Data transmission path optimization method and system based on LSTM model
By using the LSTM model -based data transmission path optimization method in the distribution network, the problem of delay and flexibility of data transmission path optimization in the existing technologies, real -time dynamic optimization and efficiency of the data transmission path.
Patent Information
- Application Number
- CN202411473251.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-22
- Publication Date
- 2025-01-28
Smart Images
Figure CN119378779A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of data processing technology, and in particular to a data transmission path optimization method and system based on an LSTM model. Background Art
[0002] Modern distribution networks have widely accepted a series of new entities including electric vehicles, distributed energy systems and energy storage equipment. The addition of these emerging entities has prompted the generation of massive real-time data. The efficient processing and proper storage of these data have become key tasks that need to be urgently solved.
[0003] The traditional data processing methods mainly used in the existing technology include batch processing, streaming computing and heuristic algorithm processing. Among them, batch processing usually requires the accumulation of data to a certain amount before processing, which has high latency and cannot meet the needs of data processing; streaming computing will encounter the problem of unbalanced processing nodes when processing real-time data at different rates. When the processing rate of a node is low, data will accumulate, which will affect the sorting and processing efficiency; heuristic algorithm processing relies on preset rules and strategies. When the amount of data surges, it lacks flexibility and adaptability, and it is difficult to dynamically optimize the data transmission path.
[0004] Application Contents
[0005] The present application provides a data transmission path optimization method and system based on the LSTM model to improve the real-time and efficiency of data processing, while dynamically optimizing the data transmission path.
[0006] In a first aspect, the present application provides a data transmission path optimization method based on an LSTM model, comprising:
[0007] Acquire system data of the distribution network in real time, wherein the system data includes production node data and communication node data;
[0008] Input the system data into a preset LSTM model, predict the data transmission probability between each node, and form a data transmission probability evaluation matrix based on the data transmission probability;
[0009] The data transmission probability evaluation matrix is calculated using the A* algorithm to determine the optimal data transmission path.
[0010] The present application obtains the system data of the distribution network in real time, which helps the subsequent LSTM model to evaluate the probability of data transmission; by inputting the system data into a preset LSTM model, the data transmission probability between nodes can be quickly evaluated, which helps to subsequently select the optimal data transmission path and adapt to the changeable data flows of various new entities in the distribution network; using the A* algorithm to calculate the data transmission probability evaluation matrix, the optimal data transmission path can be quickly perceived and calculated, and real-time data transmission path dynamic optimization can be achieved, ensuring the efficiency and timeliness of data transmission.
[0011] Furthermore, after acquiring the system data of the power distribution network in real time, the method further includes:
[0012] Performing a quality assessment on the data transmission path of the system data to obtain a quality assessment result;
[0013] Data is selected from the system data based on the quality assessment result to construct training samples for the LSTM model.
[0014] In this way, the quality of the data transmission path of the system data is evaluated, and data is selected from the system data based on the quality evaluation result so as to train the LSTM model, thereby improving the accuracy of the LSTM model in predicting the probability of data transmission between nodes.
[0015] Furthermore, the calculation formula for the quality assessment is specifically:
[0016]
[0017] Where Q(p) is the quality score of path p, T(p) is the node data processing speed, M(p) is the node memory size, and M max is the maximum memory of the node, B(p) is the data transmission capacity of the node, and B max is the maximum transmission capacity of the node, D(p) is the delay, and w1,w2,w3,w4 are weight coefficients.
[0018] In this way, by comprehensively considering multi-dimensional indicators such as node data processing speed, node memory size, node transmission capacity and delay, a transmission path quality assessment method is proposed, which provides a scientific basis for path selection and significantly improves the operating efficiency and stability of the distribution network.
[0019] Furthermore, the system data is input into a preset LSTM model to predict the data transmission probability between each node, and a data transmission probability evaluation matrix is formed based on the data transmission probability, specifically:
[0020] Inputting the system data into the LSTM model, wherein the LSTM model comprises an input layer, an LSTM layer and an output layer;
[0021] The input layer processes the system data through a time window sliding technique to obtain a corresponding time series;
[0022] Processing the time series through the LSTM layer to obtain a processing result;
[0023] The processing result is converted into the data transmission probability between each node through the output layer, and a data transmission probability evaluation matrix is formed based on the data transmission probability.
