A black phosphorus-based mid-infrared laser data processing system

Abstract

The present invention provides a black phosphorus-based mid-infrared band laser data processing system, which relates to the field of infrared band laser technology. The system includes: a laser emission module, which is used to generate and output laser pulses in the mid-infrared band; an optical path control module, which is used to control the propagation path, focusing state and polarization characteristics of the laser pulse so that the laser beam irradiates the target area; a signal detection and acquisition module, which is used to capture laser pulse data and use a camera to capture image data of the target area, and record the image data in real time through a data acquisition card. The present invention realizes the control and intelligent analysis of mid-infrared laser pulses by integrating emission, control, detection, processing, feedback and interaction modules, improves system performance and ease of operation, and ensures efficient data processing and feedback control capabilities.

Description

Technical Field

[0001] The invention relates to the technical field of infrared band lasers, and in particular to a black phosphorus-based mid-infrared band laser data processing system. Background Art

[0002] Although traditional mid-infrared lasers can meet basic laser emission requirements, they have certain limitations in terms of precise control and real-time monitoring of laser pulses. In application scenarios such as laser processing and spectral analysis, this limitation may affect the stability and accuracy of laser pulses, thereby affecting the reliability and accuracy of experimental or processing results.

[0003] In addition, traditional laser systems are often not intelligent enough in data processing. When operating these systems, users usually need to manually adjust laser parameters, and cannot make flexible adjustments based on real-time conditions. This situation may not only reduce work efficiency, but also increase the complexity of operation and the possibility of errors to a certain extent. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a black phosphorus-based mid-infrared band laser data processing system to achieve stable and efficient mid-infrared laser pulse output.

[0005] In order to solve the above technical problems, the technical solution of the present invention is as follows:

[0006] A black phosphorus-based mid-infrared band laser data processing system, comprising:

[0007] A laser emission module, used to generate and output laser pulses in the mid-infrared band;

[0008] An optical path control module is used to control the propagation path, focusing state and polarization characteristics of the laser pulse so that the laser beam irradiates the target area;

[0009] The signal detection and acquisition module is used to capture laser pulse data and image data of the target area using a camera, and to record the image data in real time through a data acquisition card. The laser pulse data includes the intensity, frequency, and phase parameter data of the laser pulse;

[0010] Data processing and analysis module, used for preprocessing, feature extraction, pattern recognition and performance evaluation of laser pulse data and image data, including pulse width, repetition frequency, energy distribution, nonlinear effect analysis, data identification and comparison, to obtain data processing and analysis results;

[0011] A feedback control module is used to automatically adjust the parameters of the laser emission module according to the data processing and analysis results, including pump power, resonant cavity length, temperature, focal length of the camera, and exposure parameters, and send adjustment instructions to the laser emission module and the camera;

[0012] Human-computer interaction interface for users to set and monitor laser status, view and analyze processing results, and export or print reports.

[0013] Further, the propagation path, focusing state and polarization characteristics of the laser pulse are regulated so that the laser beam irradiates the target area, including:

[0014] Measure the parameters of the laser pulse output by the laser emission module, and construct the optical path layout according to the location of the target area, including the propagation path of the laser pulse, the location of the focus point, and the polarization direction;

[0015] Determine the lenses, mirrors, and grating optical elements, and install the optical elements at predetermined positions according to the constructed optical path;

[0016] Use a reflector to change the propagation direction of the laser pulse, and adjust the angle and position of the reflector to make the laser pulse propagate along the optical path;

[0017] By adjusting the position and focal length of the lens, the laser pulse forms a focusing state in the target area, including point focusing and line focusing;

[0018] The polarization direction of the laser pulse is regulated by a polarization controller, and the polarization direction of the laser pulse is controlled by adjusting the parameters of the polarization controller;

[0019] By adjusting and setting the optical elements of lenses, mirrors, gratings and polarization controllers, the laser pulse propagates along the constructed optical path so that the laser beam illuminates the target area.

[0020] Further, capturing laser pulse data and using a camera to capture image data of the target area, and recording the image data in real time through a data acquisition card, includes:

[0021] Use a photoelectric detector to detect the laser pulse output by the laser emission module in real time and measure the intensity, frequency and phase parameters of the laser pulse;

[0022] The photoelectric detector converts the detected laser pulse parameters into electrical signals and transmits them to the data acquisition card;

[0023] The data acquisition card collects the electrical signal transmitted by the photoelectric detector in real time at a set sampling rate;

[0024] Use the camera to aim at the target area, convert the captured image data into video signals, and transmit them to the data acquisition card;

[0025] The data acquisition card acquires the video signal transmitted by the camera in real time at a set frame rate and records the image data in real time.

[0026] Furthermore, the laser pulse data and image data are preprocessed, feature extracted, pattern recognized and performance evaluated, including pulse width, repetition frequency, energy distribution, data identification and comparison, to obtain data processing and analysis results, including:

[0027] The laser pulse data is smoothed by Gaussian filtering, and the filtering parameters are set to perform denoising and contrast enhancement operations on the image data to obtain pre-processed laser pulse data and image data;

[0028] According to the preprocessed laser pulse data, the pulse time difference is calculated to obtain the pulse width, and the number of pulses is counted within a given time window to calculate the repetition frequency;

[0029] Analyze the pulse waveform in the laser pulse data, calculate the total energy of the laser pulse, and detect the edge information in the image based on the preprocessed image data, and extract the texture features and shape features of the image;

[0030] According to the pulse time difference and repetition frequency, the parameters of the decision tree model are set, and the decision tree model is trained using a training data set containing different types of laser pulses and corresponding labels to obtain a classifier for the laser pulse type; according to the extracted image features, the parameters of the convolutional neural network model are set, including the number of convolutional layers and the filter size, and the convolutional neural network model is trained using a training data set containing the target area and the corresponding labels to obtain a data identifier for detecting and identifying targets in the image;

[0031] Use the trained decision tree model to classify and identify the laser pulse data to be identified, output the pulse type label, and calculate the classification accuracy; use the trained convolutional neural network model to detect and identify the target image to be identified, output the target position and type label, and calculate the recognition accuracy and recall rate indicators;

[0032] According to the classification of laser pulse data features and the detection and recognition of image data targets, the processing and analysis results of laser pulse data and image data are sorted and output.

