Laser power self-adaptive control method and system for laser-assisted machining

Abstract

The invention discloses a laser power adaptive control method for laser-assisted processing. The method comprises the following steps: S1, thermal modeling and simulation experiment are carried out to obtain data; s2, designing a model prediction control algorithm; s3, controlling software by an actuator; the output power of the laser is controlled through laser control software, and the predicted output power of the model prediction control algorithm is used as the input of the control software, so that the output power of the laser is dynamically updated in real time; and S4, designing a sensor and detecting.

Description

Technical Field

[0001] The present invention belongs to the technical field of laser-assisted processing and control engineering, and in particular relates to a laser power adaptive control method and system for laser-assisted processing. Background Art

[0002] Optical components made of hard and brittle materials are widely used in the fields of lighting, imaging, and biomedical equipment. However, due to the extremely high hardness and brittleness of the material itself, severe diamond tool wear occurs during processing, making it difficult to achieve high-quality, high-precision, and high-efficiency processing of optical components made of hard and brittle materials. The Laser Assisted Diamond Turning (LADT) method can effectively improve the processing quality of hard and brittle materials and the service life of diamond tools. However, it is difficult for laser-assisted diamond turning to ensure the stability of the laser heating thermal field, which can easily lead to heat accumulation and thermal damage in areas with low processing line speeds such as the center of the workpiece, destroying the processing surface quality of optical components made of hard and brittle materials. Summary of the invention

[0003] The purpose of the present invention is to solve the above problems and provide a laser power adaptive control method which can customize the laser output power according to the position of the processing point in laser-assisted processing, thereby ensuring uniform laser heating temperature and modification effect.

[0004] To solve the above technical problems, the technical solution of the present invention is: a laser power adaptive control method for laser-assisted processing, comprising the following steps:

[0005] S1, thermal modeling simulation and experimental testing to obtain data; step S1 also includes the following sub-steps:

[0006] S11. Experimental design: In the preparation stage, a series of thermal modeling simulations and experimental objects and processes need to be designed to determine the key parameters of the thermal model;

[0007] S12. Experimental execution: Use finite element analysis software to simulate different key parameter combinations of optical elements, and derive simulation results of temperature thermal field distribution under different key parameter combinations at different time nodes, that is, time series data of laser output power-temperature distribution under different parameter combinations;

[0008] S13. Data recording and organization: Record the simulation data in the database, including the key parameters of the optical components and typical data after simulation, to ensure that the collected data set has a large number of samples and covers a wide range of variable combinations.

[0009] S2. Design model predictive control algorithm;

[0010] S3, actuator control software;

[0011] The output power of the laser is controlled by the laser control software, and the predicted output power of the model predictive control algorithm is used as the input of the control software to dynamically update the output power of the laser in real time;

[0012] S4. Design and test sensors.

[0013] Furthermore, the key parameters include the material property M, shape and size S, processing thickness T, workpiece rotation speed n, feed speed f, laser output power P, laser spot radius D and laser wavelength λ of the hard and brittle material optical element.

[0014] Furthermore, the step S2 further includes the following sub-steps:

[0015] S21. Data preprocessing: Perform data cleaning, processing missing values, and preprocessing or normalizing numerical features on the simulation data set to ensure that the data is suitable for machine learning model training;

[0016] S22, feature selection: feature selection is performed on the preprocessed simulation data, and features are extracted based on statistical probability features and time domain and frequency domain analysis to form a feature training set for model training;

[0017] S23. Model selection: Recurrent Neural Network (RNN) is a type of neural network that utilizes the ordered nature of input content. Such input can be a piece of text, a speech, a time series, or any other sequence in which an element is dependent on its previous element. The difference from a feedforward neural network is that the hidden layer has a loop connection. The existence of the loop allows the output of a neuron at a certain moment to enter the neuron again as input, that is, the state at the previous moment t-1 can be transmitted to the current moment t, and the network state at the current moment t can also be transmitted to the next moment t+1;

[0018] S24. Model training and optimization: Define the model architecture of the recurrent neural network, divide the feature data set of the simulation data into a training set and a test set in a ratio of 8:2, and use the training data set to train the model; evaluate and optimize the model, and use cross-validation to evaluate the generalization ability of the model to prevent overfitting.

