Multi-stage intelligent control method for driving optical fiber amplifier

Abstract

The invention relates to a multi-stage intelligent control method for driving an optical fiber amplifier, belongs to the technical field of optical fiber amplifiers, solves the problem of insufficient output power control precision of an optical fiber amplifier in the prior art, and comprises the following steps: S1, building a pumping source system, building an optical fiber amplifier system with multi-stage pumping sources, and setting the multi-stage pumping sources, the input seed light and an input signal of the multi-stage pumping source are amplified through multi-stage optical fibers to obtain output power; s2, building a mathematical model of a pumping source system, including building a mathematical model of pumping power of a pumping source and a mathematical model of system output power, and setting initialized input parameters; s3, based on the constructed mathematical model of the pumping source system, executing a control algorithm iteration process to obtain the optimal driving current and the maximum power of the pumping source; and S4, outputting the optimal driving current and the maximum power of the pumping source to the pumping source system, and executing multi-stage intelligent control on the pumping source system.

Description

Technical Field

[0001] The present invention relates to the technical field of optical fiber amplifiers, and in particular to a multi-level intelligent control method for driving an optical fiber amplifier. Background Art

[0002] With the rapid development of optical communication technology, fiber amplifiers have become an indispensable and critical component of modern communication networks, especially for long-distance, high-capacity data transmission. The primary function of a fiber amplifier is to amplify signal light and overcome signal attenuation caused by fiber losses during transmission. Maintaining constant output power is crucial in many applications, as fluctuations in output power can directly affect signal quality and even cause system failure or performance degradation. Therefore, maintaining output power stability and adapting to various variations in the actual operating environment is a key issue in the design and application of fiber amplifiers.

[0003] The performance of fiber amplifiers is affected by many factors, with temperature fluctuations having a particularly significant impact on the pump source and optical components. As ambient temperature fluctuates, the power output of the pump source and the performance of the optical components in the system will change, affecting the output power and overall efficiency of the fiber amplifier system.

[0004] Compared to single-stage fiber amplifiers, cascaded fiber amplifiers offer higher output power and lower noise figures and other parameters, making them widely used. However, their output power control is more complex than that of a single-stage amplifier. Therefore, two major challenges currently facing cascaded fiber amplifier systems are how to precisely control the output power of cascaded fiber amplifiers to meet system accuracy requirements; and how to further coordinate and optimize the driving current of the multi-stage pump source while maintaining a constant output power in the cascaded fiber amplifier system to improve system efficiency.

[0005] However, existing methods for fiber amplifier power control suffer from the following shortcomings: 1) Existing technologies only control the output power of single-stage fiber amplifiers, but multi-stage fiber amplifiers are widely used in practice, and precise control of their output power is still difficult to achieve. 2) They only emphasize the accuracy of fiber amplifier output power without comprehensively considering system efficiency, failing to fully exploit the potential for optimizing the drive current of the system's multi-stage pump sources. Summary of the Invention

[0006] To overcome the limitations of cascaded fiber amplifier output power control caused by the aforementioned issues, the present invention provides a multi-stage intelligent control method for fiber amplifier driving. Specifically, taking into account the need for constant output power and optimal system efficiency, a mathematical model of the pump sources at each stage and the cascaded fiber amplifier system is established based on existing data on the relationship between pump source output power, drive current, and temperature, as well as data on the relationship between pump source drive current and system output power. This method then proposes a multi-stage intelligent coordinated control method for fiber amplifiers based on output power sampling, achieving precise control of the cascaded fiber amplifier output power and optimizing system efficiency.

[0007] According to an embodiment of the present invention, a multi-stage intelligent control method for driving a fiber amplifier is provided. The output power is sampled and, through a multi-stage coordinated control algorithm, multiple currents are output to drive a multi-stage pump source, thereby maintaining a constant output power while optimizing system efficiency.

[0008] According to an embodiment of the present invention, a multi-level intelligent control method for driving an optical fiber amplifier is provided, comprising:

[0009] Step S1, building a pump source system, building a multi-stage fiber amplifier system with a multi-stage pump source, setting up the multi-stage pump source, and amplifying the input seed light and the input signal of the multi-stage pump source through the multi-stage fiber amplifier to obtain output power;

[0010] Step S2, constructing a mathematical model of the pump source system, including constructing a mathematical model of the pump power of the pump source and a mathematical model of the system output power, and setting initialization input parameters, including setting an initial ambient temperature, a target ambient temperature, and a target output power;

[0011] Step S3: Based on the constructed mathematical model of the pump source system, when the ambient temperature changes from the initial ambient temperature to the target ambient temperature, the control algorithm is iteratively executed with the system output power maintained at the preset target output power as a constraint condition and the system efficiency maximized as the optimization goal to obtain the optimal drive current and maximum efficiency of the system under the target ambient temperature and the preset target output power conditions;

[0012] Step S4: obtaining the optimal driving current and maximum efficiency of the pump source system, and performing multi-level intelligent control of the pump source system.

