Optical cross connection method and device based on wavelength division matrix

By converting network traffic data into images and extracting features using deep learning algorithms, we build a wavelength allocation strategy generator, and solving the problem of low response and operation and maintenance efficiency of traditional optical communication networks in traffic change, realizing efficient data transmission and intelligence manage.

CN119012055APending Publication Date: 2024-11-22YANCHENG POWER SUPPLY CO STATE GRID JIANGSU ELECTRIC POWER CO

Patent Information

Application Number
CN202411065651.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-05
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

When traditional wavelength division matrix optical communication networks face expansion of scale and complex traffic, it is difficult to respond to traffic changes in real time, resulting in low operation and maintenance efficiency, difficulty in positioning faults, low port utilization, and inability to monitor the status of optical fiber ports in real time.

Method used

Adopting advanced data analysis technology and deep learning algorithms, network traffic data is converted into images, timing features are extracted through image processing methods, and a classifier is used to build a wavelength allocation strategy generator to realize intelligent wavelength allocation strategy adjustment and dynamic response. Changes in traffic.

Benefits of technology

It improves the power grid data transmission capability and communication flexibility, enhances the intelligent management level of the power grid, optimizes energy configuration and utilization efficiency, reduces fault location time, and improves port resource utilization.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119012055A_ABST
    Figure CN119012055A_ABST
Patent Text Reader

Abstract

The invention provides an optical cross connection method and device based on a wavelength division matrix. The method comprises the steps that S1, a time sequence of network flow values collected by a network sniffer is acquired; s2, converting the time sequence of the network traffic value into image data to obtain a network traffic time sequence gray level image; s3, carrying out network flow local time sequence correlation feature extraction and reverse screening enhancement on the network flow time sequence grayscale image to obtain a set of enhanced network flow local time sequence correlation feature vectors; s4, inputting the set of the enhanced network flow local time sequence correlation feature vectors into the energy measurement attention fusion network based on sequence labeling to obtain a network flow time sequence attention aggregation representation vector; and S5, determining a wavelength allocation strategy based on the network flow time sequence attention aggregation representation vector. The invention provides an optical cross connection method and device based on a wavelength division matrix, which are used for judging how to dynamically adjust a wavelength allocation strategy according to a real-time network state, so that a network can quickly respond to flow change.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of wavelength division matrix, and in particular relates to an optical cross-connection method and device based on wavelength division matrix. Background Art

[0002] With the rapid development of the Internet and communication technology, the demand for data transmission has exploded. Optical communication has become a key technology to meet the demand for high-speed data transmission due to its advantages such as high bandwidth, long transmission distance and low loss.

[0003] The power communication network is currently based on optical fiber communication. The power communication optical fiber resources are widely distributed, difficult to operate and maintain, and their scale is increasing year by year. The power communication optical fiber resources carry important businesses such as relay protection, dispatching automation, production and marketing management, and the security of communication optical fiber resources is directly related to the safe and stable operation of the power grid.

[0004] In optical communication networks, wavelength division matrix, also known as wavelength division multiplexing (WDM) matrix, is a key component used to achieve wavelength-level routing and switching in optical communication networks. The wavelength division matrix can dynamically connect and switch between different wavelength channels to achieve efficient data transmission. Optical Cross-Connect (OXC) is an important device for achieving dynamic routing of optical signals within network nodes.

[0005] However, as the scale of networks continues to expand and traffic patterns become increasingly complex, traditional wavelength allocation methods usually perform static planning in the early stages of network establishment, which makes it difficult to adapt to real-time changes in network traffic and sudden demands.

[0006] As the number of optical fiber resources continues to increase, the demand for refined operation and maintenance continues to increase. The traditional method of sending operation and maintenance personnel to the station for manual fiber jumper is not only time-consuming and labor-intensive, but also has many problems in the operation and maintenance process: 1. The problem of dumb resource management. The traditional fiber jumper method requires manual operation based on the work order, and the randomness and disorder of manual fiber jumper are difficult to avoid. In addition, the phenomenon of false services on optical fiber labels or mismatch between label services and gateway services also occurs from time to time, which poses a challenge to the routine operation and maintenance of optical fiber resources. 2. Fault detection and positioning are difficult. When the optical cable fails, for example, it is damaged by external forces, causing damage to the optical fiber. The traditional fault positioning method requires operation and maintenance personnel to enter the station, carry out detection with a handheld OTDR, and determine the fault point, which is difficult to locate quickly and accurately. 3. Low timeliness. The dispatch response time is too long, often in hours. Personnel need to enter the station to perform manual fiber jumper on site, which is cumbersome and time-consuming. 4. Low port utilization. Although there are many optical cables and optical distribution in the current station, many ports on the optical distribution are idle. The ports cannot be used reasonably, and the port resource utilization rate is low. In addition, the status of the optical fiber port cannot be monitored in real time, and illegal and irregular plugging and unplugging cannot be avoided. Therefore, an optimized solution is expected. Summary of the invention

[0007] In order to solve the above problems, the present invention provides an optical cross-connection method and device based on a wavelength division matrix. By utilizing advanced data analysis technology and deep learning algorithms, network traffic data is cleverly converted into the form of images, so that image processing means can be used to perform in-depth analysis and pattern recognition on network traffic to obtain the complex time-series nonlinear transformation feature patterns of network traffic, and a classifier is used to construct a wavelength allocation strategy generator to intelligently determine how the wavelength allocation strategy should be dynamically adjusted according to the real-time network status, so that the network can quickly respond to traffic changes.

[0008] The present invention is specifically a wavelength division matrix-based optical cross-connection method, the optical cross-connection method comprising the following steps:

[0009] S1: Obtain the time series of network traffic values ​​collected by the network sniffer;

[0010] S2: converting the time series of the network traffic value into image data to obtain a network traffic time series grayscale image;

[0011] S3: extracting local time series correlation features of network traffic and performing reverse screening and enhancement on the network traffic time series grayscale image to obtain a set of enhanced local time series correlation feature vectors of network traffic;

[0012] S4: inputting the set of the enhanced network traffic local time series correlation feature vectors into the energy metric attention fusion network based on sequence labeling to obtain a network traffic time series attention aggregation representation vector;

[0013] S5: Determine a wavelength allocation strategy based on the network traffic timing attention aggregation representation vector, where the wavelength allocation strategy is used to indicate increasing the time interval between wavelengths, reducing the time interval between wavelengths, or maintaining the time interval between wavelengths.