[0019] Furthermore, the RNN forward propagation can be expressed as follows:

[0020]

[0021] Among them, W xh is the weight matrix input to the hidden, W hh is the connection weight matrix between hidden layers, W hyis the hidden to output weight matrix; b h and b y All are offsets; x t is the input quantity at time t, h t is the hidden value at time t, y t is the output value at time t; σ(x)=1 / (1+e -x ) is called Sigmoid function, which is a nonlinear activation function commonly used in machine learning. According to the recursive formula, h t The calculation of needs to rely on h t-1 ,h t-2 ,..., until the initialization value h0.

[0022] Furthermore, the use of cross-validation to evaluate the generalization ability of the model is to evaluate the model performance under different hyperparameter combinations through cross-validation, and select the configuration that minimizes the error of the validation set; adjust the hyperparameters of the model, such as learning rate, regularization parameter, number of hidden layers and number of nodes, through grid search or random search to find the optimal model configuration.

[0023] Furthermore, the step S4 also includes the following sub-steps:

[0024] S41, Sensor integration: Install various sensors at key locations of optical element processing equipment;

[0025] S42, Data collection and real-time transmission: Sensor data is collected in real time by the data collection system and transmitted to the central monitoring system by wireless or wired means for real-time analysis;

[0026] S43, data storage and update: All data collected by the sensor during the entire processing process are stored in the database, and updated to the input data set of the model prediction algorithm module to improve the prediction accuracy of the model.

[0027] Furthermore, the various types of sensors include high-precision thermal imagers and high-precision force sensors. The high-precision thermal imagers are used to detect in real time the average surface temperature of optical elements during processing. Combined with the simulation conclusion of the relationship between the average surface temperature detected by the thermal imager and the maximum center temperature, the maximum surface temperature of the processing center is obtained; the high-precision force sensors are used to detect in real time the cutting force applied by the tool on the element during the processing of the optical element.

[0028] Furthermore, the laser power adaptive control method for laser-assisted processing forms a closed-loop control system integration, comprising the following steps:

[0029] S51. According to the high-temperature physical properties of the selected optical workpiece and tool material, a high-temperature nanoindentation experiment is performed on the nanoindentation platform to set the ideal laser heating temperature Tem opt ;

[0030] S52, inputting the key processing parameters into the model prediction algorithm module to predict the output power P(t0) or P(x0, y0, z0) at the current moment or the corresponding coordinate point of the machine tool; adjusting the output power of the laser acting on the optical element in real time through the actuator control software;

[0031] S53, the high-precision sensor installed in the processing device collects the average surface temperature of the optical component during the processing, and combines the simulation conclusion of the relationship between the average surface temperature and the highest center temperature detected by the thermal imager to infer the highest temperature Tem of the processing surface max (t);

[0032] S54, feed the data back to the front-end input of the model prediction algorithm in real time to ensure the highest temperature Tem collected in real time max (t) and ideal heating temperature Tem opt The deviation is less than the set threshold Δ, that is: |Tem max (t)-Tem opt |≤Δ; At the same time, the high-precision force sensor synchronously records the cutting force change F(t) during the whole processing process, and the actuator control software records the laser power change P(t) or P(x, y, z) during the whole processing process, and performs filtering processing to obtain the cutting force change F(t) and laser power output history change P(t) or P(x, y, z) under the processing parameters, laser parameters and ideal temperature conditions;

[0033] S55, Maximum temperature Tem of optical element surface max The changing data of the cutting force F(t) and the laser power P(t) or P(x, y, z) are filtered and fitted to verify the adaptive adjustment function of the laser output power to achieve the stability of the surface temperature of the optical workpiece.

[0034] The beneficial effects of the present invention are:

[0035] 1. The present invention provides a laser power adaptive control method for laser-assisted processing, which uses a laser power adaptive controller to accurately maintain the laser heating temperature within an ideal range, which is determined by a high-temperature nanoindentation test. The experimental results show that with the assistance of adaptive laser power, a tungsten carbide (WC) free-form surface optical device without thermal defects can be obtained.

