Data analysis driven WDM and OLP collaborative optimization method and system

Abstract

The invention discloses a WDM and OLP collaborative optimization method and system driven by data analysis, and the method comprises the steps: constructing a data set system, obtaining multi-dimensional original data based on the system, completing the judgment of the degradation level of the multi-dimensional original data, obtaining the corresponding degradation level of a sample, extracting the features of the original data, carrying out the preprocessing, and obtaining a standardized core feature set; constructing a degradation trend prediction model, inputting the obtained set into the constructed model for training to obtain a trained degradation trend prediction model, inputting a degradation level corresponding to a real-time sample into the trained model for processing, substituting the obtained predicted degradation level into a defined risk level and a quantitative mapping rule for matching, and obtaining a fault risk level. And formulating a collaborative strategy and a resource configuration scheme based on the fault risk level, and executing collaborative operation. According to the invention, the situation that WDM and OLP are difficult to cooperatively adjust according to the signal degradation trend in the prior art can be effectively improved.

Description

Technical Field

[0001] This invention belongs to the field of data analysis technology, specifically relating to a data analysis-driven WDM and OLP collaborative optimization method and system. Background Technology

[0002] Long-distance backbone networks are the core hubs of communication networks, undertaking the long-distance transmission of massive amounts of data. During long-distance transmission, optical signals are susceptible to factors such as fiber loss, dispersion, environmental temperature changes, and equipment aging, resulting in gradual degradation—manifested as slow fluctuations in optical power, a continuous decrease in optical signal-to-noise ratio (OSNR), and a gradual increase in bit error rate (BER). In traditional solutions, WDM equipment passively adjusts parameters only after signal degradation reaches a threshold, while OLP equipment typically triggers switching after a fault occurs. The lack of a coordinated mechanism between the two leads to insufficient intervention in the early stages of signal degradation, potentially causing service interruptions in severe cases. Therefore, a data-driven collaborative optimization method is urgently needed to predict degradation trends and coordinate equipment operation, proactively mitigating transmission risks. Summary of the Invention

[0003] To address the problems in the related technologies, this application provides a data analysis-driven WDM and OLP co-optimization method and system, which solves the problems mentioned in the background technology.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a data analysis-driven WDM and OLP co-optimization method, comprising the following steps: Step S1: Construct a data set system based on the monitoring requirements of optical transmission networks. Based on the data set system, adopt an adaptive acquisition mechanism to obtain multi-dimensional raw data. Combine the multi-dimensional raw data with business rules to define the degradation level and complete the degradation level judgment to obtain the corresponding degradation level of the sample. Step S2: Extract features from the multi-dimensional raw data and preprocess them to obtain a core feature set. Process the core feature set using a time-series data-specific filtering algorithm to obtain a standardized core feature set. Construct a degradation trend prediction model. Use the standardized core feature set as input and the corresponding degradation level of the sample as the label to train the degradation trend prediction model and obtain the trained degradation trend prediction model. Step S3: Input real-time multi-dimensional raw data into the trained degradation trend prediction model for processing to obtain the predicted degradation level, define the fault risk level set, formulate the quantization mapping rule, substitute the predicted degradation level into the fault risk level set and the quantization mapping rule for matching, and obtain the fault risk level corresponding to the degradation level. Step S4: Develop collaborative strategies and resource allocation plans based on the fault risk level and execute collaborative operations.

[0005] Furthermore, the specific process of step S1 is as follows: Based on the monitoring needs of optical transmission networks, a data collection system is constructed, which includes: optical transmission quality data, environmental impact data, and equipment operation data. The optical transmission quality data includes optical power, optical signal-to-noise ratio, bit error rate, and cumulative dispersion value. Environmental impact data includes temperature, relative humidity, and vibration intensity in the area where the fiber optic link is located; The equipment operation data consists of the working status parameters of the wavelength division multiplexing equipment and the automatic switching protection device for fiber optic lines; the equipment operation data includes the operating current of the optical amplifier, the operating status of the dynamic dispersion compensation module, and the connectivity of the primary and backup links. An adaptive acquisition mechanism is set up to collect optical transmission quality data, environmental impact data, and equipment operation data. Data access is completed through a standardized interface compatible with mainstream optical transmission equipment protocols, ultimately obtaining multi-dimensional raw data covering the entire optical transmission link. The multi-dimensional raw data includes historical time-series data and real-time data; among which, historical data is synchronously stored in a time-series database. Set a set of degradation levels Sort by impact level from lowest to highest; n represents the number of levels. Indicates the first There are several degradation levels; when n=4... For mild degradation, For moderate degradation, For severe degradation, This is a fatal degradation; Based on the permitted business scope, a threshold for each type of optical transmission quality data is preset. Then, combined with the defined degree of deviation of a single type of data from the threshold, the superposition weight of multiple types of data anomalies, and the judgment rules for setting a set of degradation levels, the threshold, the defined degree of deviation of a single type of data from the threshold, the superposition weight of multiple types of data anomalies, and the set of degradation levels are integrated and processed together to obtain the degradation level classification standard. Based on real-time data of optical transmission quality and the degradation level classification standard, the corresponding degradation level of the sample is obtained.