[0014] Optionally, converting the time series of the network traffic value into image data to obtain a network traffic time series grayscale image includes:

[0015] S21: Arranging the time series of the network traffic values ​​according to the time dimension into a network traffic time series input vector;

[0016] S22: Input the network traffic time series input vector into a vector-to-image converter to obtain the network traffic time series grayscale image.

[0017] Optionally, inputting the network traffic time series input vector into a vector-to-image converter to obtain the network traffic time series grayscale image includes:

[0018] S221: performing vector segmentation on the network traffic timing input vector to obtain a sequence of network traffic timing input sub-vectors;

[0019] S222: Arrange the sequence of the network traffic timing input sub-vectors into a network traffic timing input matrix;

[0020] S223: normalizing the network traffic time series input matrix to obtain the network traffic time series grayscale image; wherein the value of each position in the network traffic time series grayscale image ranges from 0 to 255.

[0021] Optionally, extracting local temporal correlation features of network traffic and performing reverse screening and enhancement on the network traffic temporal grayscale image to obtain a set of enhanced local temporal correlation feature vectors of network traffic, including:

[0022] S31: performing image block processing on the network traffic time series grayscale image to obtain a set of local time series grayscale images of network traffic;

[0023] S32: each of the network traffic local time series grayscale images in the set of network traffic local time series grayscale images is respectively passed through a network traffic time series pattern feature extractor based on a hole convolutional neural network model to obtain a set of network traffic local time series correlation feature vectors;

[0024] S33: Inputting the set of local time series correlation feature vectors of the network traffic into a reverse screening enhanced network based on the feature distribution of the anchor center to obtain the set of local time series correlation feature vectors of the enhanced network traffic.

[0025] Optionally, inputting the set of local time series correlation feature vectors of the network traffic into a reverse screening and strengthening network based on the feature distribution of the anchor center to obtain the set of local time series correlation feature vectors of the strengthened network traffic includes:

[0026] S331: Calculate the semantic correlation coefficient between each network traffic local time series correlation feature vector in the set of network traffic local time series correlation feature vectors and the network traffic correlation anchor center feature vector to obtain a sequence of semantic latent correlation coefficients;

[0027] S332: taking the reciprocal of each semantic latent association coefficient in the sequence of semantic latent association coefficients to obtain a sequence of semantic anti-association coefficients;

[0028] S333: using a Softmax function to normalize the sequence of semantic anti-correlation coefficients to obtain a sequence of semantic anti-correlation energy coefficients;

[0029] S334: using each semantic anti-correlation energy coefficient in the sequence of the semantic anti-correlation energy coefficient as a weight, weighting each network traffic local temporal correlation feature vector in the set of network traffic local temporal correlation feature vectors to obtain a set of reversely inhibited network traffic local temporal correlation feature vectors;

[0030] S335: Calculate the positional difference between the set of local timing correlation feature vectors of the network traffic and the set of local timing correlation feature vectors of the reverse-suppressed network traffic to obtain the set of local timing correlation feature vectors of the enhanced network traffic.

[0031] Optionally, calculating the semantic correlation coefficient between each network traffic local time series correlation feature vector in the set of network traffic local time series correlation feature vectors and the network traffic correlation anchor center feature vector to obtain a sequence of semantic latent correlation coefficients includes:

[0032] S3311: Calculate the position mean vector of the set of local time series correlation feature vectors of the network traffic to determine the network traffic correlation anchor center feature vector;

[0033] S3312: Calculate the positional dot product between the i-th network traffic local time series correlation feature vector and the network traffic correlation anchor center feature vector, and add the vectors obtained by the positional dot product element by element to obtain the network traffic local time series correlation feature heap value constrained by the anchor center;

[0034] S3313: Calculate the element-by-element sum of the squares of the i-th network traffic local timing correlation feature vector and the element-by-element sum of the squares of the network traffic correlation anchor center feature vector, and calculate the square root of the sum of the two values ​​obtained by the element-by-element sum of the squares, and divide the network traffic local timing correlation feature heap value constrained by the anchor center by the value obtained by the square root to obtain the i-th semantic correlation coefficient.

[0035] Optionally, the set of enhanced network traffic local time series correlation feature vectors is input into an energy metric attention fusion network based on sequence labeling to obtain a network traffic time series attention aggregation representation vector, including:

[0036] S41: marking each enhanced network traffic local time series correlation feature vector in the set of enhanced network traffic local time series correlation feature vectors with a serial number to obtain a sequence of marked numbers;

[0037] S42: Calculate the energy function value of each enhanced network traffic local time series correlation feature vector in the set of enhanced network traffic local time series correlation feature vectors, wherein the energy function value is related to the mean and variance of each enhanced network traffic local time series correlation feature vector and the label number of each enhanced network traffic local time series correlation feature vector;

[0038] S43: Using the inverse of the energy function value as a weight factor, weighted modulation and aggregation are performed on each enhanced network traffic local timing correlation feature vector in the set of enhanced network traffic local timing correlation feature vectors to obtain the network traffic timing attention aggregation representation vector.

[0039] Optionally, the energy function value of each enhanced network traffic local timing correlation feature vector in the set of enhanced network traffic local timing correlation feature vectors is calculated, wherein the energy function value is related to the mean and variance of each enhanced network traffic local timing correlation feature vector and the label number of each enhanced network traffic local timing correlation feature vector, including: calculating the square of the difference between the label number of the ith enhanced network traffic local timing correlation feature vector and the mean of the ith enhanced network traffic local timing correlation feature vector to obtain a square result, and adding the variance and regularization term of the ith enhanced network traffic local timing correlation feature vector to obtain an addition result, taking four times the addition result as the dividend, taking the sum of the square result and twice the addition result as the divisor, and calculating the quotient of the dividend divided by the divisor as the energy function value of the ith enhanced network traffic local timing correlation feature vector.

[0040] Optionally, based on the network traffic timing attention aggregation representation vector, a wavelength allocation strategy is determined, and the wavelength allocation strategy is used to indicate increasing the time interval between wavelengths, reducing the time interval between wavelengths, or maintaining the time interval between wavelengths, including: inputting the network traffic timing attention aggregation representation vector into a classifier-based wavelength allocation strategy generator to obtain the wavelength allocation strategy.

[0041] In a second aspect, the present application provides an optical cross-connection device based on a wavelength division matrix, and the optical cross-connection device based on a wavelength division matrix operates according to the optical cross-connection method based on a wavelength division matrix.