[0033] Furthermore, the calculation formula of the pulse time difference is: ;

[0034] in, Indicates the pulse time difference; represents the average window size; , Respectively represent and The timestamp of each pulse; Represents a fixed offset for timestamps; , Respectively represent the and The timestamp jitter term of each pulse; Parameters representing nonlinear effects; Represents the index of the pulse train.

[0035] Furthermore, the calculation formula of the repetition frequency is: ;

[0036] in, Indicates the repetition frequency; Indicates the number of pulse pairs measured; Indicates The weight of the pulse pair; Indicates and The time difference between the pulses; represents a correction term; Represents an index variable.

[0037] Furthermore, the total energy of the laser pulse is calculated as: ;

[0038] in, represents the total energy of the laser pulse; , represents the time range of the integration; Indicates the peak power of the pulse; represents the constant pi; represents the half-width parameter of the Lorentz distribution; represents the time variable; Indicates the center time of the pulse; represents the proportionality constant; represents the laser parameters; represents the reference laser parameter value; represents the power constant of noise and background light; represents the nonlinear effect correction coefficient; Represents the dispersion correction factor.

[0039] Furthermore, the calculation formula for recognition accuracy is: ;

[0040] in, Indicates recognition accuracy; Represents the total number of samples in the test data set; Indicates samples, the number of objects correctly identified and detected by the model in the entire test dataset; Indicates samples, the number of targets that the model missed in the entire test dataset; Indicates samples, the number of targets that the model misdetects in the entire test dataset.

[0041] Furthermore, the calculation formula of recall is: ;

[0042] in, represents the recall rate; Indicates The weight of the samples; Indicates samples, the number of negative classes correctly predicted by the model; Represents the adjustment coefficient.

[0043] Furthermore, according to the data processing and analysis results, the parameters of the laser emission module are automatically adjusted, including pump power, resonant cavity length, temperature, focal length of the camera, and exposure parameters, and the adjustment instructions are sent to the laser emission module and the camera, including:

[0044] The data processing and analysis module processes and analyzes the laser pulse data and image data, and outputs key parameters, including pulse width, repetition frequency, energy distribution laser characteristic parameters, and image clarity, contrast and image quality parameters;

[0045] The feedback control module receives the parameters output by the data processing and analysis module in real time, and formulates corresponding parameter adjustment schemes according to the received parameters, including increasing the pump power, shortening the resonant cavity length or lowering the temperature. If the image clarity is insufficient, the focus or exposure parameters of the camera are adjusted;

[0046] According to the adjustment scheme, the feedback control module generates corresponding adjustment instructions and sends the adjustment instructions to the laser emission module and the camera.

[0047] The above solution of the present invention includes at least the following beneficial effects:

[0048] The laser emission module can generate and output laser pulses in the mid-infrared band, with extremely short pulse width and high repetition frequency, providing a strong light source foundation for high-precision measurement and processing. The optical path control module can accurately control the propagation path, focusing state and polarization characteristics of the laser pulse to ensure that the laser beam can accurately irradiate the target area, improving the flexibility and applicability of the system. The signal detection and acquisition module can not only capture the intensity, frequency, phase and other key parameter data of the laser pulse, but also capture the image data of the target area through the camera, realizing comprehensive monitoring of the laser pulse and the processing process.

[0049] The data processing and analysis module can preprocess, extract features, recognize patterns and evaluate performance of laser pulse data and image data, providing a wealth of data analysis methods to help users gain a deeper understanding of the characteristics of laser pulses and processing effects. The feedback control module can automatically adjust the parameters of the laser emission module and camera based on the results of data processing and analysis, realizing the intelligence and automation of the system and improving processing efficiency and accuracy. The human-computer interaction interface provides a wealth of functional options and an intuitive operation interface, which allows users to set and monitor the laser status, view and analyze processing results, and export or print reports, improving the ease of use and maintainability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a schematic diagram of a black phosphorus-based mid-infrared band laser data processing system provided by an embodiment of the present invention.

[0051] Figure 2 It is a schematic diagram of a process for regulating the propagation path, focusing state and polarization characteristics of a laser pulse in a black phosphorus-based mid-infrared band laser data processing system provided by an embodiment of the present invention so that the laser beam irradiates a target area. DETAILED DESCRIPTION

[0052] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0053] like Figure 1 and Figure 2 As shown, an embodiment of the present invention provides a black phosphorus-based mid-infrared band laser data processing system, comprising:

[0054] The laser emission module 11 is used to generate and output laser pulses in the mid-infrared band;

[0055] An optical path control module 12, used to control the propagation path, focusing state and polarization characteristics of the laser pulse so that the laser beam irradiates the target area;

[0056] The signal detection and acquisition module 13 is used to capture laser pulse data and image data of the target area using a camera, and record the image data in real time through a data acquisition card. The laser pulse data includes the intensity, frequency, and phase parameter data of the laser pulse.