[0006] Furthermore, the specific process of step S2 is as follows: Configure the historical time window length as T, and extract the time series segments within the corresponding window from the historical time series data; Features are extracted from time series segments, including trend features, fluctuation features, and correlation features, to form an initial feature set. Redundant features in the initial feature set are removed based on the feature importance assessment method to obtain the core feature set; Preprocessing of the core feature set: The core feature set is removed by a time series data-specific filtering algorithm to obtain a denoised core feature set. The denoised core feature set is then standardized or normalized to unify the dimensions and obtain a standardized core feature set. A degradation trend prediction model was built based on long short-term memory networks; The standardized core feature set is used as the input to the degradation trend prediction model, and the corresponding degradation level of the sample is used as the label. The degradation trend prediction model is trained iteratively to obtain the trained degradation trend prediction model.

[0007] Furthermore, the specific process of step S3 is as follows: Continuously feed in new real-time multi-dimensional raw data; extract, filter, and preprocess the features of the new real-time multi-dimensional raw data, and concatenate it with the core initial feature set within the historical time window to obtain the concatenation result; input the concatenation result into the trained degradation trend prediction model; The trained degradation trend prediction model outputs the probability distribution of each degradation level within a preset future time window; the degradation level corresponding to the highest probability in the probability distribution is selected as the predicted degradation level. Define and predict the set of failure risk levels that correspond one-to-one with each degradation level. Sort by risk level from low to high; when n=4, correspond This represents low risk. correspond Representing medium risk correspond This represents high risk. correspond This represents an urgent risk; The predicted degradation level is substituted into the set of failure risk levels and a quantitative mapping rule is established for matching to obtain the failure risk level corresponding to the degradation level.

[0008] Furthermore, the specific process of step S4 is as follows: Based on the fault risk level, combined with the fault risk level set, and combined with the requirements of business continuity and network stability, a differentiated collaborative operation strategy framework is formulated. By inputting the fault risk levels of different orders into the strategy framework and matching the corresponding level of operation strategy, an OLP and WDM collaborative operation strategy system adapted to the fault risk level is obtained. Based on the execution requirements of the collaborative operation strategy system and combined with the functional characteristics of optical transmission equipment, a resource allocation scheme is formulated. Based on the resource allocation scheme, the parameter adjustment boundary of the main equipment optimization, the performance adaptation standard of the backup resources, and the timing coordination rules of cross-equipment operation are preprocessed to obtain a ready-to-go collaborative execution preparatory system. Based on the collaborative operation strategy system, resource allocation scheme and collaborative execution preparation system, OLP and WDM collaborative operations are triggered according to the preset timing logic. During the collaborative operation of OLP and WDM, real-time data on the operation status, signal quality changes, and service transmission stability of the wavelength division multiplexing equipment and the automatic switching protection device for fiber optic lines are collected. Based on the collected data on operation status, signal quality changes, and service transmission stability, a complete record of the operation process is formed, resulting in the collaborative operation execution results and process data.

[0009] A data analysis-driven WDM and OLP co-optimization system, applied to the aforementioned data analysis-driven WDM and OLP co-optimization method, includes a data management module, a model building module, a risk prediction module, and a co-execution module; The data management module is used to build a data set system based on the monitoring requirements of optical transmission networks. Based on the data set system, an adaptive acquisition mechanism is used to acquire multi-dimensional raw data. The multi-dimensional raw data is combined with business rules to define the degradation level and complete the degradation level judgment to obtain the corresponding degradation level of the sample. The model building module is used to extract features from multi-dimensional raw data and preprocess them to obtain a core feature set. The core feature set is then processed using a time-series data-specific filtering algorithm to obtain a standardized core feature set. A degradation trend prediction model is then constructed. The standardized core feature set is used as input, and the corresponding degradation level of the sample is used as a label to train the degradation trend prediction model, resulting in the trained degradation trend prediction model. The risk prediction module is used to input real-time multi-dimensional raw data into the trained degradation trend prediction model for processing, obtain the predicted degradation level, define the fault risk level set, formulate the quantification mapping rule, and substitute the predicted degradation level into the fault risk level set and the quantification mapping rule for matching to obtain the fault risk level corresponding to the degradation level. The collaborative execution module is used to formulate collaborative strategies and resource allocation schemes based on the fault risk level and to execute collaborative operations.

[0010] Furthermore, the data management module includes a data acquisition unit, a data storage unit, and a level determination unit; The data acquisition unit is used to acquire and process optical transmission quality data, environmental impact data and equipment operation data through an adaptive acquisition mechanism to obtain multi-dimensional raw data, which includes historical time-series data and real-time data. The data storage unit is used to store historical time-series data using a time-series database, supports an adaptive acquisition mechanism, and manages historical time-series data and real-time data. The degradation level determination unit is used to set a degradation level set, formulate degradation level determination rules based on the set degradation level set, and complete degradation level determination based on real-time data to obtain degradation level classification standards.

[0011] Furthermore, the model building module includes a feature processing unit, a model training unit, and a model update unit; The feature processing unit is used to extract time series segments from historical time series data, extract trend, fluctuation and correlation features, screen core features, and perform preprocessing to obtain a standardized set of core features. The model training unit is used to build a degradation trend prediction model, train the degradation trend prediction model with a standardized core feature set as input, and obtain the trained degradation trend prediction model. The model update unit is used to periodically incorporate newly collected real-time multi-dimensional raw data to update the parameters of the degradation trend prediction model, and to ensure that the degradation trend prediction model adapts to the dynamic changes in the network environment through an online incremental learning mechanism.

