Laser parameter optimization method, system, equipment and medium

Abstract

The invention discloses a laser parameter optimization method, system and device and a medium, and the method comprises the steps: constructing a laser numerical model and a parameter feasible region of a target laser; performing numerical simulation on the laser numerical model to obtain a plurality of laser working parameter groups and a pulse evolution sequence of each laser working parameter group; according to all the laser working parameter groups and the pulse evolution sequence, training an initialized pulse sequence prediction model to obtain a trained pulse sequence prediction model; and according to the parameter feasible region and the trained pulse sequence prediction model, performing working parameter optimization processing to obtain a working parameter optimization combination of the target laser. According to the method, the optimization cost of the laser parameters can be effectively reduced, and the optimization efficiency and effect of the laser parameters can be improved. The invention relates to the technical field of lasers.

Description

Technical Field

[0001] This invention relates to the field of laser technology, and in particular to a method, system, device, and medium for optimizing laser parameters. Background Technology

[0002] Mamyshev oscillators, through a synergistic mechanism of self-phase modulation and misaligned filtering, constitute virtual saturable absorbers with high modulation depth, capable of generating ultrashort pulses with high energy and high peak power. They are widely used in cutting-edge scientific research fields such as precision machining and biomedical imaging. However, the superior performance of Mamyshev oscillators is highly dependent on the precise matching and synergistic optimization of several key parameters. These parameters exhibit strong nonlinear coupling relationships, forming a complex high-dimensional parameter space, making it difficult to find the global optimal solution for the laser.

[0003] Currently, the methods for optimizing the parameters of Mamyshev oscillators can be broadly categorized into two types. The first type typically uses the controlled variable method to adjust parameters one by one. This method is prone to getting trapped in local optima and relies heavily on human experience, resulting in poor optimization effectiveness and efficiency. The second type typically uses numerical simulation to assist in optimizing the parameters of the Mamyshev oscillator. However, high-fidelity numerical simulations require numerous iterations and step-by-step calculations, making each simulation extremely time-consuming. This method is not very efficient and has a high optimization cost.

[0004] Therefore, the problems with the relevant technologies still need to be solved and optimized. Summary of the Invention

[0005] The purpose of this invention is to at least partially solve one of the technical problems existing in the related art.

[0006] Therefore, one objective of this invention is to provide a method, system, device, and medium for optimizing laser parameters, wherein the method can effectively reduce the cost of laser parameter optimization and improve the efficiency and effectiveness of laser parameter optimization.

[0007] To achieve the above-mentioned technical objectives, the technical solutions adopted in the embodiments of this application include: In a first aspect, embodiments of this application provide a method for optimizing laser parameters, including: Construct a numerical model of the target laser and its feasible parameter region; Numerical simulation was performed on the laser numerical model to obtain several laser operating parameter groups and pulse evolution sequences for each laser operating parameter group; Based on all the laser operating parameter sets and the pulse evolution sequence, the initialized pulse sequence prediction model is trained to obtain the trained pulse sequence prediction model. Based on the feasible region of the parameters and the trained pulse sequence prediction model, the working parameters are optimized to obtain the optimal combination of the working parameters of the target laser.

[0008] In addition, the method according to the above embodiments of this application may also have the following additional technical features: Further, in one embodiment of this application, the step of training the initialized pulse sequence prediction model based on all the laser operating parameter sets and the pulse evolution sequence to obtain a trained pulse sequence prediction model includes: The laser operating parameter set and the pulse evolution sequence are input into the initialized pulse sequence prediction model to perform pulse prediction, thereby obtaining the first pulse prediction sequence; Based on the pulse evolution sequence, a loss analysis is performed on the first pulse prediction sequence to obtain the target loss value; Based on the target loss value, the parameters of the initialized pulse sequence prediction model are updated to obtain the trained pulse sequence prediction model.

[0009] Further, in one embodiment of this application, the step of inputting the laser operating parameter set and the pulse evolution sequence into the initialized pulse sequence prediction model for pulse prediction to obtain a first pulse prediction sequence includes: The laser operating parameter set is subjected to a nonlinear transformation to obtain the parameter feature vector; Based on the parameter feature vector, a broadcast addition operation is performed on the pulse evolution sequence to obtain a parameter enhancement sequence; The parameter enhancement sequence is subjected to temporal feature extraction and prediction processing to obtain the first pulse prediction sequence.

