A flow prediction method, device, equipment and readable storage medium

Abstract

The application discloses a traffic prediction method, device and equipment and a readable storage medium. The traffic prediction method comprises the following steps: obtaining real-time traffic data of an optical line terminal through a software-defined network; preprocessing the real-time traffic data by using a first deep learning model to obtain preprocessed real-time traffic data; performing feature extraction on the preprocessed real-time traffic data by using a convolutional neural network to obtain feature-extracted real-time traffic data, wherein the feature-extracted real-time traffic data is used to reflect a change trend of the real-time traffic data; obtaining a traffic prediction result of the optical line terminal based on a pre-trained traffic prediction model and the feature-extracted real-time traffic data, wherein the traffic prediction result comprises traffic data of the optical line terminal in a future time period. According to the application, the traffic data of the optical line terminal in the future time period can be accurately and timely predicted according to the real-time traffic data.

Description

Technical Field

[0001] This application belongs to the field of communications, and in particular relates to a traffic prediction method, apparatus, device and readable storage medium. Background Technology

[0002] Currently, there are various methods and techniques for traffic prediction in optical line terminals (OLTs).

[0003] However, statistical analysis methods based on historical data identify and predict the periodicity, trends, and peaks of traffic through statistical methods and time series techniques, and then use these characteristics to predict future traffic. However, this method cannot cope with the dynamic changes and complexity of the network environment and has high requirements for historical data. Machine learning-based methods require high-quality data and accurate feature extraction, and also require large amounts of training data and computational resources. The prediction accuracy of methods based on network topology analysis is limited by the accurate modeling of network topology and connectivity.

[0004] In summary, the above-mentioned technical solutions are common methods in the field of OLT traffic prediction. However, these methods have certain limitations in prediction accuracy and real-time performance, and therefore require further improvement and innovation. Summary of the Invention

[0005] This application provides a traffic prediction method, apparatus, device, and readable storage medium, which can accurately and in real-time predict traffic data for future time periods based on real-time traffic data.

[0006] In a first aspect, embodiments of this application provide a traffic prediction method, which includes: acquiring real-time traffic data of an optical line terminal (OLT) through a software-defined network; preprocessing the real-time traffic data using a first deep learning model to obtain preprocessed real-time traffic data; extracting features from the preprocessed real-time traffic data using a convolutional neural network to obtain feature-extracted real-time traffic data, wherein the feature-extracted real-time traffic data reflects the changing trend of the real-time traffic data; and obtaining a traffic prediction result for the OLT based on a pre-trained traffic prediction model and the feature-extracted real-time traffic data, wherein the traffic prediction result includes traffic data of the OLT in a future time period.

[0007] According to the first aspect of this application, before acquiring real-time traffic data of an optical line terminal through a software-defined network, the traffic prediction method further includes: acquiring historical traffic data of the optical line terminal; preprocessing the historical traffic data using a first deep learning model to obtain preprocessed historical traffic data; extracting features from the preprocessed historical traffic data using a convolutional neural network to obtain feature-extracted historical traffic data; and training and evaluating the traffic prediction model based on the feature-extracted historical traffic data to obtain a trained traffic prediction model.

[0008] According to any of the foregoing embodiments of the first aspect of this application, a traffic prediction model is trained and evaluated based on historical traffic data after feature extraction to obtain a trained traffic prediction model, including: dividing the historical traffic data after feature extraction into a training set, a test set, and a validation set; training the traffic prediction model using the training set; inputting the test set into the trained traffic prediction model to obtain prediction results; comparing the prediction results with the actual results in the test set to obtain evaluation results; adjusting the model parameters of the trained traffic prediction model based on the evaluation results to obtain a trained traffic prediction model; and verifying the prediction accuracy of the trained traffic prediction model based on the validation set.

[0009] According to any of the foregoing embodiments of the first aspect of this application, adjusting the model parameters of the trained traffic prediction model based on the evaluation results to obtain a trained traffic prediction model includes: expanding the labeled data of the training set when the evaluation result indicates that the labeled data of the training set needs to be expanded, and training the trained traffic prediction model using the expanded labeled data of the training set to obtain a trained traffic prediction model; adjusting the network layer parameters of the trained traffic prediction model when the evaluation result indicates that the training efficiency of the trained traffic prediction model is less than a first preset threshold to obtain a trained traffic prediction model; and reducing the network structure of the trained traffic prediction model when the evaluation result indicates that the network structure of the trained traffic prediction model is greater than or equal to a second preset threshold, and training the traffic prediction model with the reduced network structure to obtain a trained traffic prediction model.

[0010] According to any of the foregoing embodiments of the first aspect of this application, adjusting the model parameters of the trained traffic prediction model based on the evaluation results to obtain a trained traffic prediction model includes: fusing multiple traffic prediction models from different training sets under any evaluation result to obtain a trained traffic prediction model; and / or adjusting the hyperparameters of the trained traffic prediction model based on the validation set to obtain a trained traffic prediction model.