[0042] Compared with the prior art, the beneficial effects are:

[0043] 1. Improve the data transmission capacity of the power grid: Smart grids need to collect, transmit and process a large amount of power grid operation data in real time and accurately to ensure the safe, stable and efficient operation of the power grid. With its high speed and large capacity, optical cross-connect technology provides smart grids with powerful data transmission capabilities, which can meet the high requirements of smart grids for data transmission bandwidth and rate.

[0044] 2. Enhance the flexibility and reliability of grid communication: Smart grids need to cope with various complex operating scenarios and emergencies, so they need to have high flexibility and reliability. Optical cross-connect technology can realize flexible cross-connection of optical signals and support flexible access and scheduling of multiple services. At the same time, its all-optical switching characteristics also reduce the links of electrical signal processing, thereby improving the reliability of communication.

[0045] 3. Promote intelligent management of power grids: With the development of technologies such as big data, cloud computing, and artificial intelligence, smart grids are moving towards a more intelligent direction. Optical cross-connect technology can be combined with these advanced technologies to achieve automated collection, analysis, processing, and decision-making of power grid data, improve the level of intelligent management of power grids, and optimize power grid operation efficiency.

[0046] 4. Promote the access and management of new energy: With the large-scale access of renewable energy, smart grids need to manage and dispatch these energy sources more efficiently. Optical cross-connect technology can help build a flexible optical communication network, support the access and dispatch of renewable energy, optimize energy configuration, and improve energy utilization efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A flow chart of an optical cross-connection method based on a wavelength division matrix according to the present invention;

[0048] Figure 2 A block diagram of an optical cross-connect system based on a wavelength division matrix according to the present invention;

[0049] Figure 3 A block diagram of an electronic device of the present invention;

[0050] Figure 4 This is an application scenario diagram of an optical cross-connection method based on a wavelength division matrix of the present invention. DETAILED DESCRIPTION

[0051] The specific implementation of an optical cross-connection method and device based on a wavelength division matrix of the present invention is described in detail below with reference to the accompanying drawings.

[0052] In one embodiment of the present application, based on the basic principle of optical cross-connection of wavelength division matrix, in WDM (wavelength division multiplexing) network, the management and scheduling mechanism of optical signals of different wavelengths in complex network environment is deeply studied, the wavelength allocation algorithm is optimized, the wavelength conflict is reduced, the wavelength utilization rate is improved, and a more efficient optical network resource utilization system is constructed. In-depth research is carried out on the functions of optical cross-connection automatic configuration, fault prediction and recovery, resource optimization and scheduling, etc., and advanced artificial intelligence and machine learning technologies are used to improve the self-healing ability and intelligence level of the network.

[0053] In one embodiment of the present application, an innovative and practical optical cross-connect architecture is developed based on the optical cross-connect architecture technology of the wavelength division matrix to adapt to network application scenarios of different scales and requirements. Appropriate core parameters such as the number and layout of ports, cross-connect capacity planning, and optical backplane technology are formulated to ensure efficient and reliable operation of the optical cross-connect architecture.

[0054] In one embodiment of the present application, an optical cross-connection device based on a wavelength division matrix is ​​developed based on low-power, high-reliability optical switch technology. Combined with the research and development and application of new optical switch materials, the optical switch structure is optimized, the optical switch speed is increased, the insertion loss is reduced, and the quality of optical signal transmission is improved. At the same time, research on the integration and miniaturization of optical cross-connection devices is carried out to adapt to the characteristics of dense equipment and limited space in modern communication networks. Through precise design and rigorous manufacturing, an efficient, stable and reliable optical cross-connection device is developed. High-speed and large-capacity exchange of optical signals is realized, and intelligent management functions such as automatic configuration, fault prediction and recovery are provided, providing strong support for the intelligent and green development of optical networks.

[0055] like Figure 1 As shown, the optical cross-connection method of the present invention comprises the following steps:

[0056] S1: Obtain the time series of network traffic values ​​collected by the network sniffer;

[0057] S2: converting the time series of the network traffic value into image data to obtain a network traffic time series grayscale image;

[0058] S3: extracting local time series correlation features of network traffic and performing reverse screening and enhancement on the network traffic time series grayscale image to obtain a set of enhanced local time series correlation feature vectors of network traffic;

[0059] S4: inputting the set of the enhanced network traffic local time series correlation feature vectors into the energy metric attention fusion network based on sequence labeling to obtain a network traffic time series attention aggregation representation vector;

[0060] S5: Determine a wavelength allocation strategy based on the network traffic timing attention aggregation representation vector, where the wavelength allocation strategy is used to indicate increasing the time interval between wavelengths, reducing the time interval between wavelengths, or maintaining the time interval between wavelengths.

[0061] In response to the above technical problems, the technical concept of this application is: using advanced data analysis technology and deep learning algorithms to cleverly convert network traffic data into the form of images, so that image processing methods can be used to conduct in-depth analysis and pattern recognition of network traffic, so as to obtain the complex time-series nonlinear transformation feature patterns of network traffic, and use classifiers to construct a wavelength allocation strategy generator to intelligently determine how the wavelength allocation strategy should be dynamically adjusted according to the real-time network status, so that the network can quickly respond to traffic changes.

[0062] Based on this, in the technical solution of the present application, first, the time series of network traffic values ​​collected by the network sniffer is obtained. More specifically, a network sniffer is a tool for capturing and analyzing data packets transmitted in a network. It was originally designed for network administrators to monitor network traffic, detect network performance, and find security vulnerabilities. A network sniffer can capture data packets passing through a network interface in real time, helping network administrators understand the actual situation of the network, analyze network traffic, and find the source of network congestion. In addition, a network sniffer is also commonly used by program developers to perform packet capture tests when writing network programs, because the data of network programs is transmitted in the network in the form of data packets, and the definition of the protocol header can be detected by the sniffer whether it is correct. The basis of network sniffing is data capture, and the sniffing system realizes data capture by being connected in parallel in the network. For example, Wireshark is an open source network analysis software widely used in the field of network security, which can capture data packets in the network and provide detailed protocol analysis. And tcpdump is a command-line-based network sniffer, widely used in UNIX, Linux, and MacOS systems, which can capture data packets in the network and output them to a terminal or save them in a file. It should be understood that in actual application scenarios, network traffic changes dynamically. By obtaining the time series of the network traffic values, the periodicity, burstiness and nonlinear characteristics of the traffic can be learned, which are crucial for predicting traffic changes and formulating wavelength allocation strategies.