[0012] Furthermore, the risk prediction module includes a real-time prediction unit and a risk mapping unit; The real-time prediction unit is used to continuously access real-time multi-dimensional raw data and perform feature processing. The real-time multi-dimensional raw data with feature processing is input into the trained degradation trend prediction model to obtain the probability distribution of future degradation level and determine the predicted degradation level. The risk mapping unit is used to define a set of fault risk levels, formulate quantitative mapping rules, and match the predicted degradation level with the corresponding fault risk level to obtain the fault risk level corresponding to the degradation level.

[0013] Furthermore, the collaborative execution module includes a strategy formulation unit, a resource preparation unit, and an operation execution unit; The strategy formulation unit is used to formulate a differentiated collaborative operation strategy framework based on the fault risk level, and to clarify the collaborative logic corresponding to each risk level. The resource preparation unit is used to formulate a resource configuration scheme based on the collaborative operation strategy framework, and to preprocess and verify the availability of core resources. The operation execution unit is used to trigger the collaborative operation of OLP and WDM according to a preset time sequence based on the resource configuration scheme, monitor the operation status and business stability in real time, record the data of the whole process, and obtain the collaborative operation execution results and process data.

[0014] Compared with existing technologies, the present invention has the following advantages: This invention constructs a multi-dimensional data set system covering optical transmission quality, environmental impact, and equipment operation. It combines adaptive data acquisition with precise degradation level determination and relies on models such as long short-term memory networks to predict signal degradation trends in advance. Furthermore, by quantitatively mapping degradation levels to fault risk levels, it formulates differentiated WDM and OLP collaborative strategies and resource allocation schemes. This effectively overcomes the shortcomings of traditional solutions that lack collaborative linkage between the two and only passively respond to faults. It can intervene in the early stages of signal degradation, significantly improve the service continuity and stability of optical transmission networks, and reduce the risk of service interruption. Attached Figure Description

[0015] Figure 1 This is a flowchart of the method of the present invention. Figure 2 This is a system flowchart of the present invention. Detailed Implementation

[0016] like Figure 1 As shown, the present invention provides a technical solution: a data analysis-driven WDM and OLP co-optimization method, comprising: Step S1: Construct a data set system based on the monitoring requirements of optical transmission networks. Based on the data set system, adopt an adaptive acquisition mechanism to obtain multi-dimensional raw data. Combine the multi-dimensional raw data with business rules to define the degradation level and complete the degradation level judgment to obtain the corresponding degradation level of the sample. Step S2: Extract features from the multi-dimensional raw data and preprocess them to obtain a core feature set. Process the core feature set using a time-series data-specific filtering algorithm to obtain a standardized core feature set. Construct a degradation trend prediction model. Use the standardized core feature set as input and the corresponding degradation level of the sample as the label to train the degradation trend prediction model and obtain the trained degradation trend prediction model. Step S3: Input real-time multi-dimensional raw data into the trained degradation trend prediction model for processing to obtain the predicted degradation level, define the fault risk level set, formulate the quantization mapping rule, substitute the predicted degradation level into the fault risk level set and the quantization mapping rule for matching, and obtain the fault risk level corresponding to the degradation level. Step S4: Develop collaborative strategies and resource allocation plans based on the fault risk level and execute collaborative operations.

[0017] The specific process of step S1 is as follows: Based on the monitoring needs of optical transmission networks, a data set system is constructed, which includes: optical transmission quality data, environmental impact data, and equipment operation data. Among them, optical transmission quality data is the core indicator that directly reflects the signal transmission performance, and examples include optical power, optical signal-to-noise ratio (OSNR), bit error rate (BER), and cumulative dispersion value. Environmental impact data refers to external factors that affect signal stability, such as temperature, relative humidity, and vibration intensity in the area where the fiber optic link is located. The equipment operation data consists of the working status parameters of the wavelength division multiplexing equipment (WDM) and the optical fiber line automatic switching protection device (OLP), such as the operating current of the optical amplifier (EDFA), the operating status of the dynamic dispersion compensation module (DCM), and the connectivity of the OLP primary and backup links. An adaptive acquisition mechanism is set up to collect optical transmission quality data, environmental impact data, and equipment operation data (core optical transmission quality data acquisition frequency 5~10Hz, environmental impact data acquisition frequency 10Hz). Impact data and equipment operation data The system completes data access through a standardized interface compatible with mainstream optical transmission equipment protocols (such as SNMP and NETCONF), and finally obtains multi-dimensional raw data covering the entire optical transmission link (including historical time-series data and real-time data). The historical time-series data is synchronously stored in a time-series database (such as InfluxDB and TimescaleDB). Set a set of degradation levels Sort by impact level from lowest to highest; n represents the number of levels. Indicates the first There are several degradation levels; for example, when n=4... This is a minor degradation (no impact on business operations, only slight fluctuations in parameters). The degradation level is moderate (no obvious business anomalies, but parameters continue to decline). The system is severely degraded (service is approaching the threshold and there is a risk of interruption). This is a fatal degradation (business faces direct disruption). Based on the permitted scope of the business, a threshold value is preset for each type of optical transmission quality data (i.e., the safe range of values ​​for each parameter when the business is operating normally, sourced from industry standards (such as ITU-T specifications) and service level agreements (SLAs). For example: a backbone network business requires... , Optical power at Between; and then, combining the defined degree of deviation of a single type of data from the threshold, the superposition weight of multiple types of data anomalies, and the judgment rules for setting a set of degradation levels, the threshold, the defined degree of deviation of a single type of data from the threshold, the superposition weight of multiple types of data anomalies, and the set of degradation levels are integrated and processed together to obtain the degradation level classification standard. Among them, the degree of deviation of a single type of data is defined as "slight (deviation threshold 0~10%), moderate (deviation threshold 10%~30%), and severe (deviation threshold >30%)". Multiple types of data anomalies are weighted according to the degree of impact of the parameters on the service. For example, BER weight is 0.5, OSNR weight is 0.3, and optical power weight is 0.2. The weighted sum of BER weight 0.5, OSNR weight 0.3, and optical power weight 0.2 is calculated as Σ (the quantified value of the deviation of a certain optical transmission quality parameter (such as BER) × the corresponding weight). After obtaining the weighted sum, the corresponding degradation level is matched according to the interval in which the weighted sum is located. Based on real-time data on optical transmission quality and the degradation level classification standard, the corresponding degradation level of the sample is obtained.