[0010] Furthermore, in one embodiment of this application, the step of performing operational parameter optimization processing based on the feasible region of parameters and the trained pulse sequence prediction model to obtain an optimized combination of operational parameters for the target laser includes: Based on the feasible region of the parameters, an initialization population is determined. The initialization population includes a plurality of first particles, each of which is used to characterize a set of laser operating parameters of the target laser within the feasible region of the parameters. Based on the trained pulse sequence prediction model, the initial population is optimized to obtain the optimal particle of the population. Based on the optimal particles in the population, the optimal combination of operating parameters for the target laser is determined.

[0011] Furthermore, in one embodiment of this application, the step of optimizing the initial population based on the trained pulse sequence prediction model to obtain the optimal particle includes: Obtain a third particle and several second particles, wherein the second particles are either the first particles or the fourth particles from the previous optimization process; the third particle is either the middle particle of the population from the previous optimization process or the first particle with the best position among all the first particles. All the second particles are input into the trained pulse sequence prediction model to perform sequence prediction, and the second pulse prediction sequence of each second particle is obtained. Based on all the second pulse prediction sequences, the third particle is optimized to obtain the population intermediate particle in the current optimization process; Based on all the second pulse prediction sequences, perform individual optimization on all the second particles to obtain several fourth particles in the current optimization process; The optimal particle of the population is determined based on the intermediate particles of the population.

[0012] Furthermore, in one embodiment of this application, the step of individually optimizing all second particles based on all second pulse prediction sequences to obtain several fourth particles in the current optimization process includes: Obtain the particle optimization function and the historical best fitness of the second particle; Fitness analysis is performed on the second pulse prediction sequence to obtain the particle fitness of the second particle; The fitness of the particle is compared with the historical best fitness to obtain the fitness comparison result; If the fitness comparison result is that the particle fitness is greater than the historical best fitness, then the historical best fitness is updated according to the particle fitness, and the second particle is optimized according to the particle optimization function to obtain the fourth particle; or, if the fitness comparison result is that the particle fitness is less than or equal to the historical best fitness, then the second particle is optimized according to the particle optimization function to obtain the fourth particle, and the historical best fitness is retained.

[0013] Further, in one embodiment of this application, the step of performing population-particle optimization on the third particle based on all the second pulse prediction sequences to obtain the population intermediate particle in the current optimization process includes: Obtain the population-optimal fitness of the third particle; Fitness analysis was performed on all the second pulse prediction sequences to obtain the fitness of several particles; Based on the optimal fitness of the population, the fitness of all particles is screened for maximum fitness to obtain the target fitness. Based on the target fitness, the third particle is updated to obtain the intermediate particles of the population.

[0014] Secondly, embodiments of this application provide a laser parameter optimization system, comprising: The first processing unit is used to construct the laser numerical model and parameter feasible region of the target laser. The second processing unit is used to perform numerical simulation on the laser numerical model to obtain several laser operating parameter groups and pulse evolution sequences for each laser operating parameter group. The third processing unit is used to train the initialized pulse sequence prediction model based on all the laser operating parameter groups and the pulse evolution sequence to obtain the trained pulse sequence prediction model. The fourth processing unit is used to perform working parameter optimization processing based on the feasible region of the parameters and the trained pulse sequence prediction model to obtain the optimized combination of working parameters of the target laser.

[0015] Thirdly, embodiments of this application also provide an electronic device, including: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor performs the method described above.

[0016] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a processor-executable program, which, when executed by the processor, is used to implement the above-described method.