[0063] In one embodiment of the present application, S2 converts the time series of the network traffic values ​​into image data to obtain a network traffic time series grayscale image, including: S21: arranging the time series of the network traffic values ​​into a network traffic time series input vector according to the time dimension; S22: inputting the network traffic time series input vector into a vector-image converter to obtain the network traffic time series grayscale image.

[0064] Then, after arranging the time series of the network traffic values ​​according to the time dimension into a network traffic time series input vector, the network traffic time series input vector is input into a vector-image converter to obtain a network traffic time series grayscale image. In this way, the network traffic data with discrete time series distribution is converted into a two-dimensional image form, which helps to use image processing methods to identify the implicit laws and complex patterns contained therein.

[0065] In one embodiment of the present application, in S22, the network traffic timing input vector is input into a vector-image converter to obtain the network traffic timing grayscale image, including: S221: vector segmentation of the network traffic timing input vector to obtain a sequence of network traffic timing input sub-vectors; S222: arranging the sequence of network traffic timing input sub-vectors into a network traffic timing input matrix; S223: normalizing the network traffic timing input matrix to obtain the network traffic timing grayscale image; wherein the value range of each position in the network traffic timing grayscale image is 0-255.

[0066] Next, the network traffic time series grayscale image is subjected to image block processing to obtain a set of network traffic local time series grayscale images; and each network traffic local time series grayscale image in the set of network traffic local time series grayscale images is respectively passed through a network traffic time series pattern feature extractor based on a hole convolutional neural network model to obtain a set of network traffic local time series associated feature vectors. Here, considering that local changes in network traffic may contain important information, such as sudden increases or decreases in traffic, these may be early signals of changes in network status. In the technical solution of the present application, the network traffic time series grayscale image is split into a set of network traffic local time series grayscale images, which can guide the model to perform a more fine-grained analysis of local features. In this way, each network traffic local time series grayscale image is respectively passed through a network traffic time series pattern feature extractor based on a hole convolutional neural network model, in the hope of using the hole convolution (also called dilated convolution) to allow the network to capture information within a larger range of the receptive field to capture the abstract feature representation within the neighborhood pixels of each of the network traffic local time series grayscale images, that is, the local time series fluctuation pattern of the network traffic.

[0067] In one embodiment of the present application, S3 performs network traffic local time series correlation feature extraction and reverse screening and enhancement on the network traffic time series grayscale image to obtain a set of enhanced network traffic local time series correlation feature vectors, including: S31: performing image block processing on the network traffic time series grayscale image to obtain a set of network traffic local time series grayscale images; S32: passing each network traffic local time series grayscale image in the set of network traffic local time series grayscale images through a network traffic time series pattern feature extractor based on a void convolutional neural network model to obtain a set of network traffic local time series correlation feature vectors; S33: inputting the set of network traffic local time series correlation feature vectors into a reverse screening and enhancement network based on the feature distribution of the anchor center to obtain the set of enhanced network traffic local time series correlation feature vectors.

[0068] Since the collection process of the time series of the network traffic value may contain noise information and interference information, it may be retained or even amplified in the process of feature extraction, which affects the expression of the network traffic time series feature pattern. Therefore, in the technical solution of the present application, the set of local time series correlation feature vectors of the network traffic is input into the reverse screening and strengthening network based on the feature distribution of the anchor center so that the model can more accurately capture the key feature patterns of the network traffic, thereby obtaining a set of local time series correlation feature vectors of the enhanced network traffic.

[0069] In an embodiment of the present application, the reverse screening reinforcement network based on the feature distribution of the anchor center first determines the network traffic associated anchor center feature vector, for example, the position mean vector of the set of the network traffic local time series associated feature vectors is calculated to determine the network traffic associated anchor center feature vector, so as to regard the network traffic associated anchor center feature vector as a global traffic feature pattern in the set of the network traffic local time series associated feature vectors. Then, the semantic correlation coefficient between each network traffic local time series associated feature vector and the network traffic associated anchor center feature vector is calculated to quantify the semantic similarity and degree of association between the two, thereby obtaining a sequence of semantic latent association coefficients. In this way, the features with low correlation with the anchor center feature distribution are highlighted by taking the inverse, in preparation for subsequent reverse screening. Then, the Softmax function is introduced to normalize the sequence of the semantic anti-correlation coefficients to make the weight distribution smoother, avoid the appearance of extreme values, and improve the stability and rationality of the weight distribution. In the weighted reverse suppression stage, the anchor center-based feature distribution reverse screening and reinforcement network uses the normalized semantic anti-correlation energy coefficient as the weight to weight each network traffic local temporal correlation feature vector, and calculates the positional difference between the set of network traffic local temporal correlation feature vectors and the set of reverse suppression network traffic local temporal correlation feature vectors to achieve reverse screening and reinforcement of important features. That is, first select the noise component or interference component, and then use the difference method to remove these noise components or interference components from the original feature distribution, so as to achieve the optimized expression of the feature distribution.

[0070] In one embodiment of the present application, the set of local time series correlation feature vectors of the network traffic is input into a reverse screening and strengthening network based on the feature distribution of the anchor center to obtain the set of local time series correlation feature vectors of the strengthened network traffic, including:

[0071] S331: Calculate the semantic correlation coefficient between each network traffic local time series correlation feature vector in the set of network traffic local time series correlation feature vectors and the network traffic correlation anchor center feature vector to obtain a sequence of semantic latent correlation coefficients;

[0072] S332: taking the reciprocal of each semantic latent association coefficient in the sequence of semantic latent association coefficients to obtain a sequence of semantic anti-association coefficients;

[0073] S333: using a Softmax function to normalize the sequence of semantic anti-correlation coefficients to obtain a sequence of semantic anti-correlation energy coefficients;

[0074] S334: using each semantic anti-correlation energy coefficient in the sequence of the semantic anti-correlation energy coefficient as a weight, weighting each network traffic local temporal correlation feature vector in the set of network traffic local temporal correlation feature vectors to obtain a set of reversely inhibited network traffic local temporal correlation feature vectors;

[0075] S335: Calculate the positional difference between the set of local timing correlation feature vectors of the network traffic and the set of local timing correlation feature vectors of the reverse-suppressed network traffic to obtain the set of local timing correlation feature vectors of the enhanced network traffic.

