Lossless network construction method supporting multiple scenes of electric power

Abstract

The invention discloses a lossless network construction method supporting multiple scenes of electric power, which comprises the following steps of: constructing a network topological graph by acquiring node and link information of an electric power network, dynamically optimizing a data transmission path by utilizing a planning algorithm, constructing a flow prediction model to predict a future data flow change trend, and then, constructing a data flow prediction model. And finally, establishing a network performance evaluation system, and evaluating the current network performance in real time, so that the data stability and reliability in the transmission process are effectively improved, and the network performance can be analyzed and evaluated more comprehensively and deeply.

Description

Technical Field

[0001] The present invention relates to the field of power network transmission technology, and specifically to a method for constructing a lossless network that supports multiple power scenarios. Background Art

[0002] In the power system, the requirements for the network in different scenarios are becoming increasingly diverse. Traditional networks have certain limitations in data transmission and cannot meet the needs of various power scenarios. Therefore, a new lossless network construction method is needed to improve the performance and reliability of the network.

[0003] The Chinese patent publication number: CN 117596208 A discloses a method for lossless flow control suitable for deterministic networks. The core technical solution is that the switch sends a flow control frame to the upper-level network node to control the data transmission of the upper-level network node's best-effort flow queue. The upper-level network node receives the flow control frame sent by the switch, and preferentially sends the high-priority flow queue data according to the integration strategy of the deterministic network, and sends the best-effort flow queue data according to the strict priority strategy. At the same time, the switch preferentially sends the high-priority flow queue data according to the integration strategy of the deterministic network, and sends the best-effort flow queue data according to the strict priority strategy based on the flow control frame received from the lower-level network switch. The switch switches different input devices according to the waiting time to ensure that different input devices can be sent.

[0004] In the existing technology, during lossless network transmission in different scenarios, the traffic prediction model is not fully used to predict and adjust data traffic changes, and the network performance is not deeply analyzed and evaluated, so the data stability and reliability during the transmission process cannot be ensured.

[0005] The present invention proposes a solution to the above shortcomings: construct a network topology diagram by acquiring node and link information of the power network, dynamically optimize the data transmission path using a planning algorithm, build a traffic prediction model to predict future data traffic change trends, and then dynamically adjust network transmission parameters according to the prediction results. Finally, establish a network performance evaluation system to perform real-time evaluation of the current network performance, which can effectively improve the data stability and reliability during the transmission process while enabling a more comprehensive and in-depth analysis and evaluation of network performance. Summary of the invention

[0006] In order to solve the above-mentioned technical problems, the present invention provides a method for building a lossless network that supports multiple power scenarios.

[0007] The technical solution of the present invention is implemented as follows: a method for constructing a lossless network supporting multiple power scenarios, comprising:

[0008] S1. Obtain node information and link information in networks of different power scenarios, construct a network topology map based on the obtained node information and link information, and dynamically plan a data transmission path using a planning algorithm according to the network topology map;

[0009] S2. Build a traffic prediction model to predict data traffic changes in the power network;

[0010] S3, training the traffic prediction model and evaluating the prediction accuracy of the model;

[0011] S4. Dynamically adjust network transmission parameters according to the prediction results;

[0012] S5. Establish a network performance evaluation system to conduct real-time evaluation of network performance.

[0013] Furthermore, in the step S1, node information and link information in the power scenario network are obtained, a network topology diagram is constructed based on the obtained node information and link information, and a data transmission path is dynamically planned using a planning algorithm according to the network topology diagram:

[0014] Use Nmap network mapper to obtain node information in different power scenario networks, where the different power scenarios include power generation side, power transmission side and power consumption side, and the node information includes IP address, MAC address and location; use LLDP link layer discovery protocol to obtain link information in the power scenario network, where the link information includes bandwidth parameters, delay parameters and link connection relationship; use Microsoft Visio drawing software to construct a network topology diagram, and use Dijkstra algorithm to confirm the optimal transmission path based on the network topology diagram;

[0015] Furthermore, the Dijkstra algorithm is set up as:

[0016] Set initial values: Initialize the distances of all vertices to infinity and the distance of the source point to 0;

[0017] Mark the access status: the access status of all vertices is set to unvisited;

[0018] Loop execution:

[0019] Select the vertex with the shortest path estimate from the unvisited vertices and set it as the current vertex;

[0020] Mark the current vertex as visited;

[0021] Traverse all adjacent vertices of the current vertex, calculate the distance from the current vertex to the adjacent vertex, and if the distance is less than the current shortest path estimate of the adjacent vertex, update the estimate;

[0022] Repeat steps: Repeat the above steps until all vertices are visited.