[0013] Optionally, step S2 specifically includes:

[0014] Step S2.1, establish a mathematical model of the pump source system and construct a mathematical model of the pump power of the pump source:

[0015] Ppump i =g i (I i ,t)

[0016] Among them, Ppump i represents the pump power of the i-th pump source, g i (·) represents the mathematical model of the output power, driving current and ambient temperature of the i-th stage pump source, I i represents the driving current of the i-th pump source, i represents the number of pump sources and i=1,2,…n, n represents the total number of pump sources, and t represents the ambient temperature;

[0017] The mathematical model for constructing the system output power is:

[0018] Pout=f(I1,I2,…,I i ...,I n ,t)

[0019] Where Pout represents the system output power, and f(·) represents the mathematical model of the system output power, the driving current of each level of pump source, and the ambient temperature.

[0020] Step S2.2, initialization of input parameters includes setting the initial driving current value, target ambient temperature and initial ambient temperature; calculating the initial output power based on the mathematical model of the system output power; and setting the target output power and control algorithm iteration parameters.

[0021] Optionally, in step S2.2:

[0022] Set the initial driving current value of each pump source as the initial driving current value:

[0023] I init =(I 1_init ,I 2_init ,…,I i_init ...,I n_init )

[0024] Among them, I i_init is the initial value of the driving current of the i-th stage pump source.

[0025] Optionally, in step S2.2:

[0026] The initial output power calculated based on the mathematical model of the system output power is:

[0027] P ou t _ i n iti al =f(I 1_ i n i t ,I 2_ i n it,…,I i_i n it...,I n_ i n i t ,t1)

[0028] Where t1 is the initial ambient temperature.

[0029] Optionally, in step S2.2, setting parameters of the control algorithm iteration includes: setting output power convergence tolerance, efficiency convergence tolerance, gradient norm convergence tolerance, learning rate and maximum number of iterations.

[0030] Optionally, step S3 specifically includes:

[0031] Step S3.1: Based on the current driving current of the pump source and the target ambient temperature, the constructed mathematical model of the system output power and the mathematical model of the pump power of the pump source are substituted to calculate the system output power and the pump power of each level of the pump source;

[0032] Step S3.2, obtaining a system efficiency based on the obtained system output power and the pump powers of the pump sources at each stage, wherein the system efficiency is measured by the ratio of the system output power to the sum of the pump powers of the pump sources at each stage;

[0033] Step S3.3: With the system output power maintained at the preset target output power as the constraint condition and the system efficiency maximized as the optimization goal, a Lagrangian function is constructed and gradient calculation is performed to obtain the gradients of the system efficiency and system output power constraints with respect to the current drive currents of the various pump sources, thereby obtaining the gradients of the Lagrangian function with respect to the current drive currents of the various pump sources;

[0034] Step S3.4, using the obtained gradient of the system efficiency and the system output power constraint term as the gradient of the Lagrangian function, and in combination with a preset learning rate, updating the drive current using a gradient descent method to obtain an optimized drive current, and obtaining an optimized system efficiency based on the optimized drive current;

[0035] Step S3.5, convergence judgment, when the convergence conditions are met, the currently obtained optimized driving current and optimized system efficiency are used as the optimal driving current and maximum efficiency; otherwise, it is determined that the system has not converged and returns to step S3.1, and the obtained optimized driving current is used as the current driving current input for the next round of iteration to execute the next round of iteration.

[0036] Optionally, in step S3.1: the current driving current of the pump source is the optimized driving current obtained in the previous round of iteration; and for the first round of iteration, the current driving current of the pump source is the initial driving current value.

[0037] Optionally, step S3.5 specifically includes:

[0038] If the current number of iterations is equal to or greater than the preset maximum number of iterations, the currently obtained optimized driving current and optimized system efficiency are used as the optimal driving current and maximum efficiency, and a warning is issued to remind that the maximum number of iterations has been exceeded;

[0039] If the current number of iterations is less than the preset maximum number of iterations and the following three convergence conditions are met, the system is determined to have converged, and the currently obtained optimized driving current and optimized system efficiency are used as the optimal driving current and maximum efficiency: 1) the deviation between the current system output power and the preset target output power is less than the preset output power tolerance; 2) the degree of change in system efficiency is within the preset efficiency convergence tolerance; and 3) the current gradient norm is within the preset gradient norm convergence tolerance;

[0040] Otherwise, it is determined that the system has not converged, and the process returns to step S3.1, where the obtained optimized driving current is used as the current driving current input for the next round of iteration, and the next round of iteration is performed.