[0017] The advantages and beneficial effects of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application: This application discloses a method, system, device, and medium for optimizing laser parameters. The method involves constructing a numerical model and a feasible region for the parameters of a target laser; performing numerical simulation on the laser numerical model to obtain several sets of laser operating parameters and a pulse evolution sequence for each set; training an initialized pulse sequence prediction model based on all the laser operating parameter sets and the pulse evolution sequences to obtain a trained pulse sequence prediction model; and performing operating parameter optimization based on the feasible region and the trained pulse sequence prediction model to obtain an optimized combination of operating parameters for the target laser. This method, within the feasible region, applies the trained pulse sequence prediction model to the optimization process of the target laser's operating parameters, thereby obtaining an optimized combination of operating parameters. It can effectively reduce the optimization cost of laser parameters and improve the efficiency and effectiveness of laser parameter optimization. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the following description is provided with accompanying drawings of the relevant technical solutions in the embodiments of this application or the prior art. It should be understood that the accompanying drawings described below are only for the purpose of clearly illustrating some embodiments of the technical solutions in this application. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0019] Figure 1 A flowchart illustrating a method for optimizing laser parameters provided in this application embodiment; Figure 2 A schematic diagram of the framework of a laser parameter optimization system provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0020] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application. The step numbers in the following embodiments are set only for ease of explanation, and there is no limitation on the order between the steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0022] Currently, the parameter optimization methods for Mamyshev oscillators can be broadly categorized into two types. The first type typically employs the controlled variable method to adjust parameters one by one. However, due to the complex nonlinear coupling between parameters, this "isolated" optimization strategy is prone to getting trapped in local optima, making it difficult to find the globally optimal performance point, resulting in poor optimization performance. At the same time, the entire optimization process heavily relies on the operator's experience, leading to low efficiency, poor repeatability, and long-term equipment debugging, which incurs high time, resource, and manpower costs, thus resulting in poor optimization efficiency.

[0023] The second approach typically uses numerical simulation to assist in optimizing the parameters of the Mamyshev oscillator. However, high-fidelity numerical simulation requires a large number of iterations and step-by-step calculations, and a single simulation is extremely time-consuming. Furthermore, the computational cost of performing a systematic scan or optimization search of the high-level parameter space increases exponentially. This approach is not very efficient and has a high optimization cost.

[0024] It should be noted that the aforementioned related technologies are only used to assist in understanding the technical solutions of this application and do not mean that they belong to the publicly disclosed prior art.

[0025] In view of this, embodiments of this application provide a method, system, device, and medium for optimizing laser parameters. The method, within the feasible parameter domain, applies a trained pulse sequence prediction model to the optimization process of the target laser's operating parameters. Specifically, in the population optimization process, the trained pulse sequence prediction model predicts the sequence of several second particles in the current optimization round, and performs individual and / or population optimization based on the predicted second pulse sequences. This method, on the one hand, uses a trained pulse sequence prediction model to replace time-consuming single numerical simulations, reducing parameter evaluation time from hours / minutes to milliseconds, making global optimization searches in high-dimensional parameter spaces possible, achieving fully automated optimization, completely eliminating reliance on operator experience, and effectively improving the optimization efficiency of laser parameters.

[0026] On the other hand, this method combines the fast and accurate performance prediction provided by the trained pulse sequence prediction model during the population optimization process, which can effectively escape local optima, thereby helping to find a more globally optimal solution in the population optimization process, and thus fully explore the performance potential of the Mamyshev oscillator, that is, effectively improve the optimization effect of laser parameters; and can significantly reduce the time consumption, computational resource consumption and human cost in the laser parameter optimization process.

[0027] Reference Figure 1 In this application embodiment, a method for optimizing laser parameters includes: Step 110: Construct the numerical model of the target laser and the feasible region of its parameters; In this embodiment, the target laser can be a Mamyshev oscillator, whose numerical model can be accurately described by coupling the generalized nonlinear Schrödinger equation (GNLSE) with the steady-state rate equation. The generalized nonlinear Schrödinger equation is the standard governing equation describing the propagation of optical pulses in optical fibers, and its form is as follows:

[0028]

[0029] in, It is the slowly varying envelope complex amplitude of the pulse. It is the distance the pulse travels in the optical fiber. The loss coefficient is... , These are the dispersion coefficients for second- and third-order group velocities. These are nonlinear coefficients. For the center frequency, This is the gain coefficient. It is a nonlinear response function, and its expression is shown in the above equation. It simultaneously includes the instantaneous Kerr effect (self-phase modulation SPM) and the delayed Raman effect (stimulated Raman scattering SRS). It is the Raman response coefficient. It is a step function. and These are the fitting parameters for the Raman gain spectrum.

