A wavelength-manipulable automatic mode-locking method and system

Abstract

This application discloses a wavelength-controllable automatic mode-locking method and system, relating to the field of fiber laser control technology. The method acquires user-input target state parameters; loads candidate control parameters into the controllable cavity adjustment module of a passively mode-locked fiber laser, and collects time-domain pulse sequences and frequency-domain spectral data; extracts time-domain basic mode-locking features, frequency-domain target wavelength features, and frequency-domain spectral structure features; determines the current target fitness function and target mode-locking determination conditions based on the target type; calculates the target fitness based on the current target fitness function, and determines whether the current output state is a target wavelength mode-locked state based on the target fitness and target mode-locking determination conditions; when the current output state is a target wavelength mode-locked state, the current candidate control parameters are maintained. This application can achieve automatic search for single-wavelength, dual-wavelength, or multi-wavelength target wavelength mode-locking states.

Description

Technical Field

[0001] This application relates to the field of fiber laser control technology, and in particular to a wavelength-controllable automatic mode-locking method and system that enables automatic search and switching of target wavelength mode-locking state based on target state parameters, time-domain pulse sequences and frequency-domain spectral data. Background Technology

[0002] Passively mode-locked fiber lasers can generate stable ultrashort pulse outputs and have important applications in optical communication, optical frequency combs, precision measurement, nonlinear optics, spectral analysis, biomedical imaging, and ultrafast laser processing. For passively mode-locked fiber lasers, practical applications typically require not only that the laser enter a stable mode-locked state, but also that the laser output has a specified center wavelength, target wavelength range, wavelength spacing, or multi-wavelength combination. Therefore, the mode-locking control objective is not limited to determining "whether mode-locking is achieved," but may also manifest as forming single-wavelength mode-locking near the target center wavelength, forming mode-locking within the target wavelength range, or forming a dual-wavelength or multi-wavelength mode-locking state that satisfies a preset wavelength combination.

[0003] In fiber lasers, the output wavelength and mode-locking state are typically influenced by a combination of factors, including gain spectral distribution, cavity loss, birefringence filtering, nonlinear phase shift, polarization state, pump power, filtering effects, and environmental disturbances. For passively mode-locked fiber lasers based on nonlinear polarization rotation, polarization-dependent loss, or other polarization-sensitive mechanisms, changes in the driving voltage of the electrically controlled polarization controller alter the cavity polarization state and equivalent filtering conditions, causing the laser to switch between an unlocked state, a normal mode-locked state, a single-wavelength mode-locked state with different center wavelengths, a dual-wavelength mode-locked state, a multi-wavelength mode-locked state, and a transitional state. In other words, the same laser hardware, under different cavity control parameters, may produce multiple mode-locked states with different wavelength positions and spectral structures.

[0004] Traditional methods for obtaining mode-lock status mainly rely on manual adjustment of polarization controllers, pump power, or intracavity filter components. Operators typically judge whether the current state is close to the target output based on the time-domain pulse sequence displayed on an oscilloscope and the frequency-domain spectral curve displayed on a spectrometer, and repeatedly adjust the intracavity control parameters. Although this method can achieve mode-lock search under laboratory conditions, the adjustment process is time-consuming, dependent on operator experience, and has poor repeatability and automation. When the target is further extended from ordinary mode-locking to mode-locking at a specific target wavelength or a combination of specified wavelengths, the difficulty of manual adjustment increases further. For example, although a certain output state has formed a stable pulse train, its center wavelength may not fall within the target wavelength range required by the user; or, for example, a certain output state may appear as a stable pulse in the time domain, but in the frequency domain it may exhibit a single-peak, double-peak, multi-peak transition state or a broadband noise state, and it cannot be judged whether it meets the target state requirements solely based on the stability of the time-domain pulse.

[0005] In recent years, random search, genetic algorithms, particle swarm optimization, reinforcement learning, and deep learning-based image recognition methods have been increasingly used in automatic mode-locking control. These methods can reduce manual adjustments to some extent; however, existing automatic mode-locking schemes typically focus on "entering a stable mode-locking state" or "achieving a preset type of mode-locking state," lacking parameterized constraints on user-inputted target center wavelength, target wavelength range, or target wavelength combinations. Therefore, when the laser has entered stable mode-locking but the output wavelength does not meet the target requirements, existing schemes may still misclassify a stable but non-target wavelength mode-locking state as a successful state.

[0006] Furthermore, for dual-wavelength or multi-wavelength mode-locked states, target state determination not only depends on temporal pulse repetition stability but also requires consideration of information such as the number of main spectral envelope peaks, target wavelength proximity, peak spacing, power balance, valley depth, number of significant clutter peaks, and energy concentration in the frequency domain spectrum. While some automatic mode-locking schemes introduce time-domain and frequency-domain feedback or can search for fixed dual-wavelength mode-locked states, their determination rules and fitness functions are usually constructed around a preset fixed target. This makes it difficult to determine the corresponding target fitness function and target mode-locking determination conditions based on the updated target type, target wavelength constraints, and target determination constraints when the user switches the target from a single wavelength to another single wavelength, from a single wavelength to a dual wavelength, or from a dual wavelength to a multi-wavelength.

[0007] Therefore, existing automatic mode-locking methods still have the following shortcomings: First, it is difficult to form target state parameters based on the target center wavelength, target wavelength range, or target wavelength combination input by the user; second, it is difficult to determine the corresponding current target fitness function based on single-wavelength, dual-wavelength, or multi-wavelength targets; and third, it is difficult to achieve automatic search and switching between different target wavelength mode-locking states under the same laser hardware structure. Therefore, it is necessary to provide a wavelength-controllable automatic mode-locking method and system, so that a passively mode-locked fiber laser can automatically evaluate the current output state based on the target state parameters input by the user, combined with time-domain pulse sequences and frequency-domain spectral data, and achieve automatic search and switching of single-wavelength, dual-wavelength, or multi-wavelength target wavelength mode-locking states through closed-loop optimization. Summary of the Invention

[0008] This application provides a wavelength-controllable automatic mode-locking method and system to solve the problem that existing automatic mode-locking fiber lasers are unable to determine the current target fitness function and target mode-locking judgment conditions based on the target center wavelength, target wavelength range, or target wavelength combination input by the user, thus making it difficult to achieve automatic search and switching of single-wavelength, dual-wavelength, or multi-wavelength target wavelength mode-locking states.

[0009] In a first aspect, this application provides a wavelength-controllable automatic mode-locking method, comprising: Obtain target status parameters input by the user. The target status parameters include target type, target wavelength constraint and target determination constraint. The target type is a single-wavelength target, a dual-wavelength target or a multi-wavelength target. The target wavelength constraint includes at least one of the target center wavelength, target wavelength range or target wavelength combination. Candidate control parameters are loaded into the controllable cavity adjustment module of the passively mode-locked fiber laser, and the time-domain pulse sequence and frequency-domain spectral data output by the passively mode-locked fiber laser are acquired; wherein, the controllable cavity adjustment module includes at least an electrically controlled polarization controller, and the candidate control parameters include the multi-channel driving voltage vector of the electrically controlled polarization controller; The time-domain basic mode-locking features are extracted from the time-domain pulse sequence, and the frequency-domain target wavelength features and frequency-domain spectral structure features are extracted from the frequency-domain spectral data. The frequency-domain target wavelength features include at least one of the target wavelength proximity, target wavelength range matching result, or target wavelength set matching degree. The frequency-domain spectral structure features include at least one of the following: number of spectral envelope peaks, peak spacing, power balance, valley depth, number of significant impurity peaks, or energy concentration. Based on the target type, the current target fitness function is determined from the preset single-wavelength target fitness function, dual-wavelength target fitness function, or multi-wavelength target fitness function, and the target mode-locking determination condition is determined based on the target wavelength constraint and the target determination constraint. Based on the time-domain basic mode-locking characteristics, frequency-domain target wavelength characteristics, frequency-domain spectral structure characteristics, and the current target fitness function, the target fitness corresponding to the current candidate control parameters is calculated. Based on the target fitness and the target mode-locking determination conditions, it is determined whether the output state of the passively mode-locked fiber laser under the current candidate control parameters is the target wavelength mode-locked state. When the output state of the passively mode-locked fiber laser under the current candidate control parameters is the target wavelength mode-locked state, the current candidate control parameters are maintained.

[0010] In one possible implementation, the target wavelength constraint includes a set of target wavelengths and an allowable range of target wavelength deviations, and the target determination constraint includes spectral structure constraints and temporal stability constraints related to the target state; wherein, the target wavelength set includes one target center wavelength when the target type is a single-wavelength target, two target center wavelengths when the target type is a dual-wavelength target, and multiple target center wavelengths when the target type is a multi-wavelength target.

[0011] In one possible implementation, when the target wavelength is constrained to a target wavelength range, the target wavelength matching result is determined based on whether the center wavelength of the current main spectral envelope peak is within the target wavelength range; when the center wavelength of the current main spectral envelope peak deviates from the target wavelength range, the target wavelength proximity is reduced based on the distance between the center wavelength of the current main spectral envelope peak and the boundary of the target wavelength range.

