Multi-wavelength semiconductor laser variable light spot output control method and system

Abstract

The invention discloses a multi-wavelength semiconductor laser variable light spot output control method and system, and relates to the technical field of laser output control. The multi-wavelength semiconductor laser variable light spot output control method comprises the steps that S1, laser parameter data are collected for preprocessing operation, and after storage, a multi-wavelength laser output database is constructed; s2, performing beam combination and light spot shaping on the multi-wavelength laser, constructing a loss time sequence model, and realizing correction and output of estimated energy density of each wavelength; s3, executing power compensation and pulse parameter adjustment operation according to a deviation analysis result; s4, constructing an energy correction decision model and outputting a target surface energy distribution prediction result and energy distribution characteristics; and S5, performing energy distribution evaluation through energy distribution prediction and energy distribution characteristics. The problems that in an existing multi-wavelength laser system, energy distribution is not uniform, optical path loss compensation is insufficient, and nonlinear energy correction precision is low are solved.

Description

Technical Field

[0001] This invention relates to the field of laser output control technology, specifically to a method and system for output control of a variable spot size in a multi-wavelength semiconductor laser. Background Technology

[0002] Multi-wavelength semiconductor lasers, due to their advantages such as flexible wavelength selection, adjustable output power, small size, and compact structure, are widely used in medical, testing, communication, and materials processing fields. In medical settings, multi-wavelength combinations are often used for layered irradiation of superficial and deep tissues, vascular-related treatments, and intraoperative bleeding control, placing stringent requirements on spot diameter adjustment, spot shape control, and target surface energy density distribution. In materials processing and testing, as well as optical communication, multi-wavelength semiconductor lasers are used for surface defect identification, area marking, and localized heat treatment, as well as for wavelength division multiplexing (WDM) and link testing. They need to maintain stable output power, spot parameters, and spatial energy distribution across all wavelengths during continuous operation, thus supporting multifunctional laser devices for complex applications. Existing multi-wavelength semiconductor laser devices, when used for large-area target irradiation, complex contour processing, and multi-channel transmission, face even more stringent engineering requirements regarding the consistency of multi-wavelength energy distribution, the ability to adjust the deformable spot size, and long-term operational stability.

[0003] For example, the invention patent with announcement number CN110323663B discloses a device and method for generating vector ultrashort laser pulses in the mid-infrared band. This device and method employ a gain module with a high-power, high-brightness pump source and rare-earth ion-doped sesquioxide ceramic as core components to achieve laser output in the corresponding mid-infrared band. Then, a mode-locking element is used to achieve pulsed laser output, and a birefringence mode selection mechanism using a ring aperture is used to achieve the output and control of the vector beam. Furthermore, a highly nonlinear medium is used to manage intracavity nonlinearity, and a chirped mirror is used to manage dispersion. Finally, a circular grating is used as a waveguide output coupling mirror to obtain an all-solid-state ultrafast laser output with vector characteristics in the mid-infrared band. The obtained radially polarized vector beam in the mid-infrared band has a unique "polarization singularity" effect; and the obtained pulse width is on the order of optical periods, which has wide applications in cutting-edge fields such as high-precision laser medical surgery, super-resolution imaging, quantum communication, and attosecond light sources.

[0004] For example, the invention patent with publication number CN101350498B discloses a control method for a semiconductor laser. The semiconductor laser includes multiple wavelength selection units having different wavelength characteristics, and the semiconductor laser is mounted on a temperature control device. The method of the present invention includes: a first step of correcting the temperature of the temperature control device according to the detected output wavelength of the semiconductor laser; and a second step of controlling at least one of the wavelength selection units to reduce the difference in the change amount between the wavelength characteristics of the multiple wavelength selection units, the difference in change amount being caused by correcting the temperature of the temperature control device.

[0005] However, as application scenarios place increasingly higher demands on the uniformity and stability of energy distribution, methods that rely solely on static parameter configuration or empirical rules for beam spot adjustment are no longer adequate to meet the dynamic changes required under multi-wavelength and multi-parameter coupling conditions. Existing technologies generally lack the ability to jointly analyze multi-wavelength power information, beam spot parameters, and energy feedback data, making it difficult to effectively predict the energy distribution state of the target surface and to make proactive adjustments before energy distribution anomalies occur.

[0006] Therefore, in order to address the above problems, there is an urgent need for a method and system for output control of variable spot size in multi-wavelength semiconductor lasers. Summary of the Invention

[0007] Technical problems to be solved

[0008] To address the shortcomings of existing technologies, this invention provides a method and system for output control of variable spot size in multi-wavelength semiconductor lasers, which solves the problems of uneven energy distribution, insufficient optical path loss compensation, and low accuracy of nonlinear energy correction in existing multi-wavelength laser systems.

[0009] Technical solution

[0010] To achieve the above objectives, the present invention provides the following technical solution: a method and system for output control of a variable spot size in a multi-wavelength semiconductor laser, comprising: S1, acquiring laser parameter data, performing preprocessing operations, and storing the laser parameter data to construct a multi-wavelength laser output database; S2, constructing a loss timing model by combining and shaping the multi-wavelength laser beams, and realizing the correction and output of the estimated energy density for each wavelength; S3, performing energy deviation analysis using the actual energy density, estimated energy density, nominal power, and optical path loss data of the multi-wavelength laser, and performing power compensation and pulse parameter adjustment operations based on the deviation analysis results; S4, constructing an energy correction decision model by fusing multi-wavelength power, spot parameters, and energy feedback information, and outputting the target surface energy distribution prediction results and energy distribution characteristics; S5, evaluating the energy distribution based on the energy distribution prediction and energy distribution characteristics, and driving an electric zoom lens and an adjustable shape aperture to adjust the spot diameter, spot shape, and scanning mode.

[0011] Further, the specific measures for preprocessing laser parameter data acquisition are as follows: Acquiring laser parameter data: Through the laser drive unit, power detection device, and control recording interface, the nominal power value, real-time output power, and corresponding drive current, pulse frequency, and pulse width configuration of each wavelength laser are acquired in real time. The acquisition results are then bound to the current laser wavelength parameters to form wavelength-level power and drive status data. Continuous imaging of the target area is performed using a spot imaging sensor, extracting information on the spot shape, size, outline boundary, and coverage area in real time. Simultaneously, the actual energy density of each wavelength laser on the target surface is obtained using an energy density sensor. The refractive index, beam incident angle, and polarization state of the optical elements are acquired. The transmittance and reflectance at each interface are calculated using Fresnel's formula, and the transmission and reflection losses are accumulated step-by-step. The scattering coefficient of the material is obtained and multiplied by the beam propagation length in the path to obtain the scattering loss. The total optical path loss of each wavelength laser from the source to the target surface is calculated by fusing the transmission loss, reflection loss, and scattering loss.

[0012] Furthermore, the specific measures for constructing a multi-wavelength laser output database after storing laser parameter data are as follows: The numerical data of nominal power, real-time output power, drive current, pulse frequency, pulse width, spot size, actual energy density, and total optical path loss are processed to unify units. Power-related data are uniformly converted to watts, energy density is uniformly converted to energy per unit area, and spot size is uniformly assigned to a length-based unit. After unit unification, outlier detection is performed on each numerical data point. Based on statistical distribution characteristics, sampling points that significantly deviate from the normal operating range are identified and removed. Data gaps caused by communication and sampling interruptions are filled using the average of adjacent valid sampling points. Subsequently, the continuous characteristics of nominal power, real-time output power, actual energy density, total optical path loss, and drive parameters are mapped to a unified range using a maximum-minimum-value normalization method. Categorical characteristics such as spot shape are vectorized using one-hot encoding to eliminate the influence of dimensional differences. All laser data after standardization and normalization are archived into the multi-wavelength laser output database.