[0011] According to any of the foregoing embodiments of the first aspect of this application, obtaining real-time traffic data of an optical line terminal through a software-defined network includes: deploying a software-defined network in a communication transmission network; deploying a traffic monitor in the software-defined network; and obtaining real-time traffic data of the optical line terminal through the traffic monitor.

[0012] According to any of the foregoing embodiments of the first aspect of this application, the first deep learning model includes an autoencoder model; preprocessing real-time traffic data using the first deep learning model to obtain preprocessed real-time traffic data includes: reconstructing and imputing the real-time traffic data using the autoencoder model; detecting outliers in the reconstructed and imputed real-time traffic data using the autoencoder model, detecting outliers in the real-time traffic data and processing the outliers; performing dimensionality reduction and feature extraction on the outlier-detected real-time traffic data using the autoencoder model; and performing data augmentation on the dimensionality-reduced and feature-extracted real-time traffic data using the autoencoder model to obtain preprocessed real-time traffic data.

[0013] According to any of the foregoing embodiments of the first aspect of this application, feature extraction is performed on preprocessed real-time traffic data using a convolutional neural network to obtain feature-extracted real-time traffic data, including: designing the structure of the convolutional neural network; extracting features at different scales through convolution operations and activation functions of the convolutional layers; and representing the extracted features as vectors or feature maps as feature-extracted real-time traffic data.

[0014] According to any of the foregoing embodiments of the first aspect of this application, after obtaining the traffic prediction result of the optical line terminal based on the pre-trained traffic prediction model and the real-time traffic data after feature extraction, and after the traffic prediction result includes the traffic data of the optical line terminal in the future time period, the traffic prediction method further includes: outputting early warning information based on the traffic prediction result; and outputting a device board switching strategy based on the early warning information.

[0015] Secondly, embodiments of this application provide a traffic prediction device, comprising: an acquisition module for acquiring real-time traffic data of an optical line terminal via a software-defined network; a preprocessing module for preprocessing the real-time traffic data using a first deep learning model to obtain preprocessed real-time traffic data; a feature extraction module for extracting features from the preprocessed real-time traffic data using a convolutional neural network to obtain feature-extracted real-time traffic data, the feature-extracted real-time traffic data reflecting the changing trend of the real-time traffic data; and a generation module for obtaining a traffic prediction result for the optical line terminal based on a pre-trained traffic prediction model and the feature-extracted real-time traffic data, the traffic prediction result including traffic data of the optical line terminal in a future time period.

[0016] Thirdly, embodiments of this application provide an electronic device, which includes: a processor and a memory storing computer program instructions; the processor executes the computer program instructions to implement the traffic prediction method as described in the first aspect.

[0017] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the traffic prediction method as described in the first aspect.

[0018] The traffic prediction method, apparatus, device, and readable storage medium of this application can obtain traffic data for future time periods of optical line terminals by inputting real-time traffic data after feature extraction into a traffic prediction model. Since the real-time traffic data after feature extraction is obtained through preprocessing and feature extraction, it can reflect the changing patterns and trends of the extracted real-time traffic data. This allows the traffic prediction model to learn the changing patterns and trends of the real-time traffic data when it is input, thereby achieving the effect of predicting subsequent traffic data based on the real-time input traffic data. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic flowchart of a traffic prediction method provided in an embodiment of this application;

[0021] Figure 2 This is another flowchart illustrating the traffic prediction method provided in the embodiments of this application;

[0022] Figure 3 This is another flowchart illustrating the traffic prediction method provided in the embodiments of this application;

[0023] Figure 4 This is another flowchart illustrating the traffic prediction method provided in the embodiments of this application;

[0024] Figure 5 This is another flowchart illustrating the traffic prediction method provided in the embodiments of this application;

[0025] Figure 6 This is another flowchart illustrating the traffic prediction method provided in the embodiments of this application;

[0026] Figure 7This is a schematic diagram of a flow prediction device provided in an embodiment of this application;

[0027] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0028] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0029] The acquisition, storage, use, and processing of data in this application all comply with the relevant provisions of national laws and regulations.

[0030] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0031] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0032] Various modifications and variations can be made to this application without departing from its spirit or scope, which will be apparent to those skilled in the art. Therefore, this application is intended to cover modifications and variations falling within the scope of the corresponding claims (the claimed technical solutions) and their equivalents. It should be noted that the embodiments provided in this application can be combined with each other without contradiction.

[0033] Before describing the technical solutions provided in the embodiments of this application, in order to facilitate understanding of the embodiments of this application, this application first specifically explains the problems existing in the related technologies:

[0034] Currently, there are various methods and technologies for traffic prediction in optical line terminals (OLTs). The following are some typical related technical solutions:

[0035] 1. Statistical Analysis Based on Historical Data: This method analyzes historical traffic data, including input / output traffic and packet frequency, to predict future traffic trends. It uses statistical methods and time series analysis to identify and predict traffic periodicity, trends, and peak characteristics. These characteristics are then used to predict future traffic. However, this method may not be able to handle the dynamic changes and complexity of the network environment and has high requirements for historical data.

