High-integration DWDM transmission system for 5G bearer network

Abstract

The invention relates to the technical field of optical communication transmission, in particular to a high-integration DWDM transmission system for a 5G bearer network, and the system comprises a signal collection module which captures a wavelength routing structure signal sample in real time into original data; the multi-dimensional analysis module performs cross analysis on spectrum and time dimensions, establishes a wavelength channel signal correlation model and precipitates characteristic parameters; the dynamic evaluation module imports characteristic parameters to calculate a multi-dimensional quantitative result, including a channel stability and quality degradation hot spot distribution map; the regional calibration module calibrates the evaluation result according to the geographic coordinates of the hot spots to obtain calibration stability; the instruction generation module automatically generates an operation control instruction sequence according to the calibration value. According to the system, through two-dimensional cross analysis and regional calibration, the transmission channel quality evaluation accuracy and control effectiveness are improved, and the high-reliability transmission requirement of the 5G bearer network is met.

Description

Technical Field

[0001] This invention relates to the field of optical communication transmission technology, and in particular to a highly integrated DWDM transmission system for 5G bearer networks. Background Technology

[0002] Existing highly integrated DWDM transmission systems for 5G bearer networks often employ single analytical methods based on either the spectral or temporal dimensions for signal monitoring and quality assessment. Spectral analysis focuses on static characteristics such as optical power, wavelength shift, and OSNR, while the temporal dimension addresses dynamic changes like jitter and bit error rate; these two are processed independently without establishing a correlation. The calibration process relies on globally uniform parameters to adjust the assessment results, failing to consider the impact of environmental differences across different geographical regions on transmission quality.

[0003] The shortcomings of existing technologies are that single-dimensional analysis cannot capture dynamic coupling effects such as nonlinear crosstalk and power fluctuation transmission between wavelength channels, resulting in one-sided feature parameter extraction and distorted characterization of system health status; global calibration causes systematic bias in the assessment results of degradation hotspot areas that are significantly affected by the geographical environment, making it difficult to truly reflect the quality of local channels.

[0004] The core problems to be solved by this invention include: how to establish a signal correlation model between wavelength channels through cross-analysis of spectral and temporal dimensions, and extract characteristic parameters that can comprehensively characterize the health status of the system; how to perform regional deviation calibration on the quantitative evaluation results based on the geographical coordinates indicated by the hotspot distribution map of transmission channel quality degradation obtained by dynamic evaluation, eliminate evaluation errors caused by geographical environmental interference, and improve the accuracy of the quantitative values ​​of channel stability. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies and propose a highly integrated DWDM transmission system for 5G bearer networks.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a highly integrated DWDM transmission system for 5G bearer networks, comprising: The signal acquisition module captures transmitted signal samples in real time from the wavelength routing structure of the highly integrated DWDM transmission system to form a raw observation data set; The multidimensional analysis module performs cross-analysis of the original observation data set in spectral and temporal dimensions, establishes a signal correlation model between wavelength channels, and extracts several characteristic parameters characterizing the health status of the system based on the signal correlation model. The dynamic evaluation module imports the extracted feature parameters into the dynamic evaluation network for calculation, and obtains a multi-dimensional quantitative evaluation result of the transmission channel quality. The multi-dimensional quantitative evaluation result includes a channel stability quantification value and a quality degradation hotspot distribution map. The regional calibration module, based on the geographical coordinates indicated by the channel quality degradation hotspot distribution map, performs regional deviation calibration on the multi-dimensional quantitative evaluation results of the transmission channel quality to obtain the calibrated quantitative value of channel stability. The instruction generation module automatically generates and arranges a series of operation control instruction sequences for the highly integrated DWDM transmission system based on the calibrated channel stability quantification value.

[0007] As a further aspect of the present invention, the step of performing cross-analysis of the original observation data set in both spectral and temporal dimensions to establish a signal correlation model between wavelength channels, and extracting several characteristic parameters characterizing the health status of the system based on the signal correlation model, specifically includes: In the spectral dimension, each wavelength channel in the original observation dataset is traversed to extract its optical power trajectory data and spectral shift trajectory data. The optical power trajectory data includes the power mean and power variance, and the spectral shift trajectory data includes the center wavelength shift and spectral width change. In the time dimension, the original observation data set is sliced ​​according to a preset sampling period to generate a continuous time slice sequence, and the optical power correlation coefficient and spectral shift correlation coefficient of the corresponding channels between adjacent time slices are calculated. Based on the optical power correlation coefficient, a network topology for the mutual influence of optical power between channels is constructed, and a topology for the propagation of spectral perturbations between channels is constructed based on the spectral shift correlation coefficient. By integrating the network topology of the mutual influence of optical power between channels and the propagation topology of spectral perturbation between channels, a signal correlation model between the wavelength channels is formed. The node degree distribution features, edge weight distribution features, and clustering coefficient of the network topology are extracted from the signal correlation model. The node degree distribution features, edge weight distribution features, and clustering coefficient are used together as feature parameters to characterize the health status of the system.

[0008] As a further aspect of the present invention, the node degree distribution characteristics, edge weight distribution characteristics, and clustering coefficients of the network topology are extracted from the signal correlation model, including: Identify the nodes corresponding to all wavelength channels in the signal correlation model, calculate the connectivity of each node, i.e. the number of other nodes that are correlated with the node in terms of optical power or spectral shift, and statistically analyze the distribution histogram of the connectivity of all nodes to obtain the node connectivity distribution characteristics. The weight value of each connection edge in the measurement signal correlation model is the weighted sum of the optical power correlation coefficient and the spectral shift correlation coefficient between the two nodes of the connection edge. The distribution range and central tendency of all connection edge weight values ​​are statistically analyzed to obtain the edge weight distribution characteristics. For each node in the signal correlation model, find all nodes directly connected to it, and calculate the ratio of the actual number of connecting edges between directly connected nodes to the theoretical maximum number of connecting edges, which is used as the local clustering coefficient of the node. Calculate the arithmetic mean of the local clustering coefficients of all nodes, and use it as the clustering coefficient of the topology.

[0009] As a further aspect of the present invention, the step of importing the extracted feature parameters into a dynamic evaluation network for calculation to obtain a multi-dimensional quantitative evaluation result of the transmission channel quality includes the following operations: The dynamic evaluation network consists of a feature fusion layer, a spatial mapping layer, and a result generation layer; The feature fusion layer receives the node degree distribution feature, edge weight distribution feature, and clustering coefficient, converts the node degree distribution feature into a node importance vector, the edge weight distribution feature into a connection strength matrix, and the clustering coefficient into a topology compactness scalar. The node importance vector, connection strength matrix, and topology density scalar are concatenated and dimension-reduced to generate a fused feature vector; The spatial mapping layer maps the fused feature vector to a preset evaluation space, in which each coordinate position corresponds to a quality state of a transmission channel. The result generation layer calculates the distance from the fused feature vector to each preset quality state center point based on the mapping coordinates of the fused feature vector in the evaluation space, and outputs the quality level corresponding to the center point with the smallest distance as a channel stability metric value. Simultaneously, the result generation layer identifies regions with abnormally concentrated quality states based on the density of neighboring regions in the evaluation space using the mapped coordinates, and then maps the location of these regions with abnormally concentrated quality states back to the actual wavelength channels and geographic coordinates to form the quality degradation hotspot distribution map.