[0057] The data processing and analysis module 14 is used to preprocess, extract features, recognize patterns and evaluate performance of laser pulse data and image data, including pulse width, repetition frequency, energy distribution, nonlinear effect analysis, and data recognition and comparison to obtain data processing and analysis results;

[0058] The feedback control module 15 is used to automatically adjust the parameters of the laser emission module according to the data processing and analysis results, including pump power, resonant cavity length, temperature, focal length of the camera, and exposure parameters, and send adjustment instructions to the laser emission module and the camera;

[0059] The human-machine interaction interface 16 is used for the user to set and monitor the laser status, view and analyze the processing results, and export or print reports.

[0060] In an embodiment of the present invention, the laser emission module can stably and efficiently generate and output laser pulses in the mid-infrared band, providing the system with a high-precision, high-energy light source to meet various high-precision measurement and processing requirements. The short pulse width and high repetition frequency of the laser pulse enable the system to achieve more accurate time resolution and spatial resolution, and improve the accuracy of processing and measurement. The optical path control module accurately controls the propagation path, focusing state and polarization characteristics of the laser pulse to ensure that the laser beam can accurately irradiate the target area, improving the accuracy of processing and measurement. The flexible optical path control capability enables the system to adapt to different processing and measurement scenarios, improving the applicability and flexibility of the system. The signal detection and acquisition module captures the intensity, frequency, phase and other key parameter data of the laser pulse in real time, providing a reliable basis for the precise control and performance evaluation of the system. The image data of the target area is captured by the camera to realize visual monitoring of the processing process and improve the processing quality and efficiency. The data processing and analysis module performs in-depth processing and analysis of the laser pulse data and image data, extracts key features and patterns, and provides strong support for the performance evaluation and optimization of the system. By analyzing parameters such as pulse width, repetition frequency, and energy distribution, we can accurately understand the characteristics of laser pulses and processing effects, providing guidance for precise control of the system.

[0061] According to the data processing and analysis results, the parameters of the laser emission module and the camera are automatically adjusted to realize the intelligent and automatic control of the system. Timely feedback control can ensure that the system always maintains a working state, improve processing efficiency and accuracy, and reduce human intervention and errors. The human-computer interaction interface provides an intuitive and easy-to-use operation interface, which is convenient for users to set and monitor the laser status, view and analyze the processing results.

[0062] In a preferred embodiment of the present invention, generating and outputting laser pulses in the mid-infrared band includes:

[0063] A pump source (such as solid-state laser, fiber laser, etc.) is selected to provide sufficient energy to excite the electrons in the black phosphorus material. Under the stimulation of the pump light, the electrons in the black phosphorus material jump from the valence band to the conduction band, forming a population inversion, and then generating laser light. By accurately designing the length of the resonant cavity, the curvature of the lens, the coating and other parameters, the laser pulse is ensured to oscillate stably in the resonant cavity and output to the outside. The transmittance of the output mirror is used to couple part of the laser pulse out of the resonant cavity to form a stable laser output. Optical elements such as reflectors and prisms are used to accurately control the propagation path of the laser pulse to ensure that it accurately irradiates the target area. By adjusting the focal length and position of the focusing lens, the focusing state of the laser pulse is controlled to achieve different spot sizes and depths. Optical elements such as polarizers and wave plates are used to adjust the polarization characteristics of the laser pulse to meet the needs of different applications. Photoelectric detectors, spectrum analyzers and other instruments are used to detect parameters such as the intensity, frequency and phase of the laser pulse in real time. The image data of the target area is captured by a camera, and relevant features are extracted using image processing technology.

[0064] When specifically applied, the above content specifically includes:

[0065] Use laser pulses in the mid-infrared band to perform micro-nano processing on materials, such as engraving microstructures, drilling, cutting, etc. Due to the extremely short pulse width of laser pulses, precise control and low-damage processing of materials can be achieved. For example, engraving tiny circuit structures on semiconductor materials, or using laser drilling technology to prepare microfluidic chips. Laser pulses in the mid-infrared band have broad application prospects in the biomedical field, such as cell surgery, tissue cutting, and neural stimulation. For example, the precise control ability of laser pulses can be used to achieve precise cutting and manipulation of single cells, or laser stimulation of neural tissue can be used to study the mechanism of neural signal transmission. Use laser pulses in the mid-infrared band to perform high-sensitivity detection and analysis of gases, liquids and other substances in the environment. For example, the spectral information generated by the interaction between laser pulses and gas molecules can be used to achieve real-time monitoring and early warning of pollutants in the atmosphere; or laser pulses can be used to perform high-sensitivity detection and analysis of trace substances in liquids.

[0066] In a preferred embodiment of the present invention, regulating the propagation path, focusing state and polarization characteristics of the laser pulse so that the laser beam irradiates the target area includes:

[0067] Measure the parameters of the laser pulse output by the laser emission module, and construct the optical path layout according to the location of the target area, including the propagation path of the laser pulse, the location of the focus point, and the polarization direction;

[0068] Determine the lenses, mirrors, and grating optical elements, and install the optical elements at predetermined positions according to the constructed optical path;

[0069] Use a reflector to change the propagation direction of the laser pulse, and adjust the angle and position of the reflector to make the laser pulse propagate along the optical path;

[0070] By adjusting the position and focal length of the lens, the laser pulse forms a focusing state in the target area, including point focusing and line focusing;

[0071] The polarization direction of the laser pulse is regulated by a polarization controller, and the polarization direction of the laser pulse is controlled by adjusting the parameters of the polarization controller;

[0072] By adjusting and setting the optical elements of lenses, mirrors, gratings and polarization controllers, the laser pulse propagates along the constructed optical path so that the laser beam illuminates the target area.