[0023] Furthermore, in step S2, a traffic prediction model is constructed to predict the data traffic changes in the power network:

[0024] In the S1 step, the Corelight traffic analysis tool is deployed on the nodes in the power scenario network to obtain historical traffic data, including data packet size, transmission speed, and link utilization, clean the data and perform outlier processing, normalize the data, and use the LSTM long short-term memory network algorithm to build a traffic prediction model;

[0025] The collected data is processed for outliers and cleaned, including data packet size data, transmission speed data, and link utilization data. The data packet size data, transmission speed data, and link utilization data are calculated using the z-score function. The calculation formula is:

[0026] Packet size data:

[0027]

[0028] Among them, z 1 is the normalized value of the data packet size, x 1 is the original value of the packet size data, μ 1 is the average value of the packet size data, σ 1 is the standard deviation of the packet size data;

[0029] Transfer speed data:

[0030]

[0031] Among them, z 2 is the normalized value of the transmission speed data, x 2 is the original value of the transmission speed data, μ 2 is the average value of the transmission speed data, σ 2 is the standard deviation of the transmission speed data;

[0032] Link utilization data:

[0033]

[0034] Among them, z 3 is the normalized value of link utilization data, x 3 is the raw value of the link utilization data, μ 3 is the average value of link utilization data, σ 3 is the standard deviation of the link utilization data;

[0035] Define |Z|>3 as an outlier, then the normal range of Z value is (-3, 3). When the output Z value is not in the normal range, it is an outlier. The outlier is processed, including deleting or replacing the outlier;

[0036] The data packet size data, transmission speed data and link utilization data are normalized, and the minimum-maximum normalization method is used to normalize the data to [0,1]. The calculation formula is:

[0037]

[0038] Among them, X' represents the normalized value, X represents the original value, and X max Indicates the maximum value, X min Indicates the minimum value;

[0039] The output data sequence after outlier processing and normalization is A 1 ={a 1 , a 2 , …a n};

[0040] Define the model architecture and build a prediction model architecture using the LSTM algorithm. The prediction model architecture consists of multiple memory units, each of which includes a forget gate, an input gate, and an output gate. 1 ={a 1 , a 2 , …a n} as the input sequence, where a i is the traffic feature vector at time point i, and the calculation of the LSTM unit at time step t is:

[0041] The calculation formula of the input gate is:

[0042] i t =σ(x t W xi +h t-1 W hi +b i );

[0043]

[0044] Among them, i t is the output of the input gate, σ is the sigmoid activation function, W xi , W hi , W hc is the weight of the input gate, is the candidate memory cell state output, b i , b c are the bias of the input gate and candidate memory unit, respectively, and h t-1is the hidden state of the previous time step, x t is the input of the current time step, tan h represents the tangent activation function;

[0045] The calculation formula of the forget gate is:

[0046] f t =σ(W xf x t +W hf h t-1 +b f )

[0047] Among them, f t is the output of the forget gate, σ is the sigmoid activation function, W xf and W hf is the weight of the forget gate, b f is the forget gate bias, h t-1 is the hidden state of the previous time step, x t is the input of the current time step;

[0048] The calculation formula for unit status update is:

[0049]

[0050] Among them, C t is the cell state at time t, f t is the output of the forget gate, C t-1 is the cell state at the previous time step, i t is the output of the input gate, is the output of the unit updating state;

[0051] The calculation formula of the output gate is:

[0052] o t =σ(W xo x t +W ho h t-1 +b o ),

[0053] h t =o t ·tanh(c t )

[0054] Among them, t is the output of the output gate, σ is the sigmoid activation function, W xo and W ho is the weight of the output gate, b o is the bias of the output gate, h t-1 is the hidden state of the previous time step, h tis the current hidden layer state update output, x t is the input of the current time step, C t is the unit state at time t, and tanh represents the tangent activation function.