[0036] 2. The present invention provides a laser power adaptive control method for laser-assisted processing. This method can be used to customize the laser output power according to the position of the processing point in laser-assisted processing, thereby ensuring uniform laser heating temperature and modification effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a schematic diagram of finite element thermal field simulation of the laser power control method of laser-assisted processing of the present invention;

[0038] Figure 2 It is a schematic diagram of the recurrent neural network structure of the laser power control system for laser-assisted processing of the present invention;

[0039] Figure 3 It is a schematic diagram of a laser output power closed-loop system for laser-assisted processing of the present invention;

[0040] Figure 4 This is the high temperature nanoindentation test curve of tungsten carbide (WC), a typical hard and brittle material of the present invention;

[0041] Figure 5 It is a schematic diagram of output power control and temperature variation curve of the laser processing process of the laser assisted processing of the present invention. DETAILED DESCRIPTION

[0042] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments:

[0043] like Figures 1 to 5 As shown, the present invention provides a laser power adaptive control method for laser-assisted processing, comprising the following steps:

[0044] S1. Obtain data through thermal modeling simulation and experimental testing.

[0045] Step S1 also includes the following sub-steps:

[0046] S11. Experimental design: In the preparation stage, it is necessary to design a series of thermal modeling simulations and experimental objects and processes to determine the key parameters of the thermal model.

[0047] The key parameters include the material properties M, shape and size S, processing thickness T, workpiece speed n, feed speed f, laser output power P, laser spot radius D and laser wavelength λ of hard and brittle optical components, such as Figure 1 shown.

[0048] S12. Experimental execution: Use finite element analysis software to simulate different key parameter combinations of optical elements, and derive simulation results of temperature thermal field distribution under different key parameter combinations at different time nodes, that is, the time series data of laser output power-temperature distribution under different parameter combinations.

[0049] S13. Data recording and organization: The simulation data are recorded in the database, including the key parameters of the optical components and typical data after simulation, such as the surface temperature of the workpiece, to ensure that the number of collected data set samples is large and covers a wide range of variable combinations.

[0050] S2. Design model predictive control algorithm.

[0051] The step S2 also includes the following sub-steps:

[0052] S21. Data preprocessing: Perform data cleaning, processing missing values, and preprocessing or normalizing numerical features on the simulation data set to ensure that the data is suitable for machine learning model training.

[0053] S22. Feature selection: Select features of the preprocessed simulation data, extract features based on statistical probability features and time domain and frequency domain analysis to form a feature training set for model training.

[0054] S23. Model selection: Recurrent Neural Network (RNN) is a type of neural network that utilizes the ordered nature of the input content, such as Figure 2 As shown. Such input can be a piece of text, a speech, a time series, or any other sequence in which the elements are dependent on the previous elements. The difference from the feedforward neural network is that the hidden layer has a loop connection. The existence of the loop allows the output of the neuron at a certain moment to enter the neuron again as input, that is, the state of the previous moment t-1 can be transmitted to the current moment t, and the network state of the current moment t can also be transmitted to the next moment t+1.

[0055] The RNN forward propagation can be expressed as follows:

[0056]

[0057] Among them, W xh is the weight matrix input to the hidden, W hh is the connection weight matrix between hidden layers, W hy is the hidden to output weight matrix; b h and b y All are offsets; x t is the input quantity at time t, h t is the hidden value at time t, y t is the output value at time t; σ(x)=1 / (1+e -x ) is called Sigmoid function, which is a nonlinear activation function commonly used in machine learning. According to the recursive formula, h t The calculation of needs to rely on h t-1 ,h t-2 ,..., until the initialization value h0.

[0058] S24. Model training and optimization: Define the model architecture of the recurrent neural network, such as the size of the hidden layer and the activation function. Divide the feature data set of the simulation data into a training set and a test set in a ratio of 8:2. Use the training data set to train the model. Evaluate and optimize the model, and use cross-validation to evaluate the generalization ability of the model to prevent overfitting.

[0059] Cross-validation is used to evaluate the generalization ability of the model. The model performance under different hyperparameter combinations is evaluated through cross-validation, and the configuration that minimizes the error of the validation set is selected; the hyperparameters of the model, such as learning rate, regularization parameter, number of hidden layers, and number of nodes, are adjusted through grid search or random search to find the optimal model configuration.

[0060] S3, actuator control software.

[0061] The output power of the laser is controlled by the laser control software, and the predicted output power of the model predictive control algorithm is used as the input of the control software to dynamically update the output power of the laser in real time.

[0062] S4. Design and test sensors.

[0063] Wherein step S4 also comprises the following sub-steps:

[0064] S41. Sensor integration: Install various sensors at key locations of optical element processing equipment.