[0076] Further, in one embodiment of the present application, calculating the semantic correlation coefficient between each network traffic local time series correlation feature vector in the set of network traffic local time series correlation feature vectors and the network traffic correlation anchor center feature vector in S331 to obtain a sequence of semantic latent correlation coefficients includes:

[0077] S3311: Calculate the position mean vector of the set of local time series correlation feature vectors of the network traffic to determine the network traffic correlation anchor center feature vector;

[0078] S3312: Calculate the positional dot product between the i-th network traffic local time series correlation feature vector and the network traffic correlation anchor center feature vector, and add the vectors obtained by the positional dot product element by element to obtain the network traffic local time series correlation feature heap value constrained by the anchor center;

[0079] S3313: Calculate the element-by-element sum of the squares of the i-th network traffic local timing correlation feature vector and the element-by-element sum of the squares of the network traffic correlation anchor center feature vector, and calculate the square root of the sum of the two values ​​obtained by the element-by-element sum of the squares, and divide the network traffic local timing correlation feature heap value constrained by the anchor center by the value obtained by the square root to obtain the i-th semantic correlation coefficient.

[0080] Specifically, the set of local time series correlation feature vectors of the network traffic is input into the reverse screening and strengthening network based on the feature distribution of the anchor center according to the following strengthening formula to obtain the set of local time series correlation feature vectors of the strengthened network traffic; wherein the strengthening formula is:

[0081] X={x (1) ,x (2) ,...,x (N)}

[0082]

[0083] Y={y (1) ,y (2) ,...,y(N)},y (i) =w (i) ·x (i)

[0084] Z={z (1) ,z (2) ,...,z (N)}, z (i) =x (i) -y (i)

[0085] Among them, X is the set of local time series correlation feature vectors of network traffic, x (1) 、x (2) 、x (i) and x (N) are respectively the first network traffic local timing correlation feature vector, the second network traffic local timing correlation feature vector, the i-th network traffic local timing correlation feature vector and the N-th network traffic local timing correlation feature vector in the set of the network traffic local timing correlation feature vector, the value of N is the number of network traffic local timing correlation feature vectors in the set of the network traffic local timing correlation feature vector, V r is the network traffic associated anchor center feature vector, and are the jth eigenvalues ​​of the ith network traffic local time series correlation feature vector and the network traffic correlation anchor center feature vector, respectively. The value of L is the length of the network traffic local time series correlation feature vector, r (i) , (i) and w (i) are the i-th semantic latent correlation coefficient, semantic anti-correlation coefficient and semantic anti-correlation energy coefficient respectively, Y is the set of local temporal correlation feature vectors of reverse suppression network traffic, y (1) ,y (2) ,y (i) and (N) are respectively the first reverse inhibition network traffic local timing correlation feature vector, the second reverse inhibition network traffic local timing correlation feature vector, the i-th reverse inhibition network traffic local timing correlation feature vector and the N-th reverse inhibition network traffic local timing correlation feature vector in the set of the reverse inhibition network traffic local timing correlation feature vector, Z is the set of the enhanced network traffic local timing correlation feature vector, z (1) 、z (2) 、z (i) and z (N) They are the first enhanced network traffic local timing correlation feature vector, the second enhanced network traffic local timing correlation feature vector, the i-th enhanced network traffic local timing correlation feature vector and the N-th enhanced network traffic local timing correlation feature vector, e(·) represents an exponential function with a natural constant as the base, and ⊙ represents dot product.

[0086] It should be understood that each enhanced local temporal correlation feature vector of network traffic fully expresses the local temporal fluctuation feature pattern of network traffic with image blocks as spatial constraints. However, the temporal fluctuation feature pattern of network traffic within each local span does not exist independently. In other words, the lack of communication between the local temporal fluctuation feature patterns expressed by each enhanced local temporal correlation feature vector of network traffic will affect the ability to express the overall temporal fluctuation feature distribution of network traffic. If each enhanced local temporal correlation feature vector of network traffic can be fused, local information from different time windows can be integrated to form a comprehensive temporal feature representation, which helps to better understand the dynamic changes of network traffic in the actual application of the network. However, traditional fusion methods, such as cascade merging, may not be able to fully express the dynamic change behavior and complex temporal fluctuation laws of network traffic; in addition, each enhanced local temporal correlation feature vector of network traffic may have different importance for the description and characterization of the overall traffic change pattern, so that simple fusion methods cannot deeply mine and characterize these complex and interrelated hidden feature patterns.

[0087] Therefore, in the technical solution of the present application, the set of the local time series correlation feature vectors of the enhanced network traffic is further input into the energy metric attention fusion network based on sequence annotation to obtain the network traffic time series attention aggregation representation vector. In particular, the energy metric attention fusion network based on sequence annotation introduces an attention mechanism based on energy metric to realize the aggregation of the set of local time series correlation feature vectors of the enhanced network traffic. More specifically, the energy metric attention fusion network based on sequence annotation assigns a weight factor, i.e., the inverse of the energy function value, to each local time series correlation feature vector of the enhanced network traffic, and the energy function value is related to the mean and variance of each local time series correlation feature vector of the enhanced network traffic and the label number of each local time series correlation feature vector of the enhanced network traffic. Among them, the mean and variance of each local time series correlation feature vector of the enhanced network traffic are their own statistical characteristics, which can reflect the stability and variation range of the traffic pattern represented by each local time series correlation feature vector of the enhanced network traffic. On the other hand, the label numbers represent the position information of each enhanced network traffic local temporal correlation feature vector in the sequence, which helps to capture its influence and semantic correlation dependency in the time series, so as to help the model identify and track traffic patterns that change over time.

[0088] In one embodiment of the present application, in S4, the set of enhanced network traffic local time series correlation feature vectors is input into the energy metric attention fusion network based on sequence labeling to obtain the network traffic time series attention aggregation representation vector, including:

[0089] S41: marking each enhanced network traffic local time series correlation feature vector in the set of enhanced network traffic local time series correlation feature vectors with a serial number to obtain a sequence of marking numbers;

[0090] S42: Calculate the energy function value of each enhanced network traffic local time series correlation feature vector in the set of enhanced network traffic local time series correlation feature vectors, wherein the energy function value is related to the mean and variance of each enhanced network traffic local time series correlation feature vector and the label number of each enhanced network traffic local time series correlation feature vector;

[0091] S43: Using the inverse of the energy function value as a weight factor, weighted modulation and aggregation are performed on each enhanced network traffic local timing correlation feature vector in the set of enhanced network traffic local timing correlation feature vectors to obtain the network traffic timing attention aggregation representation vector.