[0055] Furthermore, in step S3, the traffic prediction model is trained and the prediction accuracy of the model is evaluated:

[0056] Build a training loop and split the preprocessed historical traffic data into 70% training set and 30% validation set. The training set is used to train the traffic prediction model, and the validation set is used to evaluate the accuracy of the model. Use the back propagation training algorithm to train the LSTM model and set hyperparameters, including setting the learning rate to 0.01, the training rounds to 100 times, the batch size to 32, and calculating the loss function in the training loop. The loss function selects the mean square error and selects the Adam optimizer to update the model weights. The specific training steps are:

[0057] S301, initializing the weights and biases of the model;

[0058] S302, selecting 32 sample data from the training set for training;

[0059] S303, performing forward propagation calculation to obtain a predicted value according to the current weight and bias;

[0060] S304, calculating the loss between the predicted value and the true value according to the MSE loss function;

[0061] S305, using the back propagation algorithm to calculate the gradient of the loss to the model parameters;

[0062] S306, using the Adam optimizer to update the model parameters according to the gradient;

[0063] S307, repeating steps S302 to S306 until the loss function reaches a preset number of training rounds, then stopping the training;

[0064] The trained model was evaluated using the validation set, and the prediction accuracy of the model was evaluated using the K-fold cross-validation method. K was preset to 10, and the validation set was divided into K equal-sized subsets, repeated K times, each time using one subset as the test set and the remaining K-1 subsets as the training set. In each iteration, the training set was used to train the logistic regression model, the number of correct samples was recorded, and the accuracy of each round was calculated: (the number of correctly predicted samples) / (the total number of samples), and then the average accuracy was obtained through the calculated accuracy of each round: (the sum of the accuracy rates of the correctly predicted samples in each round) / 10.

[0065] Furthermore, in the step S4, the network transmission parameters are dynamically adjusted according to the prediction results:

[0066] Collect the current network traffic value, and use the traffic prediction model trained in step S3 to predict the traffic change in the future. According to the prediction results, there are three results: traffic increase, traffic unchanged, and traffic decrease. By comparing the predicted value with the current value, the type of traffic change is determined as follows:

[0067] When the predicted value is greater than the current flow value, it is determined that the flow rate has increased;

[0068] When the predicted value is 70% close to the current flow value, the flow is determined to be unchanged;

[0069] When the predicted value is less than the current flow value, it is determined that the flow is reduced;

[0070] Adjust network transmission parameters according to different traffic categories. The specific measures are as follows:

[0071] When traffic increases, increase network bandwidth allocation;

[0072] When the traffic remains unchanged, the current network parameter settings are maintained;

[0073] When traffic reduction occurs, reduce network resource allocation.

[0074] Furthermore, in the step S5, a network performance evaluation system is established to evaluate network performance in real time:

[0075] Continuously collect traffic data predicted by the traffic prediction model, including packet size, transmission speed, and link utilization; calculate the average packet size, packet size standard deviation, average transmission speed, transmission speed standard deviation, average link utilization, and link utilization standard deviation based on the latest collected traffic data; define a performance evaluation function based on the calculated average and standard deviation; evaluate network performance based on the output of the performance evaluation function; and define the performance evaluation function to use a weighted average algorithm;

[0076] It is preset that flow data is collected every 5 minutes, including data packet size data P, transmission speed data V and link utilization data L. The average and standard deviation of the collected flow data are calculated as follows:

[0077] Average packet size:

[0078]

[0079] in, is the average packet size, P i is the size of the ith data packet, and n is the total number of data packets;

[0080] Packet size standard deviation:

[0081]

[0082] Among them, S P is the standard deviation of packet size, P i is the size of the ith data packet, is the average size of data packets, n is the total number of data packets;

[0083] Average transfer speed:

[0084]

[0085] in, is the average transmission speed, V i is the transmission speed of the i-th time period, and n is the total number of time periods;