[0024] In this way, by inputting the system data into the preset LSTM model, the data transmission probability between nodes can be quickly evaluated, which is helpful for the subsequent selection of the optimal data transmission path and adapting to the changeable data flows of various new entities in the distribution network.
[0025] Furthermore, the data transmission probability evaluation matrix is calculated using the A* algorithm to determine the optimal data transmission path, specifically:
[0026] Determine the starting node and target node of the distribution network;
[0027] Respectively obtaining an actual path cost and an estimated path cost between the starting node and the target node, and calculating the actual path cost and the estimated path cost by using an A* algorithm to obtain a comprehensive path cost;
[0028] An optimal data transmission path is determined based on the comprehensive cost of the path between the starting node and the target node, the path transmission quality and the data transmission efficiency.
[0029] In this way, the data transmission probability evaluation matrix is calculated through the A* algorithm, and factors such as the transmission quality of the path, data processing efficiency and comprehensive path cost are taken into consideration. The optimal data transmission path can be quickly perceived and calculated, and real-time data transmission path dynamic optimization can be achieved to ensure the efficiency and timeliness of data transmission.
[0030] In a second aspect, the present application provides a data transmission path optimization system based on an LSTM model, including: an acquisition module, a prediction module, and a calculation module;
[0031] The acquisition module is used to acquire system data of the power distribution network in real time, wherein the system data includes production node data and communication node data;
[0032] The prediction module is used to input the system data into a preset LSTM model, predict the data transmission probability between each node, and form a data transmission probability evaluation matrix based on the data transmission probability;
[0033] The calculation module is used to calculate the data transmission probability evaluation matrix using the A* algorithm to determine the optimal data transmission path.
[0034] The present application obtains the system data of the distribution network in real time, which helps the subsequent LSTM model to evaluate the probability of data transmission; by inputting the system data into a preset LSTM model, the data transmission probability between nodes can be quickly evaluated, which helps to subsequently select the optimal data transmission path and adapt to the changeable data flows of various new entities in the distribution network; using the A* algorithm to calculate the data transmission probability evaluation matrix, the optimal data transmission path can be quickly perceived and calculated, and real-time data transmission path dynamic optimization can be achieved, ensuring the efficiency and timeliness of data transmission.
[0035] Furthermore, the data transmission path optimization system based on the LSTM model also includes: a quality assessment module, the quality assessment module includes a quality assessment unit and a selection unit;
[0036] The quality assessment unit is used to perform quality assessment on the data transmission path of the system data to obtain a quality assessment result;
[0037] The selection unit is used to select data from the system data based on the quality assessment result to construct training samples for the LSTM model.
[0038] Furthermore, the calculation formula for the quality assessment is specifically:
[0039]
[0040] Where Q(p) is the quality score of path p, T(p) is the node data processing speed, M(p) is the node memory size, and M max is the maximum memory of the node, B(p) is the data transmission capacity of the node, and B max is the maximum transmission capacity of the node, D(p) is the delay, and w1,w2,w3,w4 are weight coefficients.
[0041] Further, the prediction module includes: an input unit, a first processing unit, a second processing unit and a third processing unit;
[0042] The input unit is used to input the system data into the LSTM model, wherein the LSTM model includes an input layer, an LSTM layer and an output layer;
[0043] The first processing unit is used for the input layer to process the system data through the time window sliding technology to obtain the corresponding time series;
[0044] The second processing unit is used to process the time series through the LSTM layer to obtain a processing result;
[0045] The third processing unit is used to convert the processing result into the data transmission probability between each node through the output layer, and form a data transmission probability evaluation matrix based on the data transmission probability.
[0046] Further, the calculation module includes: a determination unit, an acquisition unit and a fourth processing unit;
[0047] The determination unit is used to determine the starting node and the target node of the distribution network;
[0048] The acquisition unit is used to respectively acquire the actual path cost and the estimated path cost between the starting node and the target node, and calculate the actual path cost and the estimated path cost by using an A* algorithm to obtain a comprehensive path cost;
[0049] The fourth processing unit is used to determine the optimal data transmission path based on the comprehensive cost of the path between the starting node and the target node, the path transmission quality and the data transmission efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0050] Figure 1 It is a simplified structural diagram of a regional distribution network provided by this application;
[0051] Figure 2 This is a flowchart of an embodiment of a data transmission path optimization method based on an LSTM model provided by the present application;
[0052] Figure 3 It is a structural diagram of the LSTM model provided in this application;
[0053] Figure 4 It is a structural schematic diagram of an embodiment of a data transmission path optimization system based on an LSTM model provided by the present application; DETAILED DESCRIPTION
[0054] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0055] It should be understood that the step numbers used in this article are only for the convenience of description and are not intended to limit the order in which the steps are executed.