[0041] Optionally, the degree of change in system efficiency is the difference between the optimized system efficiency obtained in the current iteration round obtained in step S3.4 and the system efficiency obtained in step S3.2, and the difference is compared with a preset efficiency convergence tolerance.

[0042] Compared with the prior art, a multi-stage intelligent control method for optical fiber amplifier driving provided in accordance with an embodiment of the present invention has at least the following advantages.

[0043] 1) This technology overcomes the limitations of traditional power control technology, which is limited to single-stage fiber amplifiers. Compared to single-stage amplifiers, cascaded amplifiers offer greater gain while also exhibiting a lower noise figure. Consequently, they can be widely applied to more complex optical communication systems and other scenarios requiring efficient amplification, meeting the power stability requirements of multi-stage systems.

[0044] 2) Utilizing sophisticated modeling and numerical calculations, the model more accurately describes the relationship between pump source output power, system output power, temperature, and pump source drive current. By accounting for the impact of temperature changes on system performance, as well as the effect of temperature on other environmental factors, the model can reflect real-time state changes in actual operation.

[0045] 3) By comprehensively considering the system's output power and pump source power during the control process, a control scheme with optimal efficiency is provided. This not only ensures output power stability but also improves overall system efficiency by optimizing the pump source drive current. This system efficiency-oriented optimization strategy significantly reduces energy waste and improves equipment operational efficiency, providing reliable technical support for energy conservation and cost control in practical applications.

[0046] 4) The algorithm used has wide applicability and can be modeled and optimized based on experimental data. An accurate model can be constructed simply by experimentally obtaining the operating characteristics of the pump source and experimental data on the system output power, pump source drive current, and temperature. Based on this data, the optimization algorithm can dynamically adjust the pump source current during actual operation to achieve optimal efficiency control. This makes the algorithm applicable not only to erbium-doped fiber amplifiers but also to other types of fiber amplifiers or similar systems, as long as the systems have similar input-output relationships and dependence on temperature or other factors. This flexibility and adaptability give the present invention broad application prospects in a variety of optical amplification applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. By referring to the drawings, the features and advantages of the present invention can be more clearly understood. The drawings are schematic and should not be understood as limiting the present invention in any way. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0048] Figure 1 The present invention provides a flowchart of a multi-stage intelligent control method for driving an optical fiber amplifier according to an embodiment of the present invention.

[0049] Figure 2 4 is a flow chart of a multi-level intelligent coordinated control algorithm for a multi-level intelligent control method for optical fiber amplifier driving according to an embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present invention and the features therein can be combined with each other without conflict.

[0051] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0052] A multi-stage intelligent control method for driving an optical fiber amplifier according to an embodiment of the present invention is described in detail below with reference to the accompanying drawings. The optical fiber amplifier may be a cascaded optical fiber amplifier, and optionally, for example, may be a cascaded erbium-doped optical fiber amplifier.

[0053] like Figure 1 and Figure 2As shown, a multi-level intelligent control method for optical fiber amplifier driving provided according to a first embodiment of the present invention includes the following steps.

[0054] The first step is to build the pump source and system mathematical model.

[0055] Temperature affects the pump source and various optical components of a fiber amplifier, causing fluctuations in the pump source output power and the system output power. For cascaded erbium-doped fiber amplifiers (EDFAs) used in practical engineering applications, experimental data on the pump source output power versus drive current and temperature is collected. Experimental data on the system output power versus drive current and temperature at each pump source stage is also collected. This data, along with various algorithms, is then used to build laser and system models. Because the output power of a cascaded EDFA system is affected by a combination of factors, temperature alone cannot theoretically be considered the sole influencing variable. However, temperature is often the primary factor affecting system performance, as temperature changes directly impact the performance of several key components, such as the pump source output power and the characteristics of optical components. Therefore, while the system output power is affected by other factors, these factors generally have a minor impact on the system. Therefore, temperature is chosen to provide unified compensation and adjustment to simplify the problem. On the other hand, while modeling based on the fundamental mechanism of EDFAs can provide theoretical guidance, this purist approach fails to fully account for the diverse environments and complex factors present in practical applications. Mechanism modeling focuses on idealized assumptions, ignoring potential external disturbances and operational uncertainties. Therefore, relying solely on mechanistic modeling can lead to discrepancies with actual conditions and fail to effectively address the issue of optimizing system performance in complex environments. Therefore, a more comprehensive approach, modeling based on measurable and widely influential factors like temperature, ensures a solid theoretical foundation while more closely addressing the needs and challenges of real-world applications.

[0056] The second step is to optimize the multi-stage intelligent coordinated control algorithm of erbium-doped fiber amplifier.