[0010] As a further aspect of the present invention, the regional deviation calibration of the multi-dimensional quantitative evaluation results of the transmission channel quality based on the geographical coordinates indicated by the channel quality degradation hotspot distribution map includes: Extract the central geographic coordinates and influence radius of one or more hotspot areas from the hotspot distribution map of quality degradation; Based on the central geographic coordinates, locate the corresponding fiber segment, optical amplifier node, and wavelength selection switch node in the fiber optic link topology database of the highly integrated DWDM transmission system, and query the historical performance attenuation data and current environmental parameters of the network components. Input the historical performance degradation data and environmental parameters obtained from the query into a preset deviation compensation calculation function to calculate the localized compensation factor for each hotspot area. Using the calculated localized compensation factor, the channel stability quantization values ​​of all transmission channels falling within the influence radius of the corresponding hotspot area are weighted and corrected. The correction formula is to multiply the original quantization value by the corresponding localized compensation factor. All corrected channel stability metrics are globally normalized to ensure that their numerical range remains consistent with that before calibration, and finally the calibrated channel stability metrics are output.

[0011] As a further aspect of the present invention, the step of inputting the historical performance degradation data obtained from the query and environmental parameters into a preset deviation compensation calculation function to calculate the localized compensation factor for each hotspot area is specifically achieved through the following steps: The deviation compensation calculation function internally maintains a multi-factor influence lookup table, which uses the combination of historical performance decay rate and ambient temperature and humidity as input index and outputs a basic influence coefficient. The historical performance degradation data obtained from the query is converted into historical performance degradation rate, which is then combined with environmental temperature and humidity parameters and retrieved in the multi-factor influence lookup table to obtain the corresponding basic influence coefficient. Obtain the real-time load rate of fiber optic links within the influence radius of the current hotspot area, and map the real-time load rate into a load influence coefficient through a non-linear transformation curve; Multiply the basic influence coefficient by the load influence coefficient to obtain a preliminary compensation factor; The initial compensation factor is multiplied by a radius scaling factor that is dynamically adjusted according to the size of the influence radius of the hotspot area to obtain the final localized compensation factor, wherein the larger the influence radius, the smaller the radius scaling factor.

[0012] As a further aspect of the present invention, the step of automatically generating and arranging a series of operation control command sequences for the highly integrated DWDM transmission system based on the calibrated channel stability quantification value includes the following steps: Set multiple channel stability threshold ranges, and associate each threshold range with a set of preset operation control instruction templates; The calibrated channel stability quantification value is compared with the multiple channel stability threshold intervals to determine the threshold interval to which each transmission channel belongs. For each transmission channel, an operation control instruction template associated with its respective threshold range is loaded. The operation control instruction template includes one or more of the following: optical power adjustment instruction, wavelength fine-tuning instruction, and error correction coding redundancy adjustment instruction. Check if there are any conflicts in the operation control commands loaded on adjacent transmission channels. If there are any conflicts, adjust the command parameters of the low-priority channel according to the priority of the channel stability quantification value to eliminate the conflict. According to the physical order and signal transmission direction of the optical fiber link, the final operation control instructions of all transmission channels are time-sequenced to generate an ordered sequence of operation control instructions. The sequence of operation control instructions ensures that at any network node, the number of instructions executed at the same time does not exceed the processing capacity limit of the network node.

[0013] As a further aspect of the present invention, the step of checking whether there is a conflict between the operation control commands loaded on adjacent transmission channels, and if a conflict exists, adaptively adjusting the command parameters of the low-priority channel according to the priority of the channel stability quantification values ​​to eliminate the conflict, specifically includes: Conflict types are defined, including: power contention conflict, which is when the target value of the optical power adjustment command of adjacent channels exceeds the saturation power of the shared amplifier; wavelength overlap conflict, which is when the target range of wavelength fine-tuning commands of adjacent channels overlaps; and resource overload conflict, which is when the computational resources required for the combination of commands of adjacent channels exceed the node quota. For detected power contention conflicts, compare the channel stability quantization values ​​of the conflicting channels. For channels with low quantization values, reduce the optical power adjustment target value in their instruction template proportionally until the total power demand of the shared amplifier is lower than the saturation power. For detected wavelength overlap conflicts, compare the channel stability quantization values ​​of the conflicting channels. For channels with low quantization values, restrict the adjustment direction of their wavelength fine-tuning commands to keep them away from the target wavelength range of channels with high quantization values. For detected resource overload conflicts, the channel stability quantification values ​​of the conflicting channels are compared. For channels with low quantification values, the time slot allocation required for instruction execution is extended proportionally, thereby reducing instantaneous resource demand.

[0014] As a further aspect of the present invention, it also includes: A periodic self-optimization module performs a periodic self-optimization process on the dynamic evaluation network, the process including: At the end of an optimization cycle, the actual performance index data generated by the highly integrated DWDM transmission system during the optimization cycle are collected as a set of verification truth data. The verification true value dataset is compared with the multi-dimensional quantitative evaluation results of the transmission channel quality output by the dynamic evaluation network during the same period, and the evaluation error of each channel is calculated. Based on the evaluation errors of all channels, an error distribution map is constructed to analyze the clustering characteristics of errors in the spectral and geographic dimensions. Based on the error distribution map, the feature weights of the feature fusion layer and the mapping parameters of the spatial mapping layer in the dynamic evaluation network are iteratively adjusted in order to reduce the overall error entropy of the error distribution map. The dynamically evaluated network with adjusted parameters is used for the evaluation calculation in the next optimization cycle, and a new periodic self-optimization process is initiated.

[0015] As a further aspect of the present invention, the step of constructing an error distribution map based on the evaluation errors of all channels and analyzing the clustering characteristics of the errors in the spectral and geographic dimensions is achieved through the following method: A two-dimensional error distribution grid is established with wavelength channels as the horizontal axis and geographic node sequences as the vertical axis. The evaluation error value of each channel is filled into its corresponding coordinate position in the two-dimensional error distribution grid; Spatial clustering analysis is performed on the filled two-dimensional error distribution grid to identify continuous regions where the evaluation error value is higher than the threshold, and these continuous regions are marked as high error clustering areas. The wavelength coverage of high error clusters in the spectral dimension and the fiber segment coverage in the geographical dimension were statistically analyzed separately. Calculate the mean and variance of error values ​​within each high error cluster, as well as the area proportion of the high error cluster in the two-dimensional grid, and use the statistics as a quantitative description of the error clustering characteristics.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: The multidimensional analysis module simultaneously performs cross-analysis of the original observation data set along both spectral and temporal dimensions. The spectral dimension encompasses spectral characteristics such as optical power, wavelength shift, and OSNR, while the temporal dimension covers the trends in signal jitter and bit error rate over time. By combining time-frequency analysis and cross-channel correlation matrices, a wavelength channel signal correlation model is established to quantify coupling relationships such as nonlinear crosstalk and power fluctuation propagation between channels, extracting characteristic parameters such as crosstalk coefficients and correlation attenuation rates. This technology overcomes the limitations of conventional single-dimensional independent analysis, capturing dynamic correlation effects between channels, enabling a more comprehensive characterization of system health status using characteristic parameters, avoiding misjudgments due to missing correlations, and providing more accurate input for subsequent evaluation.