[0056] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, the singular forms "a", "an" and "the" are intended to include plural forms unless the context clearly indicates otherwise.
[0057] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.
[0058] The term "and / or" means and includes any and all possible combinations of one or more of the associated listed items.
[0059] The LSTM model, which stands for Long Short-Term Memory networks, is an important variant of the recurrent neural network (RNN). It aims to solve the gradient vanishing and gradient exploding problems faced by traditional RNNs when processing long sequence data, thereby more effectively capturing long-term dependencies in time series data.
[0060] A-star algorithm is a heuristic search algorithm widely used in computer science, mainly used to find the shortest path from the starting point to the end point. This algorithm combines the advantages of breadth-first search (BFS) and heuristic search, and can find shorter paths in large search spaces while maintaining low computational complexity.
[0061] Modern distribution networks have widely accepted new entities such as electric vehicles, distributed energy and energy storage equipment. The addition of these new entities has generated massive real-time data. Efficient processing and proper storage of this data has become an urgent task. Existing technologies such as batch processing, streaming computing and heuristic algorithms have their own limitations in data processing: among them, batch processing has high latency, and streaming computing is prone to node imbalance leading to data accumulation, thus affecting processing efficiency; heuristic algorithms are not flexible enough when data surges, and it is difficult to dynamically optimize the transmission path.
[0062] When various new entities are connected to the distribution network, massive real-time data will be generated. By installing a data acquisition system in the distribution network, various types of data after the new entities are connected to the distribution network can be collected. The data generated by each new entity includes the data flow and data generation rate of each new entity. When the distribution network collects the data generated by each new entity, it will perform data processing and data transmission in real time through the communication nodes in the distribution network. Figure 1It is a simplified structural diagram of a regional distribution network provided by the present application, wherein each local main distribution network will be connected to multiple new entities, and multiple local main distribution networks are connected to the cloud disaster recovery center. Taking the local main distribution network 1 as an example, when multiple new entities, namely production nodes, generate various types of data, they will be transmitted to the local main distribution network 1. After the data acquisition system in the local main distribution network 1 collects the data, it will process the data or transmit the data through its own communication nodes. When the communication node transmits the data, it will upload it to the cloud disaster recovery center. Of course, there are also communication nodes in the cloud disaster recovery center, which can also process data or transmit data to other nodes.
[0063] Embodiment 1
[0064] Please refer to Figure 2 , Figure 2 It is a flowchart of an embodiment of a data transmission path optimization method based on an LSTM model provided by the present application, including steps S1 to S3;
[0065] Step S1, acquiring system data of the distribution network in real time, wherein the system data includes production node data and communication node data;
[0066] It is understandable that a large amount of system data generated after the access of massive new entities can be obtained through sensors or data acquisition equipment in the distribution network. The collected system data includes not only the data flow and data generation rate generated by the production nodes, but also the processing capacity, storage status and data transmission performance of the communication nodes.
[0067] After obtaining the system data of the distribution network, the quality of each data transmission path can be evaluated by considering the node data processing capability, node memory size, node data transmission capability and transmission delay in the system data to obtain a quality evaluation result; and based on the quality evaluation result, data with a score greater than a preset threshold is selected from the system data, and the training sample of the LSTM model is constructed based on this data. It should be noted that the training sample includes not only the static attribute data of the node, but also the data during dynamic operation to ensure that the training sample can accurately reflect the actual performance of the distribution network under different operating conditions.
[0068] In this way, the quality of the data transmission path of the system data is evaluated, and data is selected from the system data based on the quality evaluation result so as to train the LSTM model, thereby improving the accuracy of the LSTM model in predicting the probability of data transmission between nodes.
[0069] The calculation formula for the quality assessment is specifically:
[0070]
[0071] Where Q(p) is the quality score of path p, T(p) is the node data processing speed, M(p) is the node memory size, and M max is the maximum memory of the node, B(p) is the data transmission capacity of the node, and B max is the maximum transmission capacity of the node, D(p) is the delay, and w1,w2,w3,w4 are weight coefficients.