[0030] When the gain medium of the target laser is ytterbium-doped fiber, its steady-state rate equations (RE) are used to calculate the rare-earth ions (such as Yb) in the gain medium. 3+ The population of particles in the energy level of ) is expressed by the following core expression:

[0031] in, The number density of particles in the upper energy level. The total doping concentration of ytterbium ions is given by , h is Planck's constant, and c is the speed of light in vacuum. It is the core area. It is the fill factor at wavelength K. and The wavelength-dependent absorption and emission cross sections of ytterbium-doped ions Represents the laser power of the k-th wavelength channel. This refers to the lifetime of the upper energy level.

[0032] It is understandable that there are already various implementations for the coupled solution of the generalized nonlinear Schrödinger equation and the steady-state rate variance, which will not be elaborated upon here. Furthermore, the feasible region of parameters can be a set of reasonable ranges for various operating parameters of the target laser. For example, for one type of feasible region of parameters, the feasible region of parameters of the target laser may include the pump power. The range of this parameter (e.g.) ), gain fiber length The range of this parameter (e.g.) ), passive fiber length The range of this parameter (e.g.) and filter offset The range of this parameter (e.g.) ).

[0033] Step 120: Perform numerical simulation on the laser numerical model to obtain several laser operating parameter groups and pulse evolution sequence for each laser operating parameter group; In this embodiment of the application, the numerical simulation can first be carried out by randomly sampling within the feasible domain of the target laser to generate several sets of parameter combinations, and then inputting each set of parameter combinations (i.e., the laser operating parameter set) into the laser numerical model for independent simulation. Each simulation runs until the pulse output reaches a stable mode-locked state, and the output pulse sequence of each evolution (i.e., the pulse evolution sequence) is extracted.

[0034] Step 130: Train the initialized pulse sequence prediction model based on all the laser operating parameter groups and the pulse evolution sequence to obtain the trained pulse sequence prediction model; In this embodiment, each laser operating parameter group and its corresponding pulse evolution sequence can be used as a training data group, and each training data group can be input into an initialized pulse sequence prediction model for training. The pulse sequence prediction model can be an LSTM model, thereby obtaining a trained pulse sequence prediction model.

[0035] Understandably, the pulse evolution sequence output by the original numerical simulation has extremely high temporal resolution. Directly using all data points would not only drastically increase the computational complexity and training time of the model, but may also introduce noise and make it difficult for the model to capture the key macroscopic features of the pulse. Therefore, in one feasible implementation, the pulse evolution sequence can be downsampled to ensure that key physical characteristics are not lost, and the effective output range of the pulse is truncated with the pulse peak output as the center, effectively reducing the dimensionality of the data; and the downsampled pulse evolution sequence can be resampled to a fixed dimension using interpolation to ensure that the sample input dimension of the pulse evolution sequence is consistent.

[0036] In some embodiments, training the initialized pulse sequence prediction model based on all the laser operating parameter sets and the pulse evolution sequence to obtain a trained pulse sequence prediction model includes: The laser operating parameter set and the pulse evolution sequence are input into the initialized pulse sequence prediction model to perform pulse prediction, thereby obtaining the first pulse prediction sequence; Further, the step of inputting the laser operating parameter set and the pulse evolution sequence into the initialized pulse sequence prediction model for pulse prediction to obtain a first pulse prediction sequence includes: The laser operating parameter set is subjected to a nonlinear transformation to obtain the parameter feature vector; Based on the parameter feature vector, a broadcast addition operation is performed on the pulse evolution sequence to obtain a parameter enhancement sequence; The parameter enhancement sequence is subjected to temporal feature extraction and prediction processing to obtain the first pulse prediction sequence.

[0037] In this embodiment, the parameter form of the laser operating parameter set can be input into a fully connected layer (DenseLayer) for nonlinear transformation to map its dimension to a parameter feature vector that matches the data dimension of the pulse evolution sequence. Then, the parameter feature vector and the pulse evolution sequence are subjected to a broadcast addition operation, that is, the same prior information vector is added to the data of each time step in the pulse sequence to form an enhanced sequence input that incorporates physical parameters, denoted as the parameter enhancement sequence.

[0038] Understandably, the temporal feature extraction and prediction process can first involve inputting the parameter-enhanced sequence into a two-layer Long Short-Term Memory (LSTM) network. The LSTM network's gating mechanism captures and learns the long-term temporal dependencies in the pulse evolution process, thereby achieving temporal feature extraction. Then, a fully connected layer is used to predict the output of the LSTM network, thus obtaining the first pulse prediction sequence predicted by the pulse sequence prediction model.