[0057] The system's temperature changes are monitored in real time. Based on the mathematical model of the pump source and system obtained in the first step, the output power of the pump source and system is calculated. As the temperature changes, the current component is continuously optimized using the current adjustment step size derived from the multi-level coordinated control algorithm, using the constant output power as a feedback constraint. This updates and calculates the pump power, output power, and system efficiency. Through several iterations of optimization until the efficiency meets the required level, the optimal solution for the pump source drive current under temperature changes is obtained, achieving constant output power and optimized system efficiency.

[0058] refer to Figure 1 and Figure 2The following describes a multi-stage intelligent control method for driving a fiber amplifier according to a second embodiment of the present invention. The fiber amplifier may be a cascaded fiber amplifier, or alternatively, a cascaded erbium-doped fiber amplifier. More specifically, the fiber amplifier may be a two-stage erbium-doped fiber amplifier.

[0059] Taking the intelligent coordinated control of the output power of a two-stage erbium-doped fiber amplifier used in laboratory experiments as an example, the pump source and the two-stage erbium-doped fiber amplifier are used as black boxes. Based on the experimental data of the pump source and the two-stage erbium-doped fiber amplifier's output power, temperature, and pump source driving current measured in the laboratory, a mathematical model is established using software code. The condition of constant system output power is used as a feedback constraint. Taking into account the system efficiency index, a two-stage coordinated control algorithm is used to further maximize the system efficiency and optimize its economic performance while ensuring constant output power. This provides guidance for the control of the output power of actual cascaded erbium-doped fiber amplifiers. The pump source can be a pump laser.

[0060] Optionally, the coordinated control algorithm in the multi-level intelligent control method for optical fiber amplifier driving provided in this embodiment is not limited to the specific algorithm used in specific control, and any suitable algorithm can be used as needed, such as but not limited to gradient descent optimization algorithm, particle swarm algorithm, genetic algorithm, etc.

[0061] A multi-level intelligent control method for optical fiber amplifier driving provided in accordance with a second embodiment of the present invention includes the following steps.

[0062] Step S1, building a pump source system, building a fiber amplifier system, setting up a multi-stage pump source as needed, inputting seed light and the first-stage pump light through a wavelength division multiplexer coupled into the optical fiber for primary amplification, and the signal light output after the first-stage amplification and the second-stage pump light are coupled through a wavelength division multiplexer into the second-stage optical fiber for secondary amplification to output optical power, and so on for fiber amplifiers with more stages.

[0063] Step S2: constructing a mathematical model of the pump source system.

[0064] Step S2.1: Establish a mathematical model of the pump source system based on the experimental data of the optical fiber amplifier. The mathematical model of the pump power of the pump source is constructed as follows:

[0065] Ppump i =g i (I i ,t)

[0066] Among them, Ppump i represents the pump power of the i-th pump source, g i(·) represents the mathematical model of the output power, driving current and ambient temperature of the i-th stage pump source, I i represents the driving current of the i-th pump source, i represents the number of pump sources and i = 1, 2, ... n, n represents the total number of pump sources, and t represents the ambient temperature. The pump source and fiber amplifier system are placed in a temperature chamber to simulate the ambient temperature.

[0067] The mathematical model for constructing the system output power is:

[0068] Pout=f(I1,I2,…,I i ...,I n ,t)

[0069] Where Pout represents the system output power, and f(·) represents the mathematical model of the system output power, the driving current of each pump source, and the ambient temperature. The system output power is the overall output power of the fiber amplifier system.

[0070] Step S2.2, before the algorithm starts, the input parameters are initialized. Step S2.2 specifically includes the following steps.

[0071] Step S2.2.1, initializing the input parameters. The initialization of the input parameters includes setting the initial driving current value I init , target ambient temperature t2, and initial ambient temperature t1, output power convergence tolerance, efficiency convergence tolerance, and gradient norm convergence tolerance. Among them, output power convergence tolerance, efficiency convergence tolerance, and gradient norm convergence tolerance are used to judge the system convergence, that is, the output power remains unchanged while the efficiency is maximized.

[0072] The initial current value is the initial value of the driving current of each level of pump source, I init =(I 1_init ,I 2_init ,…,I i_init ...,I n_init ), where I i_init is the initial value of the driving current of the i-th stage pump source. The target ambient temperature is the ambient temperature after the temperature change, and the change of the ambient temperature can be simulated by a temperature box.

[0073] Step S2.2.2, then, calculate the initial output power.

[0074] The initial system output power is obtained as follows:

[0075] P ou t _ i n iti al =f(I 1_ i n it ,I 2_ i n it,…,I i_ i n it...,I n_ i n i t ,t1).