[0072] In this way, by comprehensively considering multi-dimensional indicators such as node data processing speed, node memory size, node transmission capacity and delay, a transmission path quality assessment method is proposed, which provides a scientific basis for path selection and significantly improves the operating efficiency and stability of the distribution network.
[0073] It should be noted that the training samples consist of an input set and an output set. The input set is the real-time data generation of the production nodes at each moment and the remaining memory capacity of the communication nodes, and the output set is the data transmission quality between nodes at each moment.
[0074] Step S2: input the system data into a preset LSTM model, predict the data transmission probability between each node, and form a data transmission probability evaluation matrix based on the data transmission probability;
[0075] It is understandable that after the real-time system data is acquired, the system data can be input into a pre-trained LSTM model to obtain the path transmission probability.
[0076] First, the system data is processed through the input layer of the LSTM model through the time window sliding technology to process the system data into a time series suitable for the LSTM model, wherein the system data includes but is not limited to: the real-time data generation amount of the production node and the remaining memory capacity of the communication node; wherein the format of the system data is: n i (k) is the number of data generated by node i, is the remaining memory size of node i, and k is the time.
[0077] Secondly, the time series is processed by the LSTM layer of the LSTM model to obtain a processing result, in which the transmission probability value between each node is Y(k)=[0,y 12 ,…,y 1n ], where k is the time, y 1n is the probability of data transmission between the first node and the nth node.
[0078] Finally, the processing result is converted into the data transmission probability between each node through the output layer of the LSTM model through the Softmax function, and a data transmission probability evaluation matrix is formed based on the data transmission probability, wherein the data transmission probability evaluation matrix expression is as follows:
[0079] In this way, by inputting the system data into the preset LSTM model, the data transmission probability between nodes can be quickly evaluated, which is helpful for the subsequent selection of the optimal data transmission path and adapting to the changeable data flows of various new entities in the distribution network.
[0080] The LSTM model includes not only an input layer, an LSTM layer and an output layer, but also a path-aware loss function and an optimization mechanism. The LSTM layer includes at least one LSTM unit and a gating mechanism, including an input gate, a forget gate and an output gate, so as to effectively process time series data. The structural diagram of the LSTM model is shown in FIG. Figure 3 shown.
[0081] Specifically, the training process of the LSTM model is as follows: first, set the initial number of iterations and batch size of the LSTM model, train the LSTM model using the training samples formed after the system data is sorted, input the training samples into the LSTM unit for forward propagation, and propagate the output value to the next LSTM unit to obtain the predicted data transmission probability between each node. Secondly, calculate the actual data transmission probability in advance based on the system data, compare the predicted data transmission probability with the actual data transmission probability, and calculate the loss value between the predicted data transmission probability and the actual data transmission probability through the path loss function; finally, use the gradient descent method to optimize the hyperparameters of the LSTM model. Once the loss value is determined, the parameter loss gradient of the LSTM model can be calculated by back propagation, and the parameters of the LSTM model are optimized based on the parameter loss gradient until the loss value reaches the minimum value.
[0082] The path loss function is formed by comprehensively considering the accuracy of the predicted path, processing delay and other relevant performance indicators. The calculation formula of the path loss function is:
[0083] L=α·L pred +β·L delay +γ·L eff ;
[0084]
[0085] Where, L pred is the path prediction error loss, which represents the error between the predicted data transmission probability and the actual data transmission probability; P iis the probability value of the i-th path predicted by the model, A i is the probability value of the actual path, N is the number of paths; L delay To handle delay loss, we represent the impact of data transmission delay on path quality, T i To predict the transmission delay of the i-th path, T i,act is the transmission delay of the actual path; L eff is the transmission efficiency loss, which indicates the impact of path transmission efficiency on path quality. i To predict the transmission efficiency E of the i-th path i,max is the highest transmission efficiency among all paths, and α, β, and γ are weight coefficients.
[0086] It should be noted that the loss gradient represents the local slope of the loss function relative to each parameter, indicating how the loss function changes with slight changes in the parameters. The gradient calculation formula is: In the formula, is the parameter loss gradient, are LSTM model parameters, is the loss value.
[0087] In this way, by calculating the loss value between the predicted data transmission probability and the actual data transmission probability, and gradually adjusting the LSTM model parameters based on the loss value, with the goal of minimizing the path loss function, the model's prediction performance and path selection ability can be improved.