[0039] Based on the pulse evolution sequence, a loss analysis is performed on the first pulse prediction sequence to obtain the target loss value; Based on the target loss value, the parameters of the initialized pulse sequence prediction model are updated to obtain the trained pulse sequence prediction model.

[0040] In this embodiment, loss analysis can assess the degree of difference between the first predicted pulse sequence predicted by the pulse sequence prediction model and the simulated pulse evolution sequence, thereby obtaining the target loss value. This target loss value can be calculated based on a loss function. Many types of loss functions are commonly used, such as 0-1 loss function, squared loss function, absolute loss function, logarithmic loss function, cross-entropy loss function, and mean squared error loss function. This embodiment uses the mean squared error function as an example, and the target loss value can be expressed as:

[0041] in, It can be the target loss value or a functional representation of the mean squared error function; For training batch size, For sequence length, Let be the predicted value of the k-th first pulse prediction sequence at the n-th time point; This represents the true value of the corresponding pulse evolution sequence at the nth time point.

[0042] Understandably, based on the target loss value trained on the training dataset, the backpropagation algorithm is used to update the model's parameters. After several iterations, a well-trained pulse sequence prediction model can be obtained. The specific number of iterations can be preset, or training can be considered complete when the required accuracy is achieved on the test set.

[0043] Step 140: Based on the feasible region of the parameters and the trained pulse sequence prediction model, perform working parameter optimization processing to obtain the optimized combination of working parameters of the target laser.

[0044] In this embodiment, the working parameter optimization process can be based on several laser working parameter groups within the parameter feasible domain, combined with a trained pulse sequence prediction model, to optimize the parameters of the laser working parameter groups, thereby obtaining an optimized combination of working parameters for the target laser.

[0045] In some embodiments, the step of performing operating parameter optimization processing based on the feasible region of parameters and the trained pulse sequence prediction model to obtain an optimized combination of operating parameters for the target laser includes: Based on the feasible region of the parameters, an initialization population is determined. The initialization population includes a plurality of first particles, each of which is used to characterize a set of laser operating parameters of the target laser within the feasible region of the parameters. In this embodiment of the application, several laser operating parameter sets can be randomly generated within the parameter feasible region, and an initial population corresponding to each laser operating parameter set can be determined based on the Particle Swarm Optimization (PSO) algorithm. The position of each first particle in the population represents a laser operating parameter set. = .

[0046] Based on the trained pulse sequence prediction model, the initial population is optimized to obtain the optimal particle of the population. Further, the step of optimizing the initial population based on the trained pulse sequence prediction model to obtain the optimal particle includes: Obtain a third particle and several second particles, wherein the second particles are either the first particles or the fourth particles from the previous optimization process; the third particle is either the middle particle of the population from the previous optimization process or the first particle with the best position among all the first particles. All the second particles are input into the trained pulse sequence prediction model to perform sequence prediction, and the second pulse prediction sequence of each second particle is obtained. In this embodiment, population optimization can be performed iteratively on the initial population to determine the optimal particle in the final particle population. Specifically, if the current optimization round is the first population optimization process, the second particle can be the first particle, and the third particle can be the first particle with the best position among all the first particles. This optimal position can be obtained by performing fitness analysis on each first particle. Alternatively, if the current optimization round is the second population optimization process, the second particle can be the first particle, and the third particle can be the first particle with the best position among all the first particles. This optimal position can be obtained by performing fitness analysis on each first particle.

[0047] It is understandable that, for any optimization round, each second particle of the current optimization round can be input into the trained pulse sequence prediction model, and the trained pulse sequence prediction model can predict the second pulse prediction sequence of each second particle.

[0048] Based on all the second pulse prediction sequences, the third particle is optimized to obtain the population intermediate particle in the current optimization process; Further, the step of performing population-particle optimization on the third particle based on all the second pulse prediction sequences to obtain the population intermediate particle in the current optimization process includes: Obtain the population-optimal fitness of the third particle; Fitness analysis was performed on all the second pulse prediction sequences to obtain the fitness of several particles; Based on the optimal fitness of the population, the fitness of all particles is screened for maximum fitness to obtain the target fitness. Based on the target fitness, the third particle is updated to obtain the intermediate particles of the population.