[0013] Furthermore, the specific measures for beam combining and spot shaping of multi-wavelength lasers are as follows: multi-wavelength lasers are combined using a dichroic mirror group: using a multi-fiber combiner with spectral separation function and a dichroic mirror group, the lasers of each wavelength are separated and guided to the same path; each wavelength of laser is separated by the dichroic mirror and converged to a unified beam path through optical fibers and lenses, and the total optical path loss is dynamically calculated. If the total optical path loss is greater than the loss threshold, the optical path correction process is triggered; the optical shaping unit includes an adjustable shape stop and an electric zoom lens. The adjustable shape stop is used to control the shape of the laser beam to adapt to different irradiation requirements, and the electric zoom lens adjusts the spot diameter range to provide an adjustable range.

[0014] Furthermore, the specific measures for constructing a loss time series model to achieve the correction and output of estimated energy density for each wavelength are as follows: A multivariate time series feature set is constructed from laser wavelength, spot shape, spot size, and energy density data, serving as the input loss analysis unit. The causal relationship of energy loss for each wavelength is clarified by combining optical path and environmental conditions, forming an initial loss correlation set. The optical path mechanism includes the refractive index variation law of optical elements, the scattering effect in the beam path, and the influence of environmental temperature and humidity on optical path loss. This loss relationship is determined through the physical characteristics of the optical elements and transformed into prior constraints for dynamic Bayesian network structure learning: environmental temperature and humidity node strength... The system directs the refractive index node and scattering node, forcing the refractive index node to point towards the transmission loss node and the scattering node to point towards the scattering loss node. Edges that contradict known physical causal directions are prohibited to ensure the network topology conforms to a directed acyclic graph and the causal order of energy transfer. A dynamic Bayesian network is used to learn the loss structure, characterizing the evolution of energy loss across time steps for different wavelengths of laser light in the optical path, as well as the feedback relationship between the optical path state and subsequent laser energy transfer. Direct influence relationships, indirect transmission relationships, and time lag relationships are identified. A loss time series model containing multi-time-step dependent features is constructed, and the loss for each wavelength is compensated, outputting the estimated energy density for each wavelength.

[0015] Furthermore, the specific measures for energy deviation analysis using the actual energy density, estimated energy density, nominal power, and optical path loss data of multi-wavelength lasers are as follows: obtain the first... The actual energy density, estimated energy density, nominal power, and total optical path loss of the laser at each wavelength are calculated; the calculation of the energy density of the laser at the first wavelength is performed. The difference between the actual and estimated energy densities of lasers at each wavelength is used to obtain the energy deviation of that wavelength under the current operating conditions. The effective input energy is obtained by subtracting the total optical path loss from the first wavelength and multiplying by the nominal power value. The ratio of the energy deviation to the effective input energy is then calculated to obtain the energy deviation of the first wavelength. Energy correction value for each wavelength of laser.

[0016] Furthermore, the specific measures for power compensation and pulse parameter adjustment based on the deviation analysis results are as follows: By comparing the energy correction value with the correction threshold in real time, when the energy correction value is less than the correction threshold, the drive current setting, pulse frequency, and pulse width corresponding to the current wavelength remain unchanged, and only the energy density collected in this cycle is recorded, and the wavelength is marked as an energy stable state; when the energy correction value is greater than or equal to the correction threshold, power compensation is performed: the drive current is adjusted to correct the energy gap caused by optical path loss in the input power compensation, and an energy density resampling is triggered after the adjustment is completed; when the drive current adjustment reaches the maximum adjustable drive current but the energy correction value is still greater than or equal to the correction threshold, the control strategy... Switch to pulse optimization strategy: Keep the pulse frequency constant, adjust the single pulse energy by changing the pulse width, and redistribute the energy input per unit area by changing the spot diameter; when the difference between two consecutive energy correction values ​​exceeds the energy change threshold and the difference between two consecutive optical path loss values ​​is greater than the loss change threshold within m consecutive cycles, it is determined that the abnormal change in energy correction value and the synchronous increase in optical path loss are triggered, the optical path recalibration process is updated to update the path loss parameters, the power and pulse compensation process is re-executed, and the energy correction value is recalculated. If the energy correction value is still greater than or equal to the correction threshold within n consecutive cycles, the current wavelength channel is marked as an energy abnormal channel, the energy feedback parameters are recorded, and the output settings are limited to retain only the monitoring function.

[0017] Furthermore, the specific measures taken to construct an energy correction decision model by fusing multi-wavelength power, spot parameters, and energy feedback information, and to output the target surface energy distribution prediction results and energy distribution characteristics, are as follows: The nominal power value, real-time output power, spot shape, spot size, energy correction value, and energy feedback parameters for each wavelength are used as input features to construct multi-dimensional training samples; the energy distribution data of each wavelength at the same time and on the same target surface are used to construct a multi-channel two-dimensional image matrix, where each channel represents a grayscale image of a spot at an independent wavelength, and its pixel value is proportional to the energy density at that point under that wavelength; the multi-dimensional training samples and the multi-channel two-dimensional image matrix are input into a deep learning network for end-to-end feature learning and mapping; and real-time... The collected and updated data are used to train a convolutional neural network to extract the coupling relationship between each input parameter and the actual energy output. The nonlinear deviation of energy distribution under different wavelengths, different spot parameters, and dynamic conditions is observed. A support vector machine is used to establish a nonlinear mapping relationship between power adjustment amplitude, pulse adjustment mode, spot adjustment direction, and energy correction effect. During operation, new energy correction results and adjustment execution records are continuously received and incrementally updated to construct an energy correction decision model. The model outputs the predicted energy distribution values ​​and energy distribution characteristics of each wavelength in the target area. The specific energy distribution characteristics include the energy difference characteristics between the center and the edge, the energy difference characteristics in the left and right directions, the energy difference characteristics in the up and down directions, and the local energy loss characteristics.

[0018] Furthermore, energy distribution is assessed through energy distribution prediction and energy distribution characteristics. Specific measures to drive the motorized zoom lens and adjustable diaphragm to adjust the spot diameter, spot shape, and scanning mode are as follows: By comparing the predicted energy distribution value with the distribution threshold in real time, when the predicted energy distribution value is less than the distribution threshold, it is determined that the energy distribution in the target area is within the allowable deviation range. The control unit maintains the current spot diameter, diaphragm shape, and scanning mode unchanged, and records the current spot state and corresponding energy distribution as a stable operating condition reference and archives it to the multi-wavelength laser output database. When the predicted energy distribution value is greater than or equal to the distribution threshold, it is determined that there is a sparse energy distribution in the current target area. The energy distribution characteristic results are read, and a spot optimization adjustment strategy is executed: When the energy difference characteristics between the center and the edge are displayed, the motorized zoom lens is controlled to change the spot diameter, so that the energy per unit area is redistributed from the center area to the edge area, alleviating the central energy... The system addresses the issues of excessive energy concentration and insufficient edge energy. When left-right energy differences are observed, the lateral opening parameter of the adjustable aperture is adjusted to stretch and compress the spot in the left-right direction, allowing energy to be rebalanced and cover the target area horizontally. When up-down energy differences are observed, the longitudinal opening parameter of the aperture is corrected, and the focal position of the zoom lens is slightly adjusted to redistribute energy vertically. When localized energy loss is observed, the spot shape and size are adjusted in combination to extend the spot outline towards the energy-deficient area, enhancing the energy coverage intensity of the missing area. After each adjustment of the spot diameter, shape, and aperture opening, the energy correction decision model is called to update the predicted energy distribution values ​​and energy distribution characteristics of each wavelength in the target area, and the distribution threshold comparison is performed again, continuously iterating the spot optimization adjustment process until the predicted energy distribution value is less than the distribution threshold.

