Gain regulation and control precision maintenance method for programmable Raman amplifier

Abstract

The embodiment of the invention provides a gain regulation and control precision maintenance method for a programmable Raman amplifier. The method comprises the following steps: acquiring basic scene sample data and target scene sample data to form a source domain sample data set and a target domain sample data set; selecting a proper machine learning method as a basic learning algorithm according to the characteristics of the data in the data set; initializing weights occupied by a source domain sample and a target domain sample; fitting a prediction model by adopting a basic learning algorithm according to the weights of the source domain sample and the target domain sample; re-weighting the source domain sample and the target domain sample according to the total weighting error of the target domain sample so as to correct the difference between the distribution of the source domain sample and the distribution of the target domain sample; and obtaining a model after multiple iterations as a pumping prediction model of the programmable Raman amplifier. When the use scene of the programmable Raman amplifier changes and the accuracy of the regulation and control algorithm of the programmable Raman amplifier is reduced, by means of the method, maintenance of the gain regulation and control precision of the programmable Raman amplifier can be achieved by collecting a small amount of sample data, and the use scene of the programmable Raman amplifier is effectively expanded.

Description

Technical Field

[0001] This invention relates to the field of optical amplifier technology, and in particular to a method for maintaining the accuracy of gain control in a programmable Raman amplifier. Background Technology

[0002] Fiber Raman amplifiers are optical amplifiers based on nonlinear optical effects. Compared with other types of optical amplifiers, they have a very low noise figure, ensuring high-quality signal transmission. More importantly, fiber Raman amplifiers allow for flexible gain profile design by adjusting pump power and wavelength, and provide wide-spectrum gain availability when operating in multi-pump configurations. This makes them ideal for achieving any desired gain profile in optical networks in a controllable manner.

[0003] Because programmable Raman amplifiers (PRA) have highly reconfigurable operating scenarios, machine learning methods are typically used to ensure real-time control. However, the performance of the prediction models trained by machine learning is highly dependent on their training data. Therefore, when the application scenario of the PRA changes—for example, changes in the type or length of the amplifying fiber, changes in port connections leading to alterations in polarization effects, or changes in the total input power—the Raman gain curve corresponding to the same Raman pump settings will change accordingly. This results in a sharp drop in the accuracy of existing prediction models, potentially rendering them unusable. Therefore, researching methods for maintaining the accuracy of gain control in programmable Raman amplifiers is of significant practical importance. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a method for maintaining the gain control accuracy of a programmable Raman amplifier, in order to solve the problem of decreased control accuracy of the programmable Raman amplifier when the usage scenario changes, improve the adaptive capability of the gain control of the programmable Raman amplifier to changes in the scenario, and thus expand the application range of the programmable Raman amplifier.

[0005] The technical solution of the present invention is as follows:

[0006] A method for maintaining the gain control accuracy of a programmable Raman amplifier, the method comprising:

[0007] The source domain sample dataset and the target domain sample dataset are constructed by acquiring basic scene sample data and target scene sample data;

[0008] Choose an appropriate machine learning method as the basic learning algorithm based on the features of the data in the dataset;

[0009] Initialize the weights of the source domain samples and the target domain samples;

[0010] Based on the weights of the source domain samples and the target domain samples, a basic learning algorithm is used to fit the prediction model;

[0011] The source domain samples and target domain samples are reweighted based on the total weighted error of the target domain samples to correct the difference between the distributions of the source domain samples and the target domain samples.

[0012] The model obtained after multiple iterations is used as the pump prediction model for the programmable Raman amplifier to solve the problem of deterioration in the gain control accuracy of the programmable Raman amplifier caused by changes in the scene.

[0013] Optionally, the basic scenario is the scenario of establishing a programmable Raman amplifier gain control model, which requires the collection of a large number of samples; the target scenario is the scenario where the gain control accuracy of the programmable Raman amplifier needs to be maintained due to changes in relevant parameters, which only requires the collection of a small amount of data.

