A method and system for controlling output power of a carbon dioxide laser therapeutic device

Abstract

The present invention relates to the technical field of power control, and discloses a method and system for controlling the output power of a carbon dioxide laser therapeutic apparatus. The method comprises: performing finite element modeling on the optical path, laser tube and cooling system of the carbon dioxide laser therapeutic apparatus to obtain a three-dimensional temperature field model; performing continuous wave mode analysis to obtain continuous wave power fluctuation influence data; performing pulse mode analysis to obtain pulse power fluctuation influence data; performing comprehensive processing and dual-mode feedforward control analysis to obtain feedforward control parameters; performing multi-local control calculation on real-time laser output data to obtain feedback control parameters; performing multi-layer neural network analysis and laser energy distribution on the feedforward control parameters and feedback control parameters to obtain a multi-stage laser energy distribution scheme. The present invention realizes accurate, stable and adaptive control of the output power of the carbon dioxide laser therapeutic apparatus, thereby improving the safety of the carbon dioxide laser therapeutic apparatus.

Description

Technical Field

[0001] The present invention relates to the technical field of power control, and in particular to an output power control method and system for a carbon dioxide laser therapeutic apparatus. Background Art

[0002] As an important medical device, CO2 laser therapy is widely used in many clinical fields such as dermatology and otolaryngology. However, with the continuous advancement of medical technology and the increasing requirements of patients for treatment accuracy, traditional CO2 laser therapy has exposed some problems in output power control. The laser will produce a heat accumulation effect during operation, causing the output power to fluctuate, affecting the consistency and predictability of the treatment effect.

[0003] Secondly, existing control methods often only consider a single working state in continuous wave or pulse mode, which is difficult to adapt to complex and changing clinical needs. In addition, due to the nonlinear characteristics of the laser system and the interference of environmental factors, traditional fixed parameter control strategies are difficult to achieve high-precision power regulation, especially in the case of long-term operation or frequent switching of working modes, the control performance will be significantly reduced. Summary of the invention

[0004] The present invention provides an output power control method and system for a carbon dioxide laser therapeutic apparatus, which are used to achieve accurate, stable and adaptive control of the output power of the carbon dioxide laser therapeutic apparatus, thereby improving the safety of the carbon dioxide laser therapeutic apparatus.

[0005] In a first aspect, the present invention provides an output power control method for a carbon dioxide laser therapeutic apparatus, the output power control method for a carbon dioxide laser therapeutic apparatus comprising:

[0006] Finite element modeling was performed on the optical path, laser tube and cooling system of the carbon dioxide laser therapy device to obtain a three-dimensional temperature field model;

[0007] Performing continuous wave mode analysis on the three-dimensional temperature field model to obtain continuous wave power fluctuation impact data;

[0008] Performing pulse mode analysis on the three-dimensional temperature field model to obtain pulse power fluctuation impact data;

[0009] Comprehensively processing the continuous wave power fluctuation influence data and the pulse power fluctuation influence data and performing dual-mode feedforward control analysis to obtain feedforward control parameters;

[0010] Performing multi-local control calculations on the real-time laser output data of the carbon dioxide laser therapeutic apparatus to obtain feedback control parameters;

[0011] Multi-layer neural network analysis and laser energy distribution are performed on the feedforward control parameters and the feedback control parameters to obtain a multi-stage laser energy distribution scheme.

[0012] In a second aspect, the present invention provides an output power control system of a carbon dioxide laser therapeutic apparatus, the output power control system of the carbon dioxide laser therapeutic apparatus comprising:

[0013] A modeling module is used to perform finite element modeling on the optical path, laser tube and cooling system of the carbon dioxide laser therapy device to obtain a three-dimensional temperature field model;

[0014] A first mode analysis module is used to perform continuous wave mode analysis on the three-dimensional temperature field model to obtain continuous wave power fluctuation impact data;

[0015] A second mode analysis module is used to perform pulse mode analysis on the three-dimensional temperature field model to obtain pulse power fluctuation impact data;

[0016] A feedforward control module, used for performing comprehensive processing and dual-mode feedforward control analysis on the continuous wave power fluctuation influence data and the pulse power fluctuation influence data to obtain feedforward control parameters;

[0017] A calculation module, used for performing multi-local control calculation on the real-time laser output data of the carbon dioxide laser therapeutic apparatus to obtain feedback control parameters;

[0018] The allocation module is used to perform multi-layer neural network analysis and laser energy allocation on the feedforward control parameters and the feedback control parameters to obtain a multi-stage laser energy allocation scheme.

[0019] The third aspect of the present invention provides an output power control device for a carbon dioxide laser therapy device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the output power control device of the carbon dioxide laser therapy device executes the above-mentioned output power control method of the carbon dioxide laser therapy device.

[0020] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned method for controlling the output power of a carbon dioxide laser therapy device.

[0021] In the technical solution provided by the present invention, a high-precision three-dimensional temperature field model is obtained by performing finite element modeling on the optical path, laser tube and cooling system of the carbon dioxide laser therapeutic apparatus, and an in-depth analysis is performed on the continuous wave mode and the pulse mode respectively, and the power fluctuation influence data under different working states are obtained. By comprehensively processing the influence data of the continuous wave and pulse modes, dual-mode feedforward control is realized, and the power stability problem when switching between different working modes is solved. The real-time laser output data is processed by a multi-local control calculation method, and the dynamically adjusted feedback control parameters are obtained, which improves the system's resistance to external interference. The feedforward and feedback control parameters are analyzed by a multi-layer neural network, and the intelligent distribution of laser energy is realized, and the energy utilization efficiency is optimized. Through the online parameter adjustment mechanism, the adaptive optimization of the control parameters is realized, and the adaptability of the system under different working conditions is enhanced. Combined with feedforward compensation and nonlinear PID control, the control accuracy and stability of the output power are significantly improved. The safety assessment mechanism is introduced in the energy distribution process to ensure the safety and reliability of laser treatment. Through fuzzy rule mapping and multi-stage energy distribution, the control method has strong flexibility. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0023] Figure 1 A schematic diagram of a flow chart of a method for controlling the output power of a carbon dioxide laser therapeutic apparatus provided in an embodiment of the present application;

[0024] Figure 2 This is a schematic block diagram of the structure of the output power control system of the carbon dioxide laser therapy device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0025] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0026] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change based on actual conditions.

[0027] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0028] It should be further understood that the term “and / or” used in the specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0029] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0030] See also Figure 1 , Figure 1 A flow chart of the output power control method of the carbon dioxide laser therapeutic apparatus provided in the embodiment of the present application is as follows: Figure 1 As shown, the output power control method of the carbon dioxide laser therapy device provided in the embodiment of the present application includes steps S100 to S600.

[0031] Step S100, performing finite element modeling on the optical path, laser tube and cooling system of the carbon dioxide laser therapeutic apparatus to obtain a three-dimensional temperature field model;

[0032] It is understandable that the execution subject of the present invention may be an output power control system of a carbon dioxide laser therapeutic apparatus, or a terminal or a server, which is not specifically limited here. The embodiment of the present invention is described by taking a server as an execution subject as an example.

[0033] Specifically, the geometric structure of the carbon dioxide laser therapy device is parametrically modeled to obtain a complete three-dimensional geometric model including the optical path, laser tube and cooling system, reflecting the various physical components inside the laser and their interrelationships. According to the three-dimensional geometric model, the space inside the laser is meshed, and the entire space is divided into multiple finite element mesh units to form a finite element mesh composed of multiple mesh units. Each mesh unit represents a small part of the geometric model. By analyzing the collection of mesh units, the physical behavior of the entire system is approximately simulated. The thermal physical parameters of the material, including thermal conductivity, specific heat capacity and density, are assigned to each mesh unit to form a matrix containing these parameters, reflecting the thermal conduction characteristics of the material under different conditions, so that the model can simulate the thermal conduction behavior of the laser during operation. According to the operating parameters of the laser and the parameters of the cooling system, heat flow boundary conditions are applied to the boundaries of the finite element mesh to simulate the heat flow and heat dissipation process of the laser during operation, and a complete heat conduction model is obtained. A group of partial differential equations for the temperature change over time is established for each mesh unit in the heat conduction model, and these equations are used to describe the temperature change law of each unit. Time discretization is performed according to the partial differential equations to obtain the temperature distribution data corresponding to each time step. Based on the temperature distribution data, thermal stress and thermal deformation are calculated to obtain the stress-strain matrix of the laser structure change, which describes the deformation and internal stress distribution of the laser under thermal action. Thermal-optical coupling analysis is performed on the stress-strain matrix, combining thermodynamics and optical effects, analyzing how temperature changes affect the optical performance of the laser, and generating a three-dimensional temperature field model.