[0019] Furthermore, a second aspect of the present invention provides an output control system for a multi-wavelength semiconductor laser with a variable spot size, applied to an output control method for a multi-wavelength semiconductor laser with a variable spot size, comprising: a laser data acquisition and standardization module, used to acquire laser parameter data, perform preprocessing operations, and store the laser parameter data to construct a multi-wavelength laser output database; an optical path correction module, used to construct a loss timing model by combining and shaping the multi-wavelength laser beams, and to realize the correction and output of the estimated energy density of each wavelength; a laser energy calibration module, used to perform energy deviation analysis using the actual energy density, estimated energy density, nominal power, and optical path loss data of the multi-wavelength laser, and to perform power compensation and pulse parameter adjustment operations based on the deviation analysis results; an energy distribution prediction module, used to construct an energy correction decision model by fusing multi-wavelength power, spot parameters, and energy feedback information, and to output the target surface energy distribution prediction results and energy distribution characteristics; and a spot shape and size adjustment module, used to evaluate the energy distribution through energy distribution prediction and energy distribution characteristics, and to drive an electric zoom lens and an adjustable shape aperture to adjust the spot diameter, spot shape, and scanning mode.

[0020] Beneficial effects

[0021] The present invention has the following beneficial effects:

[0022] (1) This invention constructs a loss time series model by utilizing the optical path and environmental conditions of multi-wavelength lasers. Based on real-time collected laser wavelength, spot shape and energy density data, the energy density estimation of each wavelength is dynamically adjusted; thereby realizing dynamic correction and accurate compensation of optical path loss, effectively solving the problem of difficult accurate control of multi-wavelength laser energy loss in the prior art.

[0023] (2) The present invention analyzes the deviation between the actual energy density and the estimated energy density of multi-wavelength lasers in real time, and adopts power compensation and pulse parameter adjustment operations to dynamically adjust the laser output parameters; thereby achieving accurate energy correction and solving the problem of large deviation between nominal power value and actual energy in the prior art.

[0024] (3) This invention constructs an energy correction decision model by combining multi-wavelength power, spot parameters and energy feedback information, and updates and optimizes it in real time through deep learning algorithms; thereby realizing intelligent energy distribution prediction and optimization, and solving the problems of uneven spot distribution and unreasonable energy allocation in the prior art.

[0025] (4) The present invention effectively optimizes the spot shape of laser output by adjusting the spot diameter, spot shape and scanning method according to the predicted value of energy distribution and energy distribution characteristics; thereby achieving precise spot adjustment and solving the problem that the spot shape and energy distribution cannot be precisely controlled in the prior art.

[0026] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0027] Figure 1 This is a flowchart of a variable spot output control method for a multi-wavelength semiconductor laser according to the present invention;

[0028] Figure 2 This is a structural diagram of an output control system for a multi-wavelength semiconductor laser with variable spot size according to the present invention;

[0029] Figure 3 This is a flowchart of the multi-wavelength laser energy multi-level modulation and calibration process of the present invention;

[0030] Figure 4 This is a comparison diagram of the characteristics and adjustment actions of the multi-wavelength light spot energy distribution problem in this invention. Detailed Implementation

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] Please see Figures 1-4 This invention provides a technical solution: a method and system for output control of a variable spot size in a multi-wavelength semiconductor laser, comprising: S1, acquiring laser parameter data, performing preprocessing operations, and storing the laser parameter data to construct a multi-wavelength laser output database; S2, constructing a loss timing model by combining and shaping the multi-wavelength laser beams, and realizing the correction and output of the estimated energy density for each wavelength; S3, performing energy deviation analysis using the actual energy density, estimated energy density, nominal power, and optical path loss data of the multi-wavelength laser, and performing power compensation and pulse parameter adjustment operations based on the deviation analysis results; S4, constructing an energy correction decision model by fusing multi-wavelength power, spot parameters, and energy feedback information, and outputting the target surface energy distribution prediction results and energy distribution characteristics; S5, evaluating the energy distribution based on the energy distribution prediction and energy distribution characteristics, and driving the motorized zoom lens and adjustable shape aperture to adjust the spot diameter, spot shape, and scanning mode.

[0033] Specifically, the preprocessing measures for acquiring laser parameter data are as follows: Acquiring laser parameter data: Through the laser drive unit, power detection device, and control recording interface, the nominal power value, real-time output power, and corresponding drive current, pulse frequency, and pulse width configuration of each wavelength laser are acquired in real time. The acquired parameters are then precisely matched according to the wavelength dimension, and the acquisition results are bound to the current laser wavelength parameters to form structured wavelength-level power and drive status data. Continuous dynamic imaging of the core action area of ​​the target surface is performed using a high-resolution spot imaging sensor. Image recognition algorithms are used to extract the spot shape, spot size, spot outline boundary, and spot coverage area information in real time. Simultaneously, this is combined with data from a high-precision energy density sensor. The actual energy density distribution data of lasers of various wavelengths at different positions on the target surface are obtained; the key parameters of refractive index, beam incident angle and polarization state of optical elements are obtained, and the transmittance and reflectance at each interface of the optical system are calculated based on Fresnel formula. The transmission loss and reflection loss of each interface are accumulated step by step according to the beam propagation path to obtain the total transmission and reflection loss of the entire link; the scattering coefficient of the material is obtained, and the scattering loss of the corresponding propagation stage is obtained by multiplying the interaction distance between the beam and various materials in the transmission path, i.e. the propagation length of the beam in the path, and the interaction distance between the beam and various materials in the propagation path. Finally, the transmission loss, reflection loss and scattering loss data are integrated, and the total optical path loss of each wavelength laser from the source emission end to the target surface interaction end is calculated by weighted summation.

[0034] In this implementation scheme, through multi-dimensional and high-precision parameter acquisition and loss calculation, the entire link state of lasers of various wavelengths from the source to the target surface is quantitatively characterized. It not only integrates direct parameters such as laser drive configuration, power output and spot shape, but also combines the characteristics of optical components and material propagation properties to accurately calculate various losses such as transmission, reflection and scattering. Finally, a complete wavelength-level data system covering the entire process of laser emission, transmission and action is formed, providing comprehensive and accurate data support for subsequent laser performance optimization, loss control and target surface effect analysis.

[0035] Specifically, the measures taken to construct a multi-wavelength laser output database after storing laser parameter data are as follows: For the numerical data of nominal power, real-time output power, drive current, pulse frequency, pulse width, spot size, actual energy density, and total optical path loss, a unified unit processing is performed. Power-related data are uniformly converted to watts, energy density is uniformly converted to energy per unit area, and spot size is uniformly adopted using length as the standard unit. After unit unification, outlier detection is performed on each numerical data point. Based on statistical distribution characteristics, sampling points that significantly deviate from the normal operating range are identified and removed. For data gaps caused by communication and sampling interruptions, a limited sliding window length and supplementation are used. The measurement sequence is completed by averaging adjacent valid sampling points at the full scale upper limit to avoid altering the original statistical characteristics of the measurement sequence using a single averaging method. Subsequently, for continuous features such as nominal power, real-time output power, actual energy density, total optical path loss, and driving parameters, a maximum and minimum value normalization method is used to map them to a unified interval. For categorical features such as spot shape, one-hot encoding is used for vectorization processing. At the same time, the field structure specifications, data versioning, and timestamp synchronization strategies corresponding to the laser control scenario are bound to ensure the traceability and scenario adaptability of normalization and encoding operations, eliminating the impact of dimensional differences on subsequent data applications. All laser data after standardization and normalization are archived into a multi-wavelength laser output database.