[0018] The specific process of step S2 is as follows: Based on the time scale characteristics of optical signal degradation (gradual degradation of optical signals is mostly manifested as slow changes from minutes to hours, rather than instantaneous changes), the historical time window length T (T=20 minutes, which can be dynamically adjusted according to the actual network degradation rate) is configured, and the time segment within the corresponding window is extracted from the historical time series data. Based on the need to identify degradation trends, features are extracted from time-series segments. These features include trend features, fluctuation features, and correlation features, forming an initial feature set. Trend features include parameter change slope and cumulative offset (e.g., the total decrease in OSNR within 10 minutes); fluctuation features include parameter variance and peak fluctuation amplitude (e.g., the maximum fluctuation value of optical power); and correlation features include correlation coefficients between multiple parameters (e.g., the correlation between optical power and temperature). Based on feature importance assessment methods (such as analysis of variance (ANOVA), mutual information method, and random forest feature importance scoring), redundant features in the initial feature set are removed, and core features that are strongly correlated with the degradation trend are retained; thus, a core feature set is obtained. Preprocessing of the core feature set: Use time-series data-specific filtering algorithms (such as moving average filtering and Kalman filtering, used to filter instantaneous data fluctuations caused by equipment noise and electromagnetic interference) to remove noise from the core feature set, and obtain a denoised core feature set. Standardize or normalize the denoised core feature set to unify the dimensions, and handle invalid and abnormal data based on statistical criteria; thus obtaining a standardized core feature set. Based on the prediction requirements of time series data, a model architecture suitable for time series scenarios (such as Long Short-Term Memory Network (LSTM), Transformer, Gradient Boosting Tree (XGBoost / LightGBM) can be selected according to computing power and prediction accuracy requirements) to build a degradation trend prediction model. The standardized core feature set is used as the input to the degradation trend prediction model. The corresponding degradation level of the sample is used as the label. The degradation trend prediction model is iteratively trained using a classification loss function (such as cross-entropy loss function, FocalLoss, to solve the sample imbalance problem). The parameters of the degradation trend prediction model (such as learning rate, number of iterations, number of neurons in the network layer) are optimized until the prediction accuracy of the degradation trend prediction model meets the preset requirements. An online incremental learning mechanism is set up for the degradation trend prediction model. Newly collected standardized core feature sets are periodically incorporated to update the degradation trend prediction model parameters, ensuring that the model adapts to dynamic changes in the network environment (such as changes in degradation patterns caused by fiber optic aging and changes in service traffic). The trained degradation trend prediction model is used to output the probability distribution of each degradation level within a future preset time window.

[0019] The specific process of step S3 is as follows: Continuously input new real-time multi-dimensional raw data; perform feature extraction, filtering, and preprocessing on the new real-time multi-dimensional raw data (the processing flow is consistent with step 2, ensuring that the input data format is consistent with the model training data), and concatenate it with the core initial feature set within the historical time window to obtain the concatenation result; input the concatenation result into the trained degradation trend prediction model; The trained degradation trend prediction model outputs the probability distribution of each degradation level within a preset future time window; the degradation level corresponding to the highest probability in the probability distribution is selected as the predicted degradation level. Based on the impact of the predicted degradation level on service transmission quality, a set of fault risk levels corresponding one-to-one with the predicted degradation level is defined. Sort them from lowest to highest risk; for example, when n=4, correspond (Low risk, no emergency intervention required) correspond (Medium risk, parameter optimization required) correspond (High risk, protection switch required) correspond (Urgent risk, immediate switch and alarm required); Develop quantitative mapping rules to clarify the correlation logic between each predicted degradation level and risk level; such as "degradation level". →Risk Level A one-to-one mapping, or dynamically adjusted mapping relationships based on business importance, such as core businesses can be mapped... correspond ; The predicted degradation level is substituted into the set of failure risk levels and a quantitative mapping rule is established for matching to obtain the failure risk level corresponding to the degradation level.