[0034] Step S200, performing continuous wave mode analysis on the three-dimensional temperature field model to obtain continuous wave power fluctuation impact data;

[0035] Specifically, the three-dimensional temperature field model is initialized and set to obtain the initial temperature distribution in the continuous wave mode, so as to establish the temperature distribution of the laser in a stable working state. Based on the working parameters of the continuous wave laser, the initial temperature distribution is calculated by heat source loading to simulate the heat input of the laser during operation, and the steady-state temperature distribution is obtained to reflect the temperature distribution state of the laser in the continuous wave mode. The steady-state temperature distribution is Fourier transformed to obtain the spectral characteristics of the temperature field. Through spectral analysis, the variation characteristics of the temperature field at different frequencies are revealed, and then the temperature field is modally decomposed to obtain the target temperature mode. The extraction of temperature mode helps to identify the temperature distribution mode that is most critical to the performance of the laser, which directly affects the power output and beam quality of the laser. The sensitivity analysis of the target temperature mode is performed to obtain the response function of temperature to power fluctuation, which describes how temperature changes affect the power output of the laser. Based on this response function, power fluctuations of different amplitudes and frequencies are simulated to obtain dynamic temperature distribution. The dynamic temperature distribution reflects the transient response characteristics of the temperature field under power fluctuations. The dynamic temperature distribution is mapped to the optical parameters to obtain the changes in optical properties including the gain medium temperature and the resonant cavity parameters. Through the changes in optical properties, the light field of the laser output is calculated to obtain the change data of the output power, wavelength and beam quality, reflecting the optical performance of the laser under different working conditions. The change data is statistically analyzed and spectrally analyzed to obtain the amplitude spectrum and phase spectrum of the power fluctuation. The impact of power fluctuation on the laser performance is evaluated through the amplitude spectrum and phase spectrum, so as to comprehensively evaluate the sensitivity and stability of power fluctuation, and obtain the continuous wave power fluctuation impact data including fluctuation sensitivity and stability threshold.

[0036] Step S300, performing pulse mode analysis on the three-dimensional temperature field model to obtain pulse power fluctuation impact data;

[0037] Specifically, the three-dimensional temperature field model is time discretized to obtain a high-time resolution temperature field model suitable for pulse mode. The transient characteristics of pulsed lasers require that the model can accurately capture temperature changes on an extremely short time scale. According to the working parameters of the pulsed laser, the model is loaded with a pulse heat source to simulate the temperature changes within a single pulse cycle and obtain the transient temperature distribution of a single pulse cycle. The transient temperature distribution is subjected to a fast Fourier transform to obtain the time-frequency characteristics of the temperature field. The time-frequency characteristic analysis can reveal the variation law of temperature in the time and frequency dimensions, which helps to understand the thermal dynamic behavior in the pulse mode. Based on the time-frequency characteristics, the temperature field is subjected to principal component analysis to extract the target temperature change pattern in the pulse mode. The target temperature change pattern represents the most representative and influential change trend in the temperature field under pulse operation conditions. The target temperature change pattern is subjected to thermal-optical coupling analysis to obtain the influence function of temperature on pulse energy, peak power and pulse width. According to the influence function, simulations of different pulse parameters are performed to obtain the temperature accumulation effect in the pulse sequence. The temperature accumulation effect describes the gradual increase in temperature under the action of continuous pulses and the impact on subsequent pulses, which helps to understand the thermal interaction between pulses. The temperature accumulation effect is nonlinearly fitted to generate a mathematical model of heat accumulation between pulses, which describes the accumulation of thermal effects between pulses. Then, the laser output is simulated by pulse train simulation to simulate the pulse energy stability and pulse shape change data under different operating conditions. The pulse energy stability and pulse shape change data are subjected to wavelet transformation to obtain multi-scale fluctuation characteristics. Wavelet transformation can capture the detailed characteristics of pulse fluctuations at different time scales, which helps to comprehensively evaluate the impact of pulse power fluctuations. Through a comprehensive evaluation of multi-scale fluctuation characteristics, pulse power fluctuation impact data including pulse stability indicators and heat accumulation thresholds are obtained.

[0038] Step S400, performing comprehensive processing and dual-mode feedforward control analysis on the continuous wave power fluctuation influence data and the pulse power fluctuation influence data to obtain feedforward control parameters;

[0039] Specifically, the continuous wave power fluctuation influence data and pulse power fluctuation influence data are fused to obtain the power fluctuation feature matrix, which reflects the influence of continuous wave and pulse fluctuations on laser output. Based on the power fluctuation feature matrix, the continuous wave mode and pulse mode are clustered and analyzed to identify the characteristic boundaries of the two modes. The characteristic boundary reveals the differences and commonalities between the two modes under different conditions and is the key to achieving mode switching. The mode characteristic boundary is fuzzy processed to identify the transition interval between the continuous wave mode and the pulse mode. The transition interval represents the overlapping area that may appear when the laser switches from one mode to another under different working conditions. In order to ensure the smooth transition of the laser between different modes, a dual-mode switching function is constructed according to the transition interval to generate a smooth transition strategy for mode switching, ensuring that the laser will not have abrupt power fluctuations when switching modes, and ensuring the continuity and stability of the output. The smooth transition strategy is modeled in state space to obtain the state equation of the dual-mode system, which describes the dynamic behavior of the laser in continuous wave and pulse modes. Based on the state equation, a linear quadratic regulator is designed to obtain the optimal control law. The optimal control law is the core to ensure the best performance of the laser in different modes. It maximizes the control effect by balancing the stability and response speed of the system. In order to enable the optimal control law to be applied in the actual digital control system, it is discretized to obtain the differential equation suitable for digital control. A feedforward compensator is constructed based on the differential equation. The main function of the feedforward compensator is to pre-calculate and compensate for the effects caused by temperature and power fluctuations to ensure the stability of the output. The calculation of the compensation amount is realized by the compensation amount calculation function, which dynamically adjusts the compensation amount according to the real-time power fluctuation. The compensation amount calculation function is adaptively adjusted to cope with different working conditions and environmental changes. Through adaptive adjustment, an adaptive compensator is generated, which can automatically optimize the control parameters according to the actual situation. Combined with the gain matrix and the compensation coefficient, the feedforward control parameters containing these optimized parameters are generated. The feedforward control parameters are applied to the control system of the laser to ensure the precise control and stability of the laser output power in different operation modes, so as to achieve more efficient treatment effects and higher safety.

[0040] Step S500, performing multi-local control calculation on the real-time laser output data of the carbon dioxide laser therapeutic apparatus to obtain feedback control parameters;

[0041] Specifically, the real-time laser output data of the carbon dioxide laser therapy device is processed by sliding window segmentation, and the continuous time series data is divided into multiple local time series, each of which represents the change of laser output in a specific time period. Fast Fourier transform is performed on multiple local time series to extract the spectrum characteristics of each local area and reveal the frequency components of laser output in different time periods. Based on the local spectrum characteristics, each local time series is adaptively filtered. Adaptive filtering can effectively remove noise and obtain clearer and more accurate local output data. Nonlinear system identification is performed on the denoised local data to establish a dynamic model in each local time period. The local dynamic model reflects the response characteristics of the laser output in a specific time period. According to the local dynamic model, multiple local PID controllers are designed, and each controller corresponds to a local dynamic model. By designing these controllers, a set of local PID parameter sets is obtained. The local PID parameter set is mapped to fuzzy rules to obtain a global PID parameter adjustment strategy, which optimizes the control effect in a global range and ensures the collaboration and consistency between different local controllers. According to the global PID parameter adjustment strategy, a nonlinear PID control model is constructed. The model can comprehensively control the laser output on a global scale to ensure that the output power of the laser remains stable under different operating conditions. The stability analysis of the nonlinear PID control model is carried out to verify its reliability in practical applications. Through the stability analysis, the Lyapunov stability condition is obtained. According to the Lyapunov stability condition, an online parameter adjustment mechanism is generated to adjust the control parameters in real time during the operation of the laser. The online parameter adjustment mechanism is numerically discretized to obtain feedback control parameters including the gain matrix, integral time and differential time. These feedback control parameters are used in the control system of the laser to ensure the precise control of the laser output power and achieve high efficiency, safety and stability during the treatment process.

[0042] Step S600: Perform multi-layer neural network analysis and laser energy distribution on the feedforward control parameters and the feedback control parameters to obtain a multi-stage laser energy distribution scheme.