[0086] Transmission speed standard deviation:

[0087]

[0088] Among them, S V is the transmission speed standard deviation, V i is the transmission speed in the i-th time period, is the average transmission speed, n is the total number of time periods;

[0089] Average link utilization:

[0090]

[0091] in, is the average link utilization, L i is the link utilization rate in the i-th time period, and n is the total number of time periods;

[0092] Link utilization standard deviation:

[0093]

[0094] Among them, S L is the standard deviation of link utilization, L i is the link utilization rate in the i-th time period, is the average link utilization, n is the total number of time periods;

[0095] According to the calculated S P , S V , S L , define the performance evaluation function f(x) as:

[0096]

[0097] Among them, α, β, γ, δ, ε, is the weight coefficient, are the target average packet size, target average transmission speed and target average link utilization, S Pmax , S Vmax , S Lmax are the maximum values ​​of the standard deviation of packet size, transmission speed, and link utilization, respectively;

[0098] The performance evaluation threshold is preset. When the performance evaluation function value is greater than 0.8, it indicates that the network performance is good. When the performance evaluation function value is less than 0.8, it indicates that the current network performance is poor, and treatment measures are taken. The treatment measures are:

[0099] Automatically adjust the data compression ratio to split large data packets into smaller sizes;

[0100] Recalculate the optimal transmission path;

[0101] Reallocate bandwidth resources and prioritize more bandwidth for critical tasks and tasks with high transmission speed requirements.

[0102] Beneficial effects:

[0103] In view of the shortcoming of the prior art that the stability and reliability of data in the transmission process cannot be ensured, the present invention constructs a network topology diagram by acquiring the node and link information of the power network, dynamically optimizes the data transmission path by using a planning algorithm, and constructs a traffic prediction model to predict the future trend of data traffic changes. Then, the network transmission parameters are dynamically adjusted according to the prediction results. Finally, a network performance evaluation system is established to perform real-time evaluation of the current network performance. While effectively improving the data stability and reliability in the transmission process, it can also perform a more comprehensive and in-depth analysis and evaluation of network performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0104] Figure 1 This is a structural block diagram of a method for building a lossless network supporting multiple power scenarios in an embodiment of the present invention;

[0105] Figure 2 A flowchart of a method for building a lossless network supporting multiple power scenarios in an embodiment of the present invention;

[0106] Figure 3 This is a flow chart of the flow prediction model training steps for a method for building a lossless network that supports multiple power scenarios in an embodiment of the present invention. DETAILED DESCRIPTION

[0107] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0108] The preferred implementation methods of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these implementation methods are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0109] When used herein, the singular forms "a", "an", and "said / the" may also include plural forms, unless the context clearly indicates otherwise. It should also be understood that the terms "include / comprise" or "have" and the like specify the presence of stated features, wholes, steps, operations, components, parts, or combinations thereof, but do not exclude the possibility of the presence or addition of one or more other features, wholes, steps, operations, components, parts, or combinations thereof. At the same time, the term "and / or" used in this specification includes any and all combinations of the relevant listed items.

[0110] See also Figure 1-Figure 3 As shown, a method for constructing a lossless network supporting multiple power scenarios includes:

[0111] S1. Obtain node information and link information in networks of different power scenarios, construct a network topology map based on the obtained node information and link information, and dynamically plan a data transmission path using a planning algorithm according to the network topology map;

[0112] S2. Build a traffic prediction model to predict data traffic changes in the power network;

[0113] S3, training the traffic prediction model and evaluating the prediction accuracy of the model;

[0114] S4. Dynamically adjust network transmission parameters according to the prediction results;

[0115] S5. Establish a network performance evaluation system to conduct real-time evaluation of network performance.

[0116] like Figure 2 As shown, in step S1, node information and link information in different power scenario networks are obtained, a network topology diagram is constructed based on the obtained node information and link information, and a data transmission path is dynamically planned using a planning algorithm according to the network topology diagram.