[0017] The regional calibration module, based on the quality degradation hotspot distribution map output by the dynamic evaluation module, corrects the channel stability quantification values ​​for the geographical areas marked on the hotspot distribution map using pre-calibrated geographical environment calibration factors. This technique differs from conventional globally uniform parameter calibration, achieving localized, targeted calibration for areas with significant quality degradation. It eliminates evaluation biases caused by geographical differences, making the calibrated channel stability quantification values ​​closer to actual transmission quality. This improves the guidance accuracy of subsequent operation control command sequence generation, avoids control inaccuracies caused by regional evaluation biases, and enhances the ability to precisely control the health status of highly integrated DWDM transmission systems. Attached Figure Description

[0018] Figure 1 This is a timing diagram of the highly integrated DWDM transmission system for 5G bearer networks described in this invention. Figure 2 Flowchart for establishing the multidimensional analysis module - signal correlation model; Figure 3 A flowchart for dynamic evaluation network-quantitative evaluation calculation; Figure 4 The distribution of optical power versus wavelength characteristics after transmission channel adjustment; Figure 5 This is a graph showing the trend of periodic self-optimization error entropy. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0020] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0021] See Figure 1The signal acquisition module captures transmitted signal samples in real time from the wavelength routing structure of the highly integrated DWDM transmission system, forming a raw observation data set. The multidimensional analysis module performs cross-analysis of the raw observation data set in both spectral and temporal dimensions, establishing a signal correlation model between wavelength channels, and extracting several characteristic parameters representing the system's health status based on this model. The dynamic evaluation module imports the extracted characteristic parameters into a dynamic evaluation network for calculation, obtaining a multidimensional quantitative evaluation result of the transmission channel quality, including a channel stability quantification value and a quality degradation hotspot distribution map. The regional calibration module performs regional deviation calibration on the multidimensional quantitative evaluation result of the transmission channel quality based on the geographical coordinates indicated by the channel quality degradation hotspot distribution map, obtaining a calibrated channel stability quantification value. The instruction generation module automatically generates and arranges a series of operation control instruction sequences for the highly integrated DWDM transmission system based on the calibrated channel stability quantification value.

[0022] See Figure 2 The multidimensional analysis module processes the raw observation data set received from the signal acquisition module. This raw data set contains sampling data from multiple wavelength channels within a continuous time period in a highly integrated DWDM transmission system. The multidimensional analysis module performs a traversal operation along the spectral dimension, simultaneously extracting optical power trajectory data and spectral shift trajectory data for each wavelength channel in the raw data set. The optical power trajectory data is obtained by calculating the power mean and power variance of all sampling points within the observation period for that channel. The spectral shift trajectory data is obtained by analyzing the spectral profile of that channel to obtain the center wavelength shift and spectral width variation. In the temporal dimension, the multidimensional analysis module divides the complete raw observation data set into multiple continuous time segments according to a preset fixed sampling period, generating a time-slice sequence. For any two adjacent time slices, the Pearson correlation coefficient of the optical power sequence of the same wavelength channel between the two slices is calculated as the optical power correlation coefficient, and the Pearson correlation coefficient of its spectral shift data sequence is calculated as the spectral shift correlation coefficient. Based on the calculated optical power correlation coefficients among all channels, a network topology for inter-channel optical power interaction is constructed, with wavelength channels as nodes and correlation coefficients as edge weights. Simultaneously, based on the spectral shift correlation coefficients among all channels, a similar inter-channel spectral perturbation propagation topology is constructed. Subsequently, the multi-dimensional analysis module fuses the inter-channel optical power interaction network topology and the inter-channel spectral perturbation propagation topology, calculating the edge weights between corresponding nodes in both topologies to generate a comprehensive signal correlation model between wavelength channels. This signal correlation model is a weighted undirected graph structure.

[0023] In some embodiments, the process of extracting feature parameters from the signal correlation model is systematic. First, it is necessary to identify all nodes representing wavelength channels in the signal correlation model and calculate the connectivity degree of each node. The connectivity degree is defined as the total number of other nodes connected to that node by an edge. The connectivity degrees of all nodes are statistically analyzed to form a distribution histogram. The morphological parameters of this histogram, such as skewness and kurtosis, are extracted as node degree distribution features. Next, the weight value of each connection edge in the signal correlation model is measured. The weight value of a connection edge is jointly determined by the optical power correlation coefficient and the spectral shift correlation coefficient between the two nodes connected by the edge. A specific calculation method is to assign different fixed weights to the two coefficients and then sum them. The calculation formula can be expressed as: in: This represents the weight value of the edge connecting node i and node j. This represents the optical power correlation coefficient between node i and node j. This represents the spectral offset correlation coefficient between node i and node j. and It is a pre-defined non-negative weighting coefficient and satisfies The distribution range, mean, and variance of all edge weights are statistically analyzed, and these statistics are used together as edge weight distribution characteristics. The process of extracting the topology clustering coefficient can be understood as involving two steps. The first step is to calculate the local clustering coefficient of each node. For any node in the signal association model, all directly connected neighboring nodes are identified, the actual number of connecting edges between these neighboring nodes is counted, and the theoretically maximum number of connecting edges is calculated. The ratio of the actual number of connecting edges to the maximum possible number of connecting edges is the local clustering coefficient of that node. The second step is to calculate the global topology clustering coefficient. The local clustering coefficients of all nodes in the signal association model are summed, and then divided by the total number of nodes. The resulting arithmetic mean is the clustering coefficient of the entire network topology. Finally, the node degree distribution characteristics, edge weight distribution characteristics, and topology clustering coefficient are jointly output as feature parameters characterizing the system's health status.

[0024] Optionally, when constructing the network topology for the mutual influence of optical power between channels and the propagation topology of spectral perturbations between channels, a correlation threshold can be set. Only when the absolute value of the calculated optical power correlation coefficient or spectral shift correlation coefficient is greater than this threshold will a connection be established between the corresponding two wavelength channel nodes. Otherwise, it is considered that the two nodes have no significant correlation within the current analysis period, and no connection is established. This method can filter out weak correlations caused by noise, making the generated signal correlation model clearer and more representative. When statistically analyzing the distribution characteristics of edge weights, in addition to using the mean and variance, percentiles of the weight values ​​can also be calculated, such as the median, upper quartile, and lower quartile. These percentiles can describe the central tendency and dispersion of the edge weight distribution, providing richer statistical information for subsequent evaluation.