[0049] In this embodiment of the application, for any second particle in the current optimization round, the fitness analysis can be based on the second pulse prediction sequence of the second particle to determine the particle fitness of the second particle. The particle fitness can be the output pulse energy of the second particle, which can be expressed as:

[0050] in, The output pulse energy of the second particle is also known as the particle fitness of the second particle. This is a time interval parameter, which can be determined by the simulation time window and the number of sampling points; This is the predicted value of the second pulse prediction sequence at the nth time point.

[0051] Understandably, maximum fitness selection can involve comparing the population-optimal fitness of the third particle with the particle fitness of each second particle, and determining the highest fitness among all fitness values ​​(such as the population-optimal fitness or particle fitness) as the target fitness. Particle updating can be based on the particle corresponding to the target fitness to obtain the population's intermediate particles. Specifically, if the particle corresponding to the target fitness is the third particle, then particle updating can be to determine the third particle as the population's intermediate particle; or, if the particle corresponding to the target fitness is the second particle, then the corresponding second particle can be used to replace the third particle, thereby obtaining the population's intermediate particles.

[0052] Based on all the second pulse prediction sequences, perform individual optimization on all the second particles to obtain several fourth particles in the current optimization process; Further, the step of individually optimizing all second particles based on all second pulse prediction sequences to obtain several fourth particles in the current optimization process includes: Obtain the particle optimization function and the historical best fitness of the second particle; Fitness analysis is performed on the second pulse prediction sequence to obtain the particle fitness of the second particle; The fitness of the particle is compared with the historical best fitness to obtain the fitness comparison result; If the fitness comparison result is that the particle fitness is greater than the historical best fitness, then the historical best fitness is updated according to the particle fitness, and the second particle is optimized according to the particle optimization function to obtain the fourth particle; or, if the fitness comparison result is that the particle fitness is less than or equal to the historical best fitness, then the second particle is optimized according to the particle optimization function to obtain the fourth particle, and the historical best fitness is retained.

[0053] In this embodiment, for any second particle in the current optimization round, the historical best fitness can be the best fitness of that second particle in previous optimization rounds, that is, the largest particle fitness among all particle fitnesss in different previous optimization rounds; the particle optimization function is used to update the velocity and position of each second particle, thereby realizing the optimization update of the laser operating parameter group corresponding to each second particle. The particle optimization function can be expressed as:

[0054] in, To optimize the velocity of the fourth particle in round t, that is, to optimize the velocity of the second particle in round t+1; Inertial weights control the degree to which the particle's original velocity is inherited. To optimize the velocity of the second particle in round t; and These are the acceleration weights that control the particles to adjust towards the individual optimal and the global optimal directions, respectively. and It is a random number between [0,1], used to simulate the randomness of group behavior; The position of the second particle corresponding to the historical best fitness; This represents the position of the third particle corresponding to the optimal fitness of the population in the current optimization round. To optimize the second particle at round t; To optimize the fourth particle in round t, that is, to optimize the second particle in round t+1.

[0055] Understandably, for any second particle in the current optimization round, after obtaining its fitness, the particle fitness can be compared with its historical best fitness to obtain a fitness comparison result. Specifically, if the fitness comparison result shows that the second particle's fitness is greater than the historical best fitness, then the historical best fitness is replaced and updated based on the particle fitness, and the updated historical best fitness is used as the historical best fitness for the next optimization round. Furthermore, based on the particle optimization function, the velocity and position of the second particle are updated to obtain the fourth particle. Alternatively, if the fitness comparison result shows that the particle fitness is less than or equal to the historical best fitness, then the historical best fitness remains unchanged and is used as the historical best fitness for the next optimization round. Furthermore, based on the particle optimization function, the velocity and position of the second particle are updated to obtain the fourth particle.

[0056] It should be noted that, in order to achieve a good balance between global and local exploration, in one feasible implementation, the initial value of the inertia weight can be... And it decreases linearly with the increase of the number of optimization rounds. In the early stages, a larger inertia weight is beneficial for global exploration, while a smaller weight in the later stages promotes fine-grained local search. In other words, the inertia weight has a linear negative correlation with the number of optimization rounds; while the acceleration weight is set to... , to satisfy The convergence conditions are determined, and the group is ensured to have a certain level of social learning ability.

[0057] The optimal particle of the population is determined based on the intermediate particles of the population.