[0117] Specifically, the Nmap network mapper is used to obtain node information in different power scenario networks, where the different power scenarios include power generation side, power transmission side and power consumption side, and the node information includes IP address, MAC address and location. The LLDP link layer discovery protocol is used to obtain link information in the power scenario network, where the link information includes bandwidth parameters, delay parameters and link connection relationship. Microsoft Visio drawing software is used to construct a network topology diagram, and the Dijkstra algorithm is used to confirm the optimal transmission path based on the network topology diagram.

[0118] The Dijkstra algorithm is set up as:

[0119] Set initial values: Initialize the distances of all vertices to infinity and the distance of the source point to 0;

[0120] Mark the access status: the access status of all vertices is set to unvisited;

[0121] Loop execution:

[0122] Select the vertex with the shortest path estimate from the unvisited vertices and set it as the current vertex;

[0123] Mark the current vertex as visited;

[0124] Traverse all adjacent vertices of the current vertex, calculate the distance from the current vertex to the adjacent vertex, and if the distance is less than the current shortest path estimate of the adjacent vertex, update the estimate;

[0125] Repeat steps: Repeat the above steps until all vertices are visited.

[0126] like Figure 2 As shown, in step S2, a traffic prediction model is constructed to predict the data traffic changes in the power network.

[0127] Specifically, in the S1 step, the Corelight traffic analysis tool is deployed on the nodes in the power scenario network to obtain historical traffic data, including data packet size, transmission speed, and link utilization, clean the data and perform outlier processing, normalize the data, and use the LSTM long short-term memory network algorithm to build a traffic prediction model;

[0128] The collected data is processed for outliers and cleaned, including data packet size data, transmission speed data, and link utilization data. The data packet size data, transmission speed data, and link utilization data are calculated using the z-score function. The calculation formula is:

[0129] Packet size data:

[0130]

[0131] Among them, z 1 is the normalized value of the data packet size, x 1 is the original value of the packet size data, μ 1 is the average value of the packet size data, σ 1 is the standard deviation of the packet size data;

[0132] Transfer speed data:

[0133]

[0134] Among them, z 2 is the normalized value of the transmission speed data, x 2 is the original value of the transmission speed data, μ 2 is the average value of the transmission speed data, σ 2 is the standard deviation of the transmission speed data;

[0135] Link utilization data:

[0136]

[0137] Among them, z 3 is the normalized value of link utilization data, x 3 is the raw value of the link utilization data, μ 3 is the average value of link utilization data, σ 3 is the standard deviation of the link utilization data;

[0138] Define |Z|>3 as an outlier, then the normal range of Z value is (-3, 3). When the output Z value is not in the normal range, it is an outlier. The outlier is processed, including deleting or replacing the outlier;

[0139] The data packet size data, transmission speed data and link utilization data are normalized, and the minimum-maximum normalization method is used to normalize the data to [0,1]. The calculation formula is:

[0140]

[0141] Among them, X' represents the normalized value, X represents the original value, and X max Indicates the maximum value, X min Indicates the minimum value;

[0142] The output data sequence after outlier processing and normalization is A 1 ={a 1 , a 2 , …a n};

[0143] Define the model architecture and build a prediction model architecture using the LSTM algorithm. The prediction model architecture consists of multiple memory units, each of which includes a forget gate, an input gate, and an output gate. 1 ={a 1 , a 2 , …a n} as the input sequence, where a i is the traffic feature vector at time point i, and the calculation of the LSTM unit at time step t is:

[0144] The calculation formula of the input gate is:

[0145] i t =σ(x t W xi +h t-1 W hi +b i );

[0146]

[0147] Among them, i t is the output of the input gate, σ is the sigmoid activation function, W xi , W hi , W hc is the weight of the input gate, is the candidate memory cell state output, b i , b c are the bias of the input gate and candidate memory unit, respectively, and h t-1 is the hidden state of the previous time step, x t is the input of the current time step, tan h represents the tangent activation function;

[0148] The calculation formula of the forget gate is:

[0149] f t =σ(W xf x t +W hf h t-1 +b f )

[0150] Among them, f t is the output of the forget gate, σ is the sigmoid activation function, W xf and W hf is the weight of the forget gate, b f is the forget gate bias, h t-1 is the hidden state of the previous time step, x t is the input of the current time step;