[0088] Step S3: Calculate the data transmission probability evaluation matrix using the A* algorithm to determine the optimal data transmission path.
[0089] It can be understood that after the data transmission probability evaluation matrix is determined, the data transmission path can be determined; specifically: first, the starting node and the target node of the distribution network are determined; secondly, the actual path cost and the estimated path cost between the starting node and the target node are obtained respectively, and the actual path cost and the estimated path cost are calculated by the A* algorithm to obtain the comprehensive path cost; finally, based on the comprehensive path cost, path transmission quality and data transmission efficiency between the starting node and the target node, the optimal data transmission path is determined.
[0090] In this way, the data transmission probability evaluation matrix is calculated through the A* algorithm, and factors such as the transmission quality of the path, data processing efficiency and comprehensive path cost are taken into consideration. The optimal data transmission path can be quickly perceived and calculated, and real-time data transmission path dynamic optimization can be achieved to ensure the efficiency and timeliness of data transmission.
[0091] The data acquisition system in the distribution network monitors the status of production nodes and communication nodes in real time, updates node status parameters according to the monitoring data, regenerates the data transmission probability assessment matrix based on the updated status parameters, and uses the A* algorithm to determine the optimal data transmission path corresponding to the updated status parameters. By continuously iterating this process, it can ensure that the optimal data transmission path is provided in all situations. When a production node or communication node is monitored to have a fault, the second best data transmission path will be selected to ensure the stability and reliability of data transmission.
[0092] The present application obtains the system data of the distribution network in real time, which helps the subsequent LSTM model to evaluate the probability of data transmission; by inputting the system data into a preset LSTM model, the data transmission probability between nodes can be quickly evaluated, which helps to subsequently select the optimal data transmission path and adapt to the changeable data flows of various new entities in the distribution network; using the A* algorithm to calculate the data transmission probability evaluation matrix, the optimal data transmission path can be quickly perceived and calculated, and real-time data transmission path dynamic optimization can be achieved, ensuring the efficiency and timeliness of data transmission.
[0093] Embodiment 2
[0094] Please refer to Figure 4 , Figure 4 It is a flow chart of an embodiment of a data transmission path optimization system based on an LSTM model provided by the present application, comprising: an acquisition module 01, a prediction module 02 and a calculation module 03;
[0095] The acquisition module 01 is used to acquire the system data of the power distribution network in real time, wherein the system data includes production node data and communication node data;
[0096] It is understandable that a large amount of system data generated after the access of massive new entities can be obtained through sensors or data acquisition equipment in the distribution network. The collected system data includes not only the data flow and data generation rate generated by the production nodes, but also the processing capacity, storage status and data transmission performance of the communication nodes.
[0097] After the system data of the distribution network is obtained, the quality of the system data can be evaluated. Therefore, the data transmission path optimization system based on the LSTM model also includes: a quality evaluation module, which includes a quality evaluation unit and a selection unit; the quality evaluation unit is used to perform a quality evaluation on the data transmission path of the system data to obtain a quality evaluation result; the selection unit is used to select data from the system data based on the quality evaluation result to construct a training sample for the LSTM model.
[0098] After obtaining the system data of the distribution network, the quality of each data transmission path can be evaluated by considering the node data processing capability, node memory size, node data transmission capability and transmission delay in the system data to obtain a quality evaluation result; and based on the quality evaluation result, data with a score greater than a preset threshold is selected from the system data, and the training sample of the LSTM model is constructed based on this data. It should be noted that the training sample includes not only the static attribute data of the node, but also the data during dynamic operation to ensure that the training sample can accurately reflect the actual performance of the distribution network under different operating conditions.
[0099] In this way, the quality of the data transmission path of the system data is evaluated, and data is selected from the system data based on the quality evaluation result so as to train the LSTM model, thereby improving the accuracy of the LSTM model in predicting the probability of data transmission between nodes.
[0100] The calculation formula for the quality assessment is specifically:
[0101]
[0102] Where Q(p) is the quality score of path p, T(p) is the node data processing speed, M(p) is the node memory size, and M max is the maximum memory of the node, B(p) is the data transmission capacity of the node, and B max is the maximum transmission capacity of the node, D(p) is the delay, and w1,w2,w3,w4 are weight coefficients.