[0058] In this embodiment, if the population optimization process meets the termination condition (such as the current optimization round reaching a preset maximum optimization round threshold, or the rate of change of the global optimal fitness value being less than a preset fitness threshold, where the global optimal fitness value can be the difference between the population optimal fitness values ​​of two adjacent optimization rounds), then the latest intermediate particle (i.e., the intermediate particle of the current optimization round) can be determined as the population optimal particle; or, if the population optimization process does not meet the termination condition, then the population optimization process of the next optimization round can continue to be executed.

[0059] Based on the optimal particles in the population, the optimal combination of operating parameters for the target laser is determined.

[0060] In this embodiment of the application, after determining the population optimization particle, the laser operating parameter group corresponding to the position of the population optimization particle can be determined as the target laser operating parameter optimization combination.

[0061] Reference Figure 2 The laser parameter optimization system proposed in this application includes... The first processing unit 101 is used to construct the laser numerical model and parameter feasible region of the target laser. The second processing unit 102 is used to perform numerical simulation on the laser numerical model to obtain several laser operating parameter groups and pulse evolution sequence of each laser operating parameter group. The third processing unit 103 is used to train the initialized pulse sequence prediction model based on all the laser operating parameter groups and the pulse evolution sequence to obtain the trained pulse sequence prediction model. The fourth processing unit 104 is used to perform working parameter optimization processing based on the feasible region of the parameters and the trained pulse sequence prediction model to obtain the optimized combination of working parameters of the target laser.

[0062] Reference Figure 3 This application also provides an electronic device, including: At least one processor 201; At least one memory 202 is used to store at least one program; When the at least one program is executed by the at least one processor 201, the at least one processor 201 implements the method embodiment described above.

[0063] Similarly, it can be understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0064] This application also provides a computer-readable storage medium storing a program executable by a processor 201, which, when executed by the processor 201, is used to implement the above-described method embodiments.

[0065] Similarly, the content of the above method embodiments is applicable to the present computer-readable storage medium embodiments. The specific functions implemented by the present computer-readable storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0066] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0067] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods.

[0068] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order shown in the operation diagrams. For example, depending on the functions / operations involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts of this application are provided by way of example to provide a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is changed and sub-operations described as part of a larger operation are executed independently.

[0069] Furthermore, although this application is described in the context of functional modules, it should be understood that, unless otherwise stated to the contrary, one or more of the functions and / or features may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding this application. Rather, given the properties, functions, and internal relationships of the various functional modules in the apparatus disclosed herein, the actual implementation of the module will be understood within the scope of conventional technology for an engineer. Therefore, those skilled in the art can implement the application set forth in the claims using ordinary techniques without excessive experimentation. It is also understood that the specific concepts disclosed are merely illustrative and not intended to limit the scope of this application, which is determined by the full scope of the appended claims and their equivalents.

[0070] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods in the embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0071] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0072] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0073] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0074] In the foregoing description of this specification, the references to terms such as "one embodiment," "another embodiment," or "some embodiments," etc., indicate that a specific feature, structure, material, or characteristic described in connection with an embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0075] Although embodiments of this application have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the claims and their equivalents.

[0076] The above is a detailed description of the preferred embodiments of this application, but this application is not limited to the embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this application, and these equivalent modifications or substitutions are all included within the scope defined by the claims of this application.

Claims

1. A method for optimizing laser parameters, characterized in that, include: Construct a numerical model of the target laser and its feasible parameter region; Numerical simulation was performed on the laser numerical model to obtain several laser operating parameter groups and pulse evolution sequences for each laser operating parameter group; Based on all the laser operating parameter sets and the pulse evolution sequence, the initialized pulse sequence prediction model is trained to obtain the trained pulse sequence prediction model. Based on the feasible region of the parameters and the trained pulse sequence prediction model, the working parameters are optimized to obtain the optimal combination of the working parameters of the target laser.

2. The method according to claim 1, characterized in that, The step of training the initialized pulse sequence prediction model based on all the laser operating parameter sets and the pulse evolution sequence to obtain a trained pulse sequence prediction model includes: The laser operating parameter set and the pulse evolution sequence are input into the initialized pulse sequence prediction model to perform pulse prediction, thereby obtaining the first pulse prediction sequence; Based on the pulse evolution sequence, a loss analysis is performed on the first pulse prediction sequence to obtain the target loss value; Based on the target loss value, the parameters of the initialized pulse sequence prediction model are updated to obtain the trained pulse sequence prediction model.