[0014] Optionally, the source domain sample dataset and the target domain sample dataset can be obtained through mathematical operations, simulation platforms, or real devices.

[0015] Optionally, the source domain sample dataset and the target domain sample dataset can be filled by mathematical solution methods, such as solving nonlinear differential equations or Matlab operations, to construct a special gain spectrum, thereby improving the generation effect of the special gain by the method.

[0016] Optionally, the source domain sample dataset consists of programmable Raman amplifier pump setting parameters under the basic scenario and their corresponding generated gain curves; the target domain sample dataset consists of programmable Raman amplifier pump setting parameters under the target scenario and their corresponding generated gain curves. The pump setting parameters include: the power and wavelength of each pump.

[0017] Optionally, the machine learning method is an algorithm for establishing a programmable Raman amplifier gain control model in a basic scenario. The selectable range includes neural network model construction algorithms capable of solving multi-input multi-output regression problems, including but not limited to Extreme Learning Machine, Random Forest, and Multilayer Perceptron. The model trained by the machine learning is a pump prediction model, reflecting the mapping relationship between the generated gain curve and the pump setting parameters.

[0018] Optionally, the gain curve generated by the programmable Raman amplifier represents the switching gain of the programmable Raman amplifier, which is obtained by calculating the power difference between the output spectra when the Raman amplifier pumps the laser on and off.

[0019] Optionally, the calculation of the total weighted error of the target domain samples depends on the calculation and normalization of the error vectors of the source domain and target domain sample data.

[0020] Optionally, the reweighting method continuously reduces the weight of source domain sample data with poor prediction performance and increases the weight of target domain sample data in multiple iterations, thereby enabling the trained model to maintain accuracy in the target scene.

[0021] As can be seen from the above technical solutions, this application has the following effects:

[0022] Based on a large amount of sample data collected in basic scenarios and the programmable Raman amplifier gain control accuracy maintenance method proposed in this application, when the application scenario of the programmable Raman amplifier changes, the control algorithm can be quickly adjusted by collecting a small number of target domain samples, so that the prediction model fits the current application scenario of the programmable Raman amplifier, which greatly improves the practicality and generalization of the programmable Raman amplifier.

[0023] Additional advantages, objects, and features of the invention will be set forth in part in the description which follows, and will also become apparent in part to those skilled in the art upon studying the following text, or may be learned by practice of the invention. The objects and other advantages of the invention can be realized and obtained through the structures specifically pointed out in the written description, its claims, and the accompanying drawings.

[0024] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0025] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, are used together with the embodiments of the invention to explain the invention, but do not constitute a limitation thereof. In the drawings:

[0026] Figure 1 This is a flowchart of a method for maintaining the gain control accuracy of a programmable Raman amplifier according to an embodiment of the present invention;

[0027] Figure 2 This is a schematic diagram of the sample data acquisition experimental platform construction according to an embodiment of the present invention;

[0028] Figure 3 This is a schematic diagram of the structure of a programmable wide-spectrum Raman amplifier according to an embodiment of the present invention. Detailed Implementation

[0029] 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 embodiments and accompanying drawings. Here, the illustrative embodiments and descriptions of this invention are used to explain the invention, but are not intended to limit the invention.

[0030] It should also be noted that, in order to avoid obscuring the invention with unnecessary details, only the structures and / or processing steps closely related to the solution according to the invention are shown in the accompanying drawings, while other details that are not closely related to the invention are omitted.

[0031] It should be emphasized that the term "including / comprises" as used herein refers to the presence of a feature, element, step, or component, but does not exclude the presence or addition of one or more other features, elements, steps, or components.

[0032] It should also be noted that, unless otherwise specified, the term "connection" in this article can refer not only to a direct connection, but also to an indirect connection involving an intermediary.