[0036] This implementation plan standardizes the dimensional standards of laser-related numerical data through unified processing by the unit, ensuring consistency in the measurement standards of core data such as power, energy density, and spot size. By using outlier removal and missing value completion under defined conditions, the integrity and accuracy of the data are improved while preserving the original statistical characteristics of the measurement sequence. Combined with field standardization, versioning, and timestamp synchronization strategies for laser control scenarios, dimensional differences are eliminated through normalization and one-hot encoding, achieving effective adaptation and standardization of categorical and continuous features. Finally, high-quality, standardized laser data is output and archived, providing a reliable data foundation for subsequent multi-wavelength laser-related data applications, analysis, and system operation.

[0037] Specifically, the measures for beam combining and spot shaping of multi-wavelength lasers are as follows: Multi-wavelength laser beam combining is achieved using a dichroic mirror assembly. Specifically, a multi-fiber combiner integrating high-precision spectral separation is used in conjunction with a high-transmittance dichroic mirror assembly. First, the multi-fiber combiner performs preliminary screening and coupling of multiple input lasers of different wavelengths. Then, the dichroic mirror assembly performs precise separation based on the spectral characteristics of each wavelength. Subsequently, with the help of customized fiber optic links and high-refractive-index focusing lenses, each separated laser is guided and converged onto the same path, achieving efficient beam combining of multi-wavelength lasers. During the beam combining process, the system collects optical power data of each optical path in real time, dynamically calculates the total optical path loss, and detects... When the total optical path loss exceeds the preset loss threshold, the optical path correction process is immediately triggered. The spatial orientation of the dichroic mirror and focusing lens is adjusted by driving the precision displacement stage to ensure the stability and controllability of the beam combining optical path. The core of the optical shaping unit consists of an adjustable shape stop and an electric zoom lens. The adjustable shape stop is equipped with multiple sets of electrically adjustable light-shielding blades, which can flexibly switch between various beam cross-section shapes such as circles, squares, and rectangles to accurately match the illumination requirements of different application scenarios. The electric zoom lens achieves continuous adjustment of the spot diameter from 2mm to 17mm by driving the relative displacement of the lenses within the lens group by a motor, providing a variety of spot size options for subsequent laser applications.

[0038] In this implementation scheme, the precise separation and efficient beam combining of multi-wavelength lasers are achieved through the synergistic effect of multi-fiber combiners and dichroic mirror groups. At the same time, the stability and low-loss characteristics of the combined optical path are ensured by a mechanism that calculates the total optical path loss in real time and triggers a correction process when the loss exceeds the threshold. With the addition of an optical shaping unit consisting of an adjustable shape aperture and an electric zoom lens, the cross-sectional shape of the laser beam can be flexibly adjusted to adapt to different irradiation requirements, and the spot diameter can be continuously adjusted. In the end, a combined laser beam with controllable shape, adjustable spot size and stable transmission is formed to meet the diverse needs of subsequent applications.

[0039] Specifically, the measures for constructing a loss time-series model to correct and output the estimated energy density for each wavelength are as follows: Laser wavelength, spot shape, spot size, and energy density data are systematically integrated along the time dimension to form a multivariate time-series feature set, which serves as the input loss analysis unit. Combined with real-time monitoring data of the optical path and dynamic parameters of environmental conditions, the causal relationship of energy loss for each wavelength is clarified, forming an initial loss correlation set. The optical path mechanism includes the refractive index variation law of optical elements, the scattering effect in the beam path, and the influence of environmental temperature and humidity on optical path loss. This loss relationship is precisely determined through the physical characteristics of the optical elements and transformed into prior constraints for dynamic Bayesian network structure learning: environmental temperature and humidity... The degree node is forced to point to the refractive index node and the scattering node, the refractive index node is forced to point to the transmission loss node, and the scattering node is forced to point to the scattering loss node. At the same time, physical rules are used to check and prohibit edges that contradict the known physical causal direction, ensuring that the network topology strictly conforms to the directed acyclic graph and the causal order of energy transfer. A dynamic Bayesian network is used to carry out refined loss structure learning. Based on the multivariate time series feature set, the evolution path of energy loss of lasers of different wavelengths in the optical path across time steps is characterized, as well as the feedback relationship of the optical path state on the subsequent laser energy transfer. The direct impact relationship, indirect transmission relationship and time lag relationship of loss are accurately identified, and a loss time series model containing multi-time step dependent features is constructed to output the estimated energy density of each wavelength.

[0040] In this implementation scheme, multi-dimensional laser parameters are constructed as multivariable time-series features and input into a loss analysis unit. The causal logic of energy loss at each wavelength is clarified by combining optical path mechanisms and environmental conditions, forming an initial loss correlation set. The topology of the dynamic Bayesian network is standardized by implanting prior constraints that conform to physical laws, ensuring that the network and energy transmission causal order are consistent. Then, loss structure learning is carried out based on this network to accurately characterize the cross-time step evolution path of laser energy loss at different wavelengths and the feedback relationship of optical path state. The direct, indirect, and time-lag effects of loss are identified, and finally, a loss time-series model containing multi-time step dependent features is constructed to achieve accurate compensation for loss at each wavelength and output of estimated energy density.

[0041] Specifically, the specific measures for energy deviation analysis using the actual energy density, estimated energy density, nominal power, and optical path loss data of multi-wavelength lasers are as follows: obtain the first... The actual energy density, estimated energy density, nominal power, and total optical path loss of each wavelength laser under current operating conditions; by calculating point by point... The difference between the actual and estimated energy densities of a laser wavelength is used to accurately determine the energy deviation of that wavelength under current operating conditions. This deviation directly reflects the degree of deviation between the actual energy output and the theoretical estimate. Based on the energy transmission efficiency calculation formula, the difference between the actual energy density and the total optical path loss is calculated and then multiplied by the nominal power value to obtain the effective input energy of the laser wavelength under the current optical path conditions. This value eliminates the influence of optical path loss on the input energy and more closely reflects the actual energy transmission effect. Finally, by calculating the ratio of the energy deviation to the effective input energy, the energy density of the first wavelength is quantitatively determined. The energy correction value of the wavelength laser provides core parameter support for the accurate calibration of subsequent laser energy output.

[0042] The specific calculation method for the energy correction value is as follows:

[0043] ;

[0044] In the formula, Indicates the first The energy correction value for each wavelength of laser reflects the difference between the actual energy and the estimated energy; Indicates the first The actual energy density of a laser at a wavelength reflects the actual energy after the laser irradiates the target surface. Indicates the first Estimated energy density of laser at each wavelength; Indicates the first The nominal power of a laser wavelength reflects the output power value set by the drive circuit. Indicates the first The total optical path loss of a laser beam of a certain wavelength reflects the transmission loss, reflection loss, and scattering loss factors in the path.

