Multi-wavelength pulse fiber laser for treating seborrheic dermatitis

Abstract

The invention belongs to the technical field of lasers, and particularly relates to a multi-wavelength pulse fiber laser for treating seborrheic dermatitis, which comprises a data acquisition module for acquiring the skin cuticle thickness and sebum secretion rate of a patient and calculating the current impedance scattering ratio; the treatment planning module is used for triggering a wavelength recombination instruction to recalculate a weight coefficient when the current impedance scattering ratio exceeds a first threshold value; the wavelength optimization module is used for generating a weight coefficient containing a target wavelength through a multi-target optimization algorithm; the pulse modulation module is used for calculating pulse duration corresponding to each wavelength according to the weight coefficient; the energy control module is used for calculating total output energy; the skin feedback module is used for generating a temperature gradient matrix; and the laser emitting module is used for generating driving current according to the total output energy and the temperature gradient matrix and emitting multi-wavelength pulse optical fiber laser for treating seborrheic dermatitis. Precise treatment of seborrheic dermatitis is achieved, and the treatment effect is remarkably improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of lasers, and particularly relates to a multi-wavelength pulsed fiber laser for the treatment of seborrheic dermatitis. Background Art

[0002] With the continuous development of optical communication technology, the application of fiber lasers in the field of biomedicine has gradually attracted attention. Seborrheic dermatitis is a common skin disease, and traditional treatment methods have certain limitations. Therefore, it can be considered to use fiber lasers for treatment to obtain better treatment effects.

[0003] Existing laser treatment systems usually use single-wavelength lasers for treatment. Although they can relieve symptoms to a certain extent, due to the complexity of skin tissues, single-wavelength lasers are difficult to simultaneously take into account the effects on different tissue components. Therefore, they are insufficient in inhibiting the proliferation of Malassezia and regulating sebum secretion, and the treatment effect is not ideal. In addition, there is a lack of a dynamic response mechanism to skin impedance changes during the treatment process, and it is impossible to adjust treatment parameters according to the real-time condition of the skin, which easily leads to unstable treatment effects. Moreover, the laser treatment method with fixed pulse parameters easily causes uneven energy deposition, thereby causing thermal damage and affecting the safety and effectiveness of the treatment. Summary of the Invention

[0004] Based on the problems existing in the background art, the purpose of the present invention is to provide a multi-wavelength pulsed fiber laser for the treatment of seborrheic dermatitis, which realizes the precise treatment of seborrheic dermatitis and significantly improves the treatment effect.

[0005] To achieve the above purpose, the present invention provides the following technical solutions: A multi-wavelength pulsed fiber laser for the treatment of seborrheic dermatitis, comprising: A data acquisition module, configured to obtain the skin cutin thickness and sebum secretion rate of a patient, and simultaneously obtain the skin impedance and optical scattering coefficient, and calculate the current impedance scattering ratio; A treatment planning module, configured to trigger a wavelength recombination instruction to recalculate the weight coefficient when the current impedance scattering ratio exceeds a first threshold; A wavelength optimization module, configured to generate weight coefficients including target wavelengths through a multi-objective optimization algorithm according to the skin cutin thickness and sebum secretion rate of the patient; A pulse modulation module, configured to calculate the pulse duration corresponding to each wavelength according to the weight coefficients; An energy control module, configured to integrate the energies of each wavelength and calculate the total output energy; A skin feedback module, configured to obtain the temperature distribution in the treatment area in real time and generate a temperature gradient matrix; The laser emission module is used to generate a driving current according to the total output energy and the temperature gradient matrix, drive the laser to emit, and emit multi-wavelength pulsed fiber laser for the treatment of seborrheic dermatitis.

[0006] Preferably, in the data acquisition module, the skin impedance is obtained in the following way: Step S01: Use a four-electrode bio-impedance measurement device to apply a 1 mA / 50 kHz AC signal to the skin to be measured; Step S02: Separate the real part of the impedance through a phase detection circuit and the imaginary part ; Step S03: Calculate the effective skin impedance value , with a sampling frequency of 100 Hz; Step S04: Transmit Z to the data acquisition module through a wireless transmission method.