[0076] Step S2.2.3, set the target output power P out_target , that is, the system output power maintains a constant power level, and sets the learning rate learning_rate and the maximum number of iterations max_iter to ensure that there are enough steps for adjustment during the optimization process while avoiding excessive calculation of invalid iterations. Among them, the learning rate controls the step size of each update, and the maximum number of iterations ensures that the optimization process does not stop prematurely.

[0077] In step S3, based on the constructed mathematical model of the pump source system, after the ambient temperature changes from the initial ambient temperature to the target ambient temperature, the control algorithm is iteratively executed, with the system output power maintained at the preset target output power as the constraint and maximizing system efficiency as the optimization goal, to obtain the optimal system drive current and maximum efficiency under the target ambient temperature and preset target output power conditions. The purpose of the iteration is to maximize system efficiency by adjusting the drive current of each pump source. With each iteration, the algorithm recalculates the system efficiency and adjusts the optimized drive current of each pump source based on the gradient information. This step S3 specifically includes the following steps.

[0078] In step S3.1, the current driving current of the pump source and the target ambient temperature are first substituted into the mathematical model constructed in step S2.1 to calculate the current system output power and the current pump power of each level of the pump source:

[0079] Pout=f(I′1,I′2,...,I i ′…,I n ′,t2)

[0080] Ppump i =g i (I i ′,t2)

[0081] Among them, I′=(I′1,I′2,...,I i ′…,I n ′) represents the current driving current of each level pump source of the system, t2 represents the target ambient temperature, Pout represents the system output power, Ppump i represents the pump power of the i-th pump source.

[0082] The current driving current of the pump source comes from the optimized driving current obtained in step S3.4 of the previous iteration.

[0083] In the first iteration, the initial value of the driving current of each pump source obtained in step S2.2 is: I init =(I 1_init ,I 2_init ,…,I i_init ...,I n_init ) as the current driving current of the pump source, and substituted into the mathematical model constructed above along with the changed target ambient temperature t2 to calculate the system output power and the pump power of each pump source in the first iteration. In subsequent iterations, the system output power and the pump power of each pump source are calculated using the optimized driving current obtained in the previous iterative optimization round as the current driving current input of the pump source.

[0084] Step S3.2: Based on the obtained system output power and the pump power of each level of pump source, the current system efficiency is obtained:

[0085]

[0086] The system efficiency is calculated by dividing the system output power Pout by the total pump power In each iteration, the current system efficiency η is calculated. The optimization of system efficiency is the core goal of the entire algorithm.

[0087] Step S3.3, construct the Lagrangian function and perform gradient calculation to obtain the gradient of the system efficiency and output power constraint terms with respect to the current driving current of each level of pump source, thereby obtaining the gradient of the Lagrangian function with respect to the current driving current of each level of pump source. The Lagrangian function gradient calculation is a key step in the algorithm, which helps determine how to adjust the driving current I' to maximize the system efficiency η. For each driving current component, the gradient is calculated using the numerical gradient estimation method. The specific method is to slightly increase the driving current of each level of pump source in each round of iteration to obtain the optimized driving current, and recalculate parameters such as the system output power, the output power of each pump source, and the system efficiency, and estimate the gradient based on the difference in parameters such as the system efficiency. This step S3.3 specifically includes the following steps.

[0088] Step S3.3.1, construct the Lagrangian function to combine the objective function with the constraints.

[0089] The optimization goal is to maximize the system efficiency obtained from:

[0090]

[0091] Where, f represents the model of system output power, gi Represents the mathematical model of the i-th stage pump source.

[0092] The constraint condition is that the system output power is equal to the target output power:

[0093] Pout=f(I′1,I′2,...,I i ′…,I′ n ,t2)=Pout_target。

[0094] This constraint requires that the output power Pout equal the target output power Pout_target, where the current drive current is taken from the optimized drive current obtained in the previous iteration. In the first iteration, the initial values ​​of the drive currents of each pump source obtained in step S2.2 are used as the current drive currents of the pump sources.

[0095] Construct a Lagrangian function based on the optimization objective and constraints:

[0096] L=-η+λ·(f(I′1,I′2,...,I i ′…,I′ n ,t2)-Pout_target) 2

[0097] Among them, -η is the objective function, which represents the negative value of system efficiency, reflecting that the optimization goal is to maximize efficiency. λ is the Lagrange multiplier, which characterizes the impact of output power constraint on efficiency optimization. The influence of this parameter can be gradually changed to balance it.

[0098] The constraints in the Lagrangian function are:

[0099] constraint=λ·(f(I′1,I′2,...,I i ′…,I′ n ,t2)-Pout_target) 2

[0100] Here, constraint is the constraint term in the Lagrangian function, which ensures that the system output power remains at the target value Pout_target after the temperature changes.