[0036] 2. Machine Learning-Based Approaches: This approach utilizes machine learning algorithms, such as regression models, decision trees, and support vector machines, to train and model historical traffic data to predict future traffic changes. It learns patterns and regularities from historical data, thus enabling relatively accurate predictions of future traffic trends. However, this method requires high-quality data and accurate feature extraction, as well as substantial training data and computational resources.

[0037] 3. Deep Learning-Based Approach: This approach utilizes deep neural networks, such as Recurrent Neural Networks (RNNs) and Long Short-Term Memory Networks (LSTMs), to model and train on large amounts of historical traffic data to achieve more accurate traffic prediction. Deep learning models can capture complex nonlinear relationships in the data and have strong expressive power for traffic prediction. However, this method requires a large amount of labeled data and computational resources, and the training and optimization process of the model is relatively complex.

[0038] 4. Network Topology Analysis-Based Method: This method predicts future traffic changes by analyzing the topology and connectivity of the communication transmission network, as well as the communication patterns and traffic transmission paths between OLT devices. It considers the impact of network topology on traffic transmission and predicts future traffic load by establishing network models and algorithms. However, the prediction accuracy of this method is limited by the accurate modeling of the network topology and connectivity.

[0039] In summary, the aforementioned technical solutions are common methods in the field of OLT traffic prediction. However, these methods have certain limitations in terms of prediction accuracy and real-time performance, thus requiring further improvement and innovation.

[0040] The lack of accuracy is evident in the fact that traffic prediction methods in related technologies often fail to provide accurate predictions when faced with dynamic and complex network environments. Traditional statistical analysis methods rely heavily on historical data and cannot effectively cope with changes in the network environment. While machine learning and deep learning methods can handle more complex data relationships, they have high requirements for data quality and feature extraction, and require a large amount of training data and computational resources.

[0041] The poor real-time performance of traffic prediction methods in related technologies often requires offline modeling and training, making it impossible to respond to changes in the network environment in real time. For application scenarios that require timely adjustments to network resources and service failover for equipment boards, real-time performance is a critical requirement.

[0042] To address the problems in the aforementioned related technologies, embodiments of this application provide a method, apparatus, device, and readable storage medium for traffic prediction. The following is a detailed description of a traffic prediction method from an embodiment of this application.

[0043] Figure 1 This is a schematic flowchart of a traffic prediction method provided in an embodiment of this application. Figure 1 As shown, the traffic prediction method 100 may include: S110 to S140.

[0044] S110. Obtain real-time traffic data of optical line terminals through software-defined networking.

[0045] For details on how the S110 obtains real-time traffic data from optical line terminals through software-defined networking, please refer to [link to documentation]. Figure 2 Detailed explanation of some parts.

[0046] Figure 2 This is another flowchart illustrating the traffic prediction method provided in the embodiments of this application, such as... Figure 2 As shown, in some embodiments, S110 may include S111 to S113.

[0047] S111. Deploy software-defined networks in communication transmission networks.

[0048] It is important to understand that Software Defined Networking (SDN) is a novel network architecture and a method of network virtualization. Its core technology is to separate the control plane from the data plane of network devices, thereby enabling flexible control of network traffic. In the embodiments of this application, SDN acts as a central controller responsible for network management and control.

[0049] S112. Deploy traffic monitors in software-defined networks.

[0050] For example, a traffic monitor can be deployed in an SDN controller or network switch. The traffic monitor is used to monitor the traffic information of optical line terminal (OLT) devices in real time.

[0051] S113. Obtain real-time traffic data of the optical line terminal through the traffic monitor.

[0052] The real-time traffic data of the optical line terminal (OLT) includes information such as the input and output traffic of each OLT port and the message frequency.

[0053] After obtaining the real-time traffic data of the optical line terminal (OLT) via S110, S120 is executed.

[0054] S120. Use the first deep learning model to preprocess the real-time traffic data to obtain preprocessed real-time traffic data.

[0055] For details of the preprocessing process, please refer to Figure 3 Partial detailed description.

[0056] Figure 3 This is another flowchart illustrating the traffic prediction method provided in the embodiments of this application, such as... Figure 3 As shown, in some embodiments, the first deep learning model may include an autoencoder model, and when the first deep model is an autoencoder model, S120 may include S121 to S124.

[0057] S121. Use an autoencoder model to reconstruct and fill in real-time traffic data.

[0058] It is important to understand that autoencoder models are an unsupervised machine learning technique. Therefore, autoencoder models can autonomously learn the inherent patterns of real-time traffic data, and then reconstruct and fill in the data based on these patterns to restore the accuracy of missing data and improve the completeness of the real-time traffic data.

[0059] S122. Use an autoencoder model to detect outliers in the reconstructed and filled real-time traffic data, identify outliers in the real-time traffic data, and process the outliers.