[0007] Preferably, in the treatment planning module, the determination method of the first threshold includes: Step S11: Store the critical impedance scattering ratio and clinical impedance scattering ratio data under different skin types, and establish a treatment safety database; Step S12: Based on the current impedance scattering ratio Q value and the treatment safety database, predict the risk level of the current impedance scattering ratio Q value through a support vector machine classification model; Step S13: When the prediction is a non-high risk level, do not process it. When the prediction is a high risk level, automatically update the first threshold to : ; is the lowest safety threshold.

[0008] Preferably, in step S12, the training process of the support vector machine classification model includes: Step S121: Construct a training set with the clinical impedance scattering ratio data , where j is the index of the j-th training sample, represents the impedance scattering ratio value of the j-th training sample, represents the risk level label of the j-th sample, and obtain a cleaned and labeled training data set through the risk level label; Step S122: Use the grid search method to search within the range of the penalty parameter , with a step size of . For each C value, train an SVM model and calculate its performance index on the cross-validation set, and screen out the optimal penalty parameter; Step S123: According to the training data set and the optimal penalty parameter, use the 5-fold cross-validation method to train and validate the SVM model. Randomly divide the training data set into 5 subsets. Each time, use 4 subsets for training and the remaining 1 subset for validation; Step S124: Calculate the validation accuracy of each fold and take the average as the final validation accuracy; Step S125: When the final validation accuracy exceeds the target validation accuracy, output the verified optimal SVM model and save it.

[0009] Preferably, the decision function of the SVM model is: ; where, is the impedance scattering ratio of the k-th support vector, k is the index of the support vector, n represents the total number of support vectors, is the Lagrange multiplier of the k-th support vector, is the label of the k-th support vector, b is the bias term, is the RBF kernel function, , is the bandwidth parameter.

[0010] Preferably, in the wavelength optimization module, the multi-objective optimization algorithm specifically includes: Step S21: According to the skin stratum corneum thickness, sebum secretion rate, and the i-th wavelength, construct the first objective function F: ; where, is the wavelength matching weight, is the cutin attenuation weight, is the sebum inhibition weight, is the i-th wavelength, is the characteristic absorption peak of the target tissue, is the cutin layer thickness characteristic value, e is the natural constant, d is the skin stratum corneum thickness, s is the sebum secretion rate; Step S22: Perform multi-objective optimization on the wavelength matching degree, the influence of the stratum corneum thickness, and the influence of the sebum secretion rate to obtain a subpopulation containing elite individuals, and screen out the non-dominated solution set on the Pareto front from the subpopulation containing elite individuals; Step S23: According to the non-dominated solution set on the Pareto front and the skin stratum corneum thickness d, screen out the target wavelength and obtain the target wavelength and its corresponding weight coefficient.

[0011] Preferably, the specific acquisition steps of the non-dominated solution set on the Pareto front in Step S22 include: Step S221: Obtain the population size and the crossover probability , initialize the population. Each individual in the population represents a set of wavelengths and their corresponding weight coefficients. At the same time, for the weight coefficient corresponding to the i-th wavelength Introduce constraint conditions in the initialization stage , , where H is the total number of wavelengths; Step S222: Adjust the crossover probability of the initial population according to the sebum secretion rate. The adaptive crossover probability adjustment formula is , is the new crossover probability. When the sebum secretion rate is greater than the target sebum secretion rate, increase the crossover probability to obtain the offspring population after the crossover operation; Step S223: Directly retain the elite individuals in the parent population whose first objective function F value is greater than the optimization target threshold to the offspring population to obtain the offspring population containing elite individuals.

[0012] Preferably, in the pulse modulation module, the pulse duration is obtained as follows: , is the reference pulse width, is the anti-zero constant, d is the skin stratum corneum thickness, s is the sebum secretion rate, is the weight coefficient of the i-th wavelength; The pulse duration The optimization method also includes: Step S31: Establish a thermal damage constraint condition to limit the energy deposition of the laser in the skin tissue, is the safe energy density threshold, is the wavelength energy corresponding to the i-th wavelength at time t, is the pulse duration corresponding to the i-th wavelength; Step S32: Obtain the pulse duration and the constraint weight , incorporate the constraint condition into the second objective function to solve for the optimal pulse duration, and construct the second objective function , where is the total energy of the i-th wavelength within the pulse duration ; Step S33: Through iterative solution, continuously adjust the pulse duration. When the second objective function reaches the target convergence condition, obtain the optimal pulse duration that satisfies the thermal damage constraint. Among them, the correction formula obtained by iterative solution is , is the corrected pulse duration of the i-th wavelength, is the absorption coefficient of the i-th wavelength.