[0073] In an embodiment of the present invention, the parameters of the laser pulse output by the laser emission module are measured, including pulse width, repetition frequency, energy, etc., to ensure the stability of the laser pulse. According to the location of the target area and the processing requirements, the optical path layout is constructed. This includes determining the propagation path of the laser pulse, the location of the focal point, and the polarization direction. Through analysis, the rationality and feasibility of the optical path layout are ensured. According to the constructed optical path layout, the required optical elements, such as lenses, reflectors, gratings, etc., are determined. These optical elements are installed in the optical path according to the predetermined position to ensure their stability and accuracy. This includes adjusting the angle, position and height of the optical element to ensure that the laser pulse can accurately pass through and propagate along the predetermined path. The reflector is a key element for regulating the propagation direction of the laser pulse. By adjusting the angle and position of the reflector, the propagation direction of the laser pulse can be changed so that it propagates along the constructed optical path. The selection and installation of the reflector need to consider factors such as its reflectivity, flatness, stability, etc. to ensure that the loss of the laser pulse during the reflection process is minimal and the propagation direction is accurate. The lens is used to focus the laser pulse on the target area. By adjusting the position and focal length of the lens, the focusing state of the laser pulse can be changed, including point focusing and line focusing. Point focusing is suitable for scenarios that require high-precision processing, while line focusing is suitable for scenarios that require processing along a certain direction. The selection and installation of lenses need to consider factors such as their focal length, transmittance, and aberration. The polarization controller is used to control the polarization direction of the laser pulse. By adjusting the parameters of the polarization controller, the polarization state of the laser pulse can be changed to meet the needs of different applications. The choice of polarization direction needs to be optimized according to the characteristics of the processed material and the processing effect. For example, in some materials, laser pulses with a specific polarization direction may have higher processing efficiency or better processing quality. Finally, by comprehensively adjusting and setting optical elements such as lenses, reflectors, gratings, and polarization controllers, ensure that the laser pulse propagates along the constructed optical path and accurately irradiates the target area.

[0074] By accurately measuring the parameters of the laser pulse, the stability and consistency of the laser source can be ensured. By constructing a reasonable optical path layout, the propagation path, focal point and polarization direction of the laser pulse can be optimized, thereby improving the accuracy and efficiency of laser processing. Selecting suitable optical components (such as lenses, reflectors, gratings, etc.) and installing them at predetermined positions according to the optical path layout can ensure that the laser pulse is subjected to minimal loss and interference during propagation. This helps to improve the quality of the laser beam, as well as the stability and reliability of the entire laser system. The flexible use of reflectors allows the propagation direction of the laser pulse to be easily changed and adjusted. By precisely adjusting the angle and position of the reflector, it can be ensured that the laser pulse propagates along the predetermined optical path, thereby accurately irradiating the target area. This increases the flexibility and applicability of laser processing.

[0075] The focal length and position adjustment of the lens can achieve different focusing states of the laser pulse in the target area, such as point focusing and line focusing. Point focusing is suitable for processing that requires high precision and a small range, while line focusing is suitable for scenes that require continuous processing in a certain direction. This flexible focusing method improves the diversity and accuracy of laser processing. Through the polarization controller, the polarization direction of the laser pulse can be accurately adjusted to meet the needs of processing different materials. By comprehensively adjusting and setting optical components such as lenses, reflectors, gratings and polarization controllers, the entire laser system can be ensured to be in the best working state. This can not only improve the irradiation accuracy and stability of the laser beam, but also extend the service life of the laser system and reduce maintenance costs. At the same time, this comprehensive adjustment also enables the laser system to better adapt to different processing environments and needs.

[0076] When specifically applied, the above content specifically includes:

[0077] The parameters of the laser pulse output by the laser emission module are measured. This includes pulse width (needed to be in nanoseconds), repetition frequency, and energy of each pulse (to ensure sufficient processing depth while avoiding overheating of the material). According to the location of the target area to be etched on the metal surface and the processing requirements (such as the accuracy and size of the etched pattern), the optical path layout is constructed. This includes determining the propagation path of the laser pulse from the laser source to the target area, selecting the appropriate focus point position (a few millimeters to a few centimeters above the metal surface), and optimizing the polarization direction according to the processing effect (for example, for some metal materials, linear polarized light is more effective than circularly polarized light).

[0078] According to the optical path layout, determine the required optical components, such as high-reflectivity mirrors, lenses with suitable focal lengths (a focusing system consisting of multiple lenses is required to achieve the desired focusing effect), and polarization controllers (such as liquid crystal polarization controllers). These optical components are precisely installed in the optical path according to the predetermined positions to ensure their stability and accuracy. By fine-tuning the angle and position of the reflector, it is ensured that the laser pulse can be accurately reflected and propagated along the predetermined path. The position and focal length of the lens also need to be precisely adjusted to achieve the desired focusing state. The selection of the reflector takes into account its high reflectivity, flatness and stability to reduce the loss of the laser pulse during the reflection process and ensure the accuracy of the propagation direction. By precisely adjusting the angle of the reflector, the laser pulse can be guided to propagate along the constructed optical path and eventually reach the target area.

[0079] The lens is used to focus the laser pulse to the target area on the metal surface. By adjusting the position and focal length of the lens, point focusing or line focusing can be achieved. The polarization controller is used to control the polarization direction of the laser pulse. By adjusting the parameters of the polarization controller, the polarization state of the laser pulse can be changed. The polarization direction is selected according to the characteristics of the metal material and the requirements of the processing effect. For example, if it is found that the laser pulse with a specific polarization direction has higher processing efficiency or better processing quality when etching the metal surface, then the polarization controller is set to that direction. By comprehensively adjusting and setting optical elements such as lenses, reflectors, gratings and polarization controllers, it is ensured that the laser pulse propagates according to the constructed optical path and accurately irradiates the target area on the metal surface. The etching effect is monitored in real time and the optical elements are fine-tuned to achieve the processing effect.