[0033] Fiber Raman amplifiers are based on stimulated Raman scattering (SRS). The pump light interacts with the transmitted signal within the fiber, transferring energy and amplifying the signal. The gain provided by the fiber Raman amplifier is primarily determined by the power and wavelength of the pump light. Adjusting these parameters allows for flexible profile design, providing arbitrary gains. Programmable Raman amplifiers incorporate a programmable control module with an embedded pump prediction model into the basic Raman amplifier structure. This enables accurate prediction of corresponding pump settings based on the user-input target gain curve. However, the high correlation between the model training effect and the training data means that the prediction accuracy of the embedded pump prediction model in the programmable Raman amplifier is affected by factors such as the type and length of the amplifying fiber, and the splicing quality. Therefore, ensuring the prediction accuracy of the pump prediction model in different scenarios is a crucial issue that must be addressed.

[0034] This application provides a method for maintaining the gain control accuracy of a programmable Raman amplifier. The specific process of this method is as follows: Figure 1 As shown, it includes steps S101 to S106.

[0035] It should be clarified that steps S101 to S106 described in this embodiment are not a limitation on the order of the steps. They should be understood as the steps being interchangeable or running in parallel under specific application scenarios.

[0036] S101: Obtain basic scene sample data and target scene sample data to form source domain sample dataset and target domain sample dataset.

[0037] In this example, sample data is obtained by building an experimental platform, such as... Figure 2 As shown, the experimental platform includes a programmable Raman amplifier, a signal light source, an input detector, an output detector, a filter, and a transmission optical fiber. The signal light source is used to output the desired signal light, covering a wavelength range of λ. L ~λ HAn input detector is used to monitor the power value of the input signal; an output monitor is used to monitor the power value of the output signal; a filter is used to filter the pump light; and a transmission fiber is used to transmit the optical signal and perform Raman amplification. The fiber type is x and the fiber length is l.

[0038] Specifically, the programmable Raman amplifier includes an optical amplification module, a programmable control module, and various types of external interfaces. Figure 3 The diagram shows its specific structure. The optical amplification module includes: n (n≥1) pump lasers, a wavelength division multiplexer, an optical isolator, and optical fiber. The wavelengths of the N pump lasers are represented as: λ1, λ2, ..., λ... n The power is represented as: p1, p2, ..., p n Its wavelength and power are adjustable, with a wavelength adjustment range of λ. L ~λ H The adjustment accuracy is λ I The power adjustment range is p L ~p H Adjustment precision is p I The programmable control module has an embedded pump prediction model, error compensation program, and internal device control program; external interfaces include, but are not limited to, USB and HDMI interfaces; the overall insertion loss of this programmable Raman amplifier is L.

[0039] In the above case, the total power of the input signal is set to P. in A basic environment was constructed for collecting source domain sample data. A programmable control module was used to control the amplifier pump parameters, and an output spectrum was acquired using a spectrometer. The switching gain was obtained by calculating the power difference between the output spectra when the programmable Raman amplifier pumped the laser on and off. Finally, N sets (N≥1000) of sample data containing programmable Raman amplifier pump settings and their corresponding Raman switching gains were obtained, constituting the source domain sample dataset.

[0040] The polarization effect is altered by changing the port connection of the programmable Raman amplifier, thereby constructing the target environment. Sample data of M groups (M≤200) of programmable Raman amplifier pump setting parameters and their corresponding Raman switching gains are collected in the same way to form the target domain sample dataset.

[0041] S102: Select an appropriate machine learning method as the basic learning algorithm based on the features of the data in the dataset.

[0042] Specifically, this example selects a multilayer perceptron (MLP) as the basic learning algorithm to learn the mapping relationship from the Raman gain curve to the pump setting parameters. This algorithm is a type of artificial neural network with a tendency structure that can be used to solve multi-input multi-output regression problems.

[0043] S103: Initialize the weights of the source domain samples and the target domain samples.

[0044] Specifically, the following needs to be met:

[0045] ∑ω S +∑ω T =1

[0046] Where, ω S ω represents the source domain weight. T This represents the target domain weight.