[0013] Preferably, the specific implementation method of the iterative solution in step S33 is as follows: Step S331: Obtain the initial constraint weight and the target convergence condition , is the difference between the second objective function value L and the previous second objective function value; Step S332: Calculate the energy accumulation amount for each iteration ; is the energy accumulation amount calculated in the m-th iteration, is the weight coefficient of the i-th wavelength in the m-th iteration, is the pulse duration corresponding to the i-th wavelength in the m-th iteration; Step S333: Update the constraint weight to adjust the weight of the second objective function. The update formula is: , is the learning rate, is the constraint weight in the (m + 1)-th iteration, is the constraint weight in the m-th iteration.

[0014] Preferably, in the laser emission module, the calibration process of the drive current includes the following steps: Step S41: Connect a detection resistor R in parallel at the load end of the laser diode; Step S42: Measure the voltage drop V in real time and calculate the actual current value ; Step S43: Adjust the drive voltage through a PID controller to make , is the target current value.

[0015] The beneficial effects of the present invention are as follows: The present invention realizes the dynamic combination of treatment wavelengths through a multi-objective optimization algorithm, so as to accurately adjust the weight coefficients of each wavelength according to the individual skin conditions of patients (such as stratum corneum thickness and sebum secretion rate), generate the most suitable wavelength combination, and then achieve the precise treatment of seborrheic dermatitis, significantly improving the treatment effect and effectively solving the problem that traditional single-wavelength lasers cannot simultaneously inhibit the proliferation of Malassezia and regulate sebum secretion.

[0016] Through the dynamic compensation mechanism of real-time monitoring of the current impedance scattering ratio, the present invention can sense the impedance change of the skin in real time and adjust the treatment parameters accordingly, thereby establishing a set of safety compensation mechanisms during the treatment process, further strengthening the dynamic response to skin impedance changes and enhancing the safety and stability of the treatment.

[0017] By establishing thermal damage constraint conditions and an iterative optimization method, the present invention can precisely adjust the pulse duration according to the real-time feedback of the skin, thereby realizing the optimized distribution of pulse energy, further avoiding thermal damage caused by uneven energy deposition due to fixed pulse parameters, and improving the safety and effectiveness of treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a schematic diagram of the modules of the present invention; Figure 2 is a flowchart of the architecture of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following detailed description of the specific embodiments of the present invention will be given in conjunction with the accompanying drawings of the specification.

[0020] In this embodiment, a multi-wavelength pulsed fiber laser for the treatment of seborrheic dermatitis is provided, which solves the problem that traditional single-wavelength lasers cannot simultaneously inhibit the proliferation of Malassezia and regulate sebum secretion. First, the skin impedance and optical scattering coefficient are synchronously recorded and the impedance scattering ratio is calculated. When this ratio exceeds the first threshold, a trigger is generated to recalculate the weight coefficient. The skin stratum corneum thickness and sebum secretion rate of the patient are received, and a weight coefficient including the target wavelength is generated through a multi-objective optimization algorithm. The pulse duration corresponding to each wavelength is calculated according to the weight coefficient, and the total output energy is calculated by integrating the energies of each wavelength. The temperature distribution in the treatment area is obtained in real time and a temperature gradient matrix is generated. A driving current is generated according to the total output energy and the temperature gradient matrix to drive the laser emission, realizing the precise treatment of seborrheic dermatitis and significantly improving the treatment effect.

[0021] In the calculation of each formula in this embodiment, the dimension can be removed to simplify the calculation.

[0022] As Figure 1 and Figure 2 shown, the multi-wavelength pulsed fiber laser for the treatment of seborrheic dermatitis includes: a data acquisition module, a treatment planning module, a wavelength optimization module, a pulse modulation module, an energy control module, a skin feedback module, and a laser emission module; Among them, the data acquisition module obtains the skin stratum corneum thickness and sebum secretion rate of the patient, and synchronously obtains the skin impedance Z and the optical scattering coefficient σ, and calculates the current impedance scattering ratio ; The treatment planning module is used to trigger a wavelength recombination instruction to recalculate the weight coefficient when the current impedance scattering ratio exceeds the first threshold ; The wavelength optimization module is used to generate a weight coefficient including the target wavelength through a multi-objective optimization algorithm according to the received skin stratum corneum thickness d and sebum secretion rate s of the patient; A pulse modulation module, which is used to calculate the pulse duration corresponding to each wavelength according to the weight coefficient , , is the reference pulse width, is the anti-zero constant, is the weight coefficient of the i-th wavelength; An energy control module, which is used to integrate the energy of each wavelength and calculate the total output energy , is the wavelength energy corresponding to the i-th wavelength, is the pulse duration corresponding to the i-th wavelength, is the absorption coefficient of the i-th wavelength; A skin feedback module, which is used to obtain the temperature distribution T(x, y) of the treatment area in real time and generate a temperature gradient matrix ΔT; A laser emission module, which is used to generate a driving current according to the total output energy and the temperature gradient matrix ΔT , is the conversion efficiency, is the temperature compensation coefficient, drives the laser emission, and emits a multi-wavelength pulsed fiber laser for the treatment of seborrheic dermatitis.