[0045] In this implementation scheme, by obtaining the first The system uses four core parameters for each wavelength of laser: actual energy density, estimated energy density, nominal power, and total optical path loss. First, the difference between the actual and estimated energy densities is calculated to obtain the energy deviation under the corresponding operating conditions. Then, combined with the optical path loss, the effective input energy after eliminating the influence of loss is calculated. Finally, by calculating the ratio of the deviation to the effective input energy, an energy correction value that can be used to accurately calibrate the energy output of the laser at that wavelength is generated, providing a quantitative basis for the consistency and accuracy control of the laser system's energy output.

[0046] Specifically, the measures for performing power compensation and pulse parameter adjustment based on the deviation analysis results are as follows: real-time comparison of energy correction values ​​and correction thresholds, such as... Figure 3This is a flowchart of the multi-wavelength laser energy multi-level control and calibration process in this embodiment. When the energy correction value is less than the correction threshold, the energy output corresponding to the current wavelength is determined to be in a stable range. The driving current setting, pulse frequency, and pulse width corresponding to the current wavelength remain unchanged. Only the energy density data collected by the high-precision energy detection module within this cycle is accurately recorded, and the wavelength is clearly marked as an energy stable state, providing basic data for subsequent energy trend analysis. When the energy correction value is greater than or equal to the correction threshold, the power compensation operation is initiated: the driving current is finely adjusted in a stepwise manner based on the calculation results of the energy gap. The energy loss caused by optical path loss is compensated by precisely controlling the driving current. After the driving current adjustment is completed, an energy density resampling is triggered to reacquire the adjusted energy data to verify the compensation effect. When the driving current is adjusted to the maximum adjustable driving current preset by the system, and the energy correction value is still greater than or equal to the correction threshold after resampling, the control strategy automatically switches to the pulse optimization strategy: following the execution logic of keeping the pulse frequency unchanged, adjusting the pulse width, and adjusting the spot diameter, the steps are executed sequentially. The next step is triggered only when the previous step adjustment cannot meet the energy requirements. First, the pulse frequency is kept stable. The single-pulse energy is adjusted by changing the pulse width to compensate for the energy gap. If the energy correction value still does not fall below the correction threshold after adjusting the pulse width, the spot diameter is further adjusted to optimize the energy density distribution by redistributing the energy input per unit area. ; represents the number of monitoring cycles after secondary compensation. When, within m consecutive cycles, the difference between two adjacent energy correction values ​​exceeds the preset energy change threshold, and the difference between two adjacent optical path loss values ​​is simultaneously detected to be greater than the preset loss change threshold, where m is the threshold for the number of abnormal monitoring cycles, ranging from 3 to 5, it is determined that the abnormal change in energy correction value and the increase in optical path loss are synchronously correlated. The optical path recalibration process is then triggered, optimizing the optical path transmission efficiency by recalibrating the optical element attitude and updating the path loss parameters. After completing the optical path recalibration, the above power compensation and pulse optimization processes are re-executed, and the energy correction value is recalculated based on the new optical path parameters. If, after optical path recalibration and secondary compensation, the energy correction value remains greater than or equal to the correction threshold for n consecutive cycles, the current wavelength channel is marked as an energy anomaly channel. Energy feedback parameters are recorded simultaneously, including drive current adjustment records, pulse parameter change data, and optical path loss data. Here, n is the threshold number of anomaly detection cycles, ranging from 5 to 8. In this case, the laser output settings for that channel are restricted, retaining only the real-time energy monitoring function to ensure system operational safety and data traceability.

[0047] In this implementation scheme, the constructed real-time comparison mechanism, combined with power compensation for driving current adjustment and pulse optimization strategy that progressively increases pulse width and spot diameter, achieves precise closed-loop control of laser energy. At the same time, it introduces anomaly monitoring cycle number threshold m and anomaly judgment cycle number threshold n, establishes a synchronous change judgment rule for energy correction value and optical path loss, triggers optical path recalibration process to optimize transmission parameters, and marks abnormal channels and restricts output when secondary compensation still fails to meet the standard. Ultimately, it ensures the stability, controllability and safety of laser energy output, effectively adapting to the energy requirements of different application scenarios.

[0048] Specifically, the measures taken to construct an energy correction decision model by fusing multi-wavelength power, spot parameters, and energy feedback information, and to output the target surface energy distribution prediction results and energy distribution characteristics, are as follows: The nominal power value, real-time output power, spot shape, spot size, energy correction value, and energy feedback parameters for each wavelength are classified and integrated according to feature dimensions, serving as input features to construct a multi-dimensional training sample covering laser output characteristics and feedback adjustment parameters; the energy distribution data of each wavelength at the same time and on the same target surface are reconstructed hierarchically according to wavelength channels, constructing a multi-channel two-dimensional image matrix, where each channel represents a spot grayscale image of an independent wavelength, and its pixel value is proportional to the energy density at that point under that wavelength, achieving a visual feature representation of energy distribution; after data normalization preprocessing of the multi-dimensional training samples and the multi-channel two-dimensional image matrix, they are synchronously input into a deep learning network for end-to-end feature learning and mapping, mining low-energy-density energy. The intrinsic relationship between 3D parameter features and high-dimensional image features is explored. A convolutional neural network is trained using real-time acquisition and dynamic updates of end-to-end operational data to accurately extract the complex coupling relationship between each input parameter and the actual energy output. Emphasis is placed on the nonlinear deviation of energy distribution under different wavelengths, different spot parameters, and dynamic conditions to improve adaptability to complex operating conditions. A high-precision nonlinear mapping relationship between power adjustment amplitude, pulse adjustment mode, spot adjustment direction, and energy correction effect is established using a support vector machine. During system operation, new energy correction results and adjustment execution records are continuously received and incrementally iterated to construct a robust energy correction decision model. Finally, high-precision energy distribution predictions and multi-dimensional energy distribution feature results for each wavelength in the target area are output. These energy distribution features specifically include center-edge energy difference features, left-right energy difference features, up-down energy difference features, and local energy loss features.

[0049] In this implementation scheme, multi-dimensional training samples are constructed by integrating multi-dimensional laser output and feedback parameters. Simultaneously, multi-wavelength target surface energy distribution data is reconstructed into a multi-channel two-dimensional image matrix. After preprocessing, the matrix is ​​input into a deep learning network to achieve end-to-end feature learning and mapping. A convolutional neural network is trained with real-time updated data to accurately mine the coupling relationship between input parameters and actual energy output, as well as the nonlinear deviation of energy distribution under complex working conditions. A nonlinear mapping relationship between adjustment parameters and energy correction effect is established through support vector machines and continuously updated incrementally to construct a robust energy correction decision model. Finally, the predicted values ​​of target surface energy distribution for each wavelength and multi-dimensional energy distribution features including center and edge, multi-directional differences, and local missing features are output, providing comprehensive feature support and decision basis for subsequent precise energy adjustment.