3. The method according to claim 2, characterized in that, The step of inputting the laser operating parameter set and the pulse evolution sequence into the initialized pulse sequence prediction model to perform pulse prediction, and obtaining a first pulse prediction sequence, includes: The laser operating parameter set is subjected to a nonlinear transformation to obtain the parameter feature vector; Based on the parameter feature vector, a broadcast addition operation is performed on the pulse evolution sequence to obtain a parameter enhancement sequence; The parameter enhancement sequence is subjected to temporal feature extraction and prediction processing to obtain the first pulse prediction sequence.

4. The method according to claim 1, characterized in that, The step of optimizing the operating parameters of the target laser based on the feasible region of the parameters and the trained pulse sequence prediction model to obtain the optimized combination of operating parameters includes: Based on the feasible region of the parameters, an initialization population is determined. The initialization population includes a plurality of first particles, each of which is used to characterize a set of laser operating parameters of the target laser within the feasible region of the parameters. Based on the trained pulse sequence prediction model, the initial population is optimized to obtain the optimal particle of the population. Based on the optimal particles in the population, the optimal combination of operating parameters for the target laser is determined.

5. The method according to claim 4, characterized in that, The step of optimizing the initial population based on the trained pulse sequence prediction model to obtain the optimal particle includes: Obtain a third particle and several second particles, wherein the second particles are either the first particles or the fourth particles from the previous optimization process; the third particle is either the middle particle of the population from the previous optimization process or the first particle with the best position among all the first particles. All the second particles are input into the trained pulse sequence prediction model to perform sequence prediction, and the second pulse prediction sequence of each second particle is obtained. Based on all the second pulse prediction sequences, the third particle is optimized to obtain the population intermediate particle in the current optimization process; Based on all the second pulse prediction sequences, perform individual optimization on all the second particles to obtain several fourth particles in the current optimization process; The optimal particle of the population is determined based on the intermediate particles of the population.

6. The method according to claim 5, characterized in that, The step of individually optimizing all second particles based on all second pulse prediction sequences to obtain several fourth particles in the current optimization process includes: Obtain the particle optimization function and the historical best fitness of the second particle; Fitness analysis is performed on the second pulse prediction sequence to obtain the particle fitness of the second particle; The fitness of the particle is compared with the historical best fitness to obtain the fitness comparison result; If the fitness comparison result is that the particle fitness is greater than the historical best fitness, then the historical best fitness is updated according to the particle fitness, and the second particle is optimized according to the particle optimization function to obtain the fourth particle; or, if the fitness comparison result is that the particle fitness is less than or equal to the historical best fitness, then the second particle is optimized according to the particle optimization function to obtain the fourth particle, and the historical best fitness is retained.

7. The method according to claim 5, characterized in that, The step of performing population optimization on the third particle based on all the second pulse prediction sequences to obtain the intermediate particles of the population in the current optimization process includes: Obtain the population-optimal fitness of the third particle; Fitness analysis was performed on all the second pulse prediction sequences to obtain the fitness of several particles; Based on the optimal fitness of the population, the fitness of all particles is screened for maximum fitness to obtain the target fitness. Based on the target fitness, the third particle is updated to obtain the intermediate particles of the population.

8. A laser parameter optimization system, characterized in that, include: The first processing unit is used to construct the laser numerical model and parameter feasible region of the target laser. The second processing unit is used to perform numerical simulation on the laser numerical model to obtain several laser operating parameter groups and pulse evolution sequences for each laser operating parameter group. The third processing unit is used to train the initialized pulse sequence prediction model based on all the laser operating parameter groups and the pulse evolution sequence to obtain the trained pulse sequence prediction model. The fourth processing unit is used to perform working parameter optimization processing based on the feasible region of the parameters and the trained pulse sequence prediction model to obtain the optimized combination of working parameters of the target laser.

9. An electronic device, characterized in that, include: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor performs the method as described in any one of claims 1-7.

10. A computer-readable storage medium storing a processor-executable program, characterized in that, The processor-executable program, when executed by the processor, is used to implement the method as described in any one of claims 1-7.

Images