[0012] In one possible implementation, the time-domain basic mode-locking feature includes at least one of pulse interval stability, pulse peak uniformity, time-domain signal-to-noise ratio, satellite peak ratio, or time-domain noise feature; wherein, the time-domain basic mode-locking feature is used to determine whether the current output forms a stable repeating pulse train.

[0013] In one possible implementation, frequency domain target wavelength features and frequency domain spectral structure features are extracted from frequency domain spectral data, including: filtering the effective range of the frequency domain spectral data, retaining the spectral window containing the target wavelength region, and removing invalid low-power filler values; smoothing the filtered spectral data to obtain the spectral envelope; detecting the main spectral envelope peak on the spectral envelope, and extracting the center wavelength, peak power, and number of spectral envelope peaks of the main spectral envelope peak.

[0014] In one possible implementation, the calculation of target fitness also includes a noise quality gating term, which includes a time-domain noise penalty, a frequency-domain noise penalty, and a total noise penalty determined by the time-domain noise penalty and the frequency-domain noise penalty; wherein, the total noise penalty is used as a quality gating term in the calculation of target fitness for single-wavelength targets, dual-wavelength targets, and multi-wavelength targets.

[0015] In one possible implementation, when the target type is a single-wavelength target, the single-wavelength target fitness function is determined as the current target fitness function. The single-wavelength target fitness function calculates the single-wavelength target fitness based on the time-domain basic mode-locking characteristics, target wavelength proximity, main peak intensity, single main peak spectral structure and noise quality gating term, and suppresses the non-target multi-peak state based on dual-wavelength or multi-wavelength confidence.

[0016] In one possible implementation, when the target type is a dual-wavelength target, the dual-wavelength target fitness function is determined as the current target fitness function; when the target type is a multi-wavelength target, the multi-wavelength target fitness function is determined as the current target fitness function. The dual-wavelength target fitness function calculates the dual-wavelength target fitness based on the dual-target wavelength matching degree, the bi-peak spacing, power balance, valley depth, bi-peak energy ratio, and noise quality gating term. The multi-wavelength target fitness function calculates the multi-wavelength target fitness based on whether multiple dominant spectral envelope peaks fall within the corresponding target wavelength range, multi-peak power balance, target peak energy ratio, non-target clutter suppression, and temporal stability.

[0017] In one possible implementation, the method further includes: if the output state of the passively mode-locked fiber laser under the current candidate control parameters is not the target wavelength mode-locked state, then adding the current candidate control parameters and their target fitness to the historical experimental dataset; constructing the historical experimental dataset with the evaluated control parameter vectors and their corresponding target fitness values; generating multiple sets of candidate control parameters based on the historical experimental dataset using a preset optimization algorithm, the multiple sets of candidate control parameters including multiple sets of candidate electrically controlled polarization controller voltage vectors; wherein, when the preset optimization algorithm is a Bayesian optimization algorithm, establishing a probabilistic surrogate model based on the historical experimental dataset to predict the target fitness corresponding to untested control parameters, and generating multiple sets of candidate control parameters according to the probabilistic surrogate model; sequentially loading multiple sets of candidate control parameters, and returning to the steps of loading candidate control parameters and acquiring time-domain pulse sequences and frequency-domain spectral data to continue searching for the target wavelength mode-locked state corresponding to the current target state parameter; if the output state corresponding to any candidate control parameter is the target wavelength mode-locked state, then maintaining that candidate control parameter; wherein, the preset optimization algorithm includes at least one of Bayesian optimization algorithm, genetic algorithm, particle swarm optimization algorithm, reinforcement learning algorithm, or random search algorithm.

[0018] Secondly, this application also provides a wavelength-controllable automatic mode-locking system, comprising: Passive mode-locked fiber lasers are used to output optical signals under the control of a controllable cavity modulation module. A controllable intracavity adjustment module is used to receive candidate control parameters and adjust the intracavity state of a passively mode-locked fiber laser. The controllable intracavity adjustment module includes at least an electrically controlled polarization controller. The time-domain signal acquisition module is used to acquire the time-domain pulse sequence output by the passively mode-locked fiber laser; The frequency domain spectral acquisition module is used to acquire frequency domain spectral data output by a passively mode-locked fiber laser. The computer control module is used to acquire the target state parameters input by the user, determine the current target fitness function from the preset single-wavelength target fitness function, dual-wavelength target fitness function or multi-wavelength target fitness function according to the target type, calculate the target fitness based on time-domain pulse sequence and frequency-domain spectral data, and determine the target wavelength mode-locking state according to the target fitness and target mode-locking judgment conditions.

[0019] Compared with the prior art, this application has at least the following beneficial effects: This application introduces target type, target wavelength constraints, and target determination constraints into the automatic mode-locking control process by obtaining target state parameters input by the user. This enables the automatic mode-locking target to no longer be limited to determining whether to enter the normal mode-locking state, but to search for the target wavelength mode-locking state based on the target center wavelength, target wavelength range, or target wavelength combination.

[0020] This application determines the current target fitness function from preset single-wavelength target fitness functions, dual-wavelength target fitness functions, or multi-wavelength target fitness functions based on target type, target wavelength constraints, and target determination constraints, and determines the target mode-locking determination conditions. This enables the same passively mode-locked fiber laser to adopt different current target fitness functions and target mode-locking determination conditions according to different user targets, thereby improving the pertinence of single-wavelength, dual-wavelength, or multi-wavelength target mode-locking state determination.

[0021] This application utilizes both time-domain pulse sequences and frequency-domain spectral data. The time-domain basic mode-locking features are used to determine whether the current output forms a stable repeating pulse train, while the frequency-domain target wavelength features and frequency-domain spectral structure features are used to determine whether the current output meets the target wavelength and corresponding spectral structure requirements. This reduces the probability that a stable but non-target wavelength mode-locking state is misjudged as a successful target state.

[0022] This application introduces a noise quality gating term in the target fitness calculation, which can suppress interference from non-target states such as non-pulse background spikes, excessively strong satellite peaks, indistinct main peaks, excessive significant clutter peaks, excessive envelope peaks, or energy dispersion on the target mode-locking determination, thereby improving the reliability of target wavelength mode-locking status confirmation.

[0023] When the current output state does not meet the target mode-locking determination condition, this application adds the current candidate control parameters and their target fitness to the historical experimental dataset, and generates the next round of candidate control parameters based on the historical experimental dataset through a preset optimization algorithm. This can reduce invalid searches and improve the automatic search efficiency of the target wavelength mode-locking state.

[0024] When the target state parameter is updated, this application can use the updated target state parameter as the new current target state parameter and re-execute the automatic mode-locking method process to search for the single-wavelength, dual-wavelength, or multi-wavelength target wavelength mode-locking state corresponding to the updated target state parameter, thereby realizing automatic switching between different target wavelength mode-locking states. Attached Figure Description

[0025] Figure 1 This is a flowchart of the wavelength-controllable automatic mode-locking method provided in the embodiments of this application; Figure 2 This is a schematic diagram of the wavelength-controllable automatic mode-locking system provided in the embodiments of this application; Figure 3 This is a diagram showing the first single-wavelength target mode-locking state result provided in the embodiments of this application; Figure 4 This is a diagram showing the mode-locked state result of the second single-wavelength target provided in the embodiments of this application; Figure 5This is a diagram showing the dual-wavelength target mode-locking state results provided in an embodiment of this application; Figure 6 This is a graph showing the changes in fitness and confidence during the automatic search and switching process of the target state provided in the embodiments of this application; Figure 7 This is a target wavelength range and two-dimensional spectral evolution diagram during the automatic switching process of the target state provided in the embodiments of this application. Detailed Implementation

[0026] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are all within the scope of protection of this application.

[0027] It should be noted that, unless otherwise specified, the embodiments and technical features described in this application can be combined with each other. The terms "first," "second," etc., used herein are only for distinguishing different objects and are not intended to limit the order, number, or importance of the objects.

[0028] This application provides a wavelength-controllable automatic mode-locking method and system. The method is applied to a passively mode-locked fiber laser, and can analyze the time-domain pulse sequence and frequency-domain spectral data output by the laser based on user-input target state parameters. It determines the current target fitness function based on the target type and determines the target mode-locking determination conditions based on target wavelength constraints and target determination constraints, thereby searching for and determining the target wavelength mode-locking state corresponding to the current target state parameters.

[0029] In this embodiment, the target wavelength mode-locking state refers to the output state that, in addition to satisfying the basic mode-locking stability in the time domain, further satisfies the target wavelength constraints and target determination constraints defined by the target state parameters. The target state parameters include target type, target wavelength constraints, and target determination constraints. The computer control module determines the current target fitness function based on the target state parameters and searches for the target wavelength mode-locking state by combining the time-domain pulse sequence and frequency-domain spectral data.