[0050] Specifically, the energy distribution assessment, based on energy distribution prediction and characteristics, drives the motorized zoom lens and adjustable aperture to adjust the spot diameter, spot shape, and scanning mode. The specific measures are as follows: By comparing the predicted energy distribution value with the distribution threshold in real time, when the predicted energy distribution value is less than the distribution threshold, it is determined that the energy distribution in the target area is within the allowable deviation range. The control unit maintains the current spot diameter, aperture shape, and scanning mode unchanged, and records the current spot state and corresponding energy distribution as a stable operating condition reference and archives it in the multi-wavelength laser output database. When the predicted energy distribution value is greater than or equal to the distribution threshold, it is determined that there is a sparse energy distribution in the current target area. The energy distribution characteristic results are read, and a spot optimization adjustment strategy is executed: When the energy difference between the center and the edge is displayed, the motorized zoom lens is controlled to change the spot diameter, gradually adjusting the redistribution of energy per unit area from the center area to the edge area, alleviating excessive energy concentration at the center and energy imbalance at the edge. To address the issue of insufficient energy, when energy differences are observed in the left and right directions, the lateral opening parameter of the adjustable aperture is precisely controlled to stretch and compress the light spot in the left and right directions, allowing the energy to be rebalanced and cover the target area horizontally. When energy differences are observed in the up and down directions, the longitudinal opening parameter of the aperture is corrected, and the focal position is calibrated in conjunction with a minor adjustment of the zoom lens, allowing the energy to be redistributed vertically. When local energy loss is observed, the light spot shape and size are adjusted in combination to control the adjustment amount, causing the light spot outline to extend towards the energy loss area and enhance the energy coverage intensity of the loss area. After each adjustment of the light spot diameter, shape, and aperture opening, the energy correction decision model is called to update the predicted energy distribution values ​​and energy distribution characteristics of each wavelength in the target area, and the distribution threshold comparison is performed again, continuously iterating the light spot optimization adjustment process until the predicted energy distribution value is less than the distribution threshold.

[0051] In this embodiment, when a center-to-edge energy difference is detected, the center-to-edge energy ratio is 20:1, and the energy distribution non-uniformity is greater than 85%. The adjustment action is to increase the spot diameter to redistribute energy. The spot diameter is adjusted from 8mm to 16mm using a motorized zoom lens, with a zoom travel of +8mm. After execution, the energy in the center region is reduced by 40%, and the energy in the edge region is increased to 0.35 (relative value). When a left-to-right energy difference is detected, the left-to-right energy ratio is 1:2.54, and the horizontal non-uniformity is greater than 60%. The adjustment action is to adjust the lateral opening of the aperture to perform horizontal stretching / compression. The lateral opening is adjusted from 10mm to 20mm using an adjustable shape aperture, with left and right adjustments of ±5mm each. After execution, the left-to-right energy ratio is optimized to 1:1.2, and the horizontal non-uniformity is reduced to below 15%. When a vertical energy difference is detected... When the feature is heterogeneous, the vertical energy ratio is 1:2.31, and the vertical non-uniformity is greater than 55%. The adjustment action is to adjust the vertical opening of the aperture to achieve vertical redistribution. The vertical opening is adjusted from 10mm to 20mm by the adjustable shape aperture, and the vertical adjustment is ±5mm each. After execution, the vertical energy ratio is optimized to 1:1.1, and the vertical non-uniformity is reduced to below 12%. When a local energy deficiency feature is detected, the energy ratio of the deficiency area to the average area is 1:12.5, and the local deficiency area accounts for about 8%. The adjustment action is to combine the adjustment of the spot shape and size to expand the coverage of the deficiency area. By coordinating the control of the zoom lens and the adjustable aperture, the spot shape is corrected from a circle to an ellipse that expands directionally towards the deficiency area, and the overall effective diameter is slightly increased by 2mm. After execution, the energy of the deficiency area is increased to 0.45 (relative value), and the effective energy coverage area is increased by 12%.

[0052] Table 1. Characteristics of Spot Energy Distribution Problem and Correspondence with Adjustment Parameters

[0053]

[0054] like Figure 4 The diagram shown is a comparison of the characteristics and adjustment actions of the multi-wavelength light spot energy distribution problem provided in the embodiments of this application. Figure 4 The diagram consists of four sub-plots, each corresponding to one of four typical energy distribution unevenness problems and their corresponding optical adjustment strategies. The contour plot on the left visually presents the energy density distribution of the four problem modes: center-edge difference, left-right difference, up-down difference, and local energy deficiency, and clarifies the relative energy value range from 0.04 to 1.00. The schematic diagram on the right visualizes the corresponding spot morphology adjustment actions through the original spot outline, target spot outline, adjustment direction, and energy redistribution direction; Table 1 and... Figure 4Data shows that when the energy distribution of the laser spot exhibits different characteristic non-uniformities, the energy distribution on the target surface can be effectively improved through corresponding adjustment strategies. Specifically, to address the energy difference between the center and the edge, increasing the spot diameter from 8mm to 16mm increased the edge energy from 0.05 to 0.35 while reducing the center energy by 40%. To address the energy difference between the left and right sides, adjusting the lateral opening of the aperture optimized the left-right energy ratio from 1:2.54 to 1:1.2, reducing the horizontal non-uniformity to below 15%. When local energy loss is detected and the energy in the lost area is below 1 / 12.5 of the average, the system, through combined adjustment of the spot shape and size, increases the energy in the lost area from 0.04 to 0.45 and increases the coverage area by 12%. This demonstrates that the multi-wavelength semiconductor laser variable spot output control method and system provided in this application, by real-time identification of energy distribution characteristics and driving the corresponding adjustment mechanism, can dynamically correct the spot shape and energy distribution, thereby ensuring the uniformity of the target surface energy and the stability and reliability of the laser processing process.

[0055] In this implementation scheme, the energy distribution status of the target surface is determined by comparing the predicted energy distribution value with the distribution threshold in real time. When the energy distribution is within the allowable deviation range, the current spot parameters are kept unchanged and stable operating condition reference data is archived. When it is determined that there is a sparse energy distribution, the spot diameter, aperture opening and zoom lens focal position are adjusted in a targeted manner according to four different energy distribution characteristics: center and edge, left and right direction, up and down direction and local missing. After each parameter adjustment, the energy correction decision model is called to update the predicted value and feature results and the threshold comparison is repeated. Through continuous iterative optimization, the precise control and balanced coverage of the target surface energy distribution are achieved, ensuring the stability and effectiveness of multi-wavelength laser output.

[0056] Specifically, this embodiment provides an output control system for a multi-wavelength semiconductor laser with variable spot size, applied to an output control method for a multi-wavelength semiconductor laser with variable spot size. The system includes: a laser data acquisition and standardization module, equipped with a high-precision laser parameter sensor and data preprocessing algorithm, used to acquire core parameter data such as power, pulse frequency, and pulse width of the multi-wavelength laser in real time; performing preprocessing operations such as noise reduction, normalization, and outlier removal on the acquired raw data; classifying and storing the preprocessed standardized laser parameter data; and constructing a multi-wavelength laser output database covering the entire wavelength range based on a distributed database architecture, providing accurate and consistent data source support for subsequent modules; an optical path correction module, relying on the collaborative architecture of a multi-fiber combiner and a dichroic mirror group, used to efficiently combine and shape the input multiple lasers of different wavelengths; simultaneously, based on historical optical path loss data and real-time acquired loss monitoring values, constructing a loss time-series model with time-series prediction capabilities; using this model to dynamically correct the estimated energy density of each wavelength laser; and finally outputting the corrected accurate energy density data; and a laser energy calibration module, used to retrieve nominal power data and optical path calibration data from the multi-wavelength laser output database. The estimated energy density and optical path loss data output by the positive module, combined with the actual energy density data collected by the energy detection unit, are used to conduct multi-dimensional energy deviation analysis. This accurately quantifies the degree of deviation between the actual output and the theoretical value. Based on the deviation analysis results, power compensation operations such as adjusting the drive current and pulse parameter optimization operations such as adjusting the pulse width and spot diameter are executed in stages to achieve closed-loop calibration of laser energy. The energy distribution prediction module is used to deeply integrate real-time power data of multi-wavelength lasers, spot parameters after spot reshaping, and energy feedback information from the energy calibration process through a multi-source data fusion algorithm. It constructs a machine learning-based energy correction decision model to accurately predict the energy distribution state of the laser target surface, outputs the target surface energy distribution prediction results, and extracts key energy distribution features such as energy peak, energy uniformity, and energy coverage. The spot shape and size adjustment module is used to conduct quantitative energy distribution assessment based on the energy distribution prediction results and energy distribution features. Based on the assessment results, it generates targeted adjustment commands to drive the motorized zoom lens to adjust the spot diameter and drive the adjustable shape aperture to switch the spot shape. At the same time, it links with the scanning control unit to optimize the laser scanning mode to ensure that the target surface energy distribution fully matches the actual needs of the application scenario.