[0092] Further, in one embodiment of the present application, the energy function value of each enhanced network traffic local timing correlation feature vector in the set of the enhanced network traffic local timing correlation feature vectors is calculated in S42, wherein the energy function value is related to the mean and variance of each enhanced network traffic local timing correlation feature vector and the label number of each enhanced network traffic local timing correlation feature vector, including: calculating the square of the difference between the label number of the ith enhanced network traffic local timing correlation feature vector and the mean of the ith enhanced network traffic local timing correlation feature vector to obtain a square result, and adding the variance and regularization term of the ith enhanced network traffic local timing correlation feature vector to obtain an addition result, taking four times the addition result as the dividend, taking the sum of the square result and twice the addition result as the divisor, and calculating the quotient of the dividend divided by the divisor as the energy function value of the ith enhanced network traffic local timing correlation feature vector.

[0093] Specifically, the set of enhanced network traffic local time series correlation feature vectors is input into the energy metric attention fusion network based on sequence labeling to obtain the network traffic time series attention aggregation representation vector according to the following fusion formula; wherein the fusion formula is:

[0094]

[0095] Among them, μ and σ 2 are the mean and variance of the i-th enhanced network traffic local time series correlation feature vector in the set of the enhanced network traffic local time series correlation feature vectors, t is the label number of the i-th enhanced network traffic local time series correlation feature vector, λ is the regularization term, e iis the value of the ith energy function, x i is the i-th eigenvalue in the i-th enhanced network traffic local time series correlation feature vector, the value of M is the length of the i-th enhanced network traffic local time series correlation feature vector, sigmoid represents the sigmoid function, X i is the i-th enhanced network traffic local time series correlation feature vector, the value of S is the number of enhanced network traffic local time series correlation feature vectors in the set of enhanced network traffic local time series correlation feature vectors, is the network traffic temporal attention aggregation representation vector, and ⊙ represents the dot product.

[0096] Then, the network traffic time series attention aggregation representation vector is input into the classifier-based wavelength allocation strategy generator to obtain a wavelength allocation strategy, wherein the wavelength allocation strategy is used to indicate increasing the time interval between wavelengths, reducing the time interval between wavelengths, or maintaining the time interval between wavelengths. In this way, the wavelength allocation is dynamically adjusted according to the real-time network traffic data, so that the available wavelength resources can be more effectively utilized to adapt to the traffic demand in different time periods.

[0097] In one embodiment of the present application, in S5, a wavelength allocation strategy is determined based on the network traffic timing attention aggregation representation vector, and the wavelength allocation strategy is used to indicate increasing the time interval between wavelengths, reducing the time interval between wavelengths, or maintaining the time interval between wavelengths, including: inputting the network traffic timing attention aggregation representation vector into a classifier-based wavelength allocation strategy generator to obtain the wavelength allocation strategy.

[0098] Here, each network traffic local time series correlation feature vector in the set of network traffic local time series correlation feature vectors respectively represents the local time series correlation characteristics of the network traffic value within the subdivided time domain and between subdivided time domains in the local time domain. Therefore, after inputting the feature distribution reverse screening reinforcement network based on the anchor center to perform feature distribution reinforcement within the local time domain, and inputting the energy measurement attention fusion network based on sequence annotation to perform energy measurement attention reinforcement between local time domains, the network traffic time series attention aggregation representation vector will also cause the loss of prior-posteriori class probability causal association relative to the source time series distribution representation due to the difference in feature weights in different time domain spaces, thereby affecting the accuracy of the classification results.

[0099] Based on this, in a preferred embodiment, when the applicant of the present application inputs the network traffic time series attention aggregation representation vector into a classifier-based wavelength allocation strategy generator to obtain a wavelength allocation strategy, the network traffic time series attention aggregation representation vector is optimized, specifically including the following steps:

[0100] Obtaining the network traffic time series attention aggregation representation vector input based on the classifier wavelength allocation strategy generator to obtain an increase probability value, a reduction probability value, and a maintenance probability value of increasing the time interval between wavelengths, reducing the time interval between wavelengths, or maintaining the relative proportion of the time interval between wavelengths;

[0101] After deducting one third of the increase probability value, the reduction probability value, and the retention probability value respectively, calculate the square root of the sum of their squares, and add it to one third to obtain the strategy probability value;

[0102] Probabilize each eigenvalue of the network traffic time series attention aggregation representation vector based on a probability activation function, such as a sigmoid function, to obtain a probabilistic network traffic time series attention aggregation representation vector;

[0103] Determine a network traffic timing attention aggregation representation cognitive symbol value based on a comparison between each eigenvalue of the probabilistic network traffic timing attention aggregation representation vector and the strategy probability value, wherein the network traffic timing attention aggregation representation cognitive symbol value is equal to one, zero, and negative one in response to the eigenvalue of the probabilistic network traffic timing attention aggregation representation vector being greater than, equal to, and less than the strategy probability value, respectively;

[0104] Calculate the mean of all eigenvalues ​​of the probabilistic network traffic time series attention aggregation representation vector to obtain the overall phase shift value of the network traffic time series attention aggregation representation;

[0105] Multiply each eigenvalue of the probabilistic network traffic time series attention aggregation representation vector by the network traffic time series attention aggregation representation cognitive symbol value and the network traffic time series attention aggregation representation overall phase shift value to obtain a probabilistic network traffic time series attention aggregation representation cognitive eigenvector and a probabilistic network traffic time series attention aggregation representation phase shift eigenvector;

[0106] Multiplying the probabilistic network traffic time series attention aggregation representation cognitive feature vector and the probabilistic network traffic time series attention aggregation representation phase shift feature vector by a first weight and a second weight as hyperparameters respectively, and then calculating the difference thereof to obtain a probabilistic network traffic time series attention aggregation representation differential feature vector;

[0107] Take the absolute value of each eigenvalue of the probabilistic network traffic timing attention aggregation representation differential eigenvector to obtain the optimized network traffic timing attention aggregation representation eigenvector.

[0108] It is expressed as:

[0109]

[0110] Among them, fi and f′ i are respectively the i-th eigenvalue in the probabilistic network traffic time series attention aggregation representation vector and the optimized network traffic time series attention aggregation representation feature vector, p is the strategy probability value, p 1 、p 2 and p 3 They are increasing the probability value, reducing the probability value and maintaining the probability value, sgn(v i , p) is the cognitive symbol value of the network traffic time series attention aggregation representation, is the network traffic time series attention aggregation representing the overall phase shift value, α and β are the first weight and the second weight respectively, and |·| represents the absolute value.