[0030] Reference Figure 2This application provides a wavelength-controllable automatic mode-locking system. The system includes a passively mode-locked fiber laser, a controllable intracavity adjustment module, a time-domain signal acquisition module, a frequency-domain spectrum acquisition module, and a computer control module. The passively mode-locked fiber laser outputs an optical signal under the control of the controllable intracavity adjustment module. The controllable intracavity adjustment module adjusts the intracavity state of the passively mode-locked fiber laser according to candidate control parameters, and includes at least an electrically controlled polarization controller. The time-domain signal acquisition module acquires the time-domain pulse sequence output by the passively mode-locked fiber laser. The frequency-domain spectrum acquisition module acquires the frequency-domain spectral data output by the passively mode-locked fiber laser. The computer control module obtains user-input target state parameters, determines the current target fitness function from preset single-wavelength target fitness functions, dual-wavelength target fitness functions, or multi-wavelength target fitness functions based on the target type, and determines the target mode-locking determination conditions based on target wavelength constraints and target determination constraints, thereby achieving automatic search and switching of the target wavelength mode-locking state.

[0031] In one implementation, such as Figure 2 As shown, a passively mode-locked fiber laser may include a pump source 1, a wavelength division multiplexer 2, a gain fiber 3, a manual polarization controller 4, an output coupler 5, an isolator 6, an electrically controlled polarization controller 7, and a mode-locking device 8. The pump source 1 generates pump light. The wavelength division multiplexer 2 couples the pump light output from the pump source 1 into the laser cavity. The gain fiber 3 provides optical gain under the action of the pump light. The manual polarization controller 4 coarsely adjusts the initial polarization state within the laser cavity. The output coupler 5 outputs a portion of the optical signal within the laser cavity as the output optical signal. The isolator 6 ensures unidirectional transmission of the optical signal within the laser cavity along a preset direction. The mode-locking device 8 assists in establishing passive mode-locking conditions.

[0032] The electrically controlled polarization controller 7, as at least part of the controllable intracavity adjustment module, is used to receive candidate control parameters output by the computer control module and adjust the intracavity polarization state and equivalent filtering conditions of the laser according to the candidate control parameters. In this embodiment, the candidate control parameters include the multi-channel driving voltage vector of the electrically controlled polarization controller 7. By changing the multi-channel driving voltage vector, the passively mode-locked fiber laser can switch between different intracavity states, thereby generating an unlocked state, a single-wavelength mode-locked state, a dual-wavelength mode-locked state, a multi-wavelength mode-locked state, or a transitional state.

[0033] When using a multi-channel electrically controlled polarization controller, the candidate control parameters can be expressed as the following control vector:

[0034] in, This represents the control parameter vector of the electronically controlled polarization controller; Indicates the first The driving voltage of each electrically controlled polarization control channel; This indicates transpose. For other electrically controlled polarization controllers or other adjustable intracavity devices with different numbers of channels, a control vector of the corresponding dimension can also be used.

[0035] In one embodiment, the optical signal output from the output coupler 5 is split into multiple paths by a beam splitter structure. The beam splitter structure may include a first beam splitter 9 and a second beam splitter 10. The first beam splitter 9 and the second beam splitter 10 are used to introduce the optical signal output from the passively mode-locked fiber laser into the time-domain signal acquisition module and the frequency-domain spectrum acquisition module, respectively, so as to simultaneously obtain time-domain feedback information and frequency-domain feedback information under the current candidate control parameters.

[0036] The time-domain signal acquisition module may include a photodetector 11 and an oscilloscope 12. The photodetector 11 is used to convert the optical signal output by the passively mode-locked fiber laser into an electrical signal, and the oscilloscope 12 is used to acquire the electrical signal and obtain a time-domain pulse sequence. The time-domain pulse sequence is used to extract at least one of the following time-domain fundamental mode-locking features: pulse interval stability, pulse peak uniformity, time-domain signal-to-noise ratio, satellite peak ratio, or time-domain noise characteristics.

[0037] The frequency domain spectral acquisition module may include a spectrometer 13. The spectrometer 13 is used to acquire frequency domain spectral data output from a passively mode-locked fiber laser. The frequency domain spectral data is used to extract frequency domain target wavelength features and frequency domain spectral structure features. The frequency domain target wavelength features include target wavelength proximity, and the frequency domain spectral structure features include at least one of the following: number of spectral envelope peaks, peak spacing, power balance, valley depth, number of significant impurity peaks, or energy concentration.

[0038] The computer control module may include a computer 14 and a data acquisition and control module 15. The computer 14 is used to acquire the target state parameters input by the user and determine the current search target based on the target state parameters. The data acquisition and control module 15 is used, under the control of the computer 14, to load candidate control parameters into the electrically controlled polarization controller 7 and to receive the time-domain pulse sequence acquired by the oscilloscope 12 and the frequency-domain spectral data acquired by the spectrometer 13.

[0039] Furthermore, the computer 14 determines the current target fitness function from a preset single-wavelength target fitness function, dual-wavelength target fitness function, or multi-wavelength target fitness function based on the target type, and determines the target mode-locking determination condition based on the target wavelength constraint and the target determination constraint. If the current output state meets the target mode-locking determination condition, the computer 14 determines the current output state as the target wavelength mode-locking state and maintains the current candidate control parameters. If the current output state does not meet the target mode-locking determination condition, the computer 14 adds the current candidate control parameters and their target fitness to the historical experimental dataset, and generates the next round of candidate control parameters based on the historical experimental dataset using a preset optimization algorithm.

[0040] It should be noted that, Figure 2 The pump source 1, wavelength division multiplexer 2, gain fiber 3, manual polarization controller 4, output coupler 5, isolator 6, electrically controlled polarization controller 7, mode-locking device 8, first beam splitter 9, second beam splitter 10, photodetector 11, oscilloscope 12, spectrometer 13, computer 14, and data acquisition and control module 15 shown are merely exemplary system structures. For other fiber laser systems capable of providing passive mode-locked output, supporting controllable adjustment of intracavity state, and acquiring time-domain pulse sequences and frequency-domain spectral data, the wavelength-controllable automatic mode-locking method of this application embodiment can also be used.

[0041] In one specific embodiment, the controllable intracavity adjustment module may include an electrically controlled polarization controller, a pump power control unit, an adjustable filter, an adjustable delay line, or other adjustable intracavity devices; the beam splitter 9 may be 2:8, and the beam splitter 10 may be 1:9; the mode-locking device 8 may be a semiconductor saturable absorber mirror, carbon nanotube, graphene, a nonlinear polarization rotation structure, or other passive mode-locking devices; the pump source 1 is preferably a 980nm semiconductor laser; the wavelength division multiplexer 2 is preferably a 980 / 1550nm wavelength division multiplexer; the gain fiber 3 may be an erbium-doped fiber; the center operating wavelength of the isolator 6 is preferably 1550nm; and the electrically controlled polarization controller 7 may be a multi-channel squeeze-type electrically controlled polarization controller.

[0042] Based on the above system structure, referring to Figure 1 This application provides a wavelength-controllable automatic mode-locking method. This method can be executed by a computer control module, or it can be executed in conjunction with a controllable cavity adjustment module, a time-domain signal acquisition module, and a frequency-domain spectrum acquisition module. The method includes the following steps.

[0043] S101, Obtain the target state parameters input by the user.

[0044] In S101, the computer control module acquires the target state parameters input by the user. The target state parameters characterize the target wavelength mode-locking state that the user desires the passively mode-locked fiber laser to achieve.

[0045] The target state parameters include target type, target wavelength constraints, and target determination constraints. The target type includes single-wavelength targets, dual-wavelength targets, or multi-wavelength targets. The target wavelength constraints include at least one of the target center wavelength, target wavelength range, or combination of target wavelengths. The target determination constraints define the rules for determining the target wavelength mode-locked state, and may include spectral structure constraints and temporal stability constraints related to the target state.

[0046] In one implementation, the target state parameter input by the user can be represented as:

[0047] in, Indicates the target state parameters; Indicates the target type, which includes single-wavelength targets, dual-wavelength targets, or multi-wavelength targets; Represents the target wavelength set; Indicates the allowable target wavelength deviation range; This represents the spectral structure constraints and temporal stability constraints related to the target state.

[0048] In one implementation, the target wavelength constraint includes a set of target wavelengths and an allowable target wavelength deviation range. The set of target wavelengths includes one target center wavelength when the target type is a single-wavelength target; two target center wavelengths when the target type is a dual-wavelength target; and multiple target center wavelengths when the target type is a multi-wavelength target. The allowable target wavelength deviation range is used to define the permissible deviation of the current main spectral envelope peak center wavelength relative to the target center wavelength, or to define whether the current main spectral envelope peak center wavelength is within the target wavelength range.