[0025] It is understandable that the cross-analysis of the spectral and temporal dimensions is a continuous iterative process. The multidimensional analysis module periodically obtains new sets of raw observation data from the signal acquisition module and repeatedly executes the complete process of spectral feature extraction, time slice correlation calculation, topology construction and feature extraction. This enables dynamic tracking and feature updates of the health status of the highly integrated DWDM transmission system. The node degree distribution features, edge weight distribution features and topology aggregation coefficients obtained in each iteration reflect the system's operating status within the latest time window.

[0026] See Figure 3 In one embodiment of the present invention, the dynamic evaluation module receives node degree distribution features, edge weight distribution features, and clustering coefficients from the multidimensional analysis module. The dynamic evaluation module internally includes a feature fusion layer, a spatial mapping layer, and a result generation layer. The feature fusion layer first performs format conversion on the input feature parameters. The node degree distribution features are converted into a node importance vector. The conversion process involves normalizing the frequency values ​​of each interval of the node degree distribution histogram and arranging them into an ordered sequence. The edge weight distribution features are converted into a connection strength matrix, where the rows and columns of the matrix correspond to wavelength channel nodes, and the values ​​of the matrix elements correspond to the weight values ​​of the edges connecting the nodes. For node pairs without direct connections, the matrix element value is zero. The clustering coefficient is directly used as a scalar value as the topology compactness scalar. The feature fusion layer concatenates the node importance vector, the connection strength matrix, and the topology compactness scalar. The concatenated high-dimensional features undergo a linear dimensionality reduction operation, such as principal component analysis, to generate a fused feature vector with a fixed dimension. This fused feature vector comprehensively represents the overall topology and interaction strength of the transmission network.

[0027] In some embodiments, the spatial mapping layer maintains a pre-defined evaluation space, which is a multi-dimensional vector space with the same dimension as the fused feature vector. Each point in the evaluation space is associated with a specific transmission channel quality state through a pre-trained quality state classification model. For example, one region in the space may correspond to an "excellent" state, another to a "good" state, and yet another to a "poor" state. The spatial mapping layer projects the fused feature vector output by the feature fusion layer into the evaluation space through a linear transformation matrix to obtain a specific mapped coordinate. This mapped coordinate represents the position of the current network state in the evaluation space. The mapping process can be described as multiplying the fused feature vector by a weight matrix and adding a bias vector.

[0028] As can be understood, the result generation layer calculates the channel stability quantification value and generates a quality degradation hotspot distribution map based on the mapped coordinates. Internally, the result generation layer stores the coordinates of multiple preset quality state center points, each center point corresponding to a discrete quality level label. The result generation layer calculates the Euclidean distance from the mapped coordinates to each preset quality state center point, selecting the quality level associated with the center point with the smallest distance as the output. This quality level is quantified as a numerical score, i.e., the channel stability quantification value. Simultaneously, the result generation layer analyzes the point density in the vicinity of the mapped coordinate points in the evaluation space. The point density is calculated using a kernel density estimation algorithm. If the point density of a certain region is found to be significantly higher than the historical average density of the evaluation space, the result generation layer marks this region as an area of ​​abnormal quality state clustering. The boundary coordinates of this region in the evaluation space are then converted back to the actual wavelength channel index and geographical coordinates in the network topology through an inverse mapping relationship, ultimately forming a quality degradation hotspot distribution map that identifies potential problem areas.

[0029] In some embodiments, the process of calculating the distance from the mapped coordinates to each preset quality state center point uses a weighted distance formula to more accurately reflect the differences in the influence of different feature dimensions on the quality state. A specific distance calculation formula is as follows: in: This represents the weighted distance from the current mapped coordinates to the k-th preset mass state center point. This represents the total number of dimensions in the evaluation space. This represents the value of the current mapped coordinate in the d-th dimension. This represents the coordinates of the k-th preset mass state center point in the d-th dimension. It is a preset weight coefficient used to represent the importance of the d-th dimension feature to the quality state discrimination. All weight coefficients The sum of all is 1. By comparing all The value of the minimum value is used to determine the index k of the channel stability metric.

[0030] Optionally, the dimensionality reduction operation in the feature fusion layer can be implemented using an autoencoder neural network. During training, the autoencoder learns a compact fused feature representation using historical network state data. During deployment, the encoder compresses the concatenated high-dimensional features into a fused feature vector. This method preserves more non-linear feature information. For generating the quality degradation hotspot distribution map, the result generation layer can set a dynamic point density threshold. This threshold is adaptively adjusted based on the overall distribution variance of all mapped coordinate points in the evaluation space over a recent period. This makes the identification of hotspot regions insensitive to normal fluctuations in network state, only alerting to significant abnormal clusters.

[0031] It is understandable that the parameters of the dynamic evaluation network, including the dimensionality reduction matrix of the feature fusion layer, the transformation matrix of the spatial mapping layer, and the preset center point coordinates and weight coefficients of the result generation layer, need to be initialized through an offline training process. The training process uses labeled historical network state data and corresponding real performance metrics, and adjusts these parameters through optimization algorithms so that the channel stability quantification and quality degradation hotspot distribution map output by the dynamic evaluation network accurately reflect the actual transmission channel quality. During system operation, the dynamic evaluation module periodically executes the above feature fusion, spatial mapping, and result generation processes, outputting a real-time, multi-dimensional quantitative evaluation result for each evaluation cycle.

[0032] In one embodiment of the present invention, the regional calibration module receives a quality degradation hotspot distribution map and channel stability quantification values ​​from the dynamic evaluation module. The regional calibration module parses the quality degradation hotspot distribution map and extracts the center geographic coordinates and influence radius of one or more hotspot areas marked on the map. The center geographic coordinates are expressed in latitude and longitude format, and the influence radius is defined in kilometers as the circular influence range of each hotspot area. Based on the extracted center geographic coordinates, the regional calibration module accesses the fiber optic link topology database of the highly integrated DWDM transmission system. This database stores the geographic coordinates and topology connections of all fiber segments, optical amplifier nodes, and wavelength selection switch nodes. Through coordinate matching and topology search, the regional calibration module locates the specific fiber segment covered by the center point of each hotspot area, the optical amplifier nodes along the path, and the related wavelength selection switch nodes. The regional calibration module then queries the historical performance degradation data and current environmental parameters of these located network elements. The historical performance degradation data is obtained from the performance logs of the network management system and includes the historical optical power degradation history of the elements. The current environmental parameters are read from sensors deployed at the nodes and include temperature and humidity values.