[0151] The calculation formula for unit status update is:

[0152]

[0153] Among them, C t is the cell state at time t, f t is the output of the forget gate, C t-1 is the cell state at the previous time step, i t is the output of the input gate, C~ t is the output of the unit updating state;

[0154] The calculation formula of the output gate is:

[0155] o t =σ(W xo x t +W ho h t-1 +b o ),

[0156] h t =o t ·tanh(c t )

[0157] Among them, t is the output of the output gate, σ is the sigmoid activation function, W xo and W ho is the weight of the output gate, b o is the bias of the output gate, h t-1 is the hidden state of the previous time step, h t is the current hidden layer state update output, x t is the input of the current time step, C t is the unit state at time t, and tanh represents the tangent activation function.

[0158] like Figure 2 As shown, in step S3, the traffic prediction model is trained and the prediction accuracy of the model is evaluated.

[0159] Specifically, a training loop is built, and the preprocessed historical traffic data is split into 70% training set and 30% validation set. The training set is used to train the traffic prediction model, and the validation set is used to evaluate the accuracy of the model. The LSTM model is trained using the back propagation training algorithm, and hyperparameters are set, including setting the learning rate to 0.01, the training rounds to 100 times, the batch size to 32, and calculating the loss function in the training loop. The loss function selects the mean square error, and the Adam optimizer is selected to update the model weights. The specific training steps are:

[0160] S301, initializing the weights and biases of the model;

[0161] S302, selecting 32 sample data from the training set for training;

[0162] S303, performing forward propagation calculation to obtain a predicted value according to the current weight and bias;

[0163] S304, calculating the loss between the predicted value and the true value according to the MSE loss function;

[0164] S305, using the back propagation algorithm to calculate the gradient of the loss to the model parameters;

[0165] S306, using the Adam optimizer to update the model parameters according to the gradient;

[0166] S307, repeating steps S302 to S306 until the loss function reaches a preset number of training rounds, then stopping the training;

[0167] Use the validation set to evaluate the trained model, and use the K-fold cross-validation method to evaluate the prediction accuracy of the model. Preset K=10, divide the validation set into K equal-sized subsets, repeat K times, use one subset as the test set each time, and the remaining K-1 subsets as the training set. In each iteration, use the training set to train the logistic regression model, record the number of correct samples, and calculate the accuracy of each round: (the number of samples predicted correctly) / (the total number of samples), and then use the calculated accuracy of each round to obtain the average accuracy: (the sum of the accuracy of the samples predicted correctly in each round) / 10;

[0168] Among them, forward propagation is the process of data being gradually transferred from the input layer to the output layer after a series of calculations and processing, and back propagation refers to the process of updating the model parameters by calculating the gradient according to the error between the model output and the true value.

[0169] like Figure 2 As shown, in step S4, the network transmission parameters are dynamically adjusted according to the prediction results.

[0170] Specifically, the current network traffic value is collected, and the traffic prediction model trained in step S3 is used to predict the traffic changes in the future. According to the prediction results, there are three results: traffic increase, traffic unchanged, and traffic decrease. The type of traffic change is determined by comparing the predicted value with the current value.

[0171] When the predicted value is greater than the current flow value, it is determined that the flow rate has increased;

[0172] When the predicted value is 70% close to the current flow value, the flow is determined to be unchanged;

[0173] When the predicted value is less than the current flow value, it is determined that the flow is reduced;

[0174] Adjust network transmission parameters according to different traffic categories. The specific measures are as follows:

[0175] When traffic increases, increase network bandwidth allocation;

[0176] When the traffic remains unchanged, the current network parameter settings are maintained;

[0177] When traffic reduction occurs, reduce network resource allocation.

[0178] like Figure 2 As shown, in step S5, a network performance evaluation system is established to evaluate the network performance in real time.