[0049] When the target is in a single-wavelength mode-locked state, the target wavelength set can be represented as:

[0050] in, The target center wavelength is set by the user. At this time, the system determines the single-wavelength target fitness function as the current target fitness function, focusing on evaluating whether the main spectral envelope peak is located near the target center wavelength, and whether the current output satisfies the time-domain stability of single-wavelength mode-locking and the single main peak condition in the frequency domain.

[0051] When the target is in a dual-wavelength mode-locked state, the target wavelength set can be represented as:

[0052] in, and The system sets two target center wavelengths for the user. It then determines the dual-wavelength target fitness function as the current target fitness function, focusing on evaluating whether there are two dominant spectral envelope peaks that satisfy the target wavelength conditions, and further assessing conditions such as peak spacing, peak power balance, valley depth, and peak energy ratio.

[0053] When the target is in a multi-wavelength mode-locked state, the target wavelength set can be represented as:

[0054] in, This indicates the number of target wavelengths. At this point, the system determines the multi-wavelength target fitness function as the current target fitness function, judges whether the multiple dominant spectral envelope peaks fall within the target wavelength range set by the user, and judges characteristics such as multi-peak power balance, spurious peak suppression, and overall spectral energy concentration.

[0055] For example, when a user expects to obtain a single-wavelength target mode-locked state, the target state parameters may include a single-wavelength target, a target center wavelength, and the corresponding allowable deviation range; when a user expects to obtain a dual-wavelength target mode-locked state, the target state parameters may include a dual-wavelength target, a target wavelength combination consisting of two target center wavelengths, and the corresponding bimodal spectral structure constraint; when a user expects to obtain a multi-wavelength target mode-locked state, the target state parameters may include multiple wavelength targets, multiple target center wavelengths or multiple target wavelength ranges, and the corresponding multimodal spectral structure constraint.

[0056] In this way, the target state parameters input by the user can transform the ordinary "mode-locking" judgment into a target wavelength mode-locking state search task targeting a specific target wavelength, target wavelength range, or target wavelength combination.

[0057] S102, load candidate control parameters and acquire time-domain pulse sequences and frequency-domain spectral data.

[0058] In S102, the computer control module loads candidate control parameters into the controllable intracavity adjustment module of the passively mode-locked fiber laser. The controllable intracavity adjustment module includes at least an electrically controlled polarization controller. The candidate control parameters include the multi-channel drive voltage vector of the electrically controlled polarization controller.

[0059] In the initial search phase, candidate control parameters can be given randomly, or they can be derived from historical experimental data or manually set initial values. In subsequent search phases, candidate control parameters can be generated by a preset optimization algorithm based on historical experimental datasets.

[0060] In one implementation, the computer control module determines a set of candidate control parameters based on the current search round and loads these candidate control parameters into the electrically controlled polarization controller. The electrically controlled polarization controller changes the polarization state and equivalent filtering conditions within the laser cavity according to the multi-channel driving voltage vector, enabling the passively mode-locked fiber laser to output an optical signal in the corresponding cavity state.

[0061] After the candidate control parameters are loaded, the computer control module acquires the time-domain pulse sequence and frequency-domain spectral data output by the passively mode-locked fiber laser. Specifically, the time-domain signal acquisition module acquires the time-domain pulse sequence output by the laser, and the frequency-domain spectral acquisition module acquires the frequency-domain spectral data output by the laser. The time-domain pulse sequence is used for subsequent extraction of basic time-domain mode-locking features, and the frequency-domain spectral data is used for subsequent extraction of frequency-domain target wavelength features and frequency-domain spectral structure features.

[0062] It should be noted that the output state under the current candidate control parameters may be an unlocked state, a normal mode-locked state, a single-wavelength mode-locked state, a dual-wavelength mode-locked state, a multi-wavelength mode-locked state, or a transitional state. This application does not directly regard the stable pulse output as the target wavelength mode-locked state, but rather requires further evaluation based on the target state parameters, the time-domain basic mode-locking characteristics, the frequency-domain target wavelength characteristics, and the frequency-domain spectral structure characteristics, corresponding to the target state parameters.

[0063] S103 extracts the time-domain basic mode-locking features, frequency-domain target wavelength features, and frequency-domain spectral structure features.

[0064] In S103, the computer control module extracts the basic mode-locked features in the time domain based on the time domain pulse sequence, and extracts the target wavelength features and the spectral structure features in the frequency domain based on the frequency domain spectral data.

[0065] The time-domain fundamental mode-locking feature is used to determine whether the current output state forms a stable repetitive pulse train. The time-domain fundamental mode-locking feature includes at least one of the following: pulse interval stability, pulse peak uniformity, time-domain signal-to-noise ratio, satellite peak ratio, or time-domain noise feature. Specifically, pulse interval stability characterizes the consistency of time intervals between adjacent pulses; pulse peak uniformity characterizes the consistency of peak intensities between different pulses; time-domain signal-to-noise ratio characterizes the significance of the pulse signal relative to background noise; satellite peak ratio characterizes the strength of unwanted secondary peaks near the main pulse; and time-domain noise feature characterizes non-pulse background spikes or abnormal fluctuations.

[0066] In one implementation, a time-domain repeatability stability evaluation metric can be constructed:

[0067] in, Indicates pulse interval stability. Indicates pulse peak uniformity. Represents the time-domain signal-to-noise ratio. , , is the weighting coefficient. This evaluation quantity serves as the basic mode-locked stability gating shared by all target states. If If the preset threshold is not reached, the current state is determined to be unlocked, unstable mode-locked, noisy, or transitional, and the system returns to the optimization process to continue searching. Once the preset threshold is reached, the target wavelength determination in the frequency domain begins. The time-domain repeatability stability evaluation quantity is used as a basic mode-locked stability gating mechanism shared by single-wavelength targets, dual-wavelength targets, and multi-wavelength targets.

[0068] For dual-wavelength or multi-wavelength mode-locked states, the time-domain waveform may exhibit moderate periodic modulation or weak satellite peaks. To avoid misclassifying a true dual-wavelength or multi-wavelength state as an unstable state, the proportion of satellite peaks can be calculated:

[0069] in, Indicates the proportion of satellite peaks; This indicates the amplitude of the largest satellite peak outside the main pulse within the repetition period; This represents the average amplitude of the main pulse. For single-wavelength targets, stricter temporal morphological constraints can be applied; for dual-wavelength or multi-wavelength targets, moderate periodic modulation is permissible, but conditions with significant multi-pulse chaos or excessively strong noise peaks are still excluded.

[0070] Frequency domain target wavelength characteristics are used to determine whether the current output state meets the target wavelength constraint. Frequency domain target wavelength characteristics include at least one of the following: target wavelength proximity, target wavelength range matching result, or target wavelength set matching degree. Specifically, target wavelength proximity characterizes the degree of matching between the center wavelength of the current dominant spectral envelope peak and the target center wavelength, target wavelength range, or target wavelength combination; target wavelength range matching result characterizes whether the center wavelength of the current dominant spectral envelope peak is within the user-defined target wavelength range; and target wavelength set matching degree characterizes the degree of matching between multiple dominant spectral envelope peaks and each target center wavelength or target wavelength range in the target wavelength set.

[0071] Frequency domain spectral structure features are used to determine whether the current output state meets the spectral structure requirements corresponding to the current target type. Frequency domain spectral structure features include at least one of the following: number of spectral envelope peaks, peak spacing, power balance, valley depth, number of significant clutter peaks, or energy concentration. Specifically, the number of spectral envelope peaks distinguishes between single-wavelength, dual-wavelength, or multi-wavelength spectral structures; peak spacing characterizes the wavelength distance between adjacent dominant spectral envelope peaks; power balance characterizes the degree of power equilibrium among multiple target peaks; valley depth characterizes the degree of separation between dual or multiple peaks; the number of significant clutter peaks characterizes the degree of interference from non-target peaks; and energy concentration characterizes whether the spectral energy is concentrated within the target peak or target wavelength region.

[0072] In one embodiment, frequency domain target wavelength features and frequency domain spectral structure features are extracted from frequency domain spectral data, including: filtering the effective range of the frequency domain spectral data, retaining the spectral window containing the target wavelength region, and removing invalid low-power filler values; smoothing the filtered spectral data to obtain the spectral envelope; detecting the main spectral envelope peak on the spectral envelope, and extracting the center wavelength, peak power, and number of spectral envelope peaks of the main spectral envelope peak.

[0073] In one implementation, the detected spectral envelope peaks are sorted according to their peak power, and the strongest envelope peak is denoted as ( ). , The second strongest envelope peak is denoted as ,in, The central wavelength of the main spectral envelope peak The main peak power, and These are the center wavelength and power of the second envelope peak, respectively; for multi-wavelength targets, the third and more dominant spectral envelope peaks are extracted.

[0074] When the target wavelength is constrained to a target wavelength range, the target wavelength matching result is determined based on whether the center wavelength of the current main spectral envelope peak is within the target wavelength range. When the center wavelength of the current main spectral envelope peak deviates from the target wavelength range, the target wavelength proximity is reduced based on the distance between the center wavelength of the current main spectral envelope peak and the boundary of the target wavelength range. Therefore, even if a stable pulse train has been formed in the current output state, if the center wavelength of its main spectral envelope peak deviates from the user-specified target wavelength range, it will not be directly determined to be in a target wavelength mode-locked state.