[0080] In a preferred embodiment of the present invention, capturing laser pulse data and using a camera to capture image data of a target area, and recording the image data in real time through a data acquisition card, includes:

[0081] Use a photoelectric detector to detect the laser pulse output by the laser emission module in real time and measure the intensity, frequency and phase parameters of the laser pulse;

[0082] The photoelectric detector converts the detected laser pulse parameters into electrical signals and transmits them to the data acquisition card;

[0083] The data acquisition card collects the electrical signal transmitted by the photoelectric detector in real time at a set sampling rate;

[0084] Use the camera to aim at the target area, convert the captured image data into video signals, and transmit them to the data acquisition card;

[0085] The data acquisition card acquires the video signal transmitted by the camera in real time at a set frame rate and records the image data in real time.

[0086] In an embodiment of the present invention, a photodetector is used to detect the laser pulse output by the laser emission module in real time. The photodetector can respond to the change in the light intensity of the laser pulse and convert it into a corresponding electrical signal. The laser pulse parameters (intensity, frequency, phase) detected by the photodetector are converted into electrical signals in real time, and these electrical signals carry the complete information of the laser pulse. The converted electrical signal is transmitted to the data acquisition card, and the data acquisition card collects the electrical signal transmitted by the photodetector in real time at a set sampling rate. The sampling rate determines the accuracy and resolution of the data acquisition. The collected data is stored in real time and can be used for subsequent analysis, processing or display.

[0087] Use the camera to aim at the target area and capture the image data of the target area. The camera can convert the light signal into a video signal for transmission and processing. After the captured image data is converted into a video signal, it is transmitted to the data acquisition card. The data acquisition card collects the video signal transmitted by the camera in real time at a set frame rate. The frame rate determines the continuity and smoothness of the image data. The collected image data is recorded in real time. At the same time, the real-time recording function of the data acquisition card can ensure the integrity and accuracy of the image data.

[0088] Through the cooperation of photoelectric detectors and data acquisition cards, real-time and high-precision detection of laser pulse parameters can be achieved. This helps to detect abnormal changes in laser pulses in a timely manner and ensure the stability and reliability of the laser system. Using cameras and data acquisition cards for image capture and recording can ensure the clarity and consistency of image data. The collected laser pulse parameter data and image data can be stored and recorded in real time. By integrating photoelectric detectors, cameras and data acquisition cards, a highly automated laser processing detection system can be built. This helps to reduce manual intervention and improve production efficiency and processing accuracy.

[0089] When specifically applied, the above content specifically includes:

[0090] In a laser cutting machine, the laser pulse output by the laser emission module is the key to the cutting process. In order to ensure the accuracy and efficiency of cutting, the intensity, frequency and phase of the laser pulse need to be detected in real time.

[0091] The laser pulses are detected in real time using photodetectors. The photodetectors can quickly respond to changes in the light intensity of the laser pulses and convert them into corresponding electrical signals. These electrical signals carry complete information about the laser pulses, including intensity, frequency, and phase. The converted electrical signals are transmitted to the data acquisition card. The data acquisition card collects the electrical signals in real time at a set sampling rate (for example, millions of data points per second). The sampling rate determines the accuracy and resolution of the data acquisition, which affects the accuracy of subsequent analysis and processing. The collected data is stored in real time. For example, the stability of the laser pulses can be monitored by analyzing the data to detect and solve potential problems in a timely manner.

[0092] During the laser cutting process, it is necessary to monitor the cutting situation of the target area in real time to ensure the accuracy and effect of the cutting. Use a camera to aim at the target area to capture image data during the cutting process. The camera can convert light signals into video signals for transmission and processing. After the captured image data is converted into a video signal, it is transmitted to the data acquisition card. The data acquisition card collects the video signal in real time at a set frame rate (for example, 30 frames per second or higher). The frame rate determines the continuity and smoothness of the image data, which affects the real-time and accuracy of the monitoring. The collected image data is recorded in real time, and the integrity and accuracy of the image data are ensured through the real-time recording function of the data acquisition card.

[0093] In a preferred embodiment of the present invention, laser pulse data and image data are preprocessed, feature extracted, pattern recognized and performance evaluated, including pulse width, repetition frequency, energy distribution, and data recognition and comparison, to obtain data processing and analysis results, including:

[0094] The laser pulse data is smoothed by Gaussian filtering, and the filtering parameters are set to perform denoising and contrast enhancement operations on the image data to obtain pre-processed laser pulse data and image data;

[0095] According to the preprocessed laser pulse data, the pulse time difference is calculated to obtain the pulse width, and the number of pulses is counted within a given time window to calculate the repetition frequency;

[0096] Analyze the pulse waveform in the laser pulse data, calculate the total energy of the laser pulse, and detect the edge information in the image based on the preprocessed image data, and extract the texture features and shape features of the image;

[0097] According to the pulse time difference and repetition frequency, the parameters of the decision tree model are set, and the decision tree model is trained using a training data set containing different types of laser pulses and corresponding labels to obtain a classifier for the laser pulse type; according to the extracted image features, the parameters of the convolutional neural network model are set, including the number of convolutional layers and the filter size, and the convolutional neural network model is trained using a training data set containing the target area and the corresponding labels to obtain a data identifier for detecting and identifying targets in the image;

[0098] Use the trained decision tree model to classify and identify the laser pulse data to be identified, output the pulse type label, and calculate the classification accuracy; use the trained convolutional neural network model to detect and identify the target image to be identified, output the target position and type label, and calculate the recognition accuracy and recall rate indicators;

[0099] According to the classification of laser pulse data features and the detection and recognition of image data targets, the processing and analysis results of laser pulse data and image data are sorted and output.