[0103] In this way, by comprehensively considering multi-dimensional indicators such as node data processing speed, node memory size, node transmission capacity and delay, a transmission path quality assessment method is proposed, which provides a scientific basis for path selection and significantly improves the operating efficiency and stability of the distribution network.
[0104] It should be noted that the training samples consist of an input set and an output set. The input set is the real-time data generation of the production nodes at each moment and the remaining memory capacity of the communication nodes, and the output set is the data transmission quality between nodes at each moment.
[0105] The prediction module 02 is used to input the system data into a preset LSTM model, predict the data transmission probability between each node, and form a data transmission probability evaluation matrix based on the data transmission probability;
[0106] Specifically, the prediction module includes: an input unit, a first processing unit, a second processing unit and a third processing unit; the input unit is used to input the system data into the LSTM model, wherein the LSTM model includes an input layer, an LSTM layer and an output layer; the first processing unit is used for the input layer to process the system data through a time window sliding technology to obtain a corresponding time series; the second processing unit is used to process the time series through the LSTM layer to obtain a processing result; the third processing unit is used to convert the processing result into a data transmission probability between each node through the output layer, and form a data transmission probability evaluation matrix based on the data transmission probability.
[0107] It is understandable that after the real-time system data is acquired, the system data can be input into a pre-trained LSTM model to obtain the path transmission probability.
[0108] First, the system data is processed through the input layer of the LSTM model through the time window sliding technology to process the system data into a time series suitable for the LSTM model, wherein the system data includes but is not limited to: the real-time data generation amount of the production node and the remaining memory capacity of the communication node; wherein the format of the system data is: n i (k) is the number of data generated by node i, is the remaining memory size of node i, and k is the time.
[0109] Secondly, the time series is processed by the LSTM layer of the LSTM model to obtain a processing result, in which the transmission probability value between each node is Y(k)=[0,y 12 ,…,y 1n ], where k is the time, y 1n is the probability of data transmission between the first node and the nth node.
[0110] Finally, the processing result is converted into the data transmission probability between each node through the output layer of the LSTM model through the Softmax function, and a data transmission probability evaluation matrix is formed based on the data transmission probability, wherein the data transmission probability evaluation matrix expression is as follows:
[0111] In this way, by inputting the system data into the preset LSTM model, the data transmission probability between nodes can be quickly evaluated, which is helpful for the subsequent selection of the optimal data transmission path and adapting to the changeable data flows of various new entities in the distribution network.
[0112] The LSTM model includes not only an input layer, an LSTM layer and an output layer, but also a path-aware loss function and an optimization mechanism. The LSTM layer includes at least one LSTM unit and a gating mechanism, including an input gate, a forget gate and an output gate, so as to effectively process time series data. The structural diagram of the LSTM model is shown in FIG. Figure 3 shown.
[0113] Specifically, the training process of the LSTM model is as follows: first, set the initial number of iterations and batch size of the LSTM model, train the LSTM model using the training samples formed after the system data is sorted, input the training samples into the LSTM unit for forward propagation, and propagate the output value to the next LSTM unit to obtain the predicted data transmission probability between each node. Secondly, calculate the actual data transmission probability in advance based on the system data, compare the predicted data transmission probability with the actual data transmission probability, and calculate the loss value between the predicted data transmission probability and the actual data transmission probability through the path loss function; finally, use the gradient descent method to optimize the hyperparameters of the LSTM model. Once the loss value is determined, the parameter loss gradient of the LSTM model can be calculated by back propagation, and the parameters of the LSTM model are optimized based on the parameter loss gradient until the loss value reaches the minimum value.
[0114] The path loss function is formed by comprehensively considering the accuracy of the predicted path, processing delay and other relevant performance indicators. The calculation formula of the path loss function is:
[0115] L=α·L pred +β·L delay +γ·L eff ;
[0116]
[0117] Where, L pred is the path prediction error loss, which represents the error between the predicted data transmission probability and the actual data transmission probability; P i is the probability value of the i-th path predicted by the model, A i is the probability value of the actual path, N is the number of paths; L delay To handle delay loss, we represent the impact of data transmission delay on path quality, T i To predict the transmission delay of the i-th path, T i,act is the transmission delay of the actual path; L eff is the transmission efficiency loss, which indicates the impact of path transmission efficiency on path quality. i To predict the transmission efficiency E of the i-th path i,maxis the highest transmission efficiency among all paths, and α, β, and γ are weight coefficients.