[0065] The various sensors include high-precision thermal imagers and high-precision force sensors. The high-precision thermal imagers are used to detect the average surface temperature of optical components during processing in real time. Combined with the simulation conclusion of the relationship between the average surface temperature detected by the thermal imager and the maximum center temperature, the maximum surface temperature of the machining center is obtained; the high-precision force sensors are used to detect the cutting force applied by the tool on the component during the processing of optical components in real time.

[0066] S42. Data collection and real-time transmission: Sensor data is collected in real time by the data collection system and transmitted to the central monitoring system by wireless or wired means for real-time analysis.

[0067] S43, data storage and update: All data collected by the sensor during the entire processing process are stored in the database, and updated to the input data set of the model prediction algorithm module to improve the prediction accuracy of the model.

[0068] The present invention also discloses a laser power adaptive control method for laser-assisted processing to form a closed-loop control system integration, comprising the following steps:

[0069] S51. According to the high-temperature physical properties of the selected optical workpiece and tool material, a high-temperature nanoindentation experiment is performed on the nanoindentation platform to set the ideal laser heating temperature Tem opt .

[0070] S52. Input key processing parameters into the model prediction algorithm module to predict the output power P(t0) or P(x0, y0, z0) at the current moment or the corresponding coordinate point of the machine tool; adjust the output power of the laser acting on the optical element in real time through the actuator control software.

[0071] S53, the high-precision sensor installed in the processing device collects the average surface temperature of the optical component during the processing, and combines the simulation conclusion of the relationship between the average surface temperature and the highest center temperature detected by the thermal imager to infer the highest temperature Tem of the processing surface max (t).

[0072] S54, feed the data back to the front-end input of the model prediction algorithm in real time to ensure the highest temperature Tem collected in real time max (t) and ideal heating temperature Tem opt The deviation is less than the set threshold Δ, that is: |Tem max (t)-Tem opt |≤Δ; at the same time, the high-precision force sensor synchronously records the cutting force change F(t) of the whole processing process, and the actuator control software records the laser power change P(t) or P(x, y, z) of the whole processing process, and performs filtering processing to obtain the cutting force change F(t) and laser power output history change P(t) or P(x, y, z) under the processing parameters, laser parameters and ideal temperature conditions.

[0073] S55, Maximum temperature Tem of optical element surface max The changing data of the cutting force F(t) and the laser power P(t) or P(x, y, z) are filtered and fitted to verify the adaptive adjustment function of the laser output power to achieve the stability of the surface temperature of the optical workpiece.

[0074] In the end face turning processing scenario with a constant workpiece speed, as the diamond tool moves from the outside to the inside, the laser scanning linear speed gradually decreases. If the laser power is constant, the stability of the laser heating temperature must be very poor. This will introduce a serious risk of graphitization wear of the diamond tool and thermal damage to the center area of ​​the workpiece. To address this problem, the present invention proposes a laser power adaptive control method to dynamically adjust the laser power according to the tool position during laser-assisted turning.

[0075] This example takes tungsten carbide (WC) as an example. In order to accurately determine the optimal laser heating temperature of tungsten carbide (WC), a high-temperature nanoindentation experiment was performed on the nanoindentation platform. Before the indentation test, the indenter and the workpiece were heated to a higher temperature (373, 573, 773 K) at a heating rate of 1.5 K / min. Under the same test parameters, the nanoindentation test results have ideal consistency. Typical load-displacement curves are shown in Figure 2. Figure 4 As shown. With the increase of temperature, the role of thermal softening effect becomes more and more significant, resulting in a significant decrease in hardness. Among them, at temperatures below 573K, the hardness of tungsten carbide (WC) changes slightly, however, at 773K, the hardness drops sharply by 52.81% to 5.29GPa. Considering the high critical temperature of diamond graphitization, the optimal laser heating temperature of tungsten carbide (WC) is Temopt = 773K.