[0020] The specific process of step S4 is as follows: Based on the fault risk level and the set of fault risk levels, and in conjunction with the requirements for business continuity and network stability, a differentiated collaborative operation strategy framework is formulated. Among them, the business continuity requirement in the strategy framework is to ensure uninterrupted business and service quality meets the standards (such as zero interruption of core services, compliant duration of non-core interruptions, and compliance with transmission SLAs). The network stability requirement in the strategy framework is to ensure stable network operation (such as preventing fault propagation, ensuring bandwidth / link availability meets the standards, and ensuring the topology has redundancy / load balancing capabilities). The strategy framework clearly defines the collaborative logic corresponding to various fault risk levels: low-order fault risk level Focusing on dynamic optimization of individual device parameters (such as fine-tuning EDFA output power and mild dispersion compensation); intermediate sequence fault risk level. Initiate deep optimization of the main equipment and pre-configuration of backup resources (such as precise dispersion compensation, switching to low-loss wavelengths, and simultaneously monitoring the status of backup links); high-order risk level. Perform primary / standby coordinated operation and smooth business migration (synchronously optimize WDM parameters and switch OLP links); highest priority risk level R n Prioritize service interruption prevention and alarm reporting (immediately trigger OLP switchover and simultaneously push alarm information to the operation and maintenance platform); where i and j represent positive integers. ; Low-sequence fault risk level range The upper limit level; The intermediate sequence fault risk level range The starting level; The intermediate sequence fault risk level range The upper limit level; High-order fault risk level range The starting level; High-order fault risk level range The upper limit level; By inputting the fault risk levels of different orders into the strategy framework and matching the corresponding level of operation strategy, an OLP and WDM collaborative operation strategy system adapted to the fault risk level is obtained. Based on the execution requirements of the collaborative operation strategy system and combined with the functional characteristics of optical transmission equipment, a resource configuration scheme is formulated. The resource configuration scheme includes: the parameter adjustment boundary for the main equipment optimization (such as EDFA power adjustment range ±2dB, DCM dispersion compensation step size 0.5ps / nm), the performance adaptation standard for backup resources (such as backup link OSNR≥18dB, BER≤1e-12), and the timing coordination rules for cross-device operations (such as starting OLP handover preparation within 1 second after WDM parameter adjustment is completed). Based on the resource allocation scheme, the parameter adjustment boundaries for the main device optimization, the performance adaptation standards for backup resources, and the timing coordination rules for cross-device operations are preprocessed (the purpose of preprocessing is to shorten the operation response time and ensure smooth switching, including: verifying the availability of execution resources by periodically sending detection data packets to confirm link connectivity; preloading operation parameters by writing the optimized WDM parameters into the backup device cache in advance; and ensuring the connectivity of cross-device coordination links by establishing a dedicated communication channel); thus obtaining a ready-to-go collaborative execution preparatory system. The preprocessing includes: verifying the availability of execution resources, preloading operation parameters, and ensuring the connectivity of cross-device collaborative links; Based on the collaborative operation strategy system, resource allocation scheme and collaborative execution preparation system, OLP and WDM collaborative operations are triggered according to the preset timing logic. During the collaborative operation of OLP and WDM, the operational status of the WDM equipment and the OLP automatic switching protection device of the fiber optic line (such as whether the DCM adjustment is in place and whether the OLP switching is successful), signal quality changes (such as OSNR and BER values ​​after switching), and service transmission stability data (such as service interruption duration and data packet loss rate) are collected in real time. Based on the collected operational status, signal quality changes, and service transmission stability data, a full record of the operation process is formed to obtain the collaborative operation execution results and process data. The full record data is used for subsequent optimization strategy iteration and provides fault tracing basis for operation and maintenance personnel. The results and process data of the collaborative operation are obtained, and the collaborative optimization response to the progressive degradation of the optical signal is completed.

[0021] like Figure 2 As shown, a data analysis-driven WDM and OLP collaborative optimization system is applied to the aforementioned data analysis-driven WDM and OLP collaborative optimization method. The system includes a data management module, a model building module, a risk prediction module, and a collaborative execution module. The data management module is used to construct a data set system based on the monitoring requirements of optical transmission networks. Based on the data set system, an adaptive acquisition mechanism is used to acquire multi-dimensional raw data. The multi-dimensional raw data is combined with business rules to define the degradation level and complete the degradation level determination to obtain the corresponding degradation level of the sample. The model building module is used to extract features from multi-dimensional raw data and preprocess them to obtain a core feature set. The core feature set is then processed using a time-series data-specific filtering algorithm to obtain a standardized core feature set. A degradation trend prediction model is then constructed. The standardized core feature set is used as input, and the corresponding degradation level of the sample is used as a label to train the degradation trend prediction model, resulting in the trained degradation trend prediction model. The risk prediction module is used to input real-time multi-dimensional raw data into the trained degradation trend prediction model for processing, obtain the predicted degradation level, define the fault risk level set, formulate the quantification mapping rule, and substitute the predicted degradation level into the fault risk level set and the quantification mapping rule for matching to obtain the fault risk level corresponding to the degradation level. The collaborative execution module is used to formulate collaborative strategies and resource allocation schemes based on the fault risk level and to execute collaborative operations.

[0022] The data management module includes a data acquisition unit, a data storage unit, and a level determination unit. The data acquisition unit is used to acquire and process optical transmission quality data, environmental impact data and equipment operation data through an adaptive acquisition mechanism to obtain multi-dimensional raw data, which includes historical time-series data and real-time data. The data storage unit is used to store historical time-series data using a time-series database, supports an adaptive acquisition mechanism, and manages historical time-series data and real-time data. The degradation level determination unit is used to set a degradation level set, formulate degradation level determination rules based on the set degradation level set, and complete degradation level determination based on real-time data to obtain degradation level classification standards.