[0060] Based on the same principle, the autoencoder model can also learn the distribution and pattern of the acquired real-time traffic data. By setting appropriate thresholds or using anomaly detection algorithms, it can detect outliers in the reconstructed and filled real-time traffic data, identify and mark outliers that do not conform to the normal data distribution, and then delete or modify the marked outliers.

[0061] S123. Use an autoencoder model to perform dimensionality reduction and feature extraction on the real-time traffic data after outlier detection.

[0062] Using an autoencoder model to transform high-dimensional data into low-dimensional data can reduce the storage requirements for subsequent real-time traffic data and also reduce the computational complexity. Simultaneously, in some embodiments, normalization techniques (Z-score normalization) are used to normalize the real-time traffic data, ensuring that the data is within a uniform scale and improving the training performance of the deep learning model. For example, some real-time traffic data may contain maxima or minima, which can negatively impact the training performance of the deep learning model. To improve training effectiveness, normalization techniques (Z-score normalization) are necessary to normalize the real-time traffic data.

[0063] S124. Use an autoencoder model to perform data augmentation on the real-time traffic data after dimensionality reduction and feature extraction to obtain preprocessed real-time traffic data.

[0064] It's important to understand that the pre-trained traffic prediction model used in subsequent training is generated based on a deep learning model. Training a deep learning model requires a large amount of training data, necessitating data augmentation techniques. Data augmentation involves rotating, flipping, translating, and scaling the original real-time traffic data to generate new samples. This increases the diversity and quantity of training data, improving the robustness and generalization ability of the deep learning model.

[0065] In summary, the S120 preprocesses the collected real-time traffic data, making the collected real-time traffic data more accurate, more abundant, and requiring less storage space.

[0066] S130. Feature extraction is performed on the preprocessed real-time traffic data using a convolutional neural network to obtain the feature-extracted real-time traffic data.

[0067] The extracted real-time traffic data is used to reflect the changing trends of real-time traffic data. For details on the feature extraction process using a convolutional neural network, please refer to [link to documentation]. Figure 4 Detailed explanation of some parts.

[0068] Figure 4 This is another flowchart illustrating the traffic prediction method provided in the embodiments of this application, such as... Figure 4 As shown, in some embodiments, S130 may include S131 to S133.

[0069] S131. Design the structure of a convolutional neural network.

[0070] It is important to understand that convolutional neural networks are also a type of deep learning model. Convolutional neural networks include convolutional layers, pooling layers, and fully connected layers. The appropriate number of network layers, convolutional kernel size, and pooling method are selected based on the characteristics of the preprocessed real-time traffic data and the requirements of the traffic prediction task.

[0071] S132. Extract features at different scales through convolution operations and activation functions of convolutional layers.

[0072] It is important to understand that convolutional layers contain convolutional kernels, and different sizes of convolutional kernels correspond to features at different scales. When the convolutional kernel is small, it extracts local features, i.e., detailed features, while when the convolutional kernel is large, it is better able to capture overall features.

[0073] In some embodiments, the extracted features can also be mapped to a higher-level feature space. By stacking multiple convolutional and pooling layers, more abstract and advanced features can be gradually extracted to capture more complex patterns and relationships in real-time traffic data.

[0074] S133. Represent the extracted features as vectors or feature maps, which serve as real-time traffic data after feature extraction.

[0075] It's important to understand that a vector is a one-dimensional array containing multiple elements, with each element representing a feature. A feature map, on the other hand, is two-dimensional, where each element represents a local feature of the real-time traffic data. This feature-extracted real-time traffic data serves as input for subsequent traffic prediction tasks, reflecting the changing trends of the real-time traffic data.

[0076] In some embodiments, after S133, the real-time traffic data after feature extraction can also be filtered to select data including features such as timestamp, traffic volume, seasonality, periodicity and holidays, so as to reflect the changing patterns and trends of real-time traffic data.

[0077] S140. Based on the pre-trained traffic prediction model and the real-time traffic data after feature extraction, the traffic prediction result of the optical line terminal is obtained. The traffic prediction result includes the traffic data of the optical line terminal in the future time period.

[0078] S140 means that by inputting the real-time traffic data after feature extraction into a pre-trained traffic prediction model, the traffic prediction results of the optical line terminal can be generated.

[0079] For details on the traffic prediction model building and training process, please refer to [link / reference]. Figure 5 A detailed explanation of some parts.

[0080] The traffic prediction method of this application can obtain the traffic data of the optical line terminal for a future time period by inputting the real-time traffic data after feature extraction into the traffic prediction model. Since the real-time traffic data after feature extraction is obtained through preprocessing and feature extraction, it can reflect the changing patterns and trends of the extracted real-time traffic data. This allows the traffic prediction model to learn the changing patterns and trends of the real-time traffic data when it is input, thereby achieving the effect of predicting subsequent traffic data based on the real-time input traffic data.

[0081] In some embodiments, after obtaining the traffic prediction result of the optical line terminal based on the pre-trained traffic prediction model and the real-time traffic data after feature extraction in S140, the traffic prediction method 100 further includes: step one and step two.

[0082] Step 1: Output early warning information based on the traffic forecast results.