[0076] In the preparation stage of this embodiment, a series of thermal modeling and simulation experimental objects and processes were designed to determine the key parameters of the thermal model, including the material properties M (single crystal silicon, tungsten carbide) of hard and brittle optical elements, shape and size S (10mm, 15mm, 20mm), processing thickness T (2mm, 4mm, 8mm), workpiece speed n (60 rpm, 120 rpm, 180 rpm), feed speed f (0.06mm / min, 0.12mm / min, 0.18mm / min), laser output power P (5W, 10W, 20W) and laser spot radius D (170μm, 340μm). The data results after finite element simulation are used as the data set of the neural network model of this embodiment. The finite element model simultaneously considers the forced convection of the cutting fluid, natural air convection and thermal radiation boundary conditions. Through data preprocessing and feature selection of the data set, a series of simulation results of temperature thermal field distribution under different key parameter combinations at different time nodes are obtained.

[0077] In this embodiment, a P = 20W Gaussian laser spot with a diameter of D = 170μm is used to heat a tungsten carbide (M = WC) workpiece with a diameter of S = 10mm. The workpiece rotates at a speed of n = 120 rpm. The laser spot moves radially at a feed speed of f = 0.12mm / min, forming a relatively spiral laser scanning trajectory. According to the current maximum temperature Tem of the optical element processing surface max (t) and ideal heating temperature Tem opt The deviation from the set threshold Δ is used to predict the laser output power P(t) or P(x, y, z) at the next moment or at the corresponding coordinate point of the machine tool through the recurrent neural network algorithm, thereby controlling the laser to dynamically adjust the output power value in real time, affecting the surface temperature of the optical element being processed. By adjusting the adaptive laser power to change within the range of (0W, 20W), the maximum temperature Tem of the machining center of the entire workpiece is ensured.max (t) Approximates the ideal laser heating temperature Tem opt =773K.

[0078] The maximum temperature of the workpiece during laser heating is controlled at around 773K. Figure 5 As shown by the top slash in the figure. The corresponding laser input power decreases with the increase of the feed distance in the x direction. At the beginning, the recommended laser power is large, which is because most of the laser beam is located on the outside of the workpiece. Due to the reduction of the laser scanning line speed, the laser power decreases slowly when the laser heating point approaches the center of the workpiece. In addition, due to the limitation of the finite element calculation accuracy, there are high-frequency vibrations of the heating temperature and laser power in the simulation results. In order to improve the stability of laser control in the laser-assisted turning experiment, the laser power changes obtained by FEA simulation calculation are first filtered. In actual processing, the smooth laser power change curve obtained by filtering is used to control the output of the laser device, such as Figure 5 Shown in the bottom middle curve.

[0079] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A laser power adaptive control method for laser-assisted processing, characterized in that: The following steps are involved: S1, thermal modeling simulation and experimental testing to obtain data; step S1 also includes the following sub-steps: S11. Experimental design: In the preparation stage, a series of thermal modeling simulations and experimental objects and processes need to be designed to determine the key parameters of the thermal model; S12. Experimental execution: Use finite element analysis software to simulate different key parameter combinations of optical elements, and derive simulation results of temperature thermal field distribution under different key parameter combinations at different time nodes, that is, time series data of laser output power-temperature distribution under different parameter combinations; S13, data recording and organization: record the simulation data in the database, including the key parameters of the optical components and typical data after simulation, to ensure that the number of collected data set samples is large and covers a wide range of variable combinations; S2. Design model predictive control algorithm; S3, actuator control software; The output power of the laser is controlled by the laser control software, and the predicted output power of the model predictive control algorithm is used as the input of the control software to dynamically update the output power of the laser in real time; S4. Design and test sensors.

2. The laser power adaptive control method for laser-assisted processing according to claim 1, characterized in that: The key parameters include the material property M, shape and size S, processing thickness T, workpiece rotation speed n, feed speed f, laser output power P, laser spot radius D and laser wavelength λ of the hard and brittle material optical element.

3. The laser power adaptive control method for laser-assisted processing according to claim 1, characterized in that: The step S2 also includes the following sub-steps: S21. Data preprocessing: Perform data cleaning, processing missing values, and preprocessing or normalizing numerical features on the simulation data set to ensure that the data is suitable for machine learning model training; S22, feature selection: feature selection is performed on the preprocessed simulation data, and features are extracted based on statistical probability features and time domain and frequency domain analysis to form a feature training set for model training; S23. Model selection: Recurrent Neural Network (RNN) is a type of neural network that utilizes the ordered nature of input content. Such input can be a piece of text, a speech, a time series, or any other sequence in which an element is dependent on its previous element. The difference from a feedforward neural network is that the hidden layer has a loop connection. The existence of the loop allows the output of a neuron at a certain moment to enter the neuron again as input, that is, the state at the previous moment t-1 can be transmitted to the current moment t, and the network state at the current moment t can also be transmitted to the next moment t+1; S24. Model training and optimization: Define the model architecture of the recurrent neural network, divide the feature data set of the simulation data into a training set and a test set in a ratio of 8:2, and use the training data set to train the model; evaluate and optimize the model, and use cross-validation to evaluate the generalization ability of the model to prevent overfitting.