[0075] S104, determine the current target fitness function and target mode-locking determination conditions based on the target type.

[0076] The current target fitness function is determined from the preset single-wavelength target fitness function, dual-wavelength target fitness function, or multi-wavelength target fitness function according to the target type, and the target mode-locking determination condition is determined according to the target wavelength constraint and the target determination constraint.

[0077] Based on the target state parameters input by the user, the current target fitness function can be expressed as:

[0078] in, Indicates the target state The corresponding current target fitness function; This represents the control parameter vector of an electrically controlled polarization controller or other intracavity adjustable device. Represents the fitness function for a single-wavelength target; Represents the fitness function for a dual-wavelength target; Represents the fitness function for multi-wavelength targets; dual and multi These represent single-wavelength targets, dual-wavelength targets, and multi-wavelength targets, respectively.

[0079] When the target type is a single-wavelength target, the single-wavelength target fitness function is determined as the current target fitness function. The single-wavelength target fitness function calculates the single-wavelength target fitness based on temporal fundamental mode-locking characteristics, target wavelength proximity, main peak intensity, single main peak spectral structure, and noise quality gating terms, and suppresses non-target multi-peak states based on dual-wavelength or multi-wavelength confidence. Therefore, for a single-wavelength target, the current output state not only needs to satisfy temporal fundamental mode-locking stability but also needs to have a single main peak spectral structure that satisfies the target wavelength constraint.

[0080] When the target type is a single-wavelength target, let the user-defined target center wavelength be... The proximity of the target wavelength can be expressed as:

[0081] in, Indicates the proximity of the target wavelength; Indicates the center wavelength of the main spectral envelope peak; This indicates the target center wavelength set by the user. This represents the target wavelength tolerance parameter.

[0082] The corresponding single-wavelength target fitness function can be expressed as:

[0083] in, Indicates the single-wavelength target fitness; Indicates confidence level for a single wavelength; This represents a single-wavelength quality evaluation term consisting of temporal stability, target wavelength proximity, peak intensity, and noise quality gating. Indicates confidence level for dual-wavelength or multi-wavelength applications; The repulsion coefficient representing the non-target multimodal state; clip ( ·,0,1 The parentheses indicate that the value within the brackets is limited to between 0 and 1; the coefficient 100 is used to scale the fitness to a percentage range.

[0084] When the target type is a dual-wavelength target, the dual-wavelength target fitness function is determined as the current target fitness function. The dual-wavelength target fitness function is calculated based on the dual-target wavelength matching degree, peak spacing, power balance, valley depth, peak energy ratio, and noise quality gating term. Therefore, for a dual-wavelength target, the current output state not only needs to have two dominant spectral envelope peaks, but also requires that the two dominant spectral envelope peaks respectively meet the target wavelength combination requirements, and satisfy the spectral structure requirements such as peak spacing, power balance, and valley depth.

[0085] The corresponding dual-wavelength target fitness function can be expressed as:

[0086] in, Represented as dual-wavelength target fitness; Indicates the confidence level of the two wavelengths; It represents a dual-wavelength quality evaluation term consisting of temporal stability, dual-target wavelength matching degree, dual-peak spacing, power balance, valley depth, and noise quality gating; The repulsion coefficient representing the non-target single-wavelength state; clip ( ·, 0 , 1) indicates that the value in parentheses is limited to between 0 and 1; the coefficient 100 is used to scale the fitness to a percentage range.

[0087] When the target type is a multi-wavelength target, the multi-wavelength target fitness function is determined as the current target fitness function. The multi-wavelength target fitness function is calculated based on whether multiple dominant spectral envelope peaks fall within the corresponding target wavelength range, multi-peak power balance, target peak energy proportion, non-target clutter suppression, and temporal stability. Therefore, for a multi-wavelength target, the current output state requires multiple dominant spectral envelope peaks to match the corresponding target wavelength constraints, and also requires suppression of non-target clutter and energy dispersion.

[0088] The corresponding multi-wavelength target fitness function can be expressed as:

[0089] in, Represented as multi-wavelength target fitness; Indicates multi-wavelength confidence; This indicates a multi-wavelength target quality evaluation item; This represents the confidence level of non-target states. Represents the non-target state repulsion coefficient; clip ( ·, 0 , 1) indicates that the value in parentheses is limited to between 0 and 1; the coefficient 100 is used to scale the fitness to a percentage range.

[0090] The target mode-locking determination criteria are determined based on target determination constraints. In one embodiment, the target mode-locking determination criteria include at least: the time-domain basic mode-locking features satisfy the time-domain stability requirement, the frequency-domain target wavelength features satisfy the target wavelength constraint, and the frequency-domain spectral structure features satisfy the spectral structure constraint corresponding to the current target type. The target mode-locking determination criteria may also include a noise quality gating term satisfying a preset quality requirement and a target fitness reaching a preset fitness threshold. Preferably, when all necessary conditions are simultaneously satisfied, the current output state is determined as the target wavelength mode-locking state.

[0091] Through the above method, the embodiments of this application can determine the corresponding evaluation rules based on the target state parameters input by the user. That is, when the user inputs a single-wavelength target, the computer control module uses a single-wavelength target fitness function and a single-wavelength target mode-locking condition; when the user inputs a dual-wavelength target, the computer control module uses a dual-wavelength target fitness function and a dual-wavelength target mode-locking condition; and when the user inputs a multi-wavelength target, the computer control module uses a multi-wavelength target fitness function and a multi-wavelength target mode-locking condition. Therefore, it can avoid misjudgments of target state caused by using a fixed fitness function to evaluate different target types.

[0092] S105, calculate the target fitness and determine whether the current output state is the target wavelength mode-locked state.

[0093] In S105, the computer control module calculates the target fitness corresponding to the current candidate control parameters based on the time-domain basic mode-locking characteristics, frequency-domain target wavelength characteristics, frequency-domain spectral structure characteristics, noise quality gating terms, and the current target fitness function, and determines whether the current output state meets the target mode-locking judgment conditions based on the target fitness.

[0094] Specifically, after obtaining the time-domain basic mode-locking characteristics, frequency-domain target wavelength characteristics, and frequency-domain spectral structure characteristics corresponding to the current candidate control parameters, the computer control module inputs these characteristics into the current target fitness function corresponding to the current target state parameters to obtain the target fitness corresponding to the current candidate control parameters. The target fitness is used to characterize the degree of matching between the current output state and the current target state parameters.

[0095] In one implementation, the calculation of the target fitness further includes a noise quality gating term, which includes a time-domain noise penalty, a frequency-domain noise penalty, and a total noise penalty determined by the time-domain noise penalty and the frequency-domain noise penalty.

[0096] The time-domain noise penalty is used to suppress non-pulse background spikes, excessively strong satellite peaks, or time-domain waveforms with low signal-to-noise ratios. If there are obvious non-pulse background spikes in the current time-domain pulse sequence, excessively strong satellite peaks near the main pulse, or the time-domain signal-to-noise ratio is lower than the preset requirement, the computer control module reduces the target fitness corresponding to the current candidate control parameters, or determines that the current output state does not meet the time-domain stability requirements.

[0097] Frequency domain noise penalty is used to suppress spectral states where the main peak is not obvious, there are too many significant clutter peaks, too many envelope peaks, or the energy is dispersed. If the main spectral envelope peak is not obvious in the current frequency domain spectral data, or there are too many significant clutter peaks, or the spectral energy is not concentrated in the target wavelength region, the computer control module reduces the target fitness corresponding to the current candidate control parameter, or determines that the current output state does not meet the frequency domain spectral structure requirements.

[0098] The total noise penalty is used as a quality gating term in the target fitness calculation for single-wavelength, dual-wavelength, and multi-wavelength targets. In other words, even if the current output state is close to the target wavelength constraint in a certain single feature, if its time-domain noise or frequency-domain noise does not meet the quality requirements, the noise quality gating term will still reduce the probability that the current output state is judged as the target wavelength mode-locked state.

[0099] The total noise penalty can be expressed as:

[0100] in, Indicates the total noise penalty; Indicates time-domain noise penalty; Indicates frequency domain noise penalty; and The total noise penalty is not used to distinguish between single-wavelength, dual-wavelength, or multi-wavelength targets, but rather participates in the fitness calculation of all targets as a quality gating term. When the total noise penalty is too high, even if the current spectral morphology is close to the target state, it will not be determined as the target wavelength mode-locked state.

[0101] In this way, the computer control module does not judge whether mode-locking is successful based solely on whether the time-domain pulse is stable. Instead, it combines the current target state parameters, the current target fitness function, and the target mode-locking determination conditions to determine the target wavelength mode-locking status of the current output state.

[0102] S106, If the current output state is the target wavelength mode-locked state, then maintain the current candidate control parameters.