[0118] It should be noted that the loss gradient represents the local slope of the loss function relative to each parameter, indicating how the loss function changes with slight changes in the parameters. The gradient calculation formula is: In the formula, is the parameter loss gradient, are LSTM model parameters, is the loss value.
[0119] In this way, by calculating the loss value between the predicted data transmission probability and the actual data transmission probability, and gradually adjusting the LSTM model parameters based on the loss value, with the goal of minimizing the path loss function, the model's prediction performance and path selection ability can be improved.
[0120] The calculation module 03 is used to calculate the data transmission probability evaluation matrix using the A* algorithm to determine the optimal data transmission path.
[0121] Specifically, the calculation module 03 includes: a determination unit, an acquisition unit and a fourth processing unit; the determination unit is used to determine the starting node and the target node of the distribution network; the acquisition unit is used to respectively acquire the actual path cost and the estimated path cost between the starting node and the target node, and calculate the actual path cost and the estimated path cost through the A* algorithm to obtain the path comprehensive cost; the fourth processing unit is used to determine the optimal data transmission path based on the path comprehensive cost, path transmission quality and data transmission efficiency between the starting node and the target node.
[0122] It can be understood that after the data transmission probability evaluation matrix is determined, the data transmission path can be determined; specifically: first, the starting node and the target node of the distribution network are determined; secondly, the actual path cost and the estimated path cost between the starting node and the target node are obtained respectively, and the actual path cost and the estimated path cost are calculated by the A* algorithm to obtain the comprehensive path cost; finally, based on the comprehensive path cost, path transmission quality and data transmission efficiency between the starting node and the target node, the optimal data transmission path is determined.
[0123] In this way, the data transmission probability evaluation matrix is calculated through the A* algorithm, and factors such as the transmission quality of the path, data processing efficiency and comprehensive path cost are taken into consideration. The optimal data transmission path can be quickly perceived and calculated, and real-time data transmission path dynamic optimization can be achieved to ensure the efficiency and timeliness of data transmission.
[0124] The data acquisition system in the distribution network monitors the status of production nodes and communication nodes in real time, updates node status parameters according to the monitoring data, regenerates the data transmission probability assessment matrix based on the updated status parameters, and uses the A* algorithm to determine the optimal data transmission path corresponding to the updated status parameters. By continuously iterating this process, it can ensure that the optimal data transmission path is provided in all situations. When a production node or communication node is monitored to have a fault, the second best data transmission path will be selected to ensure the stability and reliability of data transmission.
[0125] The present application obtains the system data of the distribution network in real time, which helps the subsequent LSTM model to evaluate the probability of data transmission; by inputting the system data into a preset LSTM model, the data transmission probability between nodes can be quickly evaluated, which helps to subsequently select the optimal data transmission path and adapt to the changeable data flows of various new entities in the distribution network; using the A* algorithm to calculate the data transmission probability evaluation matrix, the optimal data transmission path can be quickly perceived and calculated, and real-time data transmission path dynamic optimization can be achieved, ensuring the efficiency and timeliness of data transmission.
[0126] The contents of the above-mentioned embodiment of the data transmission path optimization method based on the LSTM model are all applicable to the present storage medium embodiment. The functions specifically implemented by the present storage medium embodiment are the same as those of the above-mentioned embodiment of the data transmission path optimization method based on the LSTM model, and the beneficial effects achieved are also the same as those achieved by the above-mentioned embodiment of the data transmission path optimization method based on the LSTM model. Therefore, this embodiment can also realize real-time dynamic optimization of the data transmission path to ensure the efficiency and timeliness of data transmission.
[0127] Those skilled in the art can understand that all or part of the processes in the above embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes in the above embodiments. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.
[0128] The specific embodiments described above further describe the purpose, technical solutions and beneficial effects of the present application in detail. It should be understood that the above description is only a specific embodiment of the present application and is not intended to limit the scope of protection of the present application. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of protection of the present application.
Claims
1. A data transmission path optimization method based on LSTM model, characterized in that: include: Acquire system data of the distribution network in real time, wherein the system data includes production node data and communication node data; Input the system data into a preset LSTM model, predict the data transmission probability between each node, and form a data transmission probability evaluation matrix based on the data transmission probability; The data transmission probability evaluation matrix is calculated using the A* algorithm to determine the optimal data transmission path.