[0023] In this embodiment, the wavelength optimization module determines the weight coefficients of three wavelengths (1450nm, 2940nm, 10600nm) according to the skin parameters (stratum corneum thickness d and sebum secretion rate s) of the patient by using a multi-objective optimization algorithm to achieve the best treatment effect. For example, when the patient has a relatively thick stratum corneum and a high sebum secretion rate, the algorithm will adjust the weight coefficients to increase the weight of the wavelength that has a stronger effect on deep tissues. The pulse modulation module calculates the pulse duration of each wavelength according to the weight coefficients, and ensures that the pulse duration adapts to different skin conditions through a specific formula. The energy control module integrates the energy of each wavelength and calculates the total output energy. The laser emission module generates a driving current according to the total energy and temperature feedback, so that the laser output is stable and adapts to the real-time condition of the skin. The data acquisition module records the skin impedance and optical scattering coefficient, calculates the impedance scattering ratio, and when this ratio exceeds the threshold, the treatment planning module recalculates the wavelength weight coefficients to achieve dynamic adjustment.

[0024] The measurement method of skin impedance includes: Step S01: Use a four-electrode bioimpedance measurement device to apply a 1mA / 50kHz AC signal; Step S02: Separate the real part of the impedance through a phase detection circuit and the imaginary part ; Step S03: Calculate the effective skin impedance value , the sampling frequency is 100 Hz.

[0025] The first threshold The determination method includes: Step S11: Store the critical impedance scattering ratio and clinical impedance scattering ratio data under different skin types, and establish a treatment safety database; Step S12: Based on the current impedance scattering ratio Q value and the treatment safety database, predict the risk level of the current impedance scattering ratio Q value through a support vector machine classification model; Step S13: When the prediction is a non-high risk level, no processing is performed. When the prediction is a high risk level, automatically update the first threshold to : ; is the lowest safety threshold.

[0026] The training process of the support vector machine classification model in Step S12 includes: Step S121: Construct a training set from the clinical impedance scattering ratio data , where j is the index of the jth training sample, represents the impedance scattering ratio value, , represents the risk level label of the jth sample, and obtain the cleaned and labeled training data set; Step S122: Use the grid search method to search within the range of the penalty parameter , with a step size . For each C value, train an SVM model and calculate its performance metrics on the cross-validation set, and select the optimal penalty parameter; Step S123: According to the training data set and the optimal penalty parameter, use the 5-fold cross-validation method to train and validate the SVM model. Randomly divide the training data set into 5 subsets, use 4 subsets for training each time, and the remaining 1 subset for validation; Step S124: Calculate the validation accuracy of each fold and take the average as the final validation accuracy; Step S125: When the final validation accuracy exceeds the target validation accuracy, output the verified optimal SVM model and save it.

[0027] In this embodiment, the treatment safety database collects a large amount of critical impedance scattering ratio data of different skin types, providing a basis for threshold determination. The support vector machine classification model uses this data to predict the risk level of the current impedance scattering ratio through the RBF kernel function (σ = 0.5). For example, when the impedance scattering ratio approaches or exceeds the historical high-risk threshold, the model determines it as a high risk. At this time, the system automatically updates Threshold, fuse the current threshold and the lowest safety threshold to reduce the safety risk.

[0028] The decision function of the SVM model is : ; Among them, is the impedance scattering ratio of the k-th support vector, k is the index of the support vector, n represents the total number of support vectors, is the Lagrange multiplier of the k-th support vector, is the label of the k-th support vector, b is the bias term, is the RBF kernel function, , is the bandwidth parameter.