[0100] In the embodiment of the present invention, Gaussian filtering is used to smooth the raw laser pulse data to remove noise and burrs. The filter parameters, such as filter width (determines the degree of smoothing) and standard deviation (affects the shape of the filter), are set, and the laser pulse data after smoothing is output. The raw image data is recorded as ,in and Represent the horizontal and vertical coordinates of the image respectively. The median filter algorithm is used to denoise the original image. Median filtering is a nonlinear filtering technology that uses the median value of a pixel to replace the value of the point to remove noise. The specific steps are:

[0101] For each pixel in the image , take all the pixels in its neighborhood (the size of the neighborhood can be set as needed, such as 3×3, 5×5, etc.). Sort the pixels in these neighborhoods by grayscale value. Take the median value after sorting as the point The new gray value. The formula is:

[0102] set up Pixel Neighborhood of is the denoised image, then ;

[0103] in, Indicates the median operation; Indicated in coordinates The gray value of the pixel at ; , A variable representing the coordinates of a pixel in an image. The histogram equalization method is used to enhance the contrast of an image. It enhances the contrast of an image by adjusting the grayscale distribution of the image. The specific steps are:

[0104] Computing images The grayscale histogram of , get the number of pixels at each gray level, specifically, let is the denoised image, where and Represent the horizontal and vertical coordinates of the image respectively, Indicated in coordinates The pixel value at . Indicates grayscale The number of pixels that appear in the denoised image, where The value range of is the range of the image grayscale level (for example, for an 8-bit grayscale image, The value range is 0 to 255), the calculation formula of the grayscale histogram can be expressed as: ;in, and are the width and height of the image respectively; Represents the gray level. For each pixel , check its gray value Is it equal to gray level? If they are equal, Add 1 to the ; if they are not equal, do not add. In this way, after the traversal is completed, Includes grayscale The number of times it appears in the image. Calculate the gray level The cumulative distribution function of is: ;in, Indicates the gray level is The cumulative distribution function of Indicates the gray level is By traversing all possible gray levels , and accumulate from gray level 0 to The number of pixels can be obtained by the cumulative distribution function value of each gray level. Cumulative distribution function Provides useful information about the distribution of grayscale values ​​in an image, such as the brightness and contrast characteristics of the image.

[0105] according to Map the original grayscale to the new grayscale to get the image with enhanced contrast. Grayscale The number of pixels, is the total number of pixels in the image, is the total number of gray levels (256), Grayscale The cumulative distribution function of is the image after contrast enhancement, then . Output the image data after denoising and contrast enhancement, that is .

[0106] By detecting the start and end time of the pulse, the pulse time difference is calculated to obtain the pulse width. Within a given time window, the number of pulses is counted and then the repetition frequency is calculated. The pulse waveform is integrated to calculate the total energy of the laser pulse.

[0107] Use the Canny edge detector to detect edge information in an image. First, load the image and convert it to a grayscale image, because the Canny edge detector operates on grayscale images. Apply a Gaussian filter to smooth the grayscale image to reduce the impact of noise on edge detection. Specifically, select a Gaussian kernel. The size and standard deviation of the Gaussian kernel determine the degree and range of smoothing. After determining the Gaussian kernel, calculate the Gaussian function. The formula for the Gaussian function is; ,in, Represents the coordinates of the pixel point. Represents the standard deviation. By calculation, the Gaussian function value corresponding to each pixel can be obtained . Apply the Gaussian kernel to the grayscale image. The specific method is to align the center of the Gaussian kernel with a pixel position in the image, and then calculate the weighted sum of the kernel and the local area of ​​the image to obtain the smoothed pixel value: ;in, Indicates that after smoothing, the coordinate The pixel value at ; Represents the original image at coordinates The pixel value at ; Represents the standard deviation of the Gaussian filter, which determines the degree of smoothing. The larger the standard deviation, the more obvious the smoothing effect, but it may also lead to the loss of image details; Indicates the current pixel With surrounding pixels The square of the distance between them. This formula calculates the smoothed pixel value by taking a weighted average of each pixel in the original image, thereby achieving the purpose of reducing noise and smoothing the image. This process is repeated for each pixel in the image to obtain a complete smoothed image. When applying a Gaussian filter, adjust parameters such as the kernel size and standard deviation according to actual conditions. By adjusting these parameters, the smoothing effect can be further optimized and the impact of noise on edge detection can be reduced.

[0108] The first-order derivative (gradient) of the image is calculated to obtain the edge strength and direction of each pixel, and the specific implementation process of applying non-maximum suppression and double threshold method to determine the edge pixels is as follows:

[0109] Use the Sobel operator to calculate the image and The Sobel operator is a discrete differential operator that combines Gaussian smoothing and differential operations and can detect image edges well. , using the horizontal direction ( direction) and vertical direction ( direction) to perform convolution operation on the Sobel operator, and obtain the image in and First derivative in direction and The specific calculation formula is: ,in and are the Sobel operators in the horizontal and vertical directions respectively. Gradient amplitude Represents the strength of the edge, and the calculation formula is: ; Represents the direction of the edge, and the calculation formula is: It should be noted that in actual calculations, due to the nature of the inverse tangent function, the result needs to be properly adjusted to ensure that the direction value is within a reasonable range (converted to 0-360 degrees or -180-180 degrees) to refine the edge and ensure that each pixel only responds to the local maximum value in its gradient direction.

[0110] Traverse each pixel in the image. For each pixel, compare the gradient amplitude of the adjacent pixels in the gradient direction according to its gradient direction. If the gradient amplitude of the pixel is not the local maximum (i.e., smaller than the gradient amplitude of the adjacent pixels), set the gradient amplitude of the pixel to 0 to suppress non-maximum values. Set the high threshold and low threshold according to the actual situation of the image and the needs of edge detection. The high threshold is used to determine strong edges, and the low threshold is used to determine weak edges. Traverse each pixel in the image, and divide the pixels into three categories: strong edge, weak edge, and non-edge according to the comparison results of its gradient amplitude with the high and low thresholds.