[0057] In this implementation scheme, a modular collaborative working mechanism is constructed by setting up a laser data acquisition and standardization module, an optical path correction module, a laser energy calibration module, an energy distribution prediction module, and a spot shape and size adjustment module. First, the core laser parameters are acquired, preprocessed, and stored to build a multi-wavelength laser output database. Then, the estimated energy density of each wavelength is accurately corrected by constructing a laser beam combining, spot shaping, and loss timing model. Next, energy deviation analysis is carried out by combining actual and estimated energy density and other multi-dimensional data, and power compensation and pulse parameter adjustment are performed. Subsequently, an energy correction decision model is constructed by fusing multi-source data to output the target surface energy distribution prediction results and key characteristics. Finally, based on the energy distribution evaluation results, the relevant components are driven to adjust the spot diameter, shape, and scanning mode. This ultimately forms a closed-loop management of the entire process from laser parameter acquisition, optical path correction, energy calibration to energy distribution prediction and spot control, ensuring the stability, accuracy, and adaptability of multi-wavelength laser output.

[0058] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0059] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for controlling the output of a multi-wavelength semiconductor laser variable spot, characterized by, The method comprises the following steps: S1, collecting laser parameter data for preprocessing operation, and storing the laser parameter data to construct a multi-wavelength laser output database; S2, constructing a loss time sequence model by beam combining and spot shaping of the multi-wavelength laser, to realize correction and output of the estimated energy density of each wavelength; S3, performing energy deviation analysis by using the actual energy density, the estimated energy density, the nominal power and the optical path loss data of the multi-wavelength laser, and performing power compensation and pulse parameter adjustment operation according to the deviation analysis result; S4, constructing an energy correction decision model by fusing multi-wavelength power, spot parameter and energy feedback information, and outputting target surface energy distribution prediction result and energy distribution characteristics; S5, performing energy distribution evaluation by using the energy distribution prediction and the energy distribution characteristics, to drive the electric zoom lens and the adjustable shape diaphragm to adjust the spot diameter, the spot shape and the scanning mode.

2. The method of claim 1, wherein the method further comprises: determining a wavelength of the light beam; and selecting the wavelength of the light beam from a plurality of wavelengths of light beams. The specific measures of the collecting laser parameter data for preprocessing operation are as follows: Obtaining laser parameter data: through the laser driving unit, the power detection device and the control recording interface, the power nominal value, the real-time output power and the corresponding driving current, pulse frequency and pulse width configuration of each wavelength laser are collected in real time, and the collection result is bound with the current laser wavelength parameter to form wavelength-level power and driving state data; the spot imaging sensor is used to continuously image the target surface area, and the spot shape, spot size, spot contour boundary and spot coverage area information are extracted in real time, and the actual energy density of each wavelength laser on the target surface is obtained by combining the energy density sensor; The refractive index, beam incidence angle and polarization state of the optical element are obtained, the transmittance and reflectance at each interface are calculated by Fresnel formula, and the transmission and reflection loss are obtained by step-by-step accumulation; the scattering coefficient of the material and the propagation length of the beam in the path are multiplied to obtain the scattering loss; the transmission loss, reflection loss and scattering loss are fused to calculate the total optical path loss of each wavelength laser from the light source to the target surface.

3. The method of claim 1, wherein the method further comprises: determining a wavelength of the light beam; and selecting the wavelength of the light beam from a plurality of wavelengths of light beams. The specific measures of the storing the laser parameter data to construct a multi-wavelength laser output database are as follows: The power nominal value, the real-time output power, the driving current, the pulse frequency, the pulse width, the spot size, the actual energy density and the total optical path loss numerical data obtained by collection are subjected to unit uniform processing, the power related data is uniformly converted to watt dimension, the energy density is uniformly converted to unit area energy dimension, and the spot size is uniformly adopted as length reference unit; After the completion of unit unification, the abnormal value detection is performed on each numerical data, the sampling points obviously deviating from the normal working interval are identified based on the statistical distribution characteristics and are removed, the data missing section formed due to communication and sampling interruption is completed by using the mean value of adjacent effective sampling points, subsequently, the maximum and minimum normalization method is used to map the power nominal value, real-time output power, actual energy density, optical path total loss and continuous type features of driving parameters to the unified interval, the one-hot encoding is used to vectorize the category type features such as light spot shape, and the influence of dimension difference is eliminated, and all the laser data after the completion of standardization and normalization processing are archived to the multi-wavelength laser output database.

4. The method of claim 1, wherein the method further comprises: determining a wavelength of the light beam; and selecting the wavelength of the light beam from a plurality of wavelengths of light beams. The specific measures for the beam combination and light spot shaping of the multi-wavelength laser are as follows: The multi-wavelength laser is combined through a dichroic mirror set: a multi-fiber beam combiner with spectral separation function and a dichroic mirror set are used to separate and guide the laser of each wavelength to the same path; the laser of each wavelength is separated by the dichroic mirror and converged to the unified beam path through the optical fiber and the lens, and the total optical path loss is dynamically calculated, and if the total optical path loss is greater than the loss threshold, the optical path correction process is triggered; the optical shaping unit includes an adjustable shape diaphragm and an electric zoom lens, the adjustable shape diaphragm is used to control the shape of the laser beam to adapt to different irradiation requirements, and the electric zoom lens adjusts the light spot diameter range to provide an adjustable range.

5. The method of claim 1, wherein the method further comprises: determining a wavelength of the light beam; and selecting the wavelength of the light beam from a plurality of wavelengths of light beams. The specific measures for constructing the loss time sequence model and realizing the correction and output of the estimated energy density of each wavelength are as follows: The laser wavelength, light spot shape, light spot size and energy density data form a multivariate time sequence feature set, which is used as an input loss analysis unit, and the energy loss causal relationship of each wavelength is determined in combination with the optical path and environmental conditions to form an initial loss correlation set; the optical path mechanism includes the refractive index variation law of the optical element, the scattering effect in the beam path and the influence relationship of environmental temperature and humidity on the optical path loss, the loss relationship is determined by the physical characteristics of the optical element and is converted into a prior constraint for dynamic Bayesian network structure learning: the environmental temperature and humidity nodes are forced to point to the refractive index nodes and the scattering nodes, the refractive index nodes are forced to point to the transmission loss nodes, the scattering nodes are forced to point to the scattering loss nodes, while the edges contrary to the known physical causal direction are prohibited, to ensure that the network topology conforms to the directed acyclic graph and the causal order of energy transmission; the dynamic Bayesian network is used to carry out loss structure learning, to depict the evolution path of the energy loss of different wavelengths of laser in the optical path across time steps, and the feedback action relationship of the optical path state on the subsequent laser energy transmission, to identify the direct influence relationship, indirect transmission relationship and time lag relationship, to construct a loss time sequence model containing multiple time step dependent features and to compensate the loss of each wavelength, and to output the estimated energy density of each wavelength.