[0023] The model building module includes a feature processing unit, a model training unit, and a model update unit. The feature processing unit is used to extract time series segments from historical time series data, extract trend, fluctuation and correlation features, screen core features, and perform preprocessing to obtain a standardized set of core features. The model training unit is used to build a degradation trend prediction model, train the degradation trend prediction model with a standardized core feature set as input, and obtain the trained degradation trend prediction model. The model update unit is used to periodically incorporate newly collected real-time multi-dimensional raw data to update the parameters of the degradation trend prediction model, and to ensure that the degradation trend prediction model adapts to the dynamic changes in the network environment through an online incremental learning mechanism.

[0024] The risk prediction module includes a real-time prediction unit and a risk mapping unit. The real-time prediction unit is used to continuously access real-time multi-dimensional raw data and perform feature processing. The real-time multi-dimensional raw data with feature processing is input into the trained degradation trend prediction model to obtain the probability distribution of future degradation level and determine the predicted degradation level. The risk mapping unit is used to define a set of fault risk levels, formulate quantitative mapping rules, and match the predicted degradation level with the corresponding fault risk level to obtain the fault risk level corresponding to the degradation level.

[0025] The collaborative execution module includes a strategy formulation unit, a resource preparation unit, and an operation execution unit. The strategy formulation unit is used to formulate a differentiated collaborative operation strategy framework based on the fault risk level, and to clarify the collaborative logic corresponding to each risk level. The resource preparation unit is used to formulate a resource configuration scheme based on the collaborative operation strategy framework, and to preprocess and verify the availability of core resources. The operation execution unit is used to trigger the collaborative operation of OLP and WDM according to a preset time sequence based on the resource configuration scheme, monitor the operation status and business stability in real time, record the data of the whole process, and obtain the collaborative operation execution results and process data.

[0026] In this embodiment, a cross-regional long-distance backbone network undertakes the transmission services of government and enterprise leased lines and Internet backbone in multiple locations. The optical signal is prone to gradual degradation due to the long fiber optic link and changes in ambient temperature. It is necessary to ensure service stability through WDM and OLP collaborative optimization. Step S1: First, construct a data set system and collect optical transmission quality data (optical power, optical signal-to-noise ratio (OSNR), bit error rate (BER), cumulative dispersion), environmental impact data (link area temperature and humidity), and equipment operation data (optical amplifier EDFA working status, dynamic dispersion compensation module (DCM) status, and OLP primary and backup link connectivity). An adaptive acquisition mechanism is used to collect optical transmission quality data, environmental impact data, and equipment operation data. Data access is completed through a standardized interface compatible with mainstream optical transmission equipment protocols, ultimately obtaining multi-dimensional raw data (including historical time-series data and real-time data) covering the entire optical transmission link. Historical data is synchronously stored in the time-series database.

[0027] Define the set of degradation levels Based on the permissible range of business operations, preset thresholds for each transmission quality data type are used, combined with defined weights for the deviation of single-type data and the superposition of anomalies in multiple types of data, and rules for determining the set of degradation levels are set (e.g., a small deviation of a single-type data type is considered acceptable). The moderate deviation of multiple data types is ), forming a standard for classifying deterioration levels; Based on real-time data on optical transmission quality and degradation level classification standards, the network signal of the current long-distance backbone network is determined to be in a certain state. (Moderate degradation); Step S2: Based on the configured historical time window, extract time series segments from historical time series data, and further extract trend features (such as the continuous downward trend of OSNR), fluctuation features (such as the amplitude of optical power fluctuation), and correlation features (such as the correlation between temperature and optical power), and filter the core features to form a set; The core features are filtered and denoised, their units are standardized, and outliers are processed to obtain a standardized set of core features. An LSTM model was selected to build a degradation trend prediction model. The model was trained using a standardized core feature set as input to obtain the trained degradation trend prediction model. Step S3: Continuously input real-time multi-dimensional raw data, and after feature extraction, filtering, and preprocessing, concatenate it with the core initial feature set within the historical time window to obtain the concatenated result; input the concatenated result into the trained degradation trend prediction model to obtain the predicted degradation level. ; Define the set of fault risk levels Establish a one-to-one mapping rule between "deterioration level" and "risk level". correspond ), will predict the degradation level After substituting the fault risk level set and defining the "deterioration level - risk level", the risk level is obtained. (High risk); Step S4: Based on risk level (High risk) Develop a collaborative operation strategy framework: WDM performs in-depth optimization (switching to low-loss wavelengths and precise dispersion compensation), and OLP performs predictive switching; Based on the resource allocation scheme, the parameter adjustment boundary of the main equipment optimization, the performance adaptation standard of the backup resources, and the timing coordination rules of cross-equipment operations are preprocessed to obtain a ready-to-go collaborative execution preparatory system. Based on the collaborative operation strategy system, resource allocation scheme and collaborative execution preparation system, OLP and WDM collaborative operations are triggered according to the preset timing logic. During the collaborative operation of OLP and WDM, real-time data on the operation status, signal quality changes, and service transmission stability of the WDM equipment and the automatic switching protection device for fiber optic lines are collected. Based on the collected data on operation status, signal quality changes, and service transmission stability, a full record of the operation process is formed, resulting in the collaborative operation execution results and process data. The full record data is used for subsequent optimization strategy iterations and provides maintenance personnel with a basis for fault tracing and deterioration response.