[0103] In S106, if the current output state meets the target mode-locking determination condition, the computer control module determines the current output state as the target wavelength mode-locking state and maintains the current candidate control parameters.

[0104] Specifically, when the current output state meets the target mode-locking determination condition corresponding to the current target state parameter, it indicates that the current candidate control parameters can enable the passively mode-locked fiber laser output to meet the user's target mode-locking requirements. At this time, the computer control module stops the search process for the current target state parameter and maintains the current candidate control parameters, keeping the passively mode-locked fiber laser in the current target wavelength mode-locked state.

[0105] In one embodiment, the target wavelength mode-locking state may include a single-wavelength target mode-locking state, a dual-wavelength target mode-locking state, or a multi-wavelength target mode-locking state corresponding to the current target state parameters.

[0106] When the current target type is a single-wavelength target, if the current output state forms a stable repetitive pulse train in the time domain, and the center wavelength of the main spectral envelope peak meets the target center wavelength or target wavelength range requirements, and the current spectral structure meets the single main peak spectral structure requirements, then the current output state is determined as a single-wavelength target mode-locked state.

[0107] When the current target type is a dual-wavelength target, if the current output state forms a stable repetitive pulse train in the time domain, and the two dominant spectral envelope peaks respectively meet the target wavelength combination requirements, and the peak spacing, power balance, valley depth and peak energy ratio meet the target determination constraints, then the current output state is determined as the dual-wavelength target mode-locked state.

[0108] When the current target type is a multi-wavelength target, if the current output state forms a stable repetitive pulse train in the time domain, and the multiple dominant spectral envelope peaks respectively satisfy the corresponding target wavelength range, and the multi-peak power balance, target peak energy ratio, non-target clutter suppression and time domain stability satisfy the target determination constraints, then the current output state is determined as the multi-wavelength target mode-locked state.

[0109] In one implementation, after determining that the current output state is the target wavelength mode-locked state, the computer control module can output the current candidate control parameters, the current time-domain pulse sequence, the current frequency-domain spectral data, the current target fitness, and the target mode-locking determination result. Thus, the user can obtain the control parameters that enable the passively mode-locked fiber laser to reach the current target wavelength mode-locked state and the corresponding output results.

[0110] It should be noted that, in the embodiments of this application, "keeping the current candidate control parameters" means that, if the current target state parameters have not changed, the current candidate control parameters are continued to be loaded into the controllable cavity adjustment module, or the current candidate control parameters are saved as the control parameters corresponding to the current target wavelength mode-locking state, so as to maintain or reproduce the current target wavelength mode-locking state.

[0111] In one optional implementation, when the current output state is not the target wavelength mode-locked state, the historical experimental dataset is updated and candidate control parameters for the next round are generated. Specifically, the computer control module constructs the historical experimental dataset from the already evaluated control parameter vectors and their corresponding target fitness values. The historical experimental dataset is used to record the output evaluation results of the passively mode-locked fiber laser under different candidate control parameters and to provide a basis for the generation of subsequent candidate control parameters.

[0112] In one implementation, the historical experimental dataset may include evaluated candidate control parameters, corresponding target fitness values, and optionally, corresponding time-domain basic mode-locking features, frequency-domain target wavelength features, and frequency-domain spectral structure features. By recording the above data, the computer control module can avoid repeatedly testing candidate control parameters that clearly do not meet the target requirements based on historical search results, and prioritize generating the next round of candidate control parameters that are more likely to meet the current target state parameters.

[0113] In one implementation, the computer control module generates multiple sets of candidate control parameters based on historical experimental datasets using a preset optimization algorithm. These candidate control parameters include multiple sets of candidate voltage vectors for electrically controlled polarization controllers. Subsequently, the computer control module sequentially loads the multiple sets of candidate control parameters and executes steps S102 to S105 for each set to determine whether the corresponding output state meets the target mode-locking condition.

[0114] When the preset optimization algorithm is Bayesian optimization, the computer control module establishes a probabilistic surrogate model based on historical experimental datasets to predict the target fitness corresponding to untested control parameters, and generates multiple sets of candidate control parameters based on the probabilistic surrogate model. In this way, the computer control module can predict the target fitness that untested candidate control parameters may obtain based on existing experimental evaluation results, thereby improving the search efficiency of the target wavelength mode-locking state.

[0115] In one implementation, during each round of optimization, the computer control module does not directly output a single set of electrically controlled polarization controller voltages, but instead outputs K sets of candidate electrically controlled polarization controller voltage vectors; preferably, K=4. Each set of candidate electrically controlled polarization controller voltage vectors includes multiple electrically controlled polarization controller channel voltages. The system sequentially loads the above K sets of candidate voltage vectors and acquires the corresponding time-domain pulse sequences and frequency-domain spectral data; if any candidate voltage vector satisfies the target mode-locking determination condition, then the set of candidate voltage vectors is maintained and the current target search ends; otherwise, the K sets of candidate voltage vectors and their target fitness are added to the historical experimental dataset for the next round of model update.

[0116] In other embodiments, the preset optimization algorithm may also include at least one of genetic algorithm, particle swarm optimization algorithm, reinforcement learning algorithm, or random search algorithm. All of the above optimization algorithms can generate candidate control parameters for the next round based on historical experimental datasets or current search results; this application embodiment does not limit this.

[0117] In one optional implementation, the computer control module can generate multiple sets of candidate control parameters within a safe local search region near the current historical best control parameters, and load these sets of candidate control parameters sequentially. The safe local search region is used to limit the range of variation of the candidate control parameters, reducing the risk of drastic fluctuations in the output state or abnormal hardware response caused by excessive changes in the candidate control parameters. The number of multiple sets of candidate control parameters can be set according to the requirements of search efficiency, hardware response speed, and control accuracy.

[0118] To reduce the risk of laser loss of lock or drift away from the target wavelength region due to large jumps in control parameters, a safe local search region can be set:

[0119] in, Indicates a safe local search area; This represents the current historical best voltage vector; Indicates the safe search step size; This represents the voltage feasible region of the four-channel electrically controlled polarization controller. A smaller, safer search step size can be used during the normal fine-search phase; when reliable target candidates are lacking in the historical experimental dataset, a larger step size or global sampling can be used to expand the search range.

[0120] If the current output state corresponding to any candidate control parameter satisfies the target mode-locking condition, the computer control module retains that candidate control parameter and determines the corresponding current output state as the target wavelength mode-locking state. If none of the candidate control parameters make the current output state satisfy the target mode-locking condition, the computer control module continues to generate the next round of candidate control parameters based on the updated historical experimental dataset and repeats the above search process until a target wavelength mode-locking state that satisfies the target mode-locking condition is obtained, or a preset search termination condition is reached.

[0121] The preset search termination condition may include at least one of the following: the number of search rounds reaches a preset maximum, the search time reaches a preset maximum, the target fitness does not significantly improve in multiple consecutive search rounds, or the user actively stops the search. The preset search termination condition is used to avoid unlimited searching when the current target state parameters are difficult to satisfy. It should be noted that the preset search termination condition is only one optional implementation and does not affect the core process of automatically searching for the target wavelength mode-locking state through target state parameters, the current target fitness function, and the target mode-locking determination condition in this application embodiment.

[0122] In one implementation, after the current output state meets the target mode-locking determination condition and the current candidate control parameters are maintained, if the user re-enters or updates the target state parameters, the computer control module will use the updated target state parameters as the new current target state parameters and re-execute. Figure 1 The method flow is shown. The update of the target state parameters may include at least one of the following: target type, target wavelength constraint, or target determination constraint. For example, it may be updated from a single-wavelength target to a dual-wavelength target, or the center wavelength of the first target may be updated to the center wavelength of the second target if the target type is still a single-wavelength target.

[0123] It should be noted that the search process after updating the target state parameters can be regarded as a process of... Figure 1 The method described is executed again, instead of continuing to search for the same target after the previous target's mode-locking condition has been met. Therefore, the same passively mode-locked fiber laser can achieve automatic search and switching between different single-wavelength, dual-wavelength, or multi-wavelength targets without changing the hardware structure.

[0124] To illustrate the effectiveness of the wavelength-controllable automatic mode-locking method and system provided in the embodiments of this application, the following is combined with... Figures 3 to 7 The mode-locking states for different target wavelengths and their search process are explained. It should be noted that... Figures 3 to 7 The specific wavelengths, spectral curves, time-domain pulse sequences, fitness changes, confidence changes, and two-dimensional spectral evolution results shown are merely exemplary embodiments and do not constitute a limitation on the scope of protection of this application.

[0125] Reference Figure 3 , Figure 3 The results of the mode-locking state of the first single-wavelength target are shown. Among them, Figure 3 (a) in the figure is the frequency domain spectrum curve corresponding to the mode-locked state of the first single-wavelength target. Figure 3 (b) in the table is the time-domain pulse sequence corresponding to the mode-locked state of the first single-wavelength target.