[0029] The multi-objective optimization algorithm specifically includes: Step S21: Construct the first objective function F according to the skin cutin layer thickness, sebum secretion rate, and the i-th wavelength: ; Among them, is the wavelength matching weight, is the cutin attenuation weight, is the sebum inhibition weight, is the i-th wavelength, is the characteristic absorption peak of the target tissue, is the cutin layer thickness characteristic value, e is the natural constant; Step S22: Perform multi-objective optimization on the wavelength matching degree, the influence of the cutin layer thickness, and the influence of the sebum secretion rate to obtain a subpopulation containing elite individuals, and select the non-dominated solution set on the Pareto front from the subpopulation containing elite individuals; Step S23: Select the target wavelength according to the non-dominated solution set on the Pareto front and the skin cutin layer thickness d, and obtain the target wavelength and its corresponding weight coefficient.

[0030] In this embodiment, different weights are assigned to the three factors of wavelength matching degree, cutin layer thickness, and sebum secretion rate. For example, when the skin cutin layer thickness of the patient is large, the role of the cutin attenuation weight β is more significant, which prompts the algorithm to select a wavelength combination more suitable for thick cutin layers. Through operations such as initializing the population and setting the crossover probability, multi-objective optimization search is performed. The Pareto front screening selects solutions that perform excellently in multiple objectives from all possible solutions to form an optimal wavelength combination. It realizes adjusting the wavelength combination according to the specific skin condition of the patient, improving the stability and reliability of the treatment effect, and is especially suitable for patients with different skin types and disease severities.

[0031] The specific steps for obtaining the non-dominated solution set of the Pareto front in step S22 include: Step S221: Obtain the population size and crossover probability , initialize the population, where each individual in the population represents a set of wavelengths and their corresponding weight coefficients. At the same time, the weight coefficient corresponding to the i-th wavelength Introduce constraint conditions in the initialization stage , , where H is the total number of wavelengths, to obtain an initial population that satisfies the constraint conditions (create an initial population as large as possible, where each individual satisfies the constraint condition that the sum of the weight coefficients is 1); Step S222: Adjust the crossover probability of the initial population according to the sebum secretion rate. The adaptive crossover probability adjustment formula is , is the new crossover probability. When the sebum secretion rate is greater than the target sebum secretion rate, increase the crossover probability to obtain the offspring population after the crossover operation; Step S223: Directly retain the elite individuals in the parent population whose first objective function F value is greater than the optimization target threshold to the offspring population to obtain the offspring population containing elite individuals.

[0032] In this embodiment, for patients with a high sebum secretion rate, this constraint ensures that sufficient weights are assigned to the wavelengths that have a stronger effect on the deep sebaceous glands. The adaptive crossover rate is dynamically adjusted according to the sebum secretion rate. When s > 20 μg / cm²•h, increase the crossover probability to promote gene diversity and avoid premature convergence of the algorithm. The elite retention strategy directly transfers the excellent individuals to the next generation to maintain excellent characteristics.

[0033] Pulse duration The optimization method also includes: Step S31: Establish a thermal damage constraint condition to limit the energy deposition of the laser in the skin tissue, is the safe energy density threshold, is the wavelength energy corresponding to the i-th wavelength at time t; Step S32: Obtain the pulse duration and the constraint weight , incorporate the constraint conditions into the second objective function to solve for the optimal pulse duration, and construct the second objective function , where is the total energy of the i-th wavelength within the pulse duration ; Step S33: Through iterative solution, continuously adjust the pulse duration. When the second objective function When the target convergence condition is reached, the optimal pulse duration that satisfies the thermal damage constraint is obtained, where the iterative solution gives the correction formula , is the corrected pulse duration at the i-th wavelength.

[0034] In this embodiment, the energy deposition of the laser in the skin tissue is restricted not to exceed the safety threshold Φ_max = 15 J / cm². For example, this threshold is determined according to the tolerance of the skin tissue to prevent excessive energy from causing thermal damage. The objective function is constructed by the Lagrange multiplier method, taking into account both the energy utilization efficiency and the thermal damage constraint. During the iterative solution process, the pulse duration is continuously adjusted according to the correction formula to gradually approach the optimal value.