[0043] Specifically, the feedforward control parameters and feedback control parameters are concatenated to form a 30-dimensional input vector X containing a gain matrix, a compensation coefficient, and a PID parameter. The 30-dimensional input vector X is input to the first hidden layer of the neural network. In this hidden layer, the calculation is performed by matrix multiplication W1X+b1, where W1 is a 64×30 weight matrix and b1 is a 64-dimensional bias vector. Through calculation, a 64-dimensional intermediate result is obtained, which is then processed by the ReLU activation function. The ReLU function sets the negative part to zero and retains the positive part to obtain a 64-dimensional first-layer feature representation H1. The 64-dimensional feature representation H1 is input to the second hidden layer. In this layer, the calculation is also performed by matrix multiplication W2H1+b2, where W2 is a 128×64 weight matrix and b2 is a 128-dimensional bias vector. Through calculation, a 128-dimensional intermediate result is obtained, which is then processed by the ReLU activation function to generate a 128-dimensional second-layer feature representation H2. The 128-dimensional second-layer feature representation H2 is input to the third hidden layer, and is calculated again by matrix multiplication W3H2+b3, where W3 is a 256×128 weight matrix and b3 is a 256-dimensional bias vector. Through calculation, a 256-dimensional intermediate result is obtained, which is processed by the ReLU activation function to generate a 256-dimensional third-layer feature representation H3. In order to prevent overfitting, the third-layer feature representation H3 is dropout processed, and 50% of the elements are randomly set to zero to obtain a 256-dimensional denoised feature representation H3', which effectively reduces the complexity of the neural network and enhances the generalization ability of the model. The 256-dimensional denoised feature representation H3' is input to the output layer and calculated by matrix multiplication W4H3'+b4, where W4 is a 10×256 weight matrix and b4 is a 10-dimensional bias vector. Through calculation, the initial laser energy distribution result is obtained. The Softmax function is applied to the initial laser energy distribution result to generate the probability distribution of multi-stage energy distribution. The Softmax function converts the result into a probability distribution, which is used to represent the energy allocation ratio corresponding to each stage. According to the generated multi-stage energy allocation probability distribution, the inverse transform sampling method is used to generate multiple sets of candidate energy allocation schemes, each of which contains the specific energy values ​​of the corresponding stage. The candidate schemes are further analyzed, including safety assessment and energy utilization efficiency calculation for each set of candidate schemes, to ensure that the selected scheme is not only safe and reliable, but also maximizes energy utilization efficiency. The scheme with the highest energy utilization efficiency is selected as the final multi-stage laser energy allocation scheme to achieve the optimal allocation of laser energy and ensure the maximization and safety of the treatment effect.

[0044] In the embodiment of the present invention, by performing finite element modeling on the optical path, laser tube and cooling system of the carbon dioxide laser therapeutic apparatus, a high-precision three-dimensional temperature field model is obtained, and the continuous wave mode and the pulse mode are deeply analyzed respectively, and the power fluctuation influence data under different working states are obtained. By comprehensively processing the influence data of the continuous wave and pulse modes, dual-mode feedforward control is realized, and the power stability problem when switching between different working modes is solved. The real-time laser output data is processed by a multi-local control calculation method, and the dynamically adjusted feedback control parameters are obtained, which improves the system's resistance to external interference. The feedforward and feedback control parameters are analyzed by a multi-layer neural network, and the intelligent distribution of laser energy is realized, and the energy utilization efficiency is optimized. Through the online parameter adjustment mechanism, the adaptive optimization of the control parameters is realized, and the adaptability of the system under different working conditions is enhanced. Combined with feedforward compensation and nonlinear PID control, the control accuracy and stability of the output power are significantly improved. The safety assessment mechanism is introduced in the energy distribution process to ensure the safety and reliability of laser treatment. Through fuzzy rule mapping and multi-stage energy distribution, the control method has strong flexibility.

[0045] In a specific embodiment, the process of executing step S100 may specifically include the following steps:

[0046] The geometric structure of the carbon dioxide laser therapy device was parametrically modeled to obtain a three-dimensional geometric model including the optical path, laser tube and cooling system;

[0047] The internal space of the laser is meshed according to the three-dimensional geometric model to obtain a finite element mesh composed of multiple mesh units;

[0048] Assign material thermophysical property parameters to each grid unit to obtain a parameter matrix including thermal conductivity, specific heat capacity and density. According to the laser operating parameters and cooling system parameters, apply heat flow boundary conditions to the finite element grid boundary to obtain a heat conduction model.

[0049] A partial differential equation group of temperature variation with time is established for each grid unit in the heat conduction model, and time discretization is performed based on the partial differential equation group to obtain the temperature distribution data for each time step;

[0050] Thermal stress and thermal deformation are calculated based on the temperature distribution data to obtain the stress-strain matrix of the laser structure change. The stress-strain matrix is ​​then subjected to thermal-optical coupling analysis to generate a three-dimensional temperature field model.

[0051] Specifically, the geometric structure of the carbon dioxide laser therapy device is parametrically modeled. Using parametric design tools, the key parts of the laser, such as the optical path, laser tube, cooling system, etc., are geometrically modeled. The advantage of parametric modeling is that various parameters in the model can be flexibly adjusted and optimized, so that the design can adapt to different requirements and conditions. For example, the length and diameter of the laser tube and the size of the fluid channel in the cooling system can be flexibly adjusted through parametric modeling. The internal space of the laser is meshed. By dividing the entire geometric model into multiple small grid units, the local details in the model can be better captured. Usually, the key areas of the laser, such as the inner wall of the laser tube and the reflector area in the optical path, are meshed more finely to ensure that the thermal effects and stresses in these areas can be accurately simulated. For example, in the inner wall area of ​​the laser tube, a smaller grid size is selected to capture the slight changes in heat transfer. After the meshing is completed, the thermophysical parameters of the material are assigned to each grid unit. These parameters include thermal conductivity (k), specific heat capacity (c) and density (ρ). These parameter matrices reflect the response behavior of the material under different thermal environments. Thermal conductivity is a parameter that describes the ability of a material to conduct heat, specific heat capacity is a parameter that describes the amount of heat required to raise a material's unit temperature per unit mass, and density is the ratio of a material's mass to its volume. For example, for laser tube materials, quartz glass is selected, which has the characteristics of high thermal conductivity and low specific heat capacity, which is conducive to rapid heat dissipation. The material of the cooling system may be copper, whose high thermal conductivity can effectively conduct heat to the coolant. According to the operating parameters of the laser and the parameters of the cooling system, heat flow boundary conditions are applied to the boundaries of the finite element mesh. The heat flow boundary conditions determine how heat is transferred at different boundaries during the operation of the laser. For example, on the inner wall of the laser tube, the heat generated by the laser action will be transferred to the surrounding environment through heat conduction, so appropriate heat flow boundary conditions need to be applied in this area. Assuming that the heat flux density of the inner wall of the laser tube is q″, the heat flow boundary conditions are defined as:

[0052]

[0053] Where T is the temperature, n is the direction perpendicular to the boundary, Represents the heat flux density on the boundary. This formula describes the amount of heat conduction along the normal direction. After the heat flux boundary conditions are applied, a heat conduction model is constructed. The heat conduction model describes the heat transfer process in various regions inside the laser. In this model, a group of partial differential equations for the temperature variation over time is established for each grid unit to describe the distribution of heat at different time nodes. This group of equations usually takes the form of a heat conduction equation:

[0054]

[0055] Among them, ρ is the density of the material, c is the specific heat capacity, T is the temperature, t is the time, k is the thermal conductivity, and Q is the internal heat source term. This equation describes how heat is transferred at different time nodes. Represents the heat conduction term, while Q represents the heat source term introduced by factors such as laser irradiation. The partial differential equations are time discretized, the continuous time axis is divided into multiple discrete time steps, and the temperature distribution data is solved at each time step by numerical methods. For example, the time derivative is discretized using the implicit difference method to obtain the temperature distribution data at each time step. Thermal stress and thermal deformation are calculated based on the temperature distribution data. In thermodynamic analysis, temperature changes usually cause stress and deformation of the material, which can be described by the stress-strain relationship. Assuming that the elastic modulus of the material is E and the thermal expansion coefficient is α, the thermal stress can be expressed as:

[0056] σ=E·α·ΔT;

[0057] Among them, σ is thermal stress and ΔT is temperature change. This formula describes the magnitude of stress caused by temperature change. A stress-strain matrix is ​​constructed based on thermal stress data to simulate the deformation of the laser structure under thermal action. A thermal-optical coupling analysis is performed on the stress-strain matrix, combining thermomechanical analysis with optical performance analysis to study how temperature changes affect the optical properties of the laser, such as beam quality, output power, etc. Through analysis, a complete three-dimensional temperature field model is generated to describe the temperature distribution of the laser under different working conditions, and reflect the impact of thermal stress and thermal deformation on the optical properties of the laser.