[0083] In some embodiments, the warning information is stored in a rule engine. The rule engine in this application can be based on a traditional rule engine, defining warning rules and conditions. The rule engine in this application can judge and process the traffic prediction results according to preset rules and thresholds, further filtering and selecting the warning information.

[0084] Step 2: Based on the early warning information, output the device board replacement strategy.

[0085] Step two can be implemented by establishing an appropriate state space, action space, and reward mechanism to train an intelligent agent to recommend the best device board failover strategy, thereby improving network performance and reliability.

[0086] Here, the state space refers to the different states or scenarios the network may be in; the action space refers to the actions that the intelligent agent can take, such as recommending which device boards to replace; and the reward mechanism refers to the feedback the intelligent agent receives based on the effectiveness of its actions. The goal of step two is to enable the intelligent agent to learn through trial and error which actions (device board replacement) can maximize network performance and reliability.

[0087] Different warning messages correspond to different device board replacement strategies. For example, low traffic prediction: if the warning message indicates a significant decrease in traffic over a future period, it can be recommended to replace some high-capacity boards with low-capacity boards to save energy and reduce operating costs. High traffic prediction: if the warning message indicates a traffic peak within a specific time period, it is recommended to replace with higher-performance boards to cope with the increased traffic and ensure the network does not experience congestion or latency. Traffic fluctuation prediction: if the warning message indicates a significant fluctuation in traffic, it may be recommended to install boards with stronger dynamic processing capabilities to quickly adapt to traffic changes. Continuously stable traffic: if the warning message indicates a long-term stable traffic level, there is no need to replace boards; instead, routine maintenance and optimization can be performed. Step two can resolve potential faults and problems in the network.

[0088] Figure 5 This is another flowchart illustrating the traffic prediction method provided in the embodiments of this application, such as... Figure 5 As shown, before obtaining real-time traffic data of the optical line terminal through software-defined networking in S110, the traffic prediction method 100 may further include: S510 to S540.

[0089] S510: Obtain historical traffic data of the optical line terminal.

[0090] S520. Use the first deep learning model to preprocess the historical traffic data to obtain preprocessed historical traffic data.

[0091] Preprocessing historical traffic data is essential to ensure data quality and consistency. The preprocessed historical traffic data has been labeled. The process of preprocessing historical traffic data using the first deep learning model is the same as the process of preprocessing real-time traffic data using the first deep learning model, and will not be elaborated here. Please refer to [link to relevant documentation]. Figure 3 Detailed explanation of some parts.

[0092] In some embodiments, preprocessed historical traffic data is stored in a central database or cloud platform for subsequent data analysis and processing.

[0093] S530. Feature extraction is performed on the preprocessed historical traffic data using a convolutional neural network to obtain the feature-extracted historical traffic data.

[0094] S530 is designed to ensure that the historical traffic data after feature extraction reflects the changing patterns and trends of the traffic data. The process of extracting features from preprocessed historical traffic data using a convolutional neural network is the same as the process of extracting features from preprocessed real-time traffic data using a convolutional neural network, and will not be elaborated here; please refer to [link to relevant documentation]. Figure 4 Detailed explanation of some parts.

[0095] S540. Based on the historical traffic data after feature extraction, train and evaluate the traffic prediction model to obtain a trained traffic prediction model.

[0096] For the specific training and evaluation process in S540, please refer to [link / reference]. Figure 6 Detailed explanation of some parts.

[0097] Figure 6 This is another flowchart illustrating the traffic prediction method provided in the embodiments of this application, such as... Figure 6 As shown in S540, the traffic prediction model is trained and evaluated based on the historical traffic data after feature extraction, and the trained traffic prediction model can include S541 to S546.

[0098] S541. Divide the historical traffic data after feature extraction into training set, test set and validation set.

[0099] The training set is used to train the model, the test set is used to evaluate the model's performance and obtain the evaluation results, and the validation set is used to verify the prediction accuracy of the trained traffic prediction model.

[0100] S542. Train the traffic prediction model using the training set.

[0101] In some embodiments, the traffic prediction model is a Long Short Term Memory (LSTM) network model. LSTM is a type of deep learning model that can process sequential data and capture temporal dependencies, and has a strong ability to model complex patterns and trends in traffic data.

[0102] The training set consists of a portion of labeled historical traffic data. The traffic prediction model is trained using this historical traffic data. The model is iteratively trained using optimization algorithms (such as stochastic gradient descent) and loss functions (such as mean squared error), allowing it to gradually learn the patterns and regularities in the historical traffic data.

[0103] S543. Input the test set into the trained traffic prediction model to obtain the prediction results.

[0104] S544. Compare the predicted results with the actual results in the test set to obtain the evaluation results.

[0105] S545. Adjust the model parameters of the trained traffic prediction model based on the evaluation results to obtain a trained traffic prediction model.

[0106] Different assessment results correspond to different adjustments to the traffic prediction model.

[0107] For example, if the evaluation result indicates that the labeled data of the training set needs to be expanded, the labeled data of the training set is expanded, and the trained traffic prediction model is trained using the expanded labeled data of the training set to obtain a trained traffic prediction model.