[0101] In step S3.3.2, based on the system efficiency as the optimization target, the gradient of the system efficiency is calculated. The gradient of the system efficiency with respect to the current driving current is:

[0102]

[0103] Where, j represents the j-th pump source, g j Represents the mathematical model of the j-th pump source.

[0104] Step S3.3.3, based on the gradient of the system output power with respect to the current drive current, calculate the gradient of the output power constraint with respect to the current drive current:

[0105]

[0106] The Lagrange multiplier λ is introduced in the above formula.

[0107] In step S3.4, based on the calculated system efficiency and the gradient of the constraint conditions, as well as the learning rate preset in step S2.2.3, the drive current is updated to obtain the optimized drive current. Specifically, the gradient descent method is used to adjust each current component to obtain the optimized drive current:

[0108] I″ i =I′ i -learning_rate*grad[i]

[0109]

[0110] Among them, I i represents the optimized driving current of the i-th stage pump source in the current iteration round, and grad[i] represents the gradient of the Lagrangian function with respect to the current driving current of the i-th stage pump.

[0111] The optimized driving current I″=(I1″, I2″, ..., I″) of each level pump source of the system in the current iteration round can be obtained. i …,I n ″).

[0112] Optimized system efficiency based on optimized drive current:

[0113]

[0114] Where η″ represents the optimized system efficiency obtained in the current iteration round.

[0115] By continuously adjusting the input current through multiple rounds of iterative cycles, the algorithm gradually maximizes the system efficiency.

[0116] Step S3.5: Convergence judgment: Based on the convergence judgment result, the output result is executed or the next round of iteration is performed. The specific process of convergence judgment is as follows.

[0117] If the current number of iterations is equal to or greater than the preset maximum number of iterations, the currently obtained optimized driving current and optimized system efficiency are used as the optimal driving current and maximum efficiency, and a warning is issued to remind that the maximum number of iterations has been exceeded; if the current number of iterations is less than the preset maximum number of iterations and the following three convergence conditions are met, the system is determined to have converged, and the currently obtained optimized driving current and the system efficiency calculated based on the driving current are used as the optimal driving current and maximum efficiency: 1) the deviation between the system output power corresponding to the currently obtained optimized driving current and the preset target output power is less than the output power tolerance preset in step S2.2; 2) the degree of change in the system efficiency is within the efficiency convergence tolerance preset in step S2.2; and 3) the current gradient norm is within the gradient norm convergence tolerance preset in step S2.2; otherwise (i.e., the current number of iterations is less than the preset maximum number of iterations and the above three convergence conditions are not met at the same time), the system is determined to have not converged, and the process returns to step S3.1, and the optimized driving current obtained in the current iteration round is used as the current driving current input for the next round of iteration, and the next round of iteration is executed.

[0118] The degree of change in the system efficiency described above is the difference between the optimized system efficiency obtained in the current iteration round obtained in step S3.4 and the system efficiency obtained in step S3.2 (i.e., the optimized system efficiency obtained in the previous iteration round). The difference is compared with the preset efficiency convergence tolerance.

[0119] The current gradient norm mentioned above is the norm of the gradient of the Lagrangian function obtained in step S3.3 with respect to the current driving current.

[0120] Convergence judgment is a key link in the optimization algorithm. When the system is judged to have converged, the optimized driving current of each level of pump source and the corresponding maximum system efficiency are obtained when the output ambient temperature changes to the set target ambient temperature.

[0121] Step S4: outputting the optimal driving current and the maximum power to the pump source system, and executing multi-level intelligent control of the pump source system.

[0122] In summary, through the above steps, the multi-level coordinated control method continuously adjusts the pump source drive current in each iteration, ultimately finding the pump source drive current that maximizes efficiency while maintaining a constant output power for the fiber amplifier, thus meeting the requirements of constant output power and maximum system efficiency. The fiber amplifier can be, for example, a cascaded erbium-doped fiber amplifier.

[0123] Example 1

[0124] The following describes Example 1 of the multi-stage intelligent control method for optical fiber amplifier driving according to the second embodiment of the present invention to facilitate a better understanding of the technical solution of this embodiment. It should be understood that this example is not intended to limit the present invention.

[0125] Taking the intelligent coordinated control of the output power of a two-stage erbium-doped fiber amplifier used in laboratory experiments as an example, the design ideas of the scheme are described based on this example. The control algorithm used in the embodiment is a gradient descent algorithm, but the technical schemes using different control algorithms are all within the scope of protection of the present invention.

[0126] In Example 1, a two-stage erbium-doped fiber amplifier system is used as a fiber amplifier driver to implement intelligent coordinated control. The specific implementation process is as follows.