[0058] In a specific embodiment, the process of executing step S200 may specifically include the following steps:

[0059] Initialize the three-dimensional temperature field model to obtain the initial temperature distribution in the continuous wave mode, and perform heat source loading calculation on the initial temperature distribution according to the working parameters of the continuous wave laser to obtain the steady-state temperature distribution;

[0060] Perform Fourier transform on the steady-state temperature distribution to obtain the frequency spectrum characteristics of the temperature field, and perform modal decomposition on the temperature field based on the frequency spectrum characteristics to obtain the target temperature mode;

[0061] Conduct sensitivity analysis on the target temperature mode to obtain the temperature response function to power fluctuations. Based on the response function, simulate power fluctuations of different amplitudes and frequencies to obtain dynamic temperature distribution.

[0062] Perform optical parameter mapping on the dynamic temperature distribution to obtain the optical property changes including gain medium temperature and resonant cavity parameters. According to the optical property changes, perform optical field calculation on the laser output to obtain the output power, wavelength and beam quality change data.

[0063] Statistical analysis and spectral analysis are performed on the change data to obtain the amplitude spectrum and phase spectrum of the power fluctuation. Based on the amplitude spectrum and phase spectrum, a comprehensive evaluation is performed on the power fluctuation impact to obtain continuous wave power fluctuation impact data including fluctuation sensitivity and stability threshold.

[0064] Specifically, the three-dimensional temperature field model is initialized. The key to the initialization setting is to determine the initial conditions, which includes the starting temperature of each component and the setting of the ambient temperature. During the operation of the continuous wave laser, the laser tube and other related components will gradually heat up, so the initial temperature distribution is often not uniform. It can usually be assumed that the laser is at ambient temperature before startup, that is, the global initial temperature is the ambient temperature, such as 20°C. According to the initial temperature distribution, combined with the specific operating parameters of the continuous wave laser, such as laser power, operating frequency and running time, the initial temperature distribution is calculated by heat source loading to simulate the temperature change process of the laser under continuous operation. The core of the heat source loading calculation is to model the heat source distribution of the laser and solve the steady-state temperature distribution through the heat conduction equation. The heat conduction equation can describe the diffusion process of heat in the material, and the formula is:

[0065]

[0066] Among them, ρ is the density of the material, c is the specific heat capacity, T is the temperature, t is the time, k is the thermal conductivity, and Q is the heat source term. The term on the left side of the equation describes the change of temperature over time, the first term on the right side represents the diffusion of heat in space by conduction, and the second term is the direct heating contribution of the heat source to the system. In continuous wave lasers, the heat source term Q is usually generated by laser absorption, and its distribution can be specifically modeled according to the optical design of the laser. By solving the partial differential equation, the steady-state temperature distribution of the laser after a certain period of time is obtained, showing the temperature distribution of each part of the laser under continuous working state. The steady-state temperature distribution is Fourier transformed to extract the spectral characteristics of the temperature field. The Fourier transform can convert the temperature distribution from the time domain to the frequency domain, revealing the main frequency components of the temperature change. The spectral characteristics of the temperature field are the key to understanding the thermal response of the laser at different frequencies, and can help identify which frequency components contribute the most to the temperature change. Based on the spectral characteristics of the temperature field, the temperature field is modally decomposed to extract the main modes in the temperature field, that is, those modes that have the greatest impact on the behavior of the system. These modes can be considered as basic vibration modes in the temperature field, and each mode corresponds to a typical form of temperature distribution. Through modal decomposition, the complex temperature field is simplified into several main modes, making it easier to analyze and understand the thermal behavior of the system. Perform sensitivity analysis on the target temperature mode to determine the response of the temperature mode to laser power fluctuations. The sensitivity of temperature to power fluctuations is quantified by the response function. Assuming that the fluctuation of laser power is recorded as P(t) and the change of temperature is recorded as T(t), the response function R(t) can be expressed as:

[0067]

[0068] Among them, δT(t) represents the temperature change, and δP(t) represents the power fluctuation. Through this function, the degree of influence of different power fluctuations on the temperature field is quantified. Based on the response function, power fluctuations of different amplitudes and frequencies are simulated to obtain dynamic temperature distribution. Dynamic temperature distribution shows how the temperature of the laser changes with power fluctuations in actual operation. Perform optical parameter mapping. Temperature changes will directly affect the optical properties of the laser, such as the refractive index of the gain medium, the length of the resonant cavity, etc. By mapping the dynamic temperature distribution to the optical parameters, the changes in optical properties including the temperature of the gain medium and the parameters of the resonant cavity are obtained. These changes in optical properties will directly affect the output performance of the laser, such as output power, wavelength, and beam quality. Perform optical field calculations on the output of the laser. The propagation behavior of the laser in the cavity is simulated by optical equations to predict the changes in the output power, wavelength, and beam quality of the laser. These calculation results will reflect how temperature changes affect the output performance of the laser, especially the impact of power fluctuations on beam quality and wavelength stability under different operating conditions. Perform statistical and spectral analysis on the change data of output power, wavelength, and beam quality. Statistical analysis can help identify the fluctuation characteristics in the laser output, while spectrum analysis can reveal the main frequency components of these fluctuations. Through analysis, the amplitude spectrum and phase spectrum of power fluctuations are obtained, which describe the amplitude and phase changes of power fluctuations at different frequencies. Based on the amplitude spectrum and phase spectrum, the impact of power fluctuations is comprehensively evaluated to obtain continuous wave power fluctuation impact data including fluctuation sensitivity and stability threshold.

[0069] In a specific embodiment, the process of executing step S300 may specifically include the following steps:

[0070] The three-dimensional temperature field model is time discretized to obtain a high time resolution temperature field model suitable for the pulse mode. According to the working parameters of the pulsed laser, the high time resolution temperature field model is loaded with a pulse heat source to obtain the transient temperature distribution of a single pulse period.

[0071] Perform fast Fourier transform on transient temperature distribution to obtain the time-frequency characteristics of the temperature field, and perform principal component analysis on the temperature field based on the time-frequency characteristics to obtain the target temperature change pattern under the pulse mode;

[0072] Conduct thermal-optical coupling analysis on the target temperature variation pattern to obtain the influence function of temperature on pulse energy, peak power and pulse width. Based on the influence function, simulate different pulse parameters to obtain the temperature accumulation effect of the pulse sequence.

[0073] The temperature accumulation effect is nonlinearly fitted to obtain the mathematical model of heat accumulation between pulses. Based on the mathematical model, the laser output is simulated by pulse train to obtain the pulse energy stability and pulse shape change data.

[0074] The pulse energy stability and pulse shape change data are subjected to wavelet transform to obtain multi-scale fluctuation characteristics. Based on the multi-scale fluctuation characteristics, the influence of pulse power fluctuation is comprehensively evaluated to obtain pulse power fluctuation influence data including pulse stability index and thermal accumulation threshold.

[0075] Specifically, in pulse mode, the laser emits high-energy laser pulses at extremely short time intervals, resulting in transient temperature changes in the laser tube and the surrounding medium. This transient temperature change has an important impact on the thermal management and stability of the laser. In order to capture these rapidly changing temperature characteristics, the three-dimensional temperature field model is time-discretized so that it can simulate the working process of the pulsed laser with a sufficiently high time resolution. The core of time discretization is to divide the continuous time domain into multiple discrete time steps. At each time step, the change of the temperature field is solved. This process is usually implemented using explicit or implicit numerical methods. For pulsed lasers, due to their transient characteristics, explicit methods are usually able to better capture rapid temperature changes. Assuming the time step is Δt, at each time step, the evolution of the temperature field can be described by the heat conduction equation:

[0076]

[0077] Among them, T n is the temperature distribution at time step n, T n+1 is the temperature distribution at time step n+1, Δt is the time step, k is the thermal conductivity of the material, ρ is the density, c is the specific heat capacity, Q n is the heat source term at time step n. Through this formula, the temperature distribution is gradually updated at each time step to obtain a high time resolution temperature field model suitable for pulse mode. According to the working parameters of the pulsed laser, the high time resolution temperature field model is loaded with a pulsed heat source to simulate the transient thermal effect generated by each laser pulse in the material. Pulsed lasers usually release a large amount of energy in a very short time, resulting in a rapid rise in local temperature. The heat source term Q n It can be defined according to the specific parameters of the pulsed laser (such as pulse energy, pulse width, repetition frequency, etc.). For example, the pulsed heat source term can be expressed as a heat source distribution with a pulse function in time:

[0078] Q(t) = Q0·δ(t-t0);

[0079] Where Q0 is the pulse peak heat source intensity, and δ(t-t0) is the time pulse function of the pulse at time t0. Through this loading process, the transient temperature distribution within each pulse period is obtained. Fast Fourier transform is performed on the transient temperature distribution to analyze the time-frequency characteristics of the temperature field. Fourier transform can convert the temperature distribution from the time domain to the frequency domain, revealing the main frequency components of the temperature change. For example, Fourier transform can help identify which frequency components contribute most to temperature fluctuations, thereby guiding the design and optimization of the laser. According to the time-frequency characteristics, principal component analysis is performed on the temperature field to extract the target temperature change pattern in the pulse mode. Principal component analysis is a dimensionality reduction technique that extracts the main features by identifying the direction with the largest variance in the data. In the pulse mode, the target temperature change pattern represents the temperature distribution form that has the greatest impact on the thermal behavior of the system. The target temperature change pattern is subjected to thermal-optical coupling analysis to study how temperature changes affect the optical performance of the laser. Temperature changes affect the refractive index inside the laser, the excited state of the gain medium, and the length of the resonant cavity, thereby changing the output characteristics of the laser. Through thermal-optical coupling analysis, the influence function of temperature on pulse energy, peak power and pulse width is obtained. Assuming that the change in pulse peak power caused by temperature change ΔT is ΔP, the influence function can be expressed as:

[0080] ΔP=f(ΔT);

[0081] Among them, f(ΔT) is the influence function of temperature change on pulse power, which describes how temperature fluctuations affect the output power of the laser. Based on this influence function, different pulse parameters are simulated to obtain the temperature accumulation effect of the pulse sequence. The temperature accumulation effect describes how the temperature in the material gradually increases under the action of multiple laser pulses and affects the thermal effect of subsequent pulses. The accumulation effect is particularly important in high repetition rate pulse lasers, because the accumulation of temperature between pulses may cause the performance degradation or instability of the laser. The temperature accumulation effect is nonlinearly fitted to obtain a mathematical model of heat accumulation between pulses. This mathematical model can be used to predict the temperature increase and accumulation rate of the laser under specific working conditions. For example, assuming that the accumulation effect can be described by an exponential function, the model can be expressed as:

[0082] T acc (t) = T0 + A(1-e -βt );

[0083] Among them, T acc(t) is the cumulative temperature, T0 is the initial temperature, A is the temperature rise amplitude, and β is the cumulative rate constant. The model is used to predict the temperature trend of the laser under long-term operation, thereby guiding the design of the cooling and thermal management system. Based on the mathematical model of the cumulative effect, the pulse train simulation of the laser output is performed to obtain data on pulse energy stability and pulse shape changes. These simulation data can reflect the changes in the output characteristics of the laser under different pulse conditions and are the key basis for optimizing laser design and operating conditions. Wavelet transform is performed on the pulse energy stability and pulse shape change data. Wavelet transform can capture the changing characteristics of the signal at different time scales, which helps to identify the multi-scale fluctuation characteristics of pulse power fluctuations. By analyzing these characteristics, the impact of pulse power fluctuations on laser performance is evaluated, and pulse power fluctuation influence data including pulse stability indicators and thermal accumulation thresholds are obtained.

[0084] In a specific embodiment, the process of executing step S400 may specifically include the following steps:

[0085] The continuous wave power fluctuation influence data and the pulse power fluctuation influence data are fused to obtain the power fluctuation characteristic matrix, and the continuous wave mode and the pulse mode are clustered according to the power fluctuation characteristic matrix to obtain the mode characteristic boundary;

[0086] The mode feature boundary is fuzzy processed to obtain the transition interval between the continuous wave mode and the pulse mode. Based on the transition interval, a dual-mode switching function is constructed to obtain a smooth transition strategy for mode switching.

[0087] The smooth transition strategy is modeled in state space to obtain the state equation of the dual-mode system. Based on the state equation, a linear quadratic regulator is designed to obtain the optimal control law.

[0088] Discretize the optimal control law to obtain the difference equation of digital control, and build a feedforward compensator based on the difference equation to obtain the temperature and power compensation calculation function;

[0089] The compensation amount calculation function is adaptively adjusted to obtain a compensator, and the control parameters are optimized according to the compensator to obtain feedforward control parameters including a gain matrix and a compensation coefficient.

[0090] Specifically, the continuous wave power fluctuation influence data and the pulse power fluctuation influence data are fused to obtain the power fluctuation feature matrix. The continuous wave mode and the pulse mode show different fluctuation characteristics in the time domain and frequency domain. The data of these two modes are standardized to eliminate the difference in the numerical magnitude of different features, so that different features can have the same weight when fused. Assume that the power fluctuation feature of the continuous wave mode is X cw , the power fluctuation characteristic of the pulse mode is Xpulse , then data fusion can be expressed in the form of feature matrix X:

[0091] X=[X cw ∣X pulse ];

[0092] Among them, X is a matrix containing all power fluctuation features, each column represents a different feature, and each row represents a different observation point. In this way, the features of the continuous wave mode and the pulse mode are combined to form a complete power fluctuation feature matrix. The continuous wave mode and the pulse mode are clustered and divided into two categories according to the data in the feature matrix: one category represents the features of the continuous wave mode, and the other category represents the features of the pulse mode. Through clustering algorithms such as K-means, the feature boundaries in each mode are identified, which describe the feature differences between the two modes. For example, in K-means clustering, the number of cluster centers is set to 2 to divide the data and obtain the feature boundaries of the two modes. The mode feature boundaries are fuzzy processed to solve the transition problem of mode conversion in actual operation. By fuzzifying the feature boundaries, a transition interval between the continuous wave mode and the pulse mode is generated. This interval can be expressed as a fuzzy set, which describes the situation where the laser may simultaneously exhibit the characteristics of two modes under different working conditions. Based on the fuzzy transition interval, a dual-mode switching function is constructed to ensure that the laser can achieve a smooth transition when switching from one mode to another. Assume that the switching function is σ(t), which can be expressed by the Sigmoid function:

[0093]

[0094] Among them, α is a parameter for adjusting the transition smoothness, and t0 is the switching time. In this form, σ(t) exhibits smooth transition characteristics before and after mode switching, avoiding sudden changes in the laser during mode switching. The smooth transition strategy of mode switching is modeled in state space to describe the dynamic behavior of the laser in different modes. Assuming that the system state vector is x(t) and the control input is u(t), the state equation can be expressed as:

[0095]

[0096] Where A is the system matrix and B is the control matrix. The state equation describes the dynamic evolution of the system in continuous wave mode and pulse mode. Based on the state equation, a linear quadratic regulator (LQR) is designed to obtain the optimal control law of the system. The goal of LQR is to optimize the system performance by minimizing a quadratic performance index. The performance index can be expressed as:

[0097]

[0098] Among them, Q is the state weight matrix, and R is the control weight matrix. By solving the optimization problem, the optimal control law u(t) = -Kx(t) is obtained, where K is the feedback gain matrix. In order to make the optimal control law applicable in digital control systems, it is discretized and the continuous-time control law is converted into a control strategy that can be executed in discrete time steps. Through the difference method, the continuous-time state equation is converted into a difference equation:

[0099] x[k+1]=A d x[k]+B d u[k];

[0100] Among them, A d and B d is the discretized system matrix. Based on the difference equation, a feedforward compensator is constructed to calculate the temperature and power compensation of the system at each time step. The compensation calculation function can be expressed as:

[0101] u[k]=u0[k]+Fe[k];

[0102] Among them, u0[k] is the feedforward control quantity, e[k] is the error vector, and F is the compensation matrix. The compensation quantity calculation function is adaptively adjusted to generate an adaptive compensator, so that the compensator can dynamically adjust the compensation coefficient according to the actual operating conditions, thereby improving the robustness and adaptability of the system. For example, when the working environment of the system changes, the compensator can automatically adjust the gain matrix and compensation coefficient to ensure that the system can still maintain a stable and efficient working state. By adjusting and optimizing the compensator, the feedforward control parameters including the gain matrix and compensation coefficient are obtained. These control parameters can ensure that the laser achieves optimal performance in various operating modes, avoid the negative impact of power fluctuations on the output, and ensure the smooth operation of the system when the mode is switched.