6. The method of claim 1, wherein the method further comprises: determining a wavelength of the light beam; and selecting the wavelength of the light beam from a plurality of wavelengths of light beams. The specific measures for the energy deviation analysis of the actual energy density, estimated energy density, nominal power and optical path loss data of the multi-wavelength laser are as follows: The actual fluence, the estimated fluence, the power nominal value and the total optical path loss of the laser of the first wavelength are acquired. Calculate the first The difference between the actual and estimated energy densities of lasers at each wavelength is used to obtain the energy deviation of that wavelength under the current operating conditions. The effective input energy is obtained by subtracting the total optical path loss from the first wavelength and multiplying by the nominal power value. The ratio of the energy deviation to the effective input energy is then calculated to obtain the energy deviation of the first wavelength. Energy correction value for each wavelength of laser.

7. The method of claim 1, wherein the method is implemented in a system comprising: a plurality of semiconductor lasers; a plurality of wavelength selective elements; a plurality of optical elements; and a plurality of optical fibers. The specific measures for performing the power compensation and pulse parameter adjustment operation according to the deviation analysis result are as follows: By comparing the energy correction value with the correction threshold in real time, when the energy correction value is less than the correction threshold, the current wavelength corresponding driving current setting, pulse frequency and pulse width are kept unchanged, only the energy density collected in the current cycle is recorded, and the wavelength is marked as energy stable state; When the energy correction value is greater than or equal to the correction threshold, the power compensation operation is performed: the driving current is adjusted to correct the energy gap caused by the input power compensation optical path loss, and after the adjustment is completed, the energy density resampling is triggered once; when the driving current adjustment reaches the maximum adjustable driving current but the energy correction value is still greater than or equal to the correction threshold, the control strategy is switched to the pulse optimization strategy: the pulse frequency is kept unchanged, the single pulse energy is adjusted by changing the pulse width, and the spot diameter is changed to redistribute the unit area energy input; When the difference between the energy correction values in the last two consecutive cycles exceeds the energy change threshold and the difference between the optical path losses in the last two consecutive cycles is greater than the loss change threshold, it is determined that the energy correction value abnormally changes and the optical path loss synchronously increases, the optical path re-correction process is triggered to update the path loss parameter, the power and pulse compensation process is re-executed, and the energy correction value is re-calculated, and if the energy correction value is still greater than or equal to the correction threshold in the last n cycles, the current wavelength channel is marked as an energy abnormal channel, the energy feedback parameter is recorded, and the output setting is limited to only retain the monitoring function.

8. The method of claim 1, wherein the method further comprises: determining a wavelength of the light beam; and selecting the wavelength of the light beam from a plurality of wavelengths of light beams. The specific measures for constructing the energy correction decision model by fusing the multi-wavelength power, spot parameter and energy feedback information, and outputting the target surface energy distribution prediction result and energy distribution characteristics are: The power nominal value, real-time output power, spot shape, spot size, energy correction value and energy feedback parameter of each wavelength are used as input features to construct a multi-dimensional training sample; And the energy distribution data of each wavelength at the same time and on the same target surface is constructed into a multi-channel two-dimensional image matrix, each channel representing a gray-scale image of a spot of an independent wavelength, and the pixel value is proportional to the energy density at that point under that wavelength; the multi-dimensional training sample and the multi-channel two-dimensional image matrix are input into a deep learning network for end-to-end feature learning and mapping; the convolutional neural network is trained by real-time data collection and update to extract the coupling relationship between the input parameters and the actual energy output, and pay attention to the nonlinear deviation of the energy distribution under different wavelengths, different spot parameters and dynamic working conditions; the support vector machine is used to establish the nonlinear mapping relationship between the power adjustment amplitude, the pulse adjustment mode, the spot adjustment direction and the energy correction effect, and new energy correction results and adjustment execution records are continuously received and updated during operation to construct the energy correction decision model, and the energy distribution prediction value and the energy distribution characteristic result of each wavelength on the target surface area are output, and the energy distribution characteristics include the center and edge energy difference characteristics, the left and right direction energy difference characteristics, the up and down direction energy difference characteristics and the local energy loss characteristics.

9. The method of claim 1, wherein the method further comprises: determining a wavelength of the light beam; and selecting the wavelength of the light beam from a plurality of wavelengths of light beams. The specific measures for evaluating the energy distribution by energy distribution prediction and energy distribution characteristics, and adjusting the spot diameter, spot shape and scanning mode by driving the motorized zoom lens and the adjustable shape diaphragm are: By real-time comparison of the energy distribution prediction value and the distribution threshold, when the energy distribution prediction value is less than the distribution threshold, it is determined that the energy distribution of the target surface area is within the allowable deviation range, and the control unit keeps the current spot diameter, aperture shape and scanning mode unchanged. The current spot state and the corresponding energy distribution are recorded as a stable working condition reference and archived to the multi-wavelength laser output database; When the energy distribution prediction value is greater than or equal to the distribution threshold, it is determined that there is an energy distribution sparsity phenomenon in the current target surface area, the energy distribution characteristic result is read and the spot optimization adjustment strategy is executed: when the center and edge energy difference characteristics are displayed, the electric zoom lens is controlled to change the spot diameter, so that the unit area energy is redistributed from the center area to the edge area, relieving the problem of excessive concentration of center energy and insufficient edge energy; when the left and right direction energy difference characteristics are displayed, the horizontal opening parameter of the adjustable shape light barrier is adjusted to stretch and compress the spot in the left and right directions, so that the energy is rebalanced in the horizontal direction to cover the target area; When the up and down direction energy difference characteristics are displayed, the longitudinal opening parameter of the light barrier is corrected, and the focus position of the zoom lens is slightly adjusted to realize the redistribution of energy in the vertical direction; when the local energy loss characteristics are displayed, the spot shape and spot size are adjusted to expand the spot profile to the energy loss area, and the energy coverage intensity of the loss area is enhanced; after each adjustment of the spot diameter, the spot shape and the light barrier opening, the energy correction decision model is called to update the energy distribution prediction value and the energy distribution characteristic result of each wavelength in the target surface area, and the distribution threshold comparison is performed again. The spot optimization adjustment process is continuously iterated until the energy distribution prediction value is less than the distribution threshold.

10. A multi-wavelength semiconductor laser variable spot output control system, applying the multi-wavelength semiconductor laser variable spot output control method of any one of claims 1-9, comprising: a laser data acquisition and standardization module for acquiring laser parameter data for preprocessing operation and storing the laser parameter data to construct a multi-wavelength laser output database; an optical path correction module for correcting and outputting the estimated energy density of each wavelength by beam combining and spot shaping of the multi-wavelength laser to construct a loss time sequence model; a laser energy calibration module for performing energy deviation analysis based on the actual energy density, the estimated energy density, the nominal power and the optical path loss data of the multi-wavelength laser, and performing power compensation and pulse parameter adjustment operation according to the deviation analysis result; an energy distribution prediction module for constructing an energy correction decision model by fusing multi-wavelength power, spot parameters and energy feedback information, and outputting target surface energy distribution prediction results and energy distribution characteristics; a spot shape and size adjustment module for energy distribution evaluation by energy distribution prediction and energy distribution characteristics, driving the electric zoom lens and the adjustable shape light barrier to adjust the spot diameter, the spot shape and the scanning mode.

Images