[0028] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A data analysis-driven co-optimization method for WDM and OLP, characterized in that, Includes the following steps: Step S1: Construct a data set system based on the monitoring requirements of optical transmission networks. Based on the data set system, adopt an adaptive acquisition mechanism to obtain multi-dimensional raw data. Combine the multi-dimensional raw data with business rules to define the degradation level and complete the degradation level judgment to obtain the corresponding degradation level of the sample. Step S2: Extract features from the multi-dimensional raw data and preprocess them to obtain a core feature set. Process the core feature set using a time-series data-specific filtering algorithm to obtain a standardized core feature set. Construct a degradation trend prediction model. Use the standardized core feature set as input and the corresponding degradation level of the sample as the label to train the degradation trend prediction model and obtain the trained degradation trend prediction model. Step S3: Input real-time multi-dimensional raw data into the trained degradation trend prediction model for processing to obtain the predicted degradation level, define the fault risk level set, formulate the quantization mapping rule, substitute the predicted degradation level into the fault risk level set and the quantization mapping rule for matching, and obtain the fault risk level corresponding to the degradation level. Step S4: Develop collaborative strategies and resource allocation plans based on the fault risk level and execute collaborative operations.

2. The data analysis-driven WDM and OLP co-optimization method according to claim 1, characterized in that: The specific process of step S1 is as follows: Based on the monitoring needs of optical transmission networks, a data collection system is constructed, which includes: optical transmission quality data, environmental impact data, and equipment operation data. The optical transmission quality data includes optical power, optical signal-to-noise ratio, bit error rate, and cumulative dispersion value. Environmental impact data includes temperature, relative humidity, and vibration intensity in the area where the fiber optic link is located; The equipment operation data consists of the working status parameters of the wavelength division multiplexing equipment and the automatic switching protection device for fiber optic lines; the equipment operation data includes the operating current of the optical amplifier, the operating status of the dynamic dispersion compensation module, and the connectivity of the primary and backup links. An adaptive acquisition mechanism is set up to collect optical transmission quality data, environmental impact data, and equipment operation data. Data access is completed through a standardized interface compatible with mainstream optical transmission equipment protocols, ultimately obtaining multi-dimensional raw data covering the entire optical transmission link. The multi-dimensional raw data includes historical time-series data and real-time data; among which, historical data is synchronously stored in a time-series database. Set a set of degradation levels Sort by impact level from lowest to highest; n represents the number of levels. Indicates the first There are several degradation levels; when n=4... For mild degradation, For moderate degradation, For severe degradation, This is a fatal degradation; Based on the permitted business scope, a threshold for each type of optical transmission quality data is preset. Then, combined with the defined degree of deviation of a single type of data from the threshold, the superposition weight of multiple types of data anomalies, and the judgment rules for setting a set of degradation levels, the threshold, the defined degree of deviation of a single type of data from the threshold, the superposition weight of multiple types of data anomalies, and the set of degradation levels are integrated and processed together to obtain the degradation level classification standard. Based on real-time data of optical transmission quality and the degradation level classification standard, the corresponding degradation level of the sample is obtained.

3. The data analysis-driven WDM and OLP co-optimization method according to claim 2, characterized in that: The specific process of step S2 is as follows: Configure the historical time window length as T, and extract the time series segments within the corresponding window from the historical time series data; Features are extracted from time series segments, including trend features, fluctuation features, and correlation features, to form an initial feature set. Redundant features in the initial feature set are removed based on the feature importance assessment method to obtain the core feature set; Preprocessing of the core feature set: The core feature set is removed by a time series data-specific filtering algorithm to obtain a denoised core feature set. The denoised core feature set is then standardized or normalized to unify the dimensions and obtain a standardized core feature set. A degradation trend prediction model was built based on long short-term memory networks; The standardized core feature set is used as the input to the degradation trend prediction model, and the corresponding degradation level of the sample is used as the label. The degradation trend prediction model is trained iteratively to obtain the trained degradation trend prediction model.

4. The data analysis-driven WDM and OLP co-optimization method according to claim 3, characterized in that: The specific process of step S3 is as follows: Continuously feed in new real-time multi-dimensional raw data; extract, filter, and preprocess the features of the new real-time multi-dimensional raw data, and concatenate it with the core initial feature set within the historical time window to obtain the concatenation result; input the concatenation result into the trained degradation trend prediction model; The trained degradation trend prediction model outputs the probability distribution of each degradation level within a future preset time window. The degradation level corresponding to the highest probability in the probability distribution is selected as the predicted degradation level; Define and predict the set of failure risk levels that correspond one-to-one with each degradation level. Sort by risk level from low to high; When n=4 correspond This represents low risk. correspond , representing medium risk correspond This represents high risk. correspond This represents an urgent risk; The predicted degradation level is substituted into the set of failure risk levels and a quantitative mapping rule is established for matching to obtain the failure risk level corresponding to the degradation level.