[0111] Pixels with gradient magnitudes above a high threshold are considered strong edges.

[0112] Pixels with gradient magnitudes lower than the low threshold are suppressed and considered as non-edges.

[0113] Pixels with gradient magnitudes between the high and low thresholds are considered edges or suppressed based on connectivity. Specifically, if these weak edge pixels are connected to strong edge pixels (determined by 8-connectivity or 4-connectivity), they are considered edges; otherwise, they are suppressed. Strong edges and weak edges connected to strong edges are retained to form a complete edge image. Finally, a complete edge image is formed by connecting strong edges and qualified weak edges.

[0114] Gray-level co-occurrence matrix (GLCM) is applied to extract texture features of images. First, appropriate distance and direction of pixel pairs are selected to construct GLCM. The choice of distance and direction depends on the texture features to be analyzed. Calculate GLCM, which is a two-dimensional matrix that represents the frequency of occurrence of pixel pairs at a given distance and direction. Extract texture features such as contrast, energy, entropy, correlation, etc. from GLCM. These features can quantify the texture information of the image. Use Hough transform to extract shape features of the image. First, perform Hough transform on the image after edge detection to convert the points in the image space into curves (straight lines or circles) in Hough space. Statistically calculate the distribution of curves in Hough space to detect shape features in the image. For example, the angle of a straight line and the radius of a circle can be extracted by Hough transform. Based on the results of the Hough transform, information such as the position, direction, and size of the shape in the image can be determined. Use shape descriptors ( Hu Moment) to extract the shape features of the image. First, calculate the geometric moment of the image as: ;in, express Order geometric moment; Represents the image at coordinates The pixel value at ; and represents the order of the moment. Moments are a set of invariant moments based on geometric moments, which can quantify the shape characteristics of images, such as symmetry, compactness, etc. The moments are invariant to translation, rotation, and scale, which means that no matter how the image is translated, rotated, or scaled, The values ​​of the moments remain unchanged.

[0115] According to the characteristics of the laser pulse data (such as pulse time difference, repetition frequency, etc.), construct a feature vector. First, prepare a training data set containing different types of laser pulses and corresponding labels. Use the training data set and feature vector to train the decision tree model. During the training process, the performance of the model is optimized by adjusting the parameters of the decision tree (such as tree depth, splitting criteria, etc.). Use methods such as cross-validation to evaluate the performance of the model. Cross-validation is a method that divides the data set into multiple subsets and then trains and validates each subset separately. This can more comprehensively evaluate the performance of the model and avoid overfitting or underfitting problems caused by improper division of the data set. Randomly divide the data set into k Subsets of equal or similar size (called “folds”). Then, k Training and validation, each time one of the subsets is selected as the validation set, and the rest k -1 subset as the training set. Finally, kThe average of the validation results is used as the performance indicator of the model. When the data set is very small, leave-one-out cross validation can be used. That is, only one sample is selected as the validation set each time, and the rest of the samples are used as the training set.

[0116] The data set is randomly divided into a training set and a test set. The training set accounts for 70%-80% of the data set, and the test set accounts for 20%-30%. Then, the training set is k Divide and get k training subsets and validation subsets. For each training subset, use the selected machine learning algorithm and parameters to train the model. Use the corresponding validation subset to validate the model and calculate evaluation indicators (such as accuracy, recall, etc.). k fold cross validation, calculate k The mean and standard deviation of the validation results are used to evaluate the stability and performance of the model. Based on the results of cross-validation, the model parameters of the evaluation indicators are selected. Grid search can be used to traverse the parameter space and find parameter combinations. For continuous parameters, optimization algorithms such as gradient descent can be used to adjust the parameters.

[0117] Taking the logistic regression model as an example, suppose there is a binary classification data set and you want to find the regularization parameter C， The dataset was randomly divided into a training set (70%) and a test set (30%). The training set was divided into 5 folds to obtain 5 training subsets and a validation subset. For each training subset, the logistic regression algorithm and different C Use the corresponding validation subset to validate the model and calculate the accuracy as the evaluation indicator. Calculate the average accuracy and standard deviation of the 5 validation results. Select the model with the highest average accuracy. C The value is used as the final parameter. Use the grid search method to traverse within a certain range C Value, found C value.

[0118] Designing a convolutional neural network (CNN)-based model architecture for object detection based on image features (such as edges, textures, shapes, etc.) requires comprehensive consideration of multiple factors, including model complexity, computing resources, and the requirements of specific tasks. The specific process includes model architecture design, activation function, loss function selection, and training data set preparation.

[0119] Specifically, the size of the input image, e.g. (For RGB images). Use smaller filters (such as 3x3) to capture detailed features in the image, such as edges and textures. Set a larger number of filters (such as 32) to extract multiple features. Set the step size to 1. Gradually increase the size and number of filters to capture more complex features. You can use 5x5 or 7x7 filters and increase the number of filters to 64, 128, etc. As needed, you can add multiple convolutional layers to deepen the network. Add a pooling layer after each convolutional layer to reduce the size of the image and reduce the amount of calculation. The pooling methods include maximum pooling and average pooling. Use a 2x2 pooling window with a step size of 2. After the convolution and pooling layers, add a fully connected layer to integrate the extracted features. The number of neurons in the fully connected layer can be adjusted according to the specific task. For the target detection task, it is necessary to add multiple fully connected layers and output the coordinates or classification probability of the target area in the last fully connected layer. Use activation functions in the convolutional and fully connected layers to increase the nonlinear ability of the network; the calculation formula of the activation function is: ;in, Represents the activation function The input value of . Receive an input value . Inside the function, the comparison operation is performed, and the input value is Compare with 0. >0 o'clock, The output is This means that positive input values ​​will pass through losslessly. .when When ≤0, The output of is 0. This means that negative input values ​​will be Truncated towards zero. exist = 0, which introduces nonlinear characteristics. >0 area, The gradient (i.e. derivative) of is 1. This means that when the input is positive, the gradient can be transmitted losslessly, which helps the training of the network. ≤0 area, The gradient of is 0. This means that when the input is negative, the gradient is clipped to zero, which helps prevent the vanishing gradient problem.