2. The data transmission path optimization method based on the LSTM model according to claim 1 is characterized in that: After the real-time acquisition of the system data of the power distribution network, the method further includes: Performing a quality assessment on the data transmission path of the system data to obtain a quality assessment result; Data is selected from the system data based on the quality assessment result to construct training samples for the LSTM model.
3. The data transmission path optimization method based on the LSTM model according to claim 2 is characterized in that: The calculation formula for the quality assessment is specifically: Where Q(p) is the quality score of path p, T(p) is the node data processing speed, M(p) is the node memory size, and M max is the maximum memory of the node, B(p) is the data transmission capacity of the node, and B max is the maximum transmission capacity of the node, D(p) is the delay, and w1,w2,w3,w4 are weight coefficients.
4. The data transmission path optimization method based on the LSTM model according to claim 1 is characterized in that: The system data is input into a preset LSTM model to predict the data transmission probability between each node, and a data transmission probability evaluation matrix is formed based on the data transmission probability, specifically: Inputting the system data into the LSTM model, wherein the LSTM model comprises an input layer, an LSTM layer and an output layer; The input layer processes the system data through a time window sliding technique to obtain a corresponding time series; Processing the time series through the LSTM layer to obtain a processing result; The processing result is converted into the data transmission probability between each node through the output layer, and a data transmission probability evaluation matrix is formed based on the data transmission probability.
5. The data transmission path optimization method based on the LSTM model according to claim 1 is characterized in that: The data transmission probability evaluation matrix is calculated by using the A* algorithm to determine the optimal data transmission path, specifically: Determine the starting node and target node of the distribution network; Respectively obtaining an actual path cost and an estimated path cost between the starting node and the target node, and calculating the actual path cost and the estimated path cost by using an A* algorithm to obtain a comprehensive path cost; An optimal data transmission path is determined based on the comprehensive cost of the path between the starting node and the target node, the path transmission quality and the data transmission efficiency.
6. A data transmission path optimization system based on LSTM model, characterized in that: include: Acquisition module, prediction module and calculation module; The acquisition module is used to acquire system data of the power distribution network in real time, wherein the system data includes production node data and communication node data; The prediction module is used to input the system data into a preset LSTM model, predict the data transmission probability between each node, and form a data transmission probability evaluation matrix based on the data transmission probability; The calculation module is used to calculate the data transmission probability evaluation matrix using the A* algorithm to determine the optimal data transmission path.
7. The data transmission path optimization system based on the LSTM model according to claim 6 is characterized in that: It also includes: a quality assessment module, the quality assessment module includes a quality assessment unit and a selection unit; The quality assessment unit is used to perform quality assessment on the data transmission path of the system data to obtain a quality assessment result; The selection unit is used to select data from the system data based on the quality assessment result to construct training samples for the LSTM model.
8. The data transmission path optimization system based on the LSTM model according to claim 7 is characterized in that: The calculation formula for the quality assessment is specifically: Where Q(p) is the quality score of path p, T(p) is the node data processing speed, M(p) is the node memory size, and M max is the maximum memory of the node, B(p) is the data transmission capacity of the node, and B max is the maximum transmission capacity of the node, D(p) is the delay, and w1,w2,w3,w4 are weight coefficients.
9. The data transmission path optimization system based on the LSTM model according to claim 6 is characterized in that: The prediction module includes: an input unit, a first processing unit, a second processing unit and a third processing unit; The input unit is used to input the system data into the LSTM model, wherein the LSTM model includes an input layer, an LSTM layer and an output layer; The first processing unit is used for the input layer to process the system data through the time window sliding technology to obtain the corresponding time series; The second processing unit is used to process the time series through the LSTM layer to obtain a processing result; The third processing unit is used to convert the processing result into the data transmission probability between each node through the output layer, and form a data transmission probability evaluation matrix based on the data transmission probability.
10. The data transmission path optimization system based on the LSTM model according to claim 6, characterized in that: The calculation module includes: a determination unit, an acquisition unit and a fourth processing unit; The determination unit is used to determine the starting node and the target node of the distribution network; The acquisition unit is used to respectively acquire the actual path cost and the estimated path cost between the starting node and the target node, and calculate the actual path cost and the estimated path cost by using an A* algorithm to obtain a comprehensive path cost; The fourth processing unit is used to determine the optimal data transmission path based on the comprehensive cost of the path between the starting node and the target node, the path transmission quality and the data transmission efficiency.