[0126] In this implementation, the target state parameters input by the user correspond to a first single-wavelength target. The computer control module, based on the target type, target wavelength constraints, and target determination constraints corresponding to the first single-wavelength target, determines the single-wavelength target fitness function as the current target fitness function, and obtains candidate control parameters that satisfy the target mode-locking determination conditions through closed-loop search. Figure 3 It can be seen that under the candidate control parameters, the time-domain pulse sequence output by the passively mode-locked fiber laser forms a stable repetitive pulse train, and the frequency-domain spectral data exhibits a single-peak spectral structure that meets the requirements of the first single-wavelength target. The center wavelength of the main spectral envelope peak satisfies the corresponding target wavelength constraint. Therefore, the current output state can be determined as the first single-wavelength target mode-locking state.

[0127] Reference Figure 4 , Figure 4 The results of the second single-wavelength target mode-locking state are shown. Among them, Figure 4 (a) in the figure is the frequency domain spectrum curve corresponding to the mode-locked state of the second single-wavelength target. Figure 4 (b) in the figure is the time-domain pulse sequence corresponding to the mode-locked state of the second single-wavelength target.

[0128] In this implementation, the target state parameters input by the user correspond to a second single-wavelength target. The second single-wavelength target and the first single-wavelength target may have different target center wavelengths or different target wavelength ranges. The computer control module re-executes based on the target state parameters corresponding to the second single-wavelength target. Figure 1 The process is shown, and the single-wavelength target fitness function is determined as the current target fitness function. Figure 4 It can be seen that, under the candidate control parameters obtained through the re-search, the time-domain pulse sequence output by the passively mode-locked fiber laser meets the time-domain stability requirements, and the center wavelength of the main spectral envelope peak in the frequency-domain spectral data meets the target wavelength constraint corresponding to the second single-wavelength target. Therefore, the current output state can be determined as the mode-locked state of the second single-wavelength target.

[0129] Depend on Figure 3 and Figure 4 As can be seen, under the same passive mode-locked fiber laser hardware structure, the embodiments of this application can search for the single-wavelength target mode-locking state corresponding to different target wavelengths based on different target state parameters, rather than simply determining whether the laser has entered the normal mode-locking state.

[0130] Reference Figure 5 , Figure 5 The results of the dual-wavelength target mode-locking state are shown. Among them, Figure 5 In the image (a), the frequency domain spectrum curve corresponding to the mode-locked state of the dual-wavelength target is shown. Figure 5 (b) in the figure is the time-domain pulse sequence corresponding to the mode-locked state of the dual-wavelength target.

[0131] In this implementation, the target state parameters input by the user correspond to a dual-wavelength target, and the target wavelength constraint includes a combination of target wavelengths consisting of two target center wavelengths or two target wavelength ranges. The computer control module determines the dual-wavelength target fitness function as the current target fitness function based on the target type, target wavelength constraint, and target determination constraint corresponding to the dual-wavelength target, and calculates the dual-wavelength target fitness based on the dual-target wavelength matching degree, peak spacing, power balance, valley depth, peak energy ratio, and noise quality gating term.

[0132] Depend on Figure 5 It can be seen that, under the candidate control parameters that satisfy the dual-wavelength target mode-locking determination conditions, the time-domain pulse sequence output by the passively mode-locked fiber laser meets the basic time-domain mode-locking stability requirements, and the frequency-domain spectral data exhibits a dual-dominant spectral envelope peak structure. The two dominant spectral envelope peaks respectively satisfy the target wavelength combination requirements, and the peak spacing, power balance, and valley depth satisfy the target determination constraints corresponding to the dual-wavelength target. Therefore, the current output state can be determined as the dual-wavelength target mode-locking state.

[0133] Depend on Figure 5 As can be seen, the embodiments of this application do not simply take any bimodal spectral state as the dual-wavelength target mode-locked state, but need to combine the target wavelength combination, bimodal spectral structure and temporal stability for joint judgment, thereby reducing the probability of non-target bimodal states or noisy multimodal states being misjudged as target wavelength mode-locked states.

[0134] Reference Figure 6 , Figure 6 The results of fitness and confidence changes during the automatic search for the target state are shown. Figure 6 In the diagram, (a) represents the result of the change in the target fitness. Figure 6 (b) in the figure represents the result of the confidence level change.

[0135] In this implementation, the computer control module continuously generates candidate control parameters based on the current target state parameters and evaluates the output state corresponding to each candidate control parameter. As the search rounds increase and the historical experimental dataset is continuously updated, a preset optimization algorithm generates candidate control parameters for the next round that are more likely to satisfy the current target state parameters based on the historical experimental dataset. Figure 6It can be seen that during the search process, the target fitness corresponding to the current target state parameters gradually increases, and the current output state gradually approaches the target wavelength mode-locked state.

[0136] Furthermore, Figure 6 The confidence level changes can be used to characterize the evaluation changes between different output state categories or different target states. For example, when the current target type is a single-wavelength target, the confidence level corresponding to the single-wavelength target increases during the search process, while the confidence level corresponding to non-target states such as dual-wavelength or multi-wavelength targets is suppressed; when the current target type is a dual-wavelength target, the confidence level corresponding to the dual-wavelength target increases during the search process, while the confidence level corresponding to states that do not meet the requirements of dual-wavelength target wavelength combination or spectral structure is suppressed. Therefore, the embodiments of this application can guide the search process to converge to the corresponding target wavelength mode-locked state based on the current target state parameters.

[0137] Furthermore, Figure 6 In (a), the original fitness represents the target fitness corresponding to each candidate control parameter, the phased optimal fitness represents the highest target fitness obtained up to the current evaluation sequence number in the current target search phase, the threshold represents the preset fitness threshold used to determine whether the current output state meets the target mode-locking judgment condition, and the success mark in the figure indicates that the output state under the corresponding candidate control parameter meets the target mode-locking judgment condition. Figure 6 The initial seed in the table represents the candidate control parameters loaded in the initial search phase. The 1530 target state represents a single-wavelength target search phase with 1530nm as the target center wavelength, the 1560 target state represents a single-wavelength target search phase with 1560nm as the target center wavelength, and the dual-wavelength target state represents a dual-wavelength target search phase using a combination of 1530nm and 1560nm as target wavelengths. The 1530nm, 1560nm, and dual-wavelength target combinations are merely exemplary target wavelength settings and do not constitute a limitation on the target wavelength range or combination.

[0138] Figure 6 In (b), the single-wavelength confidence level represents the confidence level that the current output state belongs to the single-wavelength target mode-locked state, and the dual-wavelength confidence level represents the confidence level that the current output state belongs to the dual-wavelength target mode-locked state. The single-wavelength confidence level and the dual-wavelength confidence level are used to help characterize the evaluation changes of different target states during the search process, and are not used as the sole basis for judging whether the current output state meets the target mode-locking determination conditions.

[0139] Reference Figure 7 , Figure 7 The target wavelength range and two-dimensional spectral evolution results during the target state switching process are shown. Figure 7 (a) in the figure represents the result of the target wavelength range variation. Figure 7 (b) in the figure represents the two-dimensional spectral evolution result.

[0140] In this implementation, the user can sequentially input or update different target state parameters. For each input or updated target state parameter, the computer control module uses it as the new current target state parameter and re-executes the process. Figure 1 The method flow is shown. Figure 7 The target wavelength range variation results are used to illustrate the target wavelength constraint changes corresponding to different search stages, and the two-dimensional spectral evolution results are used to illustrate the evolution process of the passive mode-locked fiber laser output spectrum as the candidate control parameter search and target state parameter changes occur. Specifically, Figure 7 In (a) and These represent the wavelength trajectories of the first and second targets, respectively. When the target state parameters correspond to a single-wavelength target... It can represent the center wavelength of the current single-wavelength target. This can represent a second target wavelength that is not yet enabled or a candidate target wavelength for subsequent dual-wavelength target switching; when the target status parameter corresponds to a dual-wavelength target, and These represent the center wavelengths of the two targets in a dual-wavelength target combination. Figure 7 In (b), the horizontal axis represents wavelength, the vertical axis represents evaluation sequence number, and grayscale or color level represents optical power at the corresponding wavelength. The stage markers such as initial, Bayesian optimization, 1530nm search, 1560nm search, and dual-wavelength search are used to illustrate the different search stages corresponding to the updated target state parameters. Among them, the 1530nm, 1560nm, and dual-wavelength search stages are only illustrative examples and do not limit this application to the above-mentioned target wavelengths.

[0141] Depend on Figure 7 It can be seen that when the target state parameters change, the system can re-evaluate the current output state based on the new target wavelength constraints and target decision constraints, and search for a new target wavelength mode-locked state by generating candidate control parameters for the next round. In other words, changes in the target state parameters will cause the current target fitness function and target mode-locked decision conditions to be redefined, thereby shifting the search process from the previous target wavelength mode-locked state to the new target wavelength mode-locked state.