[0033] In some embodiments, the regional calibration module inputs historical performance degradation data and environmental parameters into a preset deviation compensation calculation function to calculate a localized compensation factor. The deviation compensation calculation function internally maintains a multi-factor influence lookup table, a three-dimensional lookup structure. Its input index dimensions are historical performance degradation rate, ambient temperature, and ambient humidity, and its output value is a basic influence coefficient. The regional calibration module first converts the historical performance degradation data into a scalarized historical performance degradation rate. This conversion can be achieved by calculating the average degradation per unit time. The historical performance degradation rate is then combined with real-time ambient temperature and humidity parameters into a query key, which is retrieved from the multi-factor influence lookup table to obtain the corresponding basic influence coefficient. Simultaneously, the regional calibration module obtains the real-time load rate of all fiber optic links within the influence radius of the current hotspot area from the network load monitoring system. The real-time load rate is the ratio of current traffic to the maximum capacity of the link. The real-time load rate is input into a preset non-linear transformation curve, which maps the load rate to a load influence coefficient between 0.5 and 2.0. The deviation compensation calculation function multiplies the basic influence coefficient by the load influence coefficient to obtain a preliminary compensation factor.

[0034] It is understandable that the initial compensation factor needs to be scaled according to the influence range of the hotspot area to obtain the final localized compensation factor. The deviation compensation calculation function dynamically adjusts a radius scaling factor based on the size of the influence radius of the hotspot area. The larger the influence radius, the smaller the radius scaling factor. One specific calculation method is to set a maximum influence radius threshold. When the influence radius is equal to or exceeds this threshold, the radius scaling factor takes the minimum value of 0.8; when the influence radius is zero, the radius scaling factor takes the maximum value of 1.2. Influence radii between these two values ​​have their radius scaling factor determined through linear interpolation. The final localized compensation factor is obtained by multiplying the initial compensation factor by the radius scaling factor. The calculation formula is as follows: in: This represents the final localization compensation factor. Indicates the initial compensation factor. Indicated by the radius of influence Let be the radius scaling factor function of the independent variable. The regional calibration module uses the calculated localized compensation factor to perform a weighted correction on the channel stability quantification values ​​of all transmission channels falling within the influence radius of the corresponding hotspot area. The correction operation is to directly multiply the original channel stability quantification value by the localized compensation factor of the corresponding hotspot area.

[0035] In some embodiments, all corrected channel stability metrics are globally normalized. The purpose of normalization is to ensure that the numerical range of the channel stability metrics after regional deviation calibration remains consistent with that before calibration, for example, falling within the range of 0 to 100. Normalization employs a linear scaling method to find the maximum and minimum values ​​among all calibrated channel stability metrics, and then maps each value to a target range using a formula. Finally, the calibrated and normalized channel stability metrics are output. Optionally, the construction of the multi-factor influence lookup table is based on statistical analysis of historical fault and performance data. Independent multi-factor influence lookup tables can be established for different types of fiber optic segments and optical node devices. When querying, the regional calibration module first determines the type of network element and then selects the corresponding lookup table for retrieval to improve the accuracy of the basic influence coefficients. The nonlinear conversion curve can be designed as a piecewise function. For example, when the real-time load rate is below 50%, the load impact coefficient is 1.0, indicating no impact; when the load rate is between 50% and 80%, the load impact coefficient increases linearly to 1.5; when the load rate exceeds 80%, the load impact coefficient increases exponentially to 2.0, so as to more sensitively reflect the potential risks brought by high load.

[0036] In one embodiment of the present invention, the instruction generation module receives calibrated channel stability quantization values ​​from the regional calibration module. The instruction generation module internally predefines multiple consecutive channel stability threshold intervals. Each channel stability threshold interval is associated with a set of preset operation control instruction templates. The operation control instruction templates define the types and basic parameters of various adjustment operations for the transmission channel. The instruction generation module compares the calibrated channel stability quantization value of each transmission channel with the boundary values ​​of all channel stability threshold intervals one by one to determine which specific channel stability threshold interval the quantization value of the transmission channel falls into, thereby assigning a corresponding stability level to each transmission channel. Referring to Table 1, an example of the correspondence between channel stability threshold intervals and operation control instruction templates is shown.

[0037] Table 1: Correspondence between Channel Stability Threshold Ranges and Operation Control Command Templates The instruction generation module loads an operation control instruction template associated with the stability threshold range of its respective channel for each transmission channel. The operation control instruction template includes one or more of the following instruction types: optical power adjustment instruction, wavelength fine-tuning instruction, and error correction coding redundancy adjustment instruction. After loading, the instruction generation module checks the operation control instructions of all transmission channels, paying particular attention to transmission channels that are physically adjacent on the fiber optic link or converge at the same network node. It checks for logical or resource conflicts between the operation control instructions loaded on these adjacent transmission channels. Defined conflict types include power contention conflicts, wavelength overlap conflicts, and resource excess conflicts. A power contention conflict occurs when the sum of the target values ​​of the optical power adjustment instructions of adjacent channels exceeds the saturation output power of the shared optical amplifier; a wavelength overlap conflict occurs when the target wavelength adjustment ranges specified by the wavelength fine-tuning instructions of adjacent channels overlap; and a resource excess conflict occurs when the total amount of computational or buffer resources required by the combination of instructions from adjacent channels exceeds the processing capacity quota of the network node.

[0038] In some embodiments, for detected power contention conflicts, the instruction generation module compares the channel stability metrics of multiple conflicting transmission channels. For transmission channels with lower channel stability metrics, the optical power adjustment instruction in their operation control instruction template is proportionally reduced. This reduction operation continues until the sum of the adjusted optical power target values ​​of all conflicting channels is lower than the saturation power of the shared optical amplifier. For wavelength overlap conflicts, the instruction generation module compares the channel stability metrics of conflicting channels. For transmission channels with lower channel stability metrics, the adjustment direction of their wavelength fine-tuning instructions is restricted, causing their target wavelength range to be far away from the target wavelength range of transmission channels with higher channel stability metrics. For resource excess conflicts, the instruction generation module compares the channel stability metrics of conflicting channels. For transmission channels with lower channel stability metrics, the system time slots required for instruction execution are proportionally extended, thereby reducing the instantaneous resource demand that network nodes need to process per unit time. A calculation formula for reducing the power target value of low-priority channels in power contention conflicts is as follows: in: This indicates the adjusted target optical power value. This represents the target optical power value in the original operation control command template. This represents the highest channel stability metric among the conflicting channels. This represents the channel stability quantification of the low-priority channel currently awaiting adjustment. It is a conflict resolution coefficient between 0 and 1.

[0039] Understandably, after completing instruction conflict checks and adaptive adjustments for all adjacent transmission channels, the instruction generation module needs to orchestrate the final operation control instructions for all transmission channels into an ordered executable sequence. The orchestration process strictly follows the physical connection order of the fiber optic links and the signal transmission direction. Starting from the source node of the network topology, the instruction generation module sequentially arranges the instructions to be executed at each network node according to the signal flow path. The instruction generation module ensures that at any network node, the total number of instructions scheduled for execution at the same time and their resource requirements do not exceed the processing capacity limit of that network node. If a momentary instruction overload is detected, some instructions will be delayed to subsequent idle time slots for execution. The final generated instruction sequence is a timestamped list of commands, which the instruction generation module then distributes to the control plane of the highly integrated DWDM transmission system for execution.