[0111] That is, in the above-mentioned preferred embodiment, the policy cognitive phase change response of the network traffic time series attention aggregation representation vector is obtained by comparing the probabilistic amplitude of the eigenvalue of the network traffic time series attention aggregation representation vector with the policy probability representation, and the phase shift response of the eigenvalue of the network traffic time series attention aggregation representation vector relative to the policy probability representation of the feature as a whole is subjected to a feature distribution sequence invariance transformation based on the policy probability differential distribution expansion, so as to realize the causal constraint of the posterior class probability of the network traffic time series attention aggregation representation vector on its prior feature distribution representation, so as to improve the accuracy of the wavelength allocation strategy obtained by inputting the network traffic time series attention aggregation representation vector into the classifier-based wavelength allocation strategy generator.

[0112] In summary, the above scheme is adopted to cleverly convert network traffic data into the form of images by utilizing advanced data analysis technology and deep learning algorithms, so that image processing methods can be used to conduct in-depth analysis and pattern recognition of network traffic to obtain the complex time series nonlinear transformation feature patterns of network traffic, and use classifiers to construct a wavelength allocation strategy generator to intelligently determine how the wavelength allocation strategy should be dynamically adjusted according to the real-time network status, so that the network can quickly respond to traffic changes.

[0113] In one embodiment of the present application, an optical cross-connect device based on a wavelength division matrix is ​​provided, and the optical cross-connect device based on a wavelength division matrix operates according to the method described above.

[0114] Figure 2 FIG. 1 is a block diagram of an optical cross-connect system based on a wavelength division matrix according to an exemplary embodiment. Figure 2 As shown, the optical cross-connect system 200 includes:

[0115] The network traffic value acquisition module 201 is used to acquire the time series of the network traffic value collected by the network sniffer;

[0116] An image data conversion module 202 is used to convert the time series of the network traffic value into image data to obtain a network traffic time series grayscale image;

[0117] A time series correlation feature extraction module 203 is used to extract local time series correlation features of network traffic and perform reverse screening and enhancement on the network traffic time series grayscale image to obtain a set of enhanced local time series correlation feature vectors of network traffic;

[0118] An attention fusion module 204 is used to input the set of enhanced network traffic local time series correlation feature vectors into the energy metric attention fusion network based on sequence annotation to obtain a network traffic time series attention aggregation representation vector;

[0119] The wavelength allocation strategy determination module 205 is used to determine a wavelength allocation strategy based on the network traffic time series attention aggregation representation vector, where the wavelength allocation strategy is used to indicate increasing the time interval between wavelengths, reducing the time interval between wavelengths, or maintaining the time interval between wavelengths.

[0120] Reference below Figure 3 , which shows a schematic diagram of the structure of an electronic device 600 suitable for implementing the embodiment of the present application. The terminal device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 3 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0121] like Figure 3 As shown, the electronic device 600 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the electronic device 600 are also stored. The processing device 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0122] Typically, the following devices may be connected to the I / O interface 605: an input device 606 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 608 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 609. The communication device 609 may allow the electronic device 600 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 3 The electronic device 600 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0123] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 609, or installed from the storage device 608, or installed from the ROM 602. When the computer program is executed by the processing device 601, the above-mentioned functions defined in the method of the embodiment of the present application are executed.

[0124] It should be noted that the computer-readable medium mentioned above in the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0125] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperTextTransferProtocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an adhoc peer-to-peer network), as well as any currently known or future developed network.

[0126] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0127] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including, but not limited to, object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0128] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0129] The modules involved in the embodiments described in this application can be implemented by software or hardware. The name of the module does not limit the module itself in some cases. For example, the test parameter acquisition module can also be described as a "module for acquiring device test parameters corresponding to the target device".

[0130] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0131] In the context of the present application, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0132] Figure 4 FIG. 1 is an application scenario diagram of an optical cross-connection method based on a wavelength division matrix according to an exemplary embodiment. Figure 4 As shown, in this application scenario, first, a time series of network traffic values ​​collected by a network sniffer is obtained (for example, Figure 4 Then, the obtained network traffic value is input to a server (for example, Figure 4 In the S) shown in FIG. 1 , the server is capable of processing the network traffic value based on an optical cross-connect algorithm of a wavelength division matrix to determine a wavelength allocation strategy, wherein the wavelength allocation strategy is used to indicate increasing the time interval between wavelengths, reducing the time interval between wavelengths, or maintaining the time interval between wavelengths.

[0133] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.

[0134] In addition, although each operation is described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or to be performed in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be construed as limiting the scope of the application. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

[0135] Although the subject matter has been described in language specific to structural features and / or method logic actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely example forms of implementing the claims. Regarding the device in the above embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.

[0136] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit the present invention. A person skilled in the art should understand that the specific implementation of the present invention can be modified or replaced by equivalents, but these modifications or changes are within the scope of protection of the pending claims.

Claims

1. An optical cross-connection method based on a wavelength division matrix, characterized in that: The optical cross-connection method comprises: S1: Obtain the time series of network traffic values ​​collected by the network sniffer; S2: converting the time series of the network traffic value into image data to obtain a network traffic time series grayscale image; S3: extracting local time series correlation features of network traffic and performing reverse screening and enhancement on the network traffic time series grayscale image to obtain a set of enhanced local time series correlation feature vectors of network traffic; S4: inputting the set of the enhanced network traffic local time series correlation feature vectors into the energy metric attention fusion network based on sequence labeling to obtain a network traffic time series attention aggregation representation vector; S5: Determine a wavelength allocation strategy based on the network traffic timing attention aggregation representation vector, where the wavelength allocation strategy is used to indicate increasing the time interval between wavelengths, reducing the time interval between wavelengths, or maintaining the time interval between wavelengths.

2. The optical cross-connection method based on wavelength division matrix according to claim 1, characterized in that: S2 converts the time series of the network traffic value into image data to obtain a network traffic time series grayscale image, including: S21: Arranging the time series of the network traffic values ​​according to the time dimension into a network traffic time series input vector; S22: Input the network traffic time series input vector into a vector-to-image converter to obtain the network traffic time series grayscale image.

3. The optical cross-connection method based on wavelength division matrix according to claim 2, characterized in that: In S22, the network traffic time series input vector is input into a vector-image converter to obtain the network traffic time series grayscale image, including: S221: performing vector segmentation on the network traffic timing input vector to obtain a sequence of network traffic timing input sub-vectors; S222: Arrange the sequence of the network traffic timing input sub-vectors into a network traffic timing input matrix; S223: performing normalization processing on the network traffic time series input matrix to obtain the network traffic time series grayscale image; The value of each position in the network traffic time series grayscale image ranges from 0 to 255.