[0108] In some embodiments, data augmentation techniques are used to expand the labeled data of the training set. By performing random transformations, rotations, and scaling on the original data, more samples can be generated, increasing the diversity of historical traffic data and thus improving the generalization ability of the traffic prediction model.

[0109] If the evaluation result shows that the training efficiency of the trained traffic prediction model is less than a first preset threshold, the network layer parameters of the trained traffic prediction model are adjusted to obtain a well-trained traffic prediction model. This addresses the issues of computational resources and training complexity. The trained traffic prediction model is used as the initial parameters, and then fine-tuning or adjusting some network layers is performed to adapt it to the specific OLT traffic prediction task.

[0110] If the evaluation result indicates that the network structure of the trained traffic prediction model is greater than or equal to a second preset threshold, the network structure of the trained traffic prediction model is reduced, and the traffic prediction model with the reduced network structure is trained again to obtain a well-trained traffic prediction model. In other words, to address the problem of high training complexity in traffic prediction models, the structural design of the traffic prediction model is optimized. A shallower neural network structure is adopted to reduce the model's training time and computational resource consumption.

[0111] Under any evaluation result, multiple traffic prediction models with different training sets are fused to obtain a trained traffic prediction model, and / or, based on the validation set, the hyperparameters (such as learning rate, batch size, and optimizer) of the trained traffic prediction model are adjusted to obtain a trained traffic prediction model.

[0112] Fusing multiple traffic prediction models can improve prediction performance and robustness. By combining the prediction results of multiple models, model errors and biases can be reduced, improving overall prediction accuracy. Systematic tuning of hyperparameters can improve model training speed and performance.

[0113] S546. Verify the prediction accuracy of the trained traffic prediction model based on the validation set.

[0114] Based on the same technical concept as the flow detection method provided in the above embodiments, this application also provides a flow detection device. Please see the following embodiments.

[0115] Figure 7 This is a schematic diagram of a flow prediction device provided in an embodiment of this application, such as... Figure 7As shown, the flow prediction device 700 may include: an acquisition module 710, a preprocessing module 720, a feature extraction module 730, and a generation module 740.

[0116] The acquisition module 710 is used to acquire real-time traffic data of optical line terminals through a software-defined network.

[0117] The preprocessing module 720 is used to preprocess the real-time traffic data using the first deep learning model to obtain preprocessed real-time traffic data.

[0118] The feature extraction module 730 is used to extract features from the preprocessed real-time traffic data through a convolutional neural network to obtain the feature-extracted real-time traffic data, which is used to reflect the changing trend of the real-time traffic data.

[0119] The generation module 740 is used to obtain the traffic prediction results of the optical line terminal based on the pre-trained traffic prediction model and the real-time traffic data after feature extraction. The traffic prediction results include the traffic data of the optical line terminal in the future time period.

[0120] In some embodiments, the acquisition module 710 is further configured to acquire historical traffic data of the optical line terminal; the preprocessing module 720 is further configured to preprocess the historical traffic data using a first deep learning model to obtain preprocessed historical traffic data; the feature extraction module 730 is further configured to extract features from the preprocessed historical traffic data using a convolutional neural network to obtain feature-extracted historical traffic data; the traffic prediction device 700 may further include a training and evaluation module, which is configured to train and evaluate the traffic prediction model based on the feature-extracted historical traffic data to obtain a trained traffic prediction model.

[0121] In some embodiments, the training and evaluation module is specifically used to divide the historical traffic data after feature extraction into a training set, a test set, and a validation set; train the traffic prediction model using the training set; input the test set into the trained traffic prediction model to obtain prediction results; compare the prediction results with the actual results in the test set to obtain evaluation results; adjust the model parameters of the trained traffic prediction model according to the evaluation results to obtain a trained traffic prediction model; and verify the prediction accuracy of the trained traffic prediction model based on the validation set.

[0122] In some embodiments, the training evaluation module is specifically used to: expand the labeled data of the training set when the evaluation result indicates that the labeled data of the training set needs to be expanded; train the trained traffic prediction model using the expanded labeled data of the training set to obtain a trained traffic prediction model; adjust the network layer parameters of the trained traffic prediction model when the evaluation result indicates that the training efficiency of the trained traffic prediction model is less than a first preset threshold to obtain a trained traffic prediction model; and reduce the network structure of the trained traffic prediction model when the evaluation result indicates that the network structure of the trained traffic prediction model is greater than or equal to a second preset threshold, and train the traffic prediction model with the reduced network structure to obtain a trained traffic prediction model.

[0123] In some embodiments, the training evaluation module is specifically used to fuse multiple traffic prediction models from different training sets under any evaluation result to obtain a trained traffic prediction model; and / or, based on the validation set, adjust the hyperparameters of the trained traffic prediction model to obtain a trained traffic prediction model.