[0040] See Figure 4 In the channel stability verification of a highly integrated DWDM transmission system, the correlation characteristics between adjusted optical power and wavelength are visualized using dual-dimensional quantization curves. Specifically, a spatial dimension benchmark is constructed using 20 transmission channels as the horizontal axis, and the vertical axis represents the adjusted optical power quantized in dB units and the adjusted wavelength quantized in nm units, forming a synchronous distribution map of the dual-channel data. The optical power trajectory data is presented as a dotted curve, and the spectral shift trajectory data as a box curve. The fluctuation characteristics of both directly map the coupling relationship between the inter-channel optical power interaction network topology and the spectral perturbation propagation topology. At the data analysis level, the optical power curve reaches a peak (over 20 dB) at channel 9, and valleys (close to 10 dB) appear at channels 5 and 17. This extreme value distribution characteristic corresponds to the high-value region of the node degree distribution in the signal correlation model. The wavelength curve shows a drift peak of 1570 nm at channels 6 and 11, and drops to around 1550 nm at channels 17 and 20. The drift amplitude directly reflects the quantization result of the center wavelength drift. By cross-analyzing the hyperbolic fluctuation trends, the core characteristic parameters of channel stability can be extracted: the inverse fluctuation range of optical power and wavelength (such as channels 4 and 17) indicates the coupled perturbation of the spectral and temporal dimensions, corresponding to potential areas of quality degradation hotspots; while the synchronous fluctuation range (such as channels 2 and 8) characterizes the aggregation coefficient of the channel topology being within the optimal range, confirming the effectiveness of the compensation and correction by the regional calibration module. The parameters of the spectrum strictly match the output requirements of the dynamic evaluation network, with an optical power resolution accurate to 0.1 dB and a wavelength resolution of 0.1 nm, ensuring that the subsequent instruction generation module can use this spectrum as a benchmark to complete threshold range matching and conflict verification when executing optical power adjustment and wavelength fine-tuning instructions.

[0041] In one embodiment of the present invention, the periodic self-optimization module is triggered at the end of each preset optimization cycle. The length of the optimization cycle can be configured to be 24 hours or one week. The periodic self-optimization module collects the actual performance index data generated by the system during the optimization cycle from the network performance management unit of the highly integrated DWDM transmission system. This performance index data includes the bit error rate, optical signal-to-noise ratio, and latency of each wavelength channel, constituting a verification true value dataset. The periodic self-optimization module obtains the multi-dimensional quantitative evaluation results of the transmission channel quality output by the dynamic evaluation network during the same period from the output log of the dynamic evaluation module, especially the predicted sequence of channel stability quantification values. It compares the actual performance indexes in the verification true value dataset with the predicted evaluation results channel by channel and time point by time, and calculates the evaluation error of each channel at each sampling time. The evaluation error is the absolute or relative difference between the predicted channel stability quantification value mapped to the actual performance index scale and the actual measured value. The periodic self-optimization module collects all evaluation errors of all channels within the current optimization cycle as the input dataset for error analysis.

[0042] In some embodiments, the periodic self-optimization module constructs an error distribution map based on the evaluation errors of all channels. The construction process uses the wavelength channel number as the abscissa and the geographical node sequence number traversed by the fiber optic link as the ordinate to establish a two-dimensional error distribution grid. Each grid cell corresponds to the error of a specific wavelength channel in a specific geographical node segment. The periodic self-optimization module fills the calculated evaluation error value of each channel into all grid cells covered by that channel in its corresponding geographical node sequence. If a channel spans multiple geographical nodes, its error value is copied to the corresponding grid cells at multiple ordinate positions. Spatial clustering analysis is performed on the filled two-dimensional error distribution grid. The spatial clustering algorithm identifies sets of grid cells with evaluation error values ​​higher than a preset threshold that are continuous or adjacent on both the abscissa and ordinate, marking each such continuous region as a high-error cluster. The periodic self-optimization module then calculates the wavelength coverage range of each high-error cluster in the spectral dimension, represented by the starting and ending wavelength channel numbers, and the fiber segment coverage range in the geographical dimension, represented by the starting and ending geographical node numbers. The periodic self-optimization module calculates the average and variance of the error values ​​of all grid cells in each high error cluster region, and calculates the proportion of the number of grid cells contained in the high error cluster region to the total number of grid cells in the entire two-dimensional error distribution grid. This proportion is used as the area proportion. The average, variance and area proportion together serve as a quantitative description of the clustering characteristics of errors in the spectral and geographic dimensions.

[0043] It is understandable that, based on the quantitative description of the error distribution map and error clustering characteristics, the periodic self-optimization module iteratively adjusts the internal parameters of the dynamic evaluation network. The adjustment primarily targets the feature weight parameters in the feature fusion layer, which convert node degree distribution features and edge weight distribution features into vectors and matrices, and the mapping parameters in the space mapping layer, which map the fused feature vectors to the evaluation space. The adjustment aims to reduce the overall error entropy reflected in the error distribution map. Error entropy is a measure of the uncertainty of the error distribution. A parameter adjustment objective function that promotes the reduction of overall error entropy can be expressed as minimizing the following expression: in: This represents the overall error entropy related quantity to be minimized. This represents the total number of high-error clusters identified. Indicates the first The average value of error values ​​within each high error cluster region Indicates the first The area ratio of high-error clusters This represents a logarithmic function with base e. The periodic self-optimization module uses optimization algorithms such as gradient descent or genetic algorithms to fine-tune the parameters of the dynamically evaluated network, making it... The value changes in a decreasing direction. After the parameter adjustment is completed, the periodic self-optimization module reloads the updated feature fusion layer and spatial mapping layer models into the dynamic evaluation module for transmission channel quality evaluation calculation in the next optimization cycle, and resets the optimization cycle timer to start a new periodic self-optimization process.

[0044] In some embodiments, when calculating the evaluation error, relative error or normalized absolute error can be used. For example, the channel stability quantification value output by the dynamic evaluation network can be mapped to a predicted value for the optical signal-to-noise ratio (OSNR) through a calibration curve, and then compared with the actual measured OSNR to calculate the relative error. For spatial clustering analysis, a density-based clustering algorithm can be used, treating the evaluation error value as the "height" of the grid cells, and searching for "plateau" regions where the error "altitude" is higher than a threshold and is connected.

[0045] Optionally, the length of the optimization period can be dynamically adjusted based on the stability of the network state. When the error entropy decreases significantly over several consecutive periods, the optimization period can be appropriately extended to save computational resources. Conversely, when the error entropy suddenly increases, the optimization period can be shortened to quickly adapt to network changes. The error threshold can also be designed to be adaptive, dynamically calculating the threshold for the current period based on the overall mean and variance of the historical error distribution. This makes the identification of high-error-aggregate areas sensitive to error levels at different times. When adjusting the dynamically evaluated network parameters, an incremental learning approach can be adopted, making small adjustments each time based only on the error data of the latest optimization period. This maintains the stability of the evaluation model and avoids drastic fluctuations in model performance due to abnormal data in a single period.