[0047] S104: Based on the weights of the source domain samples and the target domain samples, a basic learning algorithm is used to fit the prediction model.

[0048] Specifically, the source and target data (X) are labeled according to the respective weights of the source and target domains. s ,y s (X) T ,y T The fitted estimate f.

[0049] S105: Reweight the source domain samples and target domain samples based on the total weighted error of the target domain samples to correct the difference between the distributions of the source domain samples and the target domain samples.

[0050] S106: Obtain the model after multiple iterations as the pump prediction model for the programmable Raman amplifier to solve the problem that the accuracy of the prediction results cannot meet the usage requirements due to changes in the scene.

[0051] In practical use, a large amount of source domain sample data can be permanently stored after only one acquisition. When the application scenario of the programmable Raman amplifier changes and needs to be adapted to the current scenario, only a small amount of target domain sample data needs to be acquired in the current environment. By executing the programmable Raman amplifier gain control accuracy maintenance method, the accuracy of the programmable Raman amplifier pump prediction function can be guaranteed, which greatly improves the feasibility of the programmable Raman amplifier and expands its usability.

[0052] Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention.

[0053] In this invention, features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, and / or combined with or in place of features of other embodiments.

[0054] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, various modifications and variations can be made to the embodiments of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for maintaining the gain control accuracy of a programmable Raman amplifier, characterized in that, include: The source domain sample dataset and the target domain sample dataset are constructed by acquiring basic scene sample data and target scene sample data; Choose an appropriate machine learning method as the basic learning algorithm based on the features of the data in the dataset; Initialize the weights of the source domain samples and the target domain samples; Based on the weights of the source domain samples and the target domain samples, a basic learning algorithm is used to fit the prediction model; The source domain samples and target domain samples are reweighted based on the total weighted error of the target domain samples to correct the difference between the distributions of the source domain samples and the target domain samples. The model obtained after multiple iterations is used as the pump prediction model for the programmable Raman amplifier.

2. The method for maintaining the gain control accuracy of a programmable Raman amplifier according to claim 1, characterized in that, The basic scenario is the scenario of establishing a programmable Raman amplifier gain control model, which requires the collection of a large number of samples; the target scenario is the scenario where the gain control accuracy of the programmable Raman amplifier needs to be maintained due to changes in relevant parameters, which only requires the collection of a small amount of data.

3. The method for maintaining the gain control accuracy of a programmable Raman amplifier according to claim 1, characterized in that, The source domain sample dataset and target domain sample dataset are obtained through mathematical operations, simulation platforms, or real devices.

4. The method for maintaining the gain control accuracy of a programmable Raman amplifier according to claim 1, characterized in that, The source domain sample dataset consists of the programmable Raman amplifier pump setting parameters under the basic scenario and the corresponding generated gain curves; the target domain sample dataset consists of the programmable Raman amplifier pump setting parameters under the target scenario and the corresponding generated gain curves. The pump setting parameters include: the power and wavelength of each pump.

5. The method for maintaining the gain control accuracy of a programmable Raman amplifier according to claim 1, characterized in that, The machine learning method is an algorithm for establishing a programmable Raman amplifier gain control model in a basic scenario. The range of options includes neural network model construction algorithms that can be used to solve multi-input multi-output regression problems, including but not limited to Extreme Learning Machine, Random Forest, and Multilayer Perceptron.

6. The method for maintaining the gain control accuracy of a programmable Raman amplifier according to claim 5, characterized in that, The model obtained through machine learning training is a pump prediction model, which reflects the mapping relationship between the generated gain curve and the pump setting parameters.

7. The method for maintaining the gain control accuracy of a programmable Raman amplifier according to claim 1, characterized in that, The reweighting method continuously reduces the weight of source domain sample data with poor prediction performance and increases the weight of target domain sample data in multiple iterations, thereby enabling the trained model to maintain accuracy in the target scene.

Images