[0127] First, a two-stage erbium-doped fiber amplifier system is built with two pump sources. The input seed light and the first-stage pump light are coupled into the erbium-doped fiber through a wavelength division multiplexer for primary amplification. The signal light output from the first stage amplification and the second-stage pump light are coupled into the second-stage erbium-doped fiber through a wavelength division multiplexer for secondary amplification to output optical power.

[0128] The data of the output power, driving current, and ambient temperature of the two-stage pump source are collected. The data is fitted using the support vector machine (SVR) in the machine learning model to obtain the mathematical model of the pump source:

[0129] Ppump1=g1(I1,t)

[0130] Ppump2=g2(I2,t)

[0131] Among them, I1∈[I 1_min ,I 1_max ],I2∈[I 2_min ,I 2_max ], I i_min is the minimum driving current of the i-th pump source in the data, I i_max is the maximum value of the driving current of the i-th stage pump source in the data. During the subsequent system efficiency optimization process, the driving current of the pump source is limited to this range.

[0132] The output power of the two-stage erbium-doped fiber amplifier system and the driving current of each stage of the pump source are collected and fitted using a support vector machine (SVR) to obtain the mathematical model of the system:

[0133] Pout=f(I1,I2,t).

[0134] Then the system parameters are initialized and the initial current value I of the two-stage pump is generated by the random number generation function. init =(I 1_init ,I2_init ), set the target ambient temperature t2 to 25°C, the target output power Pout_target of the system to 2.5W, and the learning rate learning_rate to 1×10 -2 , the maximum number of iterations is 1×10 4 , the absolute value of the output power tolerance is 10mW, the absolute value of the efficiency variation tolerance is 1%, and the absolute value of the gradient norm convergence tolerance is 0.1%.

[0135] The initial pump power, initial output power, and initial efficiency are calculated using the initial current value, and the Lagrangian function L = -η + λ (f(I′1, I′2, t2) - Pout_target) is constructed. 2 , initially set λ = 1, and later change it to observe its effect on the optimization process. Calculate the gradient to get the gradient vector Then, the driving current is updated based on the driving current obtained in this round of iteration, and used as the current driving current input in the next round.

[0136] After each iteration, convergence is determined. The system is considered converged when the output power tolerance, efficiency variation tolerance, and gradient norm tolerance meet the requirements. The drive current and system efficiency from the last iteration are used as the optimal drive current and maximum efficiency. In Example 1, after 1500 iterations, the optimal system drive current is: I1 = 0.32A, I2 = 6.96A, the maximum system efficiency is 19%, and the output power error is 2mW.

[0137] Through this embodiment, the system achieves output power stabilization at the target value after several iterations, with an error of less than 10mW, and the efficiency is improved to close to the theoretical limit, verifying the effectiveness of the multi-level intelligent coordinated control algorithm.

[0138] For those skilled in the art, several modifications and improvements can be made to the embodiments of the present invention without departing from the inventive concept of the present invention, and all of these modifications and improvements fall within the scope of protection of the present invention.

[0139] All of the above optional technical solutions can be combined in any way to form optional embodiments of the present application, and will not be described in detail here.

[0140] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0141] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any technician familiar with this technical field within the technical scope disclosed by the present invention should be covered by the scope of protection of the present invention.

Claims

1. A multi-level intelligent control method for optical fiber amplifier driving, characterized in that: include: Step S1, building a pump source system, building a multi-stage fiber amplifier system with a multi-stage pump source, setting up the multi-stage pump source, and amplifying the input seed light and the input signal of the multi-stage pump source through the multi-stage fiber amplifier to obtain output power; Step S2, constructing a mathematical model of the pump source system, including constructing a mathematical model of the pump power of the pump source and a mathematical model of the system output power, and setting initialization input parameters, including setting an initial ambient temperature, a target ambient temperature, and a target output power; Step S3: Based on the constructed mathematical model of the pump source system, when the ambient temperature changes from the initial ambient temperature to the target ambient temperature, the control algorithm is iteratively executed with the system output power maintained at the preset target output power as a constraint condition and the system efficiency maximized as the optimization goal to obtain the optimal drive current and maximum efficiency of the system under the target ambient temperature and the preset target output power conditions; Step S4: obtaining the optimal driving current and maximum efficiency of the pump source system, and performing multi-level intelligent control of the pump source system.