[0124] In some embodiments, the acquisition module 710 is specifically used to deploy a software-defined network in a communication transmission network; deploy a traffic monitor in the software-defined network; and acquire real-time traffic data of the optical line terminal through the traffic monitor.

[0125] In some embodiments, the first deep learning model includes an autoencoder model; the preprocessing module 720 is specifically used to reconstruct and impute real-time traffic data using the autoencoder model; to detect outliers in the reconstructed and imputed real-time traffic data using the autoencoder model, and to process the outliers; to perform dimensionality reduction and feature extraction on the outlier-detected real-time traffic data using the autoencoder model; and to perform data augmentation on the dimensionality-reduced and feature-extracted real-time traffic data using the autoencoder model to obtain preprocessed real-time traffic data.

[0126] In some embodiments, the feature extraction module 730 is specifically used to design the structure of a convolutional neural network; extract features at different scales through the convolution operation and activation function of the convolutional layer; and represent the extracted features as vectors or feature maps as real-time traffic data after feature extraction.

[0127] In some embodiments, the traffic prediction device 700 further includes an output module, which is specifically used to output early warning information based on the traffic prediction results and to output a device board switching strategy based on the early warning information.

[0128] The traffic prediction device of this application embodiment can obtain the traffic data of the optical line terminal for a future time period by inputting the real-time traffic data after feature extraction into the traffic prediction model. Since the real-time traffic data after feature extraction is obtained through preprocessing and feature extraction, it can reflect the changing patterns and trends of the extracted real-time traffic data. This allows the traffic prediction model to learn the changing patterns and trends of the real-time traffic data when it is input, thereby achieving the effect of predicting subsequent traffic data based on the real-time input traffic data.

[0129] Figure 8 A schematic diagram of an electronic device provided in an embodiment of this application, such as... Figure 8 As shown, the electronic device includes a processor 801 and a memory 802 storing computer program instructions.

[0130] Specifically, the processor 801 may include a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0131] Memory 802 may include mass storage for data or instructions. For example, and not limitingly, memory 802 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where suitable, memory 802 may include removable or non-removable (or fixed) media. Where suitable, memory 802 may be internal or external to an electronic device. In a particular embodiment, memory 802 is a non-volatile solid-state memory.

[0132] Memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Therefore, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to the first aspect of this application.

[0133] The processor 801 implements any of the traffic detection methods described in the above embodiments by reading and executing computer program instructions stored in the memory 802.

[0134] In one example, an electronic device may further include a communication interface 803 and a bus 810. Wherein, as... Figure 8 As shown, the processor 801, memory 802, and communication interface 803 are connected through bus 810 and complete communication with each other.

[0135] The communication interface 803 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0136] Bus 810 includes hardware, software, or both, that couples components of a terminal service device together. For example, and not limited to, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Linear Predictive Coding (LPC) bus, a memory bus, a MicroChannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (Peripheral Component Interconnect-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VESA Local Bus, VLB) bus, or other suitable buses, or a combination of two or more of these. Where appropriate, bus 810 may include one or more buses. Although specific buses are described and illustrated in the embodiments of this application, this application considers any suitable bus or interconnection.

[0137] This device can execute the flow detection method in this application embodiment based on various units / components in the data processing system, thereby achieving a combination of Figures 1 to 6 Described traffic detection method

[0138] Furthermore, in conjunction with the traffic detection methods in the above embodiments, this application embodiment can provide a computer-readable storage medium for implementation. This computer-readable storage medium stores computer program instructions; when executed by a processor, these computer program instructions implement any of the traffic detection methods in the above embodiments.

[0139] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0140] The functional blocks shown in the above-described block diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, read-only memory (ROM), flash memory, erasable read-only memory (EROM), floppy disks, compact disc read-only memory (CD-ROM), optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0141] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0142] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0143] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A flow prediction method, characterized in that, include: Real-time traffic data of optical line terminals can be obtained through software-defined networking; The real-time traffic data is preprocessed using a first deep learning model to obtain preprocessed real-time traffic data. Feature extraction is performed on the preprocessed real-time traffic data using a convolutional neural network to obtain feature-extracted real-time traffic data, which is used to reflect the changing trend of the real-time traffic data. Based on the pre-trained traffic prediction model and the real-time traffic data after feature extraction, the traffic prediction result of the optical line terminal is obtained, and the traffic prediction result includes the traffic data of the optical line terminal in the future time period. Based on the traffic prediction results, an early warning message is output; Based on the warning information, the device board switching strategy of the optical line terminal is output. The board switching strategy includes switching to a high-performance board when the traffic is high, switching to a low-capacity board when the traffic is low, and switching to a dynamic processing capability board when the traffic fluctuates. The first deep learning model includes an autoencoder model; The step of preprocessing the real-time traffic data using a first deep learning model to obtain preprocessed real-time traffic data includes: The real-time traffic data is reconstructed and filled using an autoencoder model; The autoencoder model is used to detect outliers in the reconstructed and imputed real-time traffic data, and the outliers are then processed. The autoencoder model is used to perform dimensionality reduction and feature extraction on real-time traffic data after outlier detection. The autoencoder model is used to perform data augmentation on the real-time traffic data after dimensionality reduction and feature extraction to obtain the preprocessed real-time traffic data.