[0035] The specific implementation of the iterative solution in step S33 is as follows: Step S331: Obtain the initial constraint weight and the target convergence condition , is the difference between the second objective function value L and the second objective function value at the previous iteration; Step S332: Calculate the energy accumulation amount at each iteration ; is the energy accumulation amount calculated in the m-th iteration, is the weight coefficient of the i-th wavelength in the m-th iteration, is the pulse duration corresponding to the i-th wavelength in the m-th iteration; Step S333: Update the constraint weight to adjust the weight of the second objective function. The update formula is: , is the learning rate, is the constraint weight in the (m + 1)-th iteration, is the constraint weight in the m-th iteration.

[0036] The calibration process of the drive current includes the following steps: Step S41: Connect a detection resistor R in parallel at the load end of the laser diode; Step S42: Measure the voltage drop V in real time and calculate the actual current value ; Step S43: Adjust the drive voltage through the PID controller to make , is the target current value.

Claims

1. A multi-wavelength pulsed fiber laser for the treatment of seborrheic dermatitis, characterized in that, Including: A data acquisition module, which is used to obtain the skin cutin thickness and sebum secretion rate of a patient, synchronously obtain the skin impedance and optical scattering coefficient, and calculate the current impedance scattering ratio; A treatment planning module, which is used to trigger a wavelength recombination instruction to recalculate the weight coefficient when the current impedance scattering ratio exceeds a first threshold; A wavelength optimization module, which is used to generate weight coefficients including target wavelengths through a multi-objective optimization algorithm according to the skin cutin thickness and sebum secretion rate of the patient; A pulse modulation module, which is used to calculate the pulse duration corresponding to each wavelength according to the weight coefficients; An energy control module, which is used to integrate the energies of each wavelength and calculate the total output energy; A skin feedback module, which is used to obtain the temperature distribution in the treatment area in real time and generate a temperature gradient matrix; A laser emission module, which is used to generate a driving current according to the total output energy and the temperature gradient matrix, drive the laser to emit, and emit a multi-wavelength pulsed fiber laser for the treatment of seborrheic dermatitis.

2. The multi-wavelength pulsed fiber laser for treating seborrheic dermatitis according to claim 1, characterized in that In the data acquisition module, the skin impedance is obtained in the following way: Step S01: Use a four-electrode bio-impedance measurement device to apply a 1 mA / 50 kHz AC signal to the skin to be measured; Step S02: Separating the real part and the imaginary part of the impedance through a phase detection circuit and the imaginary part ; Step S03: Calculate the effective skin impedance value , with a sampling frequency of 100 Hz; Step S04: Transmit Z to the data acquisition module through a wireless transmission method.

3. The multi-wavelength pulsed fiber laser for the treatment of seborrheic dermatitis according to claim 1, characterized in that In the treatment planning module, the first threshold value is determined by the following method: Step S11: Store the critical impedance scattering ratio and clinical impedance scattering ratio data under different skin types, and establish a treatment safety database; Step S12: Based on the current impedance scattering ratio Q value and the treatment safety database, predict the risk level of the current impedance scattering ratio Q value through a support vector machine classification model; Step S13: When the prediction is a non-high-risk level, no processing is performed. When the prediction is a high-risk level, the first threshold is updated to : ； is the minimum safety threshold.

4. The multi-wavelength pulsed fiber laser for the treatment of seborrheic dermatitis according to claim 3, wherein, In step S12, the training process of the support vector machine classification model includes: Step S121: Construct a training set from the clinical impedance scattering ratio data , where j is the index of the j-th training sample represents the impedance scattering ratio value of the j-th training sample represents the risk level label of the j-th sample, and a cleaned and labeled training data set is obtained through the risk level label Step S122: Use the grid search method to search within the range of the penalty parameter , with a step size of . For each C value, train an SVM model and calculate its performance metrics on the cross-validation set, and select the optimal penalty parameter; Step S123: According to the training data set and the optimal penalty parameter, use the 5-fold cross-validation method to train and validate the SVM model. Randomly divide the training data set into 5 subsets, use 4 subsets for training each time, and the remaining 1 subset is used for validation; Step S124: Calculate the validation accuracy of each fold and take the average value as the final validation accuracy; Step S125: When the final validation accuracy exceeds the target validation accuracy, output the verified optimal SVM model and save it.

5. The multi-wavelength pulsed fiber laser for the treatment of seborrheic dermatitis according to claim 4, wherein Decision function of the SVM model is as follows: ； Among them, is the impedance scattering ratio of the k-th support vector, k is the index of the support vector, and n represents the total number of support vectors. is the Lagrange multiplier of the k-th support vector. is the label of the k-th support vector, and b is the bias term. is the RBF kernel function. , is the bandwidth parameter.