2. The multi-level intelligent control method for optical fiber amplifier driving according to claim 1, characterized in that: Step S2 specifically includes: Step S2.1, establish a mathematical model of the pump source system and construct a mathematical model of the pump power of the pump source: Ppump i =g i (I i ,t) Among them, Ppump i represents the pump power of the i-th pump source, g i (·) represents the mathematical model of the output power, driving current and ambient temperature of the i-th stage pump source, I i represents the driving current of the i-th pump source, i represents the number of pump sources and i=1,2,…n, n represents the total number of pump sources, and t represents the ambient temperature; The mathematical model for constructing the system output power is: Pout=f(I1,I2,…,I i ...,I n ,t) Where Pout represents the system output power, and f(·) represents the mathematical model of the system output power, the driving current of each level of pump source, and the ambient temperature. Step S2.2, initialization of input parameters includes setting the initial driving current value, target ambient temperature and initial ambient temperature; calculating the initial output power based on the mathematical model of the system output power; and setting the target output power and control algorithm iteration parameters.

3. The multi-level intelligent control method for optical fiber amplifier driving according to claim 2, characterized in that: In step S2.2: Set the initial driving current value of each pump source as the initial driving current value: I init =(I 1_init ,I 2_init ,…,I i_init ...,I n_init ) Among them, I i_init is the initial value of the driving current of the i-th stage pump source.

4. The multi-level intelligent control method for optical fiber amplifier driving according to claim 3, characterized in that: In step S2.2: The initial output power calculated based on the mathematical model of the system output power is: Q out_initial =f(I 1_init ,I 2_init ,…,I i_init ...,I n_init ,t1) Where t1 is the initial ambient temperature.

5. The multi-level intelligent control method for optical fiber amplifier driving according to claim 4, characterized in that: In step S2.2, setting the parameters of the control algorithm iteration includes setting the output power convergence tolerance, efficiency convergence tolerance, gradient norm convergence tolerance, learning rate and maximum number of iterations.

6. The multi-level intelligent control method for optical fiber amplifier driving according to claim 5, characterized in that: Step S3 specifically includes: Step S3.1: Based on the current driving current of the pump source and the target ambient temperature, the constructed mathematical model of the system output power and the mathematical model of the pump power of the pump source are substituted to calculate the system output power and the pump power of each level of the pump source; Step S3.2, obtaining a system efficiency based on the obtained system output power and the pump powers of the pump sources at each stage, wherein the system efficiency is measured by the ratio of the system output power to the sum of the pump powers of the pump sources at each stage; Step S3.3: With the system output power maintained at the preset target output power as the constraint condition and the system efficiency maximized as the optimization goal, a Lagrangian function is constructed and gradient calculation is performed to obtain the gradients of the system efficiency and system output power constraints with respect to the current drive currents of the various pump sources, thereby obtaining the gradients of the Lagrangian function with respect to the current drive currents of the various pump sources; Step S3.4, using the obtained gradient of the system efficiency and the system output power constraint term as the gradient of the Lagrangian function, and in combination with a preset learning rate, updating the drive current using a gradient descent method to obtain an optimized drive current, and obtaining an optimized system efficiency based on the optimized drive current; Step S3.5, convergence judgment, when the convergence conditions are met, the currently obtained optimized driving current and optimized system efficiency are used as the optimal driving current and maximum efficiency; otherwise, it is determined that the system has not converged and returns to step S3.1, and the obtained optimized driving current is used as the current driving current input for the next round of iteration to execute the next round of iteration.

7. The multi-level intelligent control method for optical fiber amplifier driving according to claim 6, characterized in that: In step S3.1: The current driving current of the pump source is the optimized driving current obtained in the previous iteration; and For the first round of iteration, the current driving current of the pump source is the initial driving current value.

8. The multi-level intelligent control method for optical fiber amplifier driving according to claim 6, characterized in that: Step S3.5 specifically includes: If the current number of iterations is equal to or greater than the preset maximum number of iterations, the currently obtained optimized driving current and optimized system efficiency are used as the optimal driving current and maximum efficiency, and a warning is issued to remind that the maximum number of iterations has been exceeded; If the current number of iterations is less than the preset maximum number of iterations and the following three convergence conditions are met, the system is determined to have converged, and the currently obtained optimized driving current and optimized system efficiency are used as the optimal driving current and maximum efficiency: 1) the deviation between the current system output power and the preset target output power is less than the preset output power tolerance; 2) the degree of change in system efficiency is within the preset efficiency convergence tolerance; and 3) the current gradient norm is within the preset gradient norm convergence tolerance; Otherwise, it is determined that the system has not converged, and the process returns to step S3.1, where the obtained optimized driving current is used as the current driving current input for the next round of iteration, and the next round of iteration is performed.

9. The multi-level intelligent control method for optical fiber amplifier driving according to claim 8, characterized in that: The degree of change in system efficiency is the difference between the optimized system efficiency obtained in the current iteration round obtained in step S3.4 and the system efficiency obtained in step S3.2, and the difference is compared with the preset efficiency convergence tolerance.

Images