2. The flow prediction method according to claim 1, characterized in that, Before acquiring real-time traffic data of the optical line terminal via a software-defined network, the method further includes: Obtain the historical traffic data of the optical line terminal; The historical traffic data is preprocessed using the first deep learning model to obtain preprocessed historical traffic data. The convolutional neural network is used to extract features from the preprocessed historical traffic data to obtain the historical traffic data after feature extraction. Based on the historical traffic data after feature extraction, the traffic prediction model is trained and evaluated to obtain a trained traffic prediction model.

3. The flow prediction method according to claim 2, characterized in that, The step of training and evaluating the traffic prediction model based on the historical traffic data after feature extraction to obtain a trained traffic prediction model includes: The historical traffic data after feature extraction is divided into a training set, a test set, and a validation set; The traffic prediction model is trained using the training set. The test set is input into the trained traffic prediction model to obtain the prediction results; The predicted results are compared with the actual results in the test set to obtain the evaluation results; The model parameters of the trained traffic prediction model are adjusted based on the evaluation results to obtain the trained traffic prediction model. The accuracy of the trained traffic prediction model is verified based on the validation set.

4. The flow prediction method according to claim 3, characterized in that, The step of adjusting the model parameters of the trained traffic prediction model based on the evaluation results to obtain the trained traffic prediction model includes: If the evaluation result indicates that the labeled data of the training set needs to be expanded, the labeled data of the training set is expanded, and the trained traffic prediction model is trained using the expanded labeled data of the training set to obtain a trained traffic prediction model. If the evaluation result indicates that the training efficiency of the trained traffic prediction model is less than a first preset threshold, the network layer parameters of the trained traffic prediction model are adjusted to obtain a trained traffic prediction model. If the evaluation result indicates that the network structure of the trained traffic prediction model is greater than or equal to a second preset threshold, the network structure of the trained traffic prediction model is reduced, and the traffic prediction model with the reduced network structure is trained to obtain a trained traffic prediction model.

5. The flow prediction method according to claim 3, characterized in that, The step of adjusting the model parameters of the trained traffic prediction model based on the evaluation results to obtain the trained traffic prediction model includes: Under any evaluation result, the traffic prediction models of multiple different training sets are fused to obtain a trained traffic prediction model; And / or, based on the validation set, adjust the hyperparameters of the trained traffic prediction model to obtain a trained traffic prediction model.

6. The flow prediction method according to claim 1, characterized in that, The acquisition of real-time traffic data of optical line terminals through software-defined networking includes: Deploy the software-defined network in the communication transmission network; Deploy traffic monitors in the software-defined network; The real-time traffic data of the optical line terminal is obtained through the traffic monitor.

7. The flow prediction method according to claim 1, characterized in that, The step of extracting features from the preprocessed real-time traffic data using a convolutional neural network to obtain feature-extracted real-time traffic data includes: Design the structure of the convolutional neural network; Features at different scales are extracted through convolution operations and activation functions in convolutional layers; The extracted features are represented as vectors or feature maps, serving as real-time traffic data after feature extraction.

8. A flow prediction device, characterized in that, The flow prediction device includes: The acquisition module is used to acquire real-time traffic data of optical line terminals through a software-defined network. The preprocessing module is used to preprocess the real-time traffic data using a first deep learning model to obtain preprocessed real-time traffic data. The feature extraction module is used to extract features from the preprocessed real-time traffic data through a convolutional neural network to obtain feature-extracted real-time traffic data, which is used to reflect the changing trend of the real-time traffic data. The generation module is used to obtain the traffic prediction result of the optical line terminal based on the pre-trained traffic prediction model and the real-time traffic data after feature extraction. The traffic prediction result includes the traffic data of the optical line terminal in a future time period. The output module is used to output early warning information based on the traffic prediction results; and to output a board switching strategy for the optical line terminal based on the early warning information. The board switching strategy includes switching to a high-performance board when the traffic is high, switching to a low-capacity board when the traffic is low, and switching to a dynamic processing capability board when the traffic fluctuates. The first deep learning model includes an autoencoder model; The preprocessing module is specifically used for: reconstructing and imputing the real-time traffic data using an autoencoder model; detecting outliers in the reconstructed and imputed real-time traffic data using the autoencoder model, and processing the outliers; reducing the dimensionality and extracting features from the outlier-detected real-time traffic data using the autoencoder model; and augmenting the reduced dimensionality and extracted features from the real-time traffic data using the autoencoder model to obtain the preprocessed real-time traffic data.

9. An electronic device, characterized in that, The electronic device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements the traffic prediction method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions that, when executed by a processor, implement the traffic prediction method as described in any one of claims 1-7.

11. A computer program product, characterized in that, When the instructions in the computer program product are executed by the processor of the electronic device, the electronic device performs the traffic prediction method as described in any one of claims 1-7.

Images