[0103] In a specific embodiment, the process of executing step S500 may specifically include the following steps:

[0104] Sliding window segmentation is performed on the real-time laser output data of the carbon dioxide laser therapy device to obtain multiple local time series, and fast Fourier transform is performed on the multiple local time series to obtain local spectrum features;

[0105] According to the local spectrum characteristics, each local part is adaptively filtered to obtain the denoised local output data, and the denoised local output data is subjected to nonlinear system identification to obtain the local dynamic model;

[0106] According to the local dynamic model, multiple local PID controllers are designed to obtain the local PID parameter set, and the local PID parameter set is mapped by fuzzy rules to obtain the global PID parameter adjustment strategy;

[0107] According to the global PID parameter adjustment strategy, a nonlinear PID control model is constructed, and the stability analysis of the nonlinear PID control model is performed to obtain the Lyapunov stability condition;

[0108] According to the Lyapunov stability condition, an online parameter adjustment mechanism is generated and numerically discretized to obtain feedback control parameters including gain matrix, integral time and differential time.

[0109] Specifically, the continuous laser output signal is divided into multiple local time series with a certain time length. Sliding window segmentation sets a time window of fixed length and slides the window over the entire signal sequence, gradually intercepting local sections of the signal to generate multiple local time series. Assuming that the original laser output signal is x(t), the length of the sliding window is L, and the sliding step is Δ, then the i-th local time series can be expressed as:

[0110] x i (t) = x(t+iΔ), t∈[0,L];

[0111] In this way, multiple local time series are generated on the entire time series, and these local time series can reflect the characteristic changes of the laser output signal in different time periods. Fast Fourier transform is performed on the local time series to extract local spectrum features. Fast Fourier transform can convert time domain signals into frequency domain signals, revealing the distribution of signals on different frequency components. For each local time series, fast Fourier transform is applied to obtain its spectrum representation:

[0112]

[0113] Among them, X i (f) represents the spectrum of the i-th local time series, f is the frequency, and L is the length of the time series. By performing Fourier transform on each local time series, a set of local spectrum features is obtained, which reflects the frequency distribution of the laser output signal in different time periods. According to the local spectrum features, each local time series is adaptively filtered to remove the noise components in the signal while retaining the useful frequency information. The filter coefficient of the adaptive filter will be dynamically adjusted according to the characteristics of the input signal to achieve the best denoising effect. Assume that the input signal of the filter is X i (f), the output of the filter is Y i (f), the adaptive coefficient of the filter is α i (f), the filtered signal can be expressed as:

[0114] Y i (f) = α i(f)X i (f);

[0115] Among them, α i (f) Adjust according to the local characteristics of the signal to ensure the optimization of the denoising effect. In this way, the denoised local output data is obtained. Nonlinear system identification is performed on the denoised local output data to build a local dynamic model. By analyzing the input and output data, a mathematical model that can describe the dynamic behavior of the system is established. Assume that the local output signal is y i (t), the input signal is u i (t), the local dynamic model can be expressed as a nonlinear equation:

[0116] y i (t) = f(u i (t),y i (t-1),…,y i (tn));

[0117] Where f is a nonlinear function and n is the memory length of the system. By identifying local data in different time periods, multiple local dynamic models are obtained to reflect the dynamic characteristics of the laser output in different time periods. Based on the local dynamic model, multiple local PID controllers are designed. PID controller is the most commonly used feedback controller in industrial control. Its core is to achieve precise control of the system by adjusting the three parameters of proportion (P), integration (I) and differentiation (D). Assume that the control output of the i-th local dynamic model is u i (t), the control error is e i (t) = r i (t)-y i (t), where r i (t) is the reference signal, the output of the local PID controller can be expressed as:

[0118]

[0119] Among them, K p,i , K i,i and K d,iThey are proportional, integral and differential gains respectively. By adjusting these gain parameters, the control effect is optimized. In order to achieve unified control on a global scale, fuzzy rule mapping is performed on the local PID parameter set. Fuzzy control is a control strategy based on fuzzy logic. By defining fuzzy rules, effective control can be achieved under large uncertainties. The core of fuzzy rule mapping is to convert local PID parameters into global PID parameter adjustment strategies through fuzzy logic. For example, according to the performance indicators of the local controller, such as steady-state error and response time, they are mapped into fuzzy variables, and then the global control strategy is obtained through fuzzy reasoning. According to the global PID parameter adjustment strategy, a nonlinear PID control model is constructed so that the control strategy can adapt to the nonlinear behavior of the system. Assuming that the nonlinear behavior of the system can be described by the state equation, the nonlinear PID control model can be expressed as:

[0120]

[0121] Where x(t) is the system state vector, and A and B are nonlinear matrices related to the state. The stability of the nonlinear PID control model needs to be verified by the Lyapunov stability condition. The Lyapunov stability condition is an important tool for judging the stability of nonlinear systems. By constructing a Lyapunov function V(x), if it can be proved that the function is positive definite in the global scope of the system and its time derivative is semi-negative definite, it can be proved that the system is asymptotically stable. Assume that the Lyapunov function is:

[0122] V(x)=x T Px;

[0123] Where P is a positive definite matrix, then the stability condition is After confirming the Lyapunov stability of the system, an online parameter adjustment mechanism is generated. The parameters of the PID controller are adjusted in real time according to the actual operation conditions to cope with changes in the system dynamic characteristics and external disturbances. In order to enable online adjustment in the digital control system, it is numerically discretized to obtain feedback control parameters including gain matrix, integral time and differential time.

[0124] In a specific embodiment, the process of executing step S600 may specifically include the following steps:

[0125] The feedforward control parameters and feedback control parameters are concatenated to obtain a 30-dimensional input vector X including a gain matrix, compensation coefficients, and PID parameters;

[0126] Input the input vector X into the first hidden layer and calculate it through matrix multiplication W1X+b1, where W1 is a 64×30 weight matrix and b1 is a 64-dimensional bias vector, to obtain a 64-dimensional intermediate result, which is then processed by the ReLU activation function max(0, x) to obtain a 64-dimensional first-layer feature representation H1;

[0127] The first layer feature representation H1 is input into the second hidden layer and calculated by matrix multiplication W2H1+b2, where W2 is a 128×64 weight matrix and b2 is a 128-dimensional bias vector, to obtain a 128-dimensional intermediate result, which is then processed by the ReLU activation function to obtain a 128-dimensional second layer feature representation H2;

[0128] The second layer feature representation H2 is input into the third hidden layer and calculated by matrix multiplication W3H2+b3, where W3 is a 256×128 weight matrix and b3 is a 256-dimensional bias vector, to obtain a 256-dimensional intermediate result, which is then processed by the ReLU activation function to obtain a 256-dimensional third layer feature representation H3;

[0129] Perform dropout processing on the third-layer feature representation H3, randomly set 50% of the elements to 0, and obtain a 256-dimensional denoised feature representation H3';

[0130] The denoising feature representation H3' is input into the output layer and calculated by matrix multiplication W4H3'+b4, where W4 is a 10×256 weight matrix and b4 is a 10-dimensional bias vector, to obtain the initial laser energy allocation result;

[0131] Apply the Softmax function to the initial laser energy allocation result to obtain the multi-stage energy allocation probability distribution, and use the inverse transform sampling method to generate multiple groups of candidate energy allocation schemes based on the multi-stage energy allocation probability distribution, each group of schemes contains the energy value of the corresponding stage;

[0132] The safety assessment and energy utilization efficiency calculation of multiple groups of candidate energy allocation schemes are performed, and the scheme with the highest energy utilization efficiency is selected as the multi-stage laser energy allocation scheme.

