on the multi-order neighbor node search model constructed by the fluctuation propagation adjacency matrix with the starting node as the center, the adjacent power supply nodes are expanded layer by layer; The fluctuation time when each power supply node first appears voltage fluctuation meeting the preset requirement is recorded, and the time difference between each fluctuation time and the starting time is calculated as the fluctuation arrival time of the corresponding power supply node; According to the fluctuation arrival time of each power supply node and the physical distance between the power supply nodes, the fluctuation diffusion speed is calculated, and the physical distance is determined based on the node coordinates in the power supply network topology graph; The voltage amplitude peak value of each power supply node in the fluctuation propagation process is extracted, and the ratio of the voltage amplitude peak value to the voltage amplitude peak value of the starting node is taken as the attenuation coefficient of the corresponding power supply node; The fluctuation arrival time, the fluctuation diffusion speed and the attenuation coefficient of each power supply node are arranged in the order of spatial distribution of nodes to generate the space-time propagation path characteristics representing the propagation time sequence and intensity change of the voltage fluctuation.

4. The voltage fluctuation suppression method for a subway DC power supply system according to Claim 1, wherein The matching degree of the cross-influence coefficient corresponding to the power supply node and the device operation mode identifier is combined to generate the node coupling influence feature, including: The device operation mode identifier of each power supply node is extracted to construct a mode identifier matrix; According to whether the device operation mode identifiers of any two power supply nodes in the mode identifier matrix are the same, the mode matching degree between the any two power supply nodes is determined; According to the mode matching degree of the any two power supply nodes, the mode influence coefficient of the any two power supply nodes is determined; The cross-influence coefficient and the mode influence coefficient are multiplied to obtain the comprehensive influence coefficient; According to the topology of the power supply network, the comprehensive influence coefficients are arranged according to the node connection relationship to form a coupling influence matrix between power supply nodes; The elements in the coupling influence matrix between the power supply nodes are deleted, and the elements with the comprehensive influence coefficients greater than the preset threshold are retained to generate a node coupling influence feature.

5. The voltage fluctuation suppression method for a subway DC power supply system according to Claim 1, wherein The pre-trained fluctuation suppression decision model is called to perform dynamic suppression strategy generation processing on the fluctuation correlation feature set, and voltage regulation parameter sequences for each power supply node are obtained, including: The fluctuation correlation feature set is input into the feature encoding layer of the fluctuation suppression decision model for high-dimensional feature mapping processing to generate an encoded feature vector with time sequence dependence; The node importance weight distribution processing of the encoded feature vector is performed through the node attention module of the fluctuation suppression decision model to generate an attention coefficient of each power supply node; The encoded feature vector is weighted and aggregated based on the attention coefficient to obtain an aggregated feature vector; The aggregated feature vector is input into the time sequence prediction layer of the fluctuation suppression decision model for multi-step voltage fluctuation prediction processing to generate a voltage fluctuation trend prediction sequence for predicting future time periods; According to the voltage fluctuation trend prediction sequence and the node coupling influence feature in the fluctuation correlation feature set, a regulation amount dynamic change curve for voltage regulation of each power supply node is determined; The regulation amount dynamic change curve is discretized according to a preset time interval to generate a voltage regulation parameter sequence for each power supply node.

6. The voltage fluctuation suppression method for a subway DC power supply system according to Claim 5, wherein The node importance weight distribution processing of the encoded feature vector is performed through the node attention module of the fluctuation suppression decision model to generate an attention coefficient of each power supply node, including: The feature components of each power supply node in the encoded feature vector are analyzed, and the fluctuation sensitivity index of each feature component is extracted, wherein the fluctuation sensitivity index is positively correlated with the change rate of the voltage fluctuation amplitude; An initial attention vector is constructed according to the fluctuation sensitivity index, and each element in the initial attention vector corresponds to an initial attention value of a power supply node; The deviation degree of the load change trend in the operating state feature of each power supply node from the average load change trend of the subway DC power supply system is calculated to generate a load deviation coefficient; The load deviation coefficient is introduced into an attention calculation function to dynamically adjust the initial attention vector, and the adjusted attention vector is subjected to neighborhood propagation processing in combination with the cross-influence coefficient in the node coupling influence feature; The attention coefficient of each power supply node is generated by normalizing the attention vector after neighborhood propagation processing.

7. The voltage fluctuation suppression method for a subway DC power supply system according to claim 5, characterized by, According to the voltage fluctuation trend prediction sequence and the node coupling influence feature in the fluctuation correlation feature set, a regulation amount dynamic change curve for voltage regulation of each power supply node is determined, including: According to the time point at which the voltage prediction value in the voltage fluctuation trend prediction sequence is greater than a preset voltage stability threshold, a fluctuation interval for voltage regulation is determined; Calculate the deviation of the voltage prediction value at each time point in each fluctuation interval from the preset voltage stability threshold to obtain the basic adjustment amount corresponding to each time point; According to the coupling influence characteristics between nodes in the fluctuation correlation characteristic set, determine the associated node set which has significant influence on the voltage fluctuation of each power supply node; Extract the adjustment amount in the voltage adjustment parameter sequence of the associated node set of each power supply node, and calculate the coupling adjustment contribution value of the corresponding power supply node; Superimpose the basic adjustment amount at the time point and the corresponding coupling adjustment contribution value to obtain the total adjustment amount of the corresponding power supply node; Based on the total adjustment amount, determine the adjustment amount dynamic change curve of each power supply node for voltage adjustment through a curve fitting algorithm.

8. The voltage fluctuation suppression method for a subway DC power supply system according to any one of claims 1 to 7, characterized by, The generation of the node-coordinated fluctuation suppression execution instruction set according to the voltage adjustment parameter sequence and the operating state characteristics of each power supply node includes: Analyze the voltage adjustment parameter sequence of each power supply node, and extract the voltage adjustment amount and adjustment duration parameter at each time point; According to the device operating mode identifier in the operating state characteristics of each power supply node, determine the adjustment permission level of each power supply node, which is used to limit the maximum range of the voltage adjustment amount; Based on the adjustment permission level, perform rationality checking and processing on the voltage adjustment amount, and cut off and correct the adjustment amount that exceeds the permission range; Analyze the synergy between the voltage adjustment amounts of different power supply nodes at the same time point, calculate the consistency coefficient of the adjustment direction, and the consistency coefficient is used to judge the coordination degree of the adjustment operation between power supply nodes; When the consistency coefficient is lower than a preset threshold, call a synergy optimization algorithm to generate a node-coordinated adjustment scheme for adjusting the voltage adjustment amount of a target power supply node, the node-coordinated adjustment scheme makes the consistency coefficient between the target power supply nodes greater than the preset threshold, wherein the target power supply node includes at least one of the power supply nodes whose consistency coefficient is lower than the preset threshold; Encapsulate the modified and adjusted voltage adjustment amount, adjustment duration parameter, and corresponding power supply node identifier to generate a node-coordinated fluctuation suppression execution instruction set.

9. A voltage fluctuation suppressing device for a subway DC power supply system, characterized by comprising: The device includes: A memory having a computer program stored thereon; A processor configured to execute the computer program stored in the memory to implement the voltage fluctuation suppression method of the subway DC power supply system according to any one of claims 1-8.

Images