[0179] Specifically, the traffic data predicted by the traffic prediction model is continuously collected, including packet size, transmission speed and link utilization; based on the latest collected traffic data, the average packet size, packet size standard deviation, average transmission speed, transmission speed standard deviation, average link utilization and link utilization standard deviation are calculated; a performance evaluation function is defined based on the calculated average and standard deviation; network performance is evaluated based on the output of the performance evaluation function; and a performance evaluation function is defined that uses a weighted average algorithm;

[0180] It is preset that flow data is collected every 5 minutes, including data packet size data P, transmission speed data V and link utilization data L. The average and standard deviation of the collected flow data are calculated as follows:

[0181] Average packet size:

[0182]

[0183] in, is the average packet size, P i is the size of the ith data packet, and n is the total number of data packets;

[0184] Packet size standard deviation:

[0185]

[0186] Among them, S P is the standard deviation of packet size, P i is the size of the ith data packet, is the average size of data packets, n is the total number of data packets;

[0187] Average transfer speed:

[0188]

[0189] in, is the average transmission speed, V iis the transmission speed of the i-th time period, and n is the total number of time periods;

[0190] Transmission speed standard deviation:

[0191]

[0192] Among them, S V is the transmission speed standard deviation, V i is the transmission speed in the i-th time period, is the average transmission speed, n is the total number of time periods;

[0193] Average link utilization:

[0194]

[0195] in, is the average link utilization, L i is the link utilization rate in the i-th time period, and n is the total number of time periods;

[0196] Link utilization standard deviation:

[0197]

[0198] Among them, S L is the standard deviation of link utilization, L i is the link utilization rate in the i-th time period, is the average link utilization, n is the total number of time periods;

[0199] According to the calculated S P , S V , S L , define the performance evaluation function f(x) as:

[0200]

[0201] Among them, α, β, γ, δ, ε, is the weight coefficient, are the target average packet size, target average transmission speed and target average link utilization, S Pmax , S Vmax , S Lmax are the maximum values ​​of the standard deviation of packet size, transmission speed, and link utilization, respectively;

[0202] The performance evaluation threshold is preset. When the performance evaluation function value is greater than 0.8, it indicates that the network performance is good. When the performance evaluation function value is less than 0.8, it indicates that the current network performance is poor, and treatment measures are taken. The treatment measures are:

[0203] Automatically adjust the data compression ratio to split large data packets into smaller sizes;

[0204] Recalculate the optimal transmission path;

[0205] Reallocate bandwidth resources and prioritize more bandwidth for critical tasks and tasks with high transmission speed requirements.

[0206] like Figure 1-Figure 3 As shown, the present invention works as follows in the application scenario of solving the disadvantage that the prior art cannot guarantee the accuracy of data during the simulation analysis process:

[0207] Acquire node information and link information in networks of different power scenarios, construct a network topology map based on the acquired node information and link information, and dynamically plan a data transmission path using a planning algorithm according to the network topology map;

[0208] Build a traffic prediction model to predict data traffic changes in the power network;

[0209] Training the traffic prediction model and evaluating the prediction accuracy of the model;

[0210] Dynamically adjust network transmission parameters based on prediction results;

[0211] Establish a network performance evaluation system to conduct real-time evaluation of network performance.

[0212] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0213] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and rules of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for constructing a lossless network supporting multiple power scenarios, characterized by: S1. Obtain node information and link information in networks of different power scenarios, construct a network topology map based on the obtained node information and link information, and dynamically plan a data transmission path using a planning algorithm according to the network topology map; S2. Build a traffic prediction model to predict data traffic changes in the power network; S3, training the traffic prediction model and evaluating the prediction accuracy of the model; S4. Dynamically adjust network transmission parameters according to the prediction results; S5. Establish a network performance evaluation system to conduct real-time evaluation of network performance.

2. According to claim 1, a method for constructing a lossless network supporting multiple power scenarios is characterized by: In the step S1, node information and link information in different power scenario networks are obtained, a network topology diagram is constructed based on the obtained node information and link information, and a planning algorithm is used according to the network topology diagram to dynamically plan the data transmission path as follows: A network mapper is used to obtain node information in different power scenario networks, where the different power scenarios include the power generation side, the power transmission side, and the power consumption side. The node information includes IP address, MAC address, and location. A link layer discovery protocol is used to obtain link information in the power scenario network, where the link information includes bandwidth parameters, delay parameters, and link connection relationships. A drawing software is used to construct a network topology diagram, and the Dijkstra algorithm is used to confirm the optimal transmission path based on the network topology diagram.