[0133] Specifically, the feedforward control parameters usually include the system's response time, target output, etc., which are used to predict the system's behavior; the feedback control parameters include error information, actual output, and the gain of the PID controller. These parameters need to be arranged in a certain order and converted to the same numerical range through data preprocessing methods such as standardization or normalization. Assuming that the feedforward control parameter vector is f and the feedback control parameter vector is b, the concatenated 30-dimensional input vector X can be expressed as:

[0134] X = [f | b];

[0135] Among them, X includes the relevant information of all control parameters. Input vector X is input into the first hidden layer of the neural network. The first hidden layer obtains a 64-dimensional intermediate result through matrix multiplication and addition. The input vector X is multiplied by the weight matrix W1 and the bias vector b1 is added. The calculation formula is:

[0136] Z1=W1X+b1;

[0137] Among them, W1 is a 64×30 weight matrix, which represents the connection strength between each input dimension and each hidden layer neuron; b1 is a 64-dimensional bias vector, which provides translation adjustment for each neuron. The calculated Z1 is a 64-dimensional intermediate result, each element of which represents the weighted sum of the input vector and the corresponding weight. In order to introduce nonlinearity, the ReLU activation function is applied to Z1. The calculation formula of ReLU is:

[0138] H1=ReLU(Z1)=max(0,Z1);

[0139] The ReLU function truncates all negative values ​​to 0, leaves positive values ​​unchanged, and generates a 64-dimensional first-layer feature representation H1. Nonlinear processing can help the network learn more complex patterns and features. The first-layer feature representation H1 is input into the second hidden layer, and the calculation process is similar to the first layer. The 128-dimensional intermediate result is obtained through matrix multiplication and addition:

[0140] Z2=W2H1+b2;

[0141] Among them, W2 is a 128×64 weight matrix, and b2 is a 128-dimensional bias vector. The ReLU activation function is applied to Z2 again to obtain the 128-dimensional second-layer feature representation H2. This process gradually extracts deep features in the input vector X, allowing the neural network to capture more complex patterns and relationships. The second-layer feature representation H2 is input to the third hidden layer, and through similar calculation steps, a 256-dimensional intermediate result Z3 is obtained:

[0142] Z3=W3H2+b3;

[0143] Among them, W3 is a 256×128 weight matrix, and b3 is a 256-dimensional bias vector. After applying the ReLU activation function to Z3, a 256-dimensional third-layer feature representation H3 is generated. In order to prevent overfitting of the neural network, the third-layer feature representation H3 is subjected to dropout processing. In the dropout processing, 50% of the elements are randomly set to 0, which means that in each training, only half of the features will be activated, thereby reducing the model's over-reliance on specific features. Noise reduction processing helps to improve the generalization ability of the model and obtains a 256-dimensional denoised feature representation H′3. The denoised feature representation H′3 is input into the output layer. The calculation of the output layer is similar to that of the previous layers. The initial laser energy distribution result is obtained by matrix multiplication and addition:

[0144] Z4=W4H′3+b4;

[0145] Among them, W4 is a 10×256 weight matrix, and b4 is a 10-dimensional bias vector. At this time, Z4 represents the initial value of 10 different energy allocation schemes. In order to convert the initial value into a probability distribution, the Softmax function is applied next. The Softmax function can convert any real number vector into a probability vector so that the sum of each component is 1. The specific calculation formula is:

[0146]

[0147] Among them, z i is the i-th element in Z4, Softmax(z i ) represents the probability of the i-th energy allocation scheme. The probability distribution of multi-stage energy allocation is obtained through the Softmax function, which represents the selection probability among different energy allocation schemes. According to the probability distribution, the inverse transform sampling method is used to generate multiple sets of candidate energy allocation schemes. The inverse transform sampling method is an effective method to generate random samples from a given probability distribution. The specific steps are to first generate a uniformly distributed random number, and then find the corresponding allocation scheme according to the cumulative distribution function (CDF). Each set of candidate schemes contains the energy value of the corresponding stage, indicating the energy allocation that may be selected in actual operation. After generating multiple sets of candidate energy allocation schemes, these schemes are subjected to safety evaluation and energy utilization efficiency calculation. The safety evaluation ensures that each candidate scheme will not produce unsafe laser output in clinical applications, while the energy utilization efficiency calculation measures the efficiency of each scheme under given conditions. By comparing these schemes, the scheme with the highest energy utilization efficiency is selected as the final multi-stage laser energy allocation scheme.

[0148] See also Figure 2 , Figure 2 The structure schematic block diagram of the output power control system 200 of the carbon dioxide laser therapy device provided in the embodiment of the present application is as follows: Figure 2 As shown, the output power control system 200 of the carbon dioxide laser therapeutic apparatus comprises:

[0149] Modeling module 210, used to perform finite element modeling on the optical path, laser tube and cooling system of the carbon dioxide laser therapy device to obtain a three-dimensional temperature field model;

[0150] A first mode analysis module 220 is used to perform continuous wave mode analysis on the three-dimensional temperature field model to obtain continuous wave power fluctuation impact data;

[0151] The second mode analysis module 230 is used to perform pulse mode analysis on the three-dimensional temperature field model to obtain pulse power fluctuation impact data;

[0152] A feedforward control module 240 is used to perform comprehensive processing and dual-mode feedforward control analysis on the continuous wave power fluctuation influence data and the pulse power fluctuation influence data to obtain feedforward control parameters;

[0153] The calculation module 250 is used to perform multi-local control calculations on the real-time laser output data of the carbon dioxide laser therapeutic apparatus to obtain feedback control parameters;

[0154] The allocation module 260 is used to perform multi-layer neural network analysis and laser energy allocation on the feedforward control parameters and the feedback control parameters to obtain a multi-stage laser energy allocation scheme.

[0155] Through the cooperation of the above components, a high-precision three-dimensional temperature field model was obtained by finite element modeling of the optical path, laser tube and cooling system of the carbon dioxide laser therapy instrument. The continuous wave mode and pulse mode were deeply analyzed respectively, and the power fluctuation influence data under different working conditions were obtained. By comprehensively processing the influence data of the continuous wave and pulse modes, dual-mode feedforward control was realized, and the power stability problem when switching between different working modes was solved. The multi-local control calculation method was used to process the real-time laser output data, and the dynamically adjusted feedback control parameters were obtained, which improved the system's resistance to external interference. The multi-layer neural network was used to analyze the feedforward and feedback control parameters, and the intelligent distribution of laser energy was realized, optimizing the energy utilization efficiency. Through the online parameter adjustment mechanism, the adaptive optimization of the control parameters was realized, and the adaptability of the system under different working conditions was enhanced. Combined with feedforward compensation and nonlinear PID control, the control accuracy and stability of the output power were significantly improved. The safety assessment mechanism was introduced in the energy distribution process to ensure the safety and reliability of laser treatment. Through fuzzy rule mapping and multi-stage energy distribution, the control method has strong flexibility.

[0156] The present application also provides an output power control device for a carbon dioxide laser therapy device, the output power control device for the carbon dioxide laser therapy device comprising a memory and a processor, the memory storing computer-readable instructions, and when the computer-readable instructions are executed by the processor, the processor executes the steps of the output power control method for the carbon dioxide laser therapy device in the above-mentioned embodiments.

[0157] The present application also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the output power control method of the carbon dioxide laser therapy device.

[0158] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0159] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0160] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for controlling the output power of a carbon dioxide laser therapeutic apparatus, characterized in that: include: Finite element modeling was performed on the optical path, laser tube and cooling system of the carbon dioxide laser therapy device to obtain a three-dimensional temperature field model; Performing continuous wave mode analysis on the three-dimensional temperature field model to obtain continuous wave power fluctuation impact data; Performing pulse mode analysis on the three-dimensional temperature field model to obtain pulse power fluctuation impact data; Comprehensively processing the continuous wave power fluctuation influence data and the pulse power fluctuation influence data and performing dual-mode feedforward control analysis to obtain feedforward control parameters; Performing multi-local control calculations on the real-time laser output data of the carbon dioxide laser therapeutic apparatus to obtain feedback control parameters; Multi-layer neural network analysis and laser energy distribution are performed on the feedforward control parameters and the feedback control parameters to obtain a multi-stage laser energy distribution scheme.

2. The output power control method of the carbon dioxide laser therapeutic apparatus according to claim 1, characterized in that: The finite element modeling of the optical path, laser tube and cooling system of the carbon dioxide laser therapeutic apparatus is performed to obtain a three-dimensional temperature field model, including: The geometric structure of the carbon dioxide laser therapy device was parametrically modeled to obtain a three-dimensional geometric model including the optical path, laser tube and cooling system; Meshing the internal space of the laser according to the three-dimensional geometric model to obtain a finite element mesh composed of a plurality of mesh units; Assign material thermophysical property parameters to each grid unit to obtain a parameter matrix including thermal conductivity, specific heat capacity and density. According to the laser operating parameters and cooling system parameters, apply heat flow boundary conditions to the finite element grid boundary to obtain a heat conduction model. Establishing a partial differential equation group of temperature variation with time for each grid unit in the heat conduction model, and performing time discretization according to the partial differential equation group to obtain temperature distribution data for each time step; Thermal stress and thermal deformation are calculated according to the temperature distribution data to obtain a stress-strain matrix of laser structure change, and a thermal-optical coupling analysis is performed on the stress-strain matrix to generate a three-dimensional temperature field model.