4. The laser power adaptive control method for laser-assisted processing according to claim 3, characterized in that: The RNN forward propagation can be expressed as follows: Among them, W xh is the weight matrix input to the hidden, W hh is the connection weight matrix between hidden layers, W hy is the hidden to output weight matrix; b h and b y All are offsets; x t is the input quantity at time t, h t is the hidden value at time t, y t is the output value at time t; σ(x)=1 / (1+e -x ) is called Sigmoid function, which is a nonlinear activation function commonly used in machine learning. According to the recursive formula, h t The calculation of needs to rely on h t-1 ,h t-2 ,..., until the initialization value h0.

5. The laser power adaptive control method for laser-assisted processing according to claim 3, characterized in that: The use of cross-validation to evaluate the generalization ability of the model is to evaluate the model performance under different hyperparameter combinations through cross-validation, and select the configuration that minimizes the error of the validation set; adjust the hyperparameters of the model, such as learning rate, regularization parameter, number of hidden layers, and number of nodes, through grid search or random search to find the optimal model configuration.

6. The laser power adaptive control method for laser-assisted processing according to claim 1, characterized in that: The step S4 also includes the following sub-steps: S41, Sensor integration: Install various sensors at key locations of optical element processing equipment; S42, Data collection and real-time transmission: Sensor data is collected in real time by the data collection system and transmitted to the central monitoring system by wireless or wired means for real-time analysis; S43, data storage and update: All data collected by the sensor during the entire processing process are stored in the database, and updated to the input data set of the model prediction algorithm module to improve the prediction accuracy of the model.

7. The laser power adaptive control method for laser-assisted processing according to claim 6, characterized in that: The various sensors include high-precision thermal imagers and high-precision force sensors. The high-precision thermal imagers are used to detect the average surface temperature of optical components during processing in real time. Combined with the simulation conclusion of the relationship between the average surface temperature detected by the thermal imager and the highest center temperature, the highest surface temperature of the processing center is obtained. High-precision force sensors are used to detect in real time the cutting force applied by the tool on the component during the optical component processing process.

8. The laser power adaptive control method for laser-assisted processing according to claim 1, characterized in that: The laser power adaptive control method for laser-assisted processing forms a closed-loop control system integration, comprising the following steps: S51. According to the high-temperature physical properties of the selected optical workpiece and tool material, a high-temperature nanoindentation experiment is performed on the nanoindentation platform to set the ideal laser heating temperature Tem opt ; S52, inputting the key processing parameters into the model prediction algorithm module to predict the output power P(t0) or P(x0, y0, z0) at the current moment or the corresponding coordinate point of the machine tool; adjusting the output power of the laser acting on the optical element in real time through the actuator control software; S53, the high-precision sensor installed in the processing device collects the average surface temperature of the optical component during the processing, and combines the simulation conclusion of the relationship between the average surface temperature and the highest center temperature detected by the thermal imager to infer the highest temperature Tem of the processing surface max (t); S54, feed the data back to the front-end input of the model prediction algorithm in real time to ensure the highest temperature Tem collected in real time max (t) and ideal heating temperature Tem opt The deviation is less than the set threshold Δ, that is: |Tem max (t)-Tem opt |≤Δ; At the same time, the high-precision force sensor synchronously records the cutting force change F(t) during the whole processing process, and the actuator control software records the laser power change P(t) or P(x, y, z) during the whole processing process, and performs filtering processing to obtain the cutting force change F(t) and laser power output history change P(t) or P(x, y, z) under the processing parameters, laser parameters and ideal temperature conditions; S55, Maximum temperature Tem of optical element surface max The changing data of the cutting force F(t) and the laser power P(t) or P(x, y, z) are filtered and fitted to verify the adaptive adjustment function of the laser output power to achieve the stability of the surface temperature of the optical workpiece.

Images