3. According to claim 1, a method for constructing a lossless network supporting multiple power scenarios is characterized in that: In step S2, a flow prediction model is constructed to predict the data flow change in the power network as follows: In the S1 step, a traffic analysis tool is deployed on nodes in the power scenario network to obtain historical traffic data, including packet size, transmission speed, and link utilization. The data is cleaned and outliers are processed, the data is normalized, and a traffic prediction model is constructed using an LSTM long short-term memory network algorithm.

4. According to claim 1, a method for constructing a lossless network supporting multiple power scenarios is characterized by: In step S3, the traffic prediction model is trained and the prediction accuracy of the model is evaluated as follows: Build a training loop and split the preprocessed historical traffic data into 70% training set and 30% validation set. The training set is used to train the traffic prediction model, and the validation set is used to evaluate the accuracy of the model. Use the backpropagation training algorithm to train the LSTM model, set hyperparameters, calculate the loss function in the training loop, and select the Adam optimizer to update the model weights.

5. According to claim 1, a method for constructing a lossless network supporting multiple power scenarios is characterized in that: In step S4, the network transmission parameters are dynamically adjusted according to the prediction results: Collect the current network traffic value, and use the traffic prediction model trained in step S3 to predict the traffic changes in the future. According to the prediction results, there are three results: traffic increase, traffic unchanged, and traffic decrease. The category of traffic change is determined by comparing the predicted value with the current value.

6. According to claim 1, a method for constructing a lossless network supporting multiple power scenarios is characterized in that: In the step S5, a network performance evaluation system is established to evaluate network performance in real time as follows: Continuously collect traffic data predicted by the traffic prediction model, including packet size, transmission speed, and link utilization; calculate the average packet size, packet size standard deviation, average transmission speed, transmission speed standard deviation, average link utilization, and link utilization standard deviation based on the latest collected traffic data; define a performance evaluation function based on the calculated average and standard deviation; evaluate network performance based on the output of the performance evaluation function; and define the performance evaluation function using a weighted average algorithm.

7. According to claim 2, a method for constructing a lossless network supporting multiple power scenarios is characterized in that: The Dijkstra algorithm is set up as: Set initial values: Initialize the distances of all vertices to infinity and the distance of the source point to 0; Mark the access status: the access status of all vertices is set to unvisited; Loop execution: Select the vertex with the shortest path estimate from the unvisited vertices and set it as the current vertex; Mark the current vertex as visited; Traverse all adjacent vertices of the current vertex, calculate the distance from the current vertex to the adjacent vertex, and if the distance is less than the current shortest path estimate of the adjacent vertex, update the estimate; Repeat steps: Repeat the above steps until all vertices are visited.

8. According to claim 4, a method for constructing a lossless network supporting multiple power scenarios is characterized in that: The training steps are: S301, initializing the weights and biases of the model; S302, selecting sample data from the training set for training; S303, performing forward propagation calculation to obtain a predicted value according to the current weight and bias; S304, calculating the loss between the predicted value and the true value according to the loss function; S305, using the back propagation algorithm to calculate the gradient of the loss to the model parameters; S306, using the Adam optimizer to update the model parameters according to the gradient; S307, repeating steps S302 to S306 until the loss function reaches a preset number of training rounds, then stopping the training.

9. According to claim 5, a method for constructing a lossless network supporting multiple power scenarios is characterized in that: The flow rate is determined as follows: When the predicted value is greater than the current flow value, it is determined that the flow rate has increased; When the predicted value is 70% close to the current flow value, the flow is determined to be unchanged; When the predicted value is less than the current flow value, it is determined that the flow is decreasing.

10. The method for constructing a lossless network supporting multiple power scenarios according to claim 6, characterized in that: The weighted average algorithm is: Among them, α, β, γ, δ, ε, is the weight coefficient, are the target average packet size, target average transmission speed and target average link utilization, S Pmax , S Vmax , S Lmax They are the maximum values ​​of the standard deviation of packet size, transmission speed and link utilization, respectively.

Images