3. The output power control method of the carbon dioxide laser therapeutic apparatus according to claim 1, characterized in that: The continuous wave mode analysis is performed on the three-dimensional temperature field model to obtain continuous wave power fluctuation impact data, including: Initializing the three-dimensional temperature field model to obtain an initial temperature distribution in a continuous wave mode, and performing a heat source loading calculation on the initial temperature distribution according to the working parameters of the continuous wave laser to obtain a steady-state temperature distribution; Performing Fourier transformation on the steady-state temperature distribution to obtain the frequency spectrum characteristics of the temperature field, and performing modal decomposition on the temperature field according to the frequency spectrum characteristics to obtain the target temperature mode; Performing sensitivity analysis on the target temperature mode to obtain a response function of temperature to power fluctuation, and simulating power fluctuations of different amplitudes and frequencies based on the response function to obtain a dynamic temperature distribution; Performing optical parameter mapping on the dynamic temperature distribution to obtain optical property changes including gain medium temperature and resonant cavity parameters, and performing optical field calculation on the laser output according to the optical property changes to obtain output power, wavelength and beam quality change data; Statistical analysis and spectrum analysis are performed on the change data to obtain the amplitude spectrum and phase spectrum of the power fluctuation, and based on the amplitude spectrum and phase spectrum, a comprehensive evaluation is performed on the power fluctuation impact to obtain continuous wave power fluctuation impact data including fluctuation sensitivity and stability threshold.

4. The output power control method of the carbon dioxide laser therapeutic apparatus according to claim 1, characterized in that: The pulse mode analysis of the three-dimensional temperature field model is performed to obtain pulse power fluctuation impact data, including: Performing time discretization processing on the three-dimensional temperature field model to obtain a high time resolution temperature field model suitable for the pulse mode, and performing pulse heat source loading on the high time resolution temperature field model according to the working parameters of the pulsed laser to obtain the transient temperature distribution of a single pulse period; Performing a fast Fourier transform on the transient temperature distribution to obtain the time-frequency characteristics of the temperature field, and performing a principal component analysis on the temperature field based on the time-frequency characteristics to obtain a target temperature change pattern under a pulse mode; Performing a thermal-optical coupling analysis on the target temperature variation pattern to obtain an influence function of temperature on pulse energy, peak power and pulse width, and simulating different pulse parameters based on the influence function to obtain a temperature accumulation effect of a pulse sequence; Performing nonlinear fitting on the temperature accumulation effect to obtain a mathematical model of inter-pulse heat accumulation, and performing pulse train simulation on the laser output based on the mathematical model to obtain pulse energy stability and pulse shape change data; The pulse energy stability and pulse shape change data are subjected to wavelet transformation to obtain multi-scale fluctuation characteristics, and based on the multi-scale fluctuation characteristics, the pulse power fluctuation influence is comprehensively evaluated to obtain pulse power fluctuation influence data including pulse stability index and heat accumulation threshold.

5. The output power control method of the carbon dioxide laser therapeutic apparatus according to claim 1, characterized in that: The comprehensive processing and dual-mode feedforward control analysis of the continuous wave power fluctuation influence data and the pulse power fluctuation influence data to obtain feedforward control parameters include: Performing data fusion on the continuous wave power fluctuation influence data and the pulse power fluctuation influence data to obtain a power fluctuation feature matrix, and performing cluster analysis on the continuous wave mode and the pulse mode according to the power fluctuation feature matrix to obtain a mode feature boundary; Performing fuzzy processing on the mode feature boundary to obtain a transition interval between a continuous wave mode and a pulse mode, and constructing a dual-mode switching function based on the transition interval to obtain a smooth transition strategy for mode switching; Performing state space modeling on the smooth transition strategy to obtain a state equation of the dual-mode system, and designing a linear quadratic regulator based on the state equation to obtain an optimal control law; Discretizing the optimal control law to obtain a differential equation for digital control, and constructing a feedforward compensator based on the differential equation to obtain a temperature and power compensation calculation function; The compensation amount calculation function is adaptively adjusted to obtain a compensator, and the control parameters are optimized according to the compensator to obtain feedforward control parameters including a gain matrix and a compensation coefficient.

6. The output power control method of the carbon dioxide laser therapeutic apparatus according to claim 1, characterized in that: The multi-local control calculation is performed on the real-time laser output data of the carbon dioxide laser therapeutic apparatus to obtain feedback control parameters, including: Segmenting the real-time laser output data of the carbon dioxide laser therapeutic apparatus using a sliding window to obtain a plurality of local time series, and performing fast Fourier transform on the plurality of local time series to obtain local spectrum features; According to the local frequency spectrum characteristics, each local part is adaptively filtered to obtain denoised local output data, and nonlinear system identification is performed on the denoised local output data to obtain a local dynamic model; According to the local dynamic model, multiple local PID controllers are designed to obtain a local PID parameter set, and fuzzy rule mapping is performed on the local PID parameter set to obtain a global PID parameter adjustment strategy; According to the global PID parameter adjustment strategy, a nonlinear PID control model is constructed, and stability analysis is performed on the nonlinear PID control model to obtain a Lyapunov stability condition; According to the Lyapunov stability condition, an online parameter adjustment mechanism is generated, and the online parameter adjustment mechanism is numerically discretized to obtain feedback control parameters including a gain matrix, an integral time, and a differential time.

7. The output power control method of the carbon dioxide laser therapeutic apparatus according to claim 1, characterized in that: The multi-layer neural network analysis and laser energy distribution are performed on the feedforward control parameters and the feedback control parameters to obtain a multi-stage laser energy distribution scheme, including: The feedforward control parameters and the feedback control parameters are concatenated to obtain a 30-dimensional input vector X including a gain matrix, a compensation coefficient, and a PID parameter; The input vector X is input into the first hidden layer and calculated by matrix multiplication W1X+b1, where W1 is 64 30, b1 is a 64-dimensional bias vector, and a 64-dimensional intermediate result is obtained. Then, it is processed by the ReLU activation function max(0, x) to obtain a 64-dimensional first-layer feature representation H1. The first layer feature representation H1 is input into the second hidden layer and calculated by matrix multiplication W2H1+b2, where W2 is 128 64 weight matrix, b2 is a 128-dimensional bias vector, and a 128-dimensional intermediate result is obtained. Then, it is processed by the ReLU activation function to obtain a 128-dimensional second-layer feature representation H2; The second layer feature representation H2 is input into the third hidden layer and calculated by matrix multiplication W3H2+b3, where W3 is 256 128 weight matrix, b3 is a 256-dimensional bias vector, and a 256-dimensional intermediate result is obtained. Then, it is processed by the ReLU activation function to obtain a 256-dimensional third-layer feature representation H3; Perform dropout processing on the third-layer feature representation H3, randomly set 50% of the elements to 0, and obtain a 256-dimensional denoised feature representation H3'; The noise reduction feature representation H3' is input into the output layer and calculated by matrix multiplication W4H3'+b4, where W4 is 10 The weight matrix of 256, b4 is a 10-dimensional bias vector, and the initial laser energy distribution result is obtained; Applying a Softmax function to the initial laser energy allocation result to obtain a multi-stage energy allocation probability distribution, and generating multiple groups of candidate energy allocation schemes using an inverse transform sampling method based on the multi-stage energy allocation probability distribution, each group of schemes including energy values ​​of a corresponding stage; The multiple groups of candidate energy allocation schemes are evaluated for safety and calculated for energy utilization efficiency, and the scheme with the highest energy utilization efficiency is selected as the multi-stage laser energy allocation scheme.

8. An output power control system for a carbon dioxide laser therapeutic apparatus, characterized in that: A method for controlling the output power of a carbon dioxide laser therapeutic apparatus according to any one of claims 1 to 7, comprising: A modeling module is used to perform finite element modeling on the optical path, laser tube and cooling system of the carbon dioxide laser therapy device to obtain a three-dimensional temperature field model; A first mode analysis module is used to perform continuous wave mode analysis on the three-dimensional temperature field model to obtain continuous wave power fluctuation impact data; A second mode analysis module is used to perform pulse mode analysis on the three-dimensional temperature field model to obtain pulse power fluctuation impact data; A feedforward control module, used for performing comprehensive processing and dual-mode feedforward control analysis on the continuous wave power fluctuation influence data and the pulse power fluctuation influence data to obtain feedforward control parameters; A calculation module, used for performing multi-local control calculation on the real-time laser output data of the carbon dioxide laser therapeutic apparatus to obtain feedback control parameters; The allocation module is used to perform multi-layer neural network analysis and laser energy allocation on the feedforward control parameters and the feedback control parameters to obtain a multi-stage laser energy allocation scheme.

Images