[0120] For the target detection task, the output is the coordinates of the target area, and the mean square error loss function can be used: ;in, represents mean square error; Indicates the number of samples; Represents the model for The predicted values ​​of the target area coordinates; Indicates In the training phase, a dataset containing target and background images is prepared, and the real region coordinates of each target are annotated. The image is input into the model, and the model predicts the region coordinates of each target in the image based on its internal parameters and learned features. For each target, calculate the predicted coordinates With real coordinates The squared error is obtained by summing the squared errors of all objects in the image. The total error is divided by the number of objects. , thus obtaining the mean square error . This value reflects the average prediction performance of the model on the current image or dataset.

[0121] Collect image data containing the target area and ensure the diversity and representativeness of the images. Annotate the images, including the coordinates of the target area and the corresponding labels. Scale the images to the size required by the input layer and normalize them to accelerate training and improve model performance. Perform data augmentation as needed, such as rotation, translation, flipping, etc., to increase data diversity and robustness. Divide the dataset into training set, validation set, and test set. The training set accounts for 70%, the validation set accounts for 15%, and the test set accounts for 15%. Ensure that the data distribution of the training set, validation set, and test set is consistent to avoid performance degradation caused by data bias. Use a deep learning framework to build a CNN model architecture. Add network layers such as convolutional layers, pooling layers, and fully connected layers according to the design, and set the corresponding parameters. Select activation functions, loss functions, and optimization algorithms, and compile the model. Set monitoring indicators during training, such as accuracy, loss, etc. Use the training set to train the model, and select and adjust the model based on the performance of the validation set. Use the test set to evaluate the trained model and calculate indicators such as accuracy and recall. Deploy the trained model to actual applications for target detection tasks.

[0122] Through data preprocessing and feature extraction, noise and interference in the data can be removed, and effective feature information can be extracted, thereby improving the accuracy of data processing and analysis. By training the decision tree model, the automatic classification and identification of laser pulse types can be realized, reducing the time of manual intervention and judgment, and improving work efficiency. By training the convolutional neural network model, the automatic detection and identification of targets in the image can be realized, and the accuracy and efficiency of image processing can be improved. At the same time, it can handle complex image scenes and variable target types. Through the processing and analysis of laser pulse data and image data, it can provide strong support for laser processing and detection. For example, anomalies and problems in the laser processing process can be discovered in time, and processing parameters and process flows can be optimized; the quality of the processing results can be evaluated and tested to ensure the qualification rate and stability of the product.

[0123] When applied specifically, the above content specifically includes: collecting laser pulse signals through the built-in sensor of the laser cutting machine, including information such as the amplitude, width, and repetition frequency of the pulse. Using an industrial camera to capture images during the laser cutting process to monitor the cutting quality and workpiece status. Using Gaussian filtering to smooth the laser pulse data, filter out high-frequency noise, and set filtering parameters such as the standard deviation to 2 to ensure that the data is smooth and undistorted. De-noising the collected images, using the median filter algorithm to reduce random noise in the image. Enhancing the image contrast, using the histogram equalization method to make the cutting edge clearer. Calculating the pulse time difference to obtain the pulse width, which is used to evaluate the stability of the laser pulse. Counting the number of pulses within a given time window (such as within 1 second), calculating the repetition frequency, to monitor the operating frequency of the laser cutting machine. Analyzing the pulse waveform, calculating the total energy of the laser pulse, which is used to evaluate the cutting ability. Detecting edge information in the image, using the Canny edge detection algorithm to extract the contour of the cutting edge. Extracting the texture features and shape features of the image, such as the grayscale co-occurrence matrix and contour area, to evaluate the cutting quality and workpiece shape.

[0124] According to the characteristics of pulse time difference and repetition frequency, the parameters of the decision tree model are set, such as the depth of the tree is 5 and the splitting standard is Gini impurity. The decision tree model is trained using a training data set containing different types of laser pulses (such as normal pulses, abnormal pulses) and corresponding labels to obtain a classifier for the laser pulse type. According to the extracted image features, the parameters of the convolutional neural network model are set, including the number of convolution layers as 3, the filter size as 3×3, and the learning rate as 0.001. The convolutional neural network model is trained using a training data set containing target areas (such as cutting edges, workpiece shapes) and corresponding labels to obtain a data identifier for detecting and identifying targets in images. The trained decision tree model is used to classify and identify the laser pulse data to be identified, and the pulse type label (such as normal, abnormal) is output. The classification accuracy is calculated, such as 95% accuracy, indicating that the model can correctly identify most laser pulse types. The trained convolutional neural network model is used to detect and identify the image to be identified, and the target position (such as cutting edge position) and type label (such as workpiece type) are output. Calculate the recognition accuracy and recall rate indicators. For example, a recognition accuracy of 90% and a recall rate of 85% indicate that the model can detect and recognize targets in the image more accurately.

[0125] According to the classification of laser pulse data features and the results of image data target detection and recognition, the processing and analysis results of laser pulse data and image data are sorted and output, including pulse type classification report, image target detection and recognition report