[0046] It is understandable that the operation of the periodic self-optimization module is independent of the monitoring and control of the main service flow. Its computing resources are allocated during system idle periods or on dedicated processing units, ensuring that it does not interfere with the performance of real-time signal acquisition, multi-dimensional analysis, dynamic evaluation, regional calibration, and instruction generation processes. Through periodic error collection, spectral analysis, and parameter iteration, the dynamic evaluation network can continuously adapt to the aging of the highly integrated DWDM transmission system itself, environmental changes, and changes in service models, enabling the multi-dimensional quantitative evaluation results to maintain high accuracy over a long period of time.

[0047] See Figure 5 The figure, with the optimization period (days) on the x-axis and the overall error entropy on the y-axis, clearly shows the trend of error entropy change and the quadratic fitting effect during the iterative optimization phase of the model parameters. As can be seen from the figure, as the optimization period progresses from day 1 to day 8, the error entropy (blue solid line) exhibits a continuous monotonically decreasing trend: the initial value is approximately 1.85, decreasing to approximately 1.25 after 8 iterations, indicating that the parameter adjustment of the dynamic evaluation network effectively reduces the uncertainty of the evaluation results. The red dashed line represents the quadratic fitting curve of the error entropy sequence, whose trend closely matches the measured data, verifying that the error entropy exhibits an approximately quadratic decay pattern with the optimization period, reflecting the convergence effect of parameter iteration on the error distribution. The core significance of the figure lies in: Quantifying the convergence effect: The continuous decrease in error entropy directly proves that the periodic self-optimization module effectively reduces the uncertainty of the evaluation error distribution in the spectral and geographical dimensions by iteratively adjusting the feature fusion layer weights and spatial mapping layer parameters. Verifying the fitting trend: The high fitting degree of the quadratic fitting curve indicates that the decay process of error entropy is predictable, providing a quantitative basis for the dynamic adjustment of the optimization period (such as extending or shortening the period). Optimization of directional anchoring: The monotonically decreasing characteristic of the curve clarifies the effectiveness of the objective function for parameter adjustment, namely, by minimizing the overall error entropy, continuously improving the accuracy of the dynamic evaluation network's assessment of the transmission channel quality.

[0048] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A highly integrated DWDM transmission system for 5G bearer networks, characterized in that, The system includes: The signal acquisition module captures transmitted signal samples in real time from the wavelength routing structure of the highly integrated DWDM transmission system to form a raw observation data set; The multidimensional analysis module performs cross-analysis of the original observation data set in spectral and temporal dimensions, establishes a signal correlation model between wavelength channels, and extracts several characteristic parameters characterizing the health status of the system based on the signal correlation model. The dynamic evaluation module imports the extracted feature parameters into the dynamic evaluation network for calculation, and obtains a multi-dimensional quantitative evaluation result of the transmission channel quality. The multi-dimensional quantitative evaluation result includes a channel stability quantification value and a quality degradation hotspot distribution map. The regional calibration module, based on the geographical coordinates indicated by the channel quality degradation hotspot distribution map, performs regional deviation calibration on the multi-dimensional quantitative evaluation results of the transmission channel quality to obtain the calibrated quantitative value of channel stability. The instruction generation module automatically generates and arranges a series of operation control instruction sequences for the highly integrated DWDM transmission system based on the calibrated channel stability quantification value.

2. The highly integrated DWDM transmission system for 5G bearer networks according to claim 1, characterized in that, The process involves cross-analysis of the original observation data set along both spectral and temporal dimensions to establish a signal correlation model between wavelength channels. Based on this model, several characteristic parameters characterizing the system's health status are extracted, specifically including: In the spectral dimension, each wavelength channel in the original observation dataset is traversed to extract its optical power trajectory data and spectral shift trajectory data. The optical power trajectory data includes the power mean and power variance, and the spectral shift trajectory data includes the center wavelength shift and spectral width change. In the time dimension, the original observation data set is sliced ​​according to a preset sampling period to generate a continuous time slice sequence, and the optical power correlation coefficient and spectral shift correlation coefficient of the corresponding channels between adjacent time slices are calculated. Based on the optical power correlation coefficient, a network topology for the mutual influence of optical power between channels is constructed, and a topology for the propagation of spectral perturbations between channels is constructed based on the spectral shift correlation coefficient. By integrating the network topology of the mutual influence of optical power between channels and the propagation topology of spectral perturbation between channels, a signal correlation model between the wavelength channels is formed. The node degree distribution features, edge weight distribution features, and clustering coefficient of the network topology are extracted from the signal correlation model. The node degree distribution features, edge weight distribution features, and clustering coefficient are used together as feature parameters to characterize the health status of the system.

3. The highly integrated DWDM transmission system for 5G bearer networks according to claim 2, characterized in that, The node degree distribution features, edge weight distribution features, and clustering coefficients of the network topology are extracted from the signal correlation model, including: Identify the nodes corresponding to all wavelength channels in the signal correlation model, calculate the connectivity of each node, i.e. the number of other nodes that are correlated with the node in terms of optical power or spectral shift, and statistically analyze the distribution histogram of the connectivity of all nodes to obtain the node connectivity distribution characteristics. The weight value of each connection edge in the measurement signal correlation model is the weighted sum of the optical power correlation coefficient and the spectral shift correlation coefficient between the two nodes of the connection edge. The distribution range and central tendency of all connection edge weight values ​​are statistically analyzed to obtain the edge weight distribution characteristics. For each node in the signal correlation model, find all nodes directly connected to it, and calculate the ratio of the actual number of connecting edges between directly connected nodes to the theoretical maximum number of connecting edges, which is used as the local clustering coefficient of the node. Calculate the arithmetic mean of the local clustering coefficients of all nodes, and use it as the clustering coefficient of the topology.

4. The highly integrated DWDM transmission system for 5G bearer networks according to claim 1, characterized in that, The process of importing the extracted feature parameters into the dynamic evaluation network for calculation to obtain a multi-dimensional quantitative evaluation result of the transmission channel quality includes the following operations: The dynamic evaluation network consists of a feature fusion layer, a spatial mapping layer, and a result generation layer; The feature fusion layer receives the node degree distribution feature, edge weight distribution feature, and clustering coefficient, converts the node degree distribution feature into a node importance vector, the edge weight distribution feature into a connection strength matrix, and the clustering coefficient into a topology compactness scalar. The node importance vector, connection strength matrix, and topology density scalar are concatenated and dimension-reduced to generate a fused feature vector; The spatial mapping layer maps the fused feature vector to a preset evaluation space, in which each coordinate position corresponds to a quality state of a transmission channel. The result generation layer calculates the distance from the fused feature vector to each preset quality state center point based on the mapping coordinates of the fused feature vector in the evaluation space, and outputs the quality level corresponding to the center point with the smallest distance as a channel stability metric value. Simultaneously, the result generation layer identifies regions with abnormally concentrated quality states based on the density of neighboring regions in the evaluation space using the mapped coordinates, and then maps the location of these regions with abnormally concentrated quality states back to the actual wavelength channels and geographic coordinates to form the quality degradation hotspot distribution map.