5. The data analysis-driven WDM and OLP co-optimization method according to claim 4, characterized in that: The specific process of step S4 is as follows: Based on the fault risk level, combined with the fault risk level set, and combined with the requirements of business continuity and network stability, a differentiated collaborative operation strategy framework is formulated. By inputting the fault risk levels of different orders into the strategy framework and matching the corresponding level of operation strategy, an OLP and WDM collaborative operation strategy system adapted to the fault risk level is obtained. Based on the execution requirements of the collaborative operation strategy system and combined with the functional characteristics of optical transmission equipment, a resource allocation scheme is formulated. Based on the resource allocation scheme, the parameter adjustment boundary of the main equipment optimization, the performance adaptation standard of the backup resources, and the timing coordination rules of cross-equipment operation are preprocessed to obtain a ready-to-go collaborative execution preparatory system. Based on the collaborative operation strategy system, resource allocation scheme and collaborative execution preparation system, OLP and WDM collaborative operations are triggered according to the preset timing logic. During the collaborative operation of OLP and WDM, real-time data on the operation status, signal quality changes, and service transmission stability of the wavelength division multiplexing equipment and the automatic switching protection device for fiber optic lines are collected. Based on the collected data on operation status, signal quality changes, and service transmission stability, a complete record of the operation process is formed, resulting in the collaborative operation execution results and process data.

6. A data analysis-driven WDM and OLP co-optimization system, applied to the data analysis-driven WDM and OLP co-optimization method according to any one of claims 1-5, characterized in that, It includes a data management module, a model building module, a risk prediction module, and a collaborative execution module; The data management module is used to build a data set system based on the monitoring requirements of optical transmission networks. Based on the data set system, an adaptive acquisition mechanism is used to acquire multi-dimensional raw data. The multi-dimensional raw data is combined with business rules to define the degradation level and complete the degradation level judgment to obtain the corresponding degradation level of the sample. The model building module is used to extract features from multi-dimensional raw data and preprocess them to obtain a core feature set. The core feature set is then processed using a time-series data-specific filtering algorithm to obtain a standardized core feature set. A degradation trend prediction model is then constructed. The standardized core feature set is used as input, and the corresponding degradation level of the sample is used as a label to train the degradation trend prediction model, resulting in the trained degradation trend prediction model. The risk prediction module is used to input real-time multi-dimensional raw data into the trained degradation trend prediction model for processing, obtain the predicted degradation level, define the fault risk level set, formulate the quantification mapping rule, and substitute the predicted degradation level into the fault risk level set and the quantification mapping rule for matching to obtain the fault risk level corresponding to the degradation level. The collaborative execution module is used to formulate collaborative strategies and resource allocation schemes based on the fault risk level and to execute collaborative operations.

7. The data analysis-driven WDM and OLP collaborative optimization system according to claim 6, characterized in that: The data management module includes a data acquisition unit, a data storage unit, and a level determination unit; The data acquisition unit is used to acquire and process optical transmission quality data, environmental impact data and equipment operation data through an adaptive acquisition mechanism to obtain multi-dimensional raw data, which includes historical time-series data and real-time data. The data storage unit is used to store historical time-series data using a time-series database, supports an adaptive acquisition mechanism, and manages historical time-series data and real-time data. The degradation level determination unit is used to set a degradation level set, formulate degradation level determination rules based on the set degradation level set, and complete degradation level determination based on real-time data to obtain degradation level classification standards.

8. The data analysis-driven WDM and OLP collaborative optimization system according to claim 7, characterized in that: The model building module includes a feature processing unit, a model training unit, and a model update unit; The feature processing unit is used to extract time series segments from historical time series data, extract trend, fluctuation and correlation features, screen core features, and perform preprocessing to obtain a standardized set of core features. The model training unit is used to build a degradation trend prediction model, train the degradation trend prediction model with a standardized core feature set as input, and obtain the trained degradation trend prediction model. The model update unit is used to periodically incorporate newly collected real-time multi-dimensional raw data to update the parameters of the degradation trend prediction model, and to ensure that the degradation trend prediction model adapts to the dynamic changes in the network environment through an online incremental learning mechanism.

9. A data analysis-driven WDM and OLP collaborative optimization system according to claim 8, characterized in that: The risk prediction module includes a real-time prediction unit and a risk mapping unit; The real-time prediction unit is used to continuously access real-time multi-dimensional raw data and perform feature processing. The real-time multi-dimensional raw data with feature processing is input into the trained degradation trend prediction model to obtain the probability distribution of future degradation level and determine the predicted degradation level. The risk mapping unit is used to define a set of fault risk levels, formulate quantitative mapping rules, and match the predicted degradation level with the corresponding fault risk level to obtain the fault risk level corresponding to the degradation level.

10. The data analysis-driven WDM and OLP collaborative optimization system according to claim 9, characterized in that: The collaborative execution module includes a strategy formulation unit, a resource preparation unit, and an operation execution unit; The strategy formulation unit is used to formulate a differentiated collaborative operation strategy framework based on the fault risk level, and to clarify the collaborative logic corresponding to each risk level. The resource preparation unit is used to formulate a resource configuration scheme based on the collaborative operation strategy framework, and to preprocess and verify the availability of core resources. The operation execution unit is used to trigger the collaborative operation of OLP and WDM according to a preset time sequence based on the resource configuration scheme, monitor the operation status and business stability in real time, record the data of the whole process, and obtain the collaborative operation execution results and process data.

Images