6. The multi-wavelength pulsed fiber laser for the treatment of seborrheic dermatitis according to claim 1, characterized in that, In the wavelength optimization module, the multi-objective optimization algorithm specifically includes: Step S21: According to the skin cutin thickness, sebum secretion rate, and the i-th wavelength, construct a first objective function F: ； Among them, is the wavelength matching weight, is the cutin attenuation weight, is the sebum inhibition weight, is the i-th wavelength, is the characteristic absorption peak of the target tissue, is the characteristic value of the stratum corneum thickness, e is the natural constant, d is the skin stratum corneum thickness, and s is the sebum secretion rate; Step S22: Perform multi-objective optimization on the wavelength matching degree, the influence of the cutin thickness, and the influence of the sebum secretion rate to obtain an offspring population containing elite individuals, and select the non-dominated solution set of the Pareto front from the offspring population containing elite individuals; Step S23: According to the non-dominated solution set of the Pareto front and the skin cutin thickness d, select the target wavelength and obtain the target wavelength and its corresponding weight coefficient.

7. The multi-wavelength pulsed fiber laser for treating seborrheic dermatitis according to claim 6, wherein The specific acquisition steps of the non-dominated solution set of the Pareto front in step S22 include: Step S221: Obtain the population size and the crossover probability , initialize the population, where each individual in the population represents a set of wavelengths and their corresponding weight coefficients. At the same time, for the weight coefficient corresponding to the i-th wavelength Introduce constraint conditions in the initialization stage , , where H is the total number of wavelengths; Step S222: Adjust the crossover probability of the initial population according to the sebum secretion rate. The adaptive crossover probability adjustment formula is , is the new crossover probability. When the sebum secretion rate is greater than the target sebum secretion rate, increase the crossover probability to obtain the offspring population after the crossover operation; Step S223: Directly retain the elite individuals in the parent population whose first objective function F value is greater than the optimization target threshold to the offspring population to obtain an offspring population containing elite individuals.

8. The multi-wavelength pulsed fiber laser for treating seborrheic dermatitis according to claim 1, wherein, In the pulse modulation module, the pulse duration is obtained by the following method: , is the reference pulse width, is the anti-zero constant, d is the skin cutin layer thickness, s is the sebum secretion rate, is the weight coefficient of the i-th wavelength; Pulse duration The optimization method also includes: Step S31: Establish thermal damage constraint conditions to limit the energy deposition of the laser in the skin tissue, is the safe energy density threshold, is the wavelength energy corresponding to the i-th wavelength at time t, is the pulse duration corresponding to the i-th wavelength; Step S32: Obtain the pulse duration and the constraint weight , incorporate the constraint condition into the second objective function to solve for the optimal pulse duration, and construct the second objective function , where is the total energy of the i-th wavelength within the pulse duration ; Step S33: Through iterative solution, continuously adjust the pulse duration. When the second objective function reaches the target convergence condition, obtain the optimal pulse duration that satisfies the thermal damage constraint. Among them, the correction formula obtained by iterative solution , is the corrected pulse duration of the i-th wavelength, is the absorption coefficient of the i-th wavelength.

9. The multi-wavelength pulsed fiber laser for the treatment of seborrheic dermatitis according to claim 8, wherein The specific implementation method of the iterative solution in step S33 is: Step S331: Obtain the initial constraint weights and the target convergence condition , which is the difference between the second objective function value L and the previous second objective function value; Step S332: Calculate the energy accumulation amount for each iteration ; is the energy accumulation amount calculated in the m-th iteration, is the weight coefficient of the i-th wavelength in the m-th iteration, is the pulse duration corresponding to the i-th wavelength in the m-th iteration; Step S333: Update the constraint weights to adjust the weights of the second objective function. The update formula is as follows: , is the learning rate, is the constraint weight in the (m + 1)-th iteration, is the constraint weight in the m-th iteration.

10. The multi-wavelength pulsed fiber laser for treating seborrheic dermatitis according to claim 1, wherein In the laser emission module, the calibration process of the driving current includes the following steps: Step S41: Connect a detection resistor R in parallel at the load end of the laser diode; Step S42: Measure the voltage drop V in real time and calculate the actual current value ; Step S43: Adjust the driving voltage through a PID controller to make , the target current value.

Images