[0142] It should be noted that, Figure 7 The target state switching process shown is used to illustrate how the same system can switch states multiple times. Figure 1The method flow shown can realize automatic search and switching between different target wavelength mode-locking states. For other target center wavelengths, target wavelength ranges, target wavelength combinations, or multi-wavelength targets, the embodiments of this application can also redetermine the current target fitness function and target mode-locking determination conditions according to the updated target state parameters, and perform a closed-loop search process to obtain the corresponding target wavelength mode-locking state.

[0143] In summary, this embodiment of the application can determine the current target fitness function based on the target type after the user inputs the target state parameters, determine the target mode-locking determination conditions based on the target wavelength constraints and target decision constraints, and evaluate the current output state in conjunction with time-domain pulse sequence and frequency-domain spectral data. When the current output state meets the target mode-locking determination conditions, the system determines the target wavelength mode-locking state and maintains the current candidate control parameters; when the current output state does not meet the target mode-locking determination conditions, the system generates the next round of candidate control parameters based on historical experimental datasets and continues the search. Therefore, this embodiment of the application can realize automatic search for single-wavelength, dual-wavelength, or multi-wavelength target wavelength mode-locking states, and can switch between different target wavelength mode-locking states by re-executing the method flow when the target state parameters change.

[0144] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A wavelength-controllable automatic mode-locking method, characterized in that, include: The system obtains target state parameters input by the user. The target state parameters include target type, target wavelength constraint, and target determination constraint. The target type is a single-wavelength target, a dual-wavelength target, or a multi-wavelength target. The target wavelength constraint includes at least one of the target center wavelength, target wavelength range, or target wavelength combination. Candidate control parameters are loaded into the controllable cavity adjustment module of the passively mode-locked fiber laser, and the time-domain pulse sequence and frequency-domain spectral data output by the passively mode-locked fiber laser are acquired; wherein, the controllable cavity adjustment module includes at least an electrically controlled polarization controller, and the candidate control parameters include the multi-channel driving voltage vector of the electrically controlled polarization controller; Based on the time-domain pulse sequence, time-domain basic mode-locking features are extracted, and based on the frequency-domain spectral data, frequency-domain target wavelength features and frequency-domain spectral structure features are extracted. The frequency-domain target wavelength features include at least one of target wavelength proximity, target wavelength range matching result, or target wavelength set matching degree. The frequency-domain spectral structure features include at least one of spectral envelope peak number, peak spacing, power balance, valley depth, number of significant impurity peaks, or energy concentration. Based on the target type, the current target fitness function is determined from the preset single-wavelength target fitness function, dual-wavelength target fitness function, or multi-wavelength target fitness function, and the target mode-locking determination condition is determined based on the target wavelength constraint and the target determination constraint. Based on the time-domain basic mode-locking characteristics, the frequency-domain target wavelength characteristics, the frequency-domain spectral structure characteristics, and the current target fitness function, the target fitness corresponding to the current candidate control parameters is calculated. Based on the target fitness and the target mode-locking determination conditions, it is determined whether the output state of the passively mode-locked fiber laser under the current candidate control parameters is the target wavelength mode-locked state. When the output state of the passively mode-locked fiber laser under the current candidate control parameters is the target wavelength mode-locked state, the current candidate control parameters are maintained.

2. The wavelength-controllable automatic mode-locking method according to claim 1, characterized in that, The target wavelength constraint includes a set of target wavelengths and an allowable range of target wavelength deviations. The target determination constraint includes spectral structure constraints and temporal stability constraints related to the target state. The target wavelength set includes one target center wavelength when the target type is a single-wavelength target, two target center wavelengths when the target type is a dual-wavelength target, and multiple target center wavelengths when the target type is a multi-wavelength target.

3. The wavelength-controllable automatic mode-locking method according to claim 1, characterized in that, When the target wavelength constraint is a target wavelength range, the target wavelength matching result is determined based on whether the center wavelength of the current main spectral envelope peak is within the target wavelength range; when the center wavelength of the current main spectral envelope peak deviates from the target wavelength range, the target wavelength proximity is reduced based on the distance between the center wavelength of the current main spectral envelope peak and the boundary of the target wavelength range.

4. The wavelength-controllable automatic mode-locking method according to claim 1, characterized in that, The time-domain basic mode-locking features include at least one of pulse interval stability, pulse peak uniformity, time-domain signal-to-noise ratio, satellite peak ratio, or time-domain noise features; wherein, the time-domain basic mode-locking features are used to determine whether the current output forms a stable repeating pulse train.

5. The wavelength-controllable automatic mode-locking method according to claim 1, characterized in that, Extracting frequency domain target wavelength features and frequency domain spectral structure features from the frequency domain spectral data, including: The frequency domain spectral data is filtered for effective range, retaining the spectral window that contains the target wavelength region and removing invalid low-power fill values; The filtered spectral data are smoothed to obtain the spectral envelope; The main spectral envelope peak is detected on the spectral envelope, and the center wavelength, peak power, and number of spectral envelope peaks of the main spectral envelope peak are extracted.

6. The wavelength-controllable automatic mode-locking method according to claim 1, characterized in that, The calculation of the target fitness also includes a noise quality gating term, which includes a time-domain noise penalty, a frequency-domain noise penalty, and a total noise penalty determined by the time-domain noise penalty and the frequency-domain noise penalty; wherein, the total noise penalty is used as a quality gating term in the calculation of the target fitness for single-wavelength targets, dual-wavelength targets, and multi-wavelength targets.

7. The wavelength-controllable automatic mode-locking method according to claim 1, characterized in that, When the target type is a single-wavelength target, the single-wavelength target fitness function is determined as the current target fitness function; the single-wavelength target fitness function calculates the single-wavelength target fitness based on the time-domain basic mode-locking characteristics, target wavelength proximity, main peak intensity, single main peak spectral structure and noise quality gating term, and suppresses the non-target multi-peak state based on dual-wavelength or multi-wavelength confidence.

8. The wavelength-controllable automatic mode-locking method according to claim 1, characterized in that, When the target type is a dual-wavelength target, the dual-wavelength target fitness function is determined as the current target fitness function; when the target type is a multi-wavelength target, the multi-wavelength target fitness function is determined as the current target fitness function; wherein, the dual-wavelength target fitness function calculates the dual-wavelength target fitness based on the dual-target wavelength matching degree, the bi-peak interval, power balance, valley depth, bi-peak energy ratio, and noise quality gating term, and the multi-wavelength target fitness function calculates the multi-wavelength target fitness based on whether multiple dominant spectral envelope peaks fall into the corresponding target wavelength range, multi-peak power balance, target peak energy ratio, non-target clutter suppression, and temporal stability.

9. The wavelength-controllable automatic mode-locking method according to claim 1, characterized in that, The method further includes: If the output state of the passively mode-locked fiber laser under the current candidate control parameters is not the target wavelength mode-locked state, then the current candidate control parameters and their target fitness are added to the historical experimental dataset. The historical experimental dataset is constructed by the evaluated control parameter vectors and their corresponding target fitness values. Multiple sets of candidate control parameters are generated based on the historical experimental dataset using a preset optimization algorithm. The multiple sets of candidate control parameters include multiple sets of candidate voltage vectors for electrically controlled polarization controllers. Wherein, when the preset optimization algorithm is the Bayesian optimization algorithm, a probabilistic surrogate model is established based on the historical experimental dataset to predict the target fitness corresponding to the untested control parameters, and the multiple sets of candidate control parameters are generated according to the probabilistic surrogate model; The multiple sets of candidate control parameters are loaded sequentially, and the process of loading candidate control parameters and acquiring the time-domain pulse sequence and the frequency-domain spectral data is returned to continue searching for the target wavelength mode-locked state corresponding to the current target state parameters. If the output state corresponding to any candidate control parameter is the target wavelength mode-locked state, then the candidate control parameter is maintained. The preset optimization algorithm includes at least one of Bayesian optimization algorithm, genetic algorithm, particle swarm optimization algorithm, reinforcement learning algorithm or random search algorithm.

10. A wavelength-controllable automatic mode-locking system, characterized in that, include: Passive mode-locked fiber lasers are used to output optical signals under the control of a controllable cavity modulation module. A controllable intracavity adjustment module is used to receive candidate control parameters and adjust the intracavity state of the passive mode-locked fiber laser. The controllable intracavity adjustment module includes at least an electrically controlled polarization controller. The time-domain signal acquisition module is used to acquire the time-domain pulse sequence output by the passively mode-locked fiber laser; A frequency domain spectral acquisition module is used to acquire the frequency domain spectral data output by the passively mode-locked fiber laser; The computer control module is used to acquire the target state parameters input by the user, determine the current target fitness function from the preset single-wavelength target fitness function, dual-wavelength target fitness function or multi-wavelength target fitness function according to the target type, calculate the target fitness based on the time-domain pulse sequence and the frequency-domain spectral data, and determine the target wavelength mode-locking state according to the target fitness and the target mode-locking determination condition.

Images