4. The optical cross-connection method based on wavelength division matrix according to claim 3, characterized in that: In S3, the network traffic local time series correlation feature extraction and reverse screening enhancement are performed on the network traffic time series grayscale image to obtain a set of enhanced network traffic local time series correlation feature vectors, including: S31: performing image block processing on the network traffic time series grayscale image to obtain a set of local time series grayscale images of network traffic; S32: each of the network traffic local time series grayscale images in the set of network traffic local time series grayscale images is respectively passed through a network traffic time series pattern feature extractor based on a hole convolutional neural network model to obtain a set of network traffic local time series correlation feature vectors; S33: Inputting the set of local time series correlation feature vectors of the network traffic into a reverse screening enhanced network based on the feature distribution of the anchor center to obtain the set of local time series correlation feature vectors of the enhanced network traffic.

5. The optical cross-connection method based on wavelength division matrix according to claim 4, characterized in that: In S33, the set of local time series correlation feature vectors of network traffic is input into a reverse screening and strengthening network based on the feature distribution of the anchor center to obtain the set of local time series correlation feature vectors of the strengthened network traffic, including: S331: Calculate the semantic correlation coefficient between each network traffic local time series correlation feature vector in the set of network traffic local time series correlation feature vectors and the network traffic correlation anchor center feature vector to obtain a sequence of semantic latent correlation coefficients; S332: taking the reciprocal of each semantic latent association coefficient in the sequence of semantic latent association coefficients to obtain a sequence of semantic anti-association coefficients; S333: using a Softmax function to normalize the sequence of semantic anti-correlation coefficients to obtain a sequence of semantic anti-correlation energy coefficients; S334: using each semantic anti-correlation energy coefficient in the sequence of the semantic anti-correlation energy coefficient as a weight, weighting each network traffic local temporal correlation feature vector in the set of network traffic local temporal correlation feature vectors to obtain a set of reversely inhibited network traffic local temporal correlation feature vectors; S335: Calculate the positional difference between the set of local timing correlation feature vectors of the network traffic and the set of local timing correlation feature vectors of the reverse-suppressed network traffic to obtain the set of local timing correlation feature vectors of the enhanced network traffic.

6. The optical cross-connection method based on wavelength division matrix according to claim 5, characterized in that: S331 calculates the semantic correlation coefficient between each network traffic local time series correlation feature vector in the set of network traffic local time series correlation feature vectors and the network traffic correlation anchor center feature vector to obtain a sequence of semantic latent correlation coefficients, including: S3311: Calculate the position mean vector of the set of local time series correlation feature vectors of the network traffic to determine the network traffic correlation anchor center feature vector; S3312: Calculate the positional dot product between the i-th network traffic local time series correlation feature vector and the network traffic correlation anchor center feature vector, and add the vectors obtained by the positional dot product element by element to obtain the network traffic local time series correlation feature heap value constrained by the anchor center; S3313: Calculate the element-by-element sum of the squares of the i-th network traffic local timing correlation feature vector and the element-by-element sum of the squares of the network traffic correlation anchor center feature vector, and calculate the square root of the sum of the two values ​​obtained by the element-by-element sum of the squares, and divide the network traffic local timing correlation feature heap value constrained by the anchor center by the value obtained by the square root to obtain the i-th semantic correlation coefficient.

7. The optical cross-connection method based on wavelength division matrix according to claim 6, characterized in that: In S4, the set of the enhanced network traffic local time series correlation feature vectors is input into the energy metric attention fusion network based on sequence annotation to obtain the network traffic time series attention aggregation representation vector, including: S41: marking each enhanced network traffic local time series correlation feature vector in the set of enhanced network traffic local time series correlation feature vectors with a serial number to obtain a sequence of marked numbers; S42: Calculate the energy function value of each enhanced network traffic local time series correlation feature vector in the set of enhanced network traffic local time series correlation feature vectors, wherein the energy function value is related to the mean and variance of each enhanced network traffic local time series correlation feature vector and the label number of each enhanced network traffic local time series correlation feature vector; S43: Using the inverse of the energy function value as a weight factor, weighted modulation and aggregation are performed on each enhanced network traffic local timing correlation feature vector in the set of enhanced network traffic local timing correlation feature vectors to obtain the network traffic timing attention aggregation representation vector.

8. The optical cross-connection method based on wavelength division matrix according to claim 7, characterized in that: In S42, the energy function value of each enhanced network traffic local time series correlation feature vector in the set of the enhanced network traffic local time series correlation feature vector is calculated, wherein the energy function value is related to the mean and variance of each enhanced network traffic local time series correlation feature vector and the label number of each enhanced network traffic local time series correlation feature vector, including: Calculate the square of the difference between the label number of the i-th enhanced network traffic local timing correlation feature vector and the mean of the i-th enhanced network traffic local timing correlation feature vector to obtain a square result, add the variance and the regularization term of the i-th enhanced network traffic local timing correlation feature vector to obtain an addition result, take four times the addition result as the dividend, take the sum of the square result and twice the addition result as the divisor, and calculate the quotient of the dividend divided by the divisor as the energy function value of the i-th enhanced network traffic local timing correlation feature vector.

9. The optical cross-connection method based on wavelength division matrix according to claim 8, characterized in that: In S5, based on the network traffic time series attention aggregation representation vector, a wavelength allocation strategy is determined, where the wavelength allocation strategy is used to indicate increasing the time interval between wavelengths, reducing the time interval between wavelengths, or maintaining the time interval between wavelengths, including: The network traffic time series attention aggregation representation vector is input into a classifier-based wavelength allocation strategy generator to obtain the wavelength allocation strategy.

10. An optical cross-connection device based on a wavelength division matrix, characterized in that: The wavelength division matrix-based optical cross-connection device operates according to the optical cross-connection method as claimed in claims 1 to 9.

Citation Information

Patent Citations

  • Respiration sleep monitor based on millimeter wave radar

    CN118592924A

  • Method and system for controlling microwave and millimeter wave communication signals

    CN118971849A

  • Intelligent management system for chemical production based on data analysis

    CN119179982A

  • Silicon imine production method adopting continuous method

    CN119390023A

Cited By

  • Generation method of network flow test model and network flow anomaly detection method

    CN119892521A