5. The highly integrated DWDM transmission system for 5G bearer networks according to claim 4, characterized in that, The regional bias calibration of the multi-dimensional quantitative evaluation results of the transmission channel quality based on the geographical coordinates indicated by the channel quality degradation hotspot distribution map includes: Extract the central geographic coordinates and influence radius of one or more hotspot areas from the hotspot distribution map of quality degradation; Based on the central geographic coordinates, locate the corresponding fiber segment, optical amplifier node, and wavelength selection switch node in the fiber optic link topology database of the highly integrated DWDM transmission system, and query the historical performance attenuation data and current environmental parameters of the network components. Input the historical performance degradation data and environmental parameters obtained from the query into a preset deviation compensation calculation function to calculate the localized compensation factor for each hotspot area. Using the calculated localized compensation factor, the channel stability quantization values ​​of all transmission channels falling within the influence radius of the corresponding hotspot area are weighted and corrected. The correction formula is to multiply the original quantization value by the corresponding localized compensation factor. All corrected channel stability metrics are globally normalized to ensure that their numerical range remains consistent with that before calibration, and finally the calibrated channel stability metrics are output.

6. The highly integrated DWDM transmission system for 5G bearer networks according to claim 5, characterized in that, The process involves inputting the historical performance degradation data and environmental parameters obtained from the query into a preset deviation compensation calculation function to calculate a localized compensation factor for each hotspot area. This is achieved through the following steps: The deviation compensation calculation function internally maintains a multi-factor influence lookup table, which uses the combination of historical performance decay rate and ambient temperature and humidity as input index and outputs a basic influence coefficient. The historical performance degradation data obtained from the query is converted into historical performance degradation rate, which is then combined with environmental temperature and humidity parameters and retrieved in the multi-factor influence lookup table to obtain the corresponding basic influence coefficient. Obtain the real-time load rate of fiber optic links within the influence radius of the current hotspot area, and map the real-time load rate into a load influence coefficient through a non-linear transformation curve; Multiply the basic influence coefficient by the load influence coefficient to obtain a preliminary compensation factor; The initial compensation factor is multiplied by a radius scaling factor that is dynamically adjusted according to the size of the influence radius of the hotspot area to obtain the final localized compensation factor, wherein the larger the influence radius, the smaller the radius scaling factor.

7. The highly integrated DWDM transmission system for 5G bearer networks according to claim 1, characterized in that, The step of automatically generating and arranging a series of operation control command sequences for the highly integrated DWDM transmission system based on the calibrated channel stability quantification value includes the following steps: Set multiple channel stability threshold ranges, and associate each threshold range with a set of preset operation control instruction templates; The calibrated channel stability quantification value is compared with the multiple channel stability threshold intervals to determine the threshold interval to which each transmission channel belongs. For each transmission channel, an operation control instruction template associated with its respective threshold range is loaded. The operation control instruction template includes one or more of the following: optical power adjustment instruction, wavelength fine-tuning instruction, and error correction coding redundancy adjustment instruction. Check if there are any conflicts in the operation control commands loaded on adjacent transmission channels. If there are any conflicts, adjust the command parameters of the low-priority channel according to the priority of the channel stability quantification value to eliminate the conflict. According to the physical order and signal transmission direction of the optical fiber link, the final operation control instructions of all transmission channels are time-sequenced to generate an ordered sequence of operation control instructions. The sequence of operation control instructions ensures that at any network node, the number of instructions executed at the same time does not exceed the processing capacity limit of the network node.

8. The highly integrated DWDM transmission system for 5G bearer networks according to claim 7, characterized in that, The process involves checking whether there are conflicts in the operation control commands loaded on adjacent transmission channels. If conflicts exist, the command parameters of the lower-priority channels are adaptively adjusted to eliminate the conflicts based on the priority of the channel stability quantification values. Specifically, this includes: Conflict types are defined, including: power contention conflict, which is when the target value of the optical power adjustment command of adjacent channels exceeds the saturation power of the shared amplifier; wavelength overlap conflict, which is when the target range of wavelength fine-tuning commands of adjacent channels overlaps; and resource overload conflict, which is when the computational resources required for the combination of commands of adjacent channels exceed the node quota. For detected power contention conflicts, compare the channel stability quantization values ​​of the conflicting channels. For channels with low quantization values, reduce the optical power adjustment target value in their instruction template proportionally until the total power demand of the shared amplifier is lower than the saturation power. For detected wavelength overlap conflicts, compare the channel stability quantization values ​​of the conflicting channels. For channels with low quantization values, restrict the adjustment direction of their wavelength fine-tuning commands to keep them away from the target wavelength range of channels with high quantization values. For detected resource overload conflicts, the channel stability quantification values ​​of the conflicting channels are compared. For channels with low quantification values, the time slot allocation required for instruction execution is extended proportionally, thereby reducing instantaneous resource demand.

9. The highly integrated DWDM transmission system for 5G bearer networks according to claim 8, characterized in that, Also includes: A periodic self-optimization module performs a periodic self-optimization process on the dynamic evaluation network, the process including: At the end of an optimization cycle, the actual performance index data generated by the highly integrated DWDM transmission system during the optimization cycle are collected as a set of verification truth data. The verification true value dataset is compared with the multi-dimensional quantitative evaluation results of the transmission channel quality output by the dynamic evaluation network during the same period, and the evaluation error of each channel is calculated. Based on the evaluation errors of all channels, an error distribution map is constructed to analyze the clustering characteristics of errors in the spectral and geographic dimensions. Based on the error distribution map, the feature weights of the feature fusion layer and the mapping parameters of the spatial mapping layer in the dynamic evaluation network are iteratively adjusted in order to reduce the overall error entropy of the error distribution map. The dynamically evaluated network with adjusted parameters is used for the evaluation calculation in the next optimization cycle, and a new periodic self-optimization process is initiated.

10. The highly integrated DWDM transmission system for 5G bearer networks according to claim 9, characterized in that, Based on the evaluation errors of all channels, an error distribution map is constructed, and the clustering characteristics of the errors in the spectral and geographic dimensions are analyzed. This is achieved through the following method: A two-dimensional error distribution grid is established with wavelength channels as the horizontal axis and geographic node sequences as the vertical axis. The evaluation error value of each channel is filled into its corresponding coordinate position in the two-dimensional error distribution grid; Spatial clustering analysis is performed on the filled two-dimensional error distribution grid to identify continuous regions where the evaluation error value is higher than the threshold, and these continuous regions are marked as high error clustering areas. The wavelength coverage of high-error clusters in the spectral dimension and the fiber segment coverage in the geographical dimension were statistically analyzed separately. Calculate the mean and variance of error values ​​within each high error cluster, as well as the area proportion of the high error cluster in the two-dimensional grid, and use the statistics as a quantitative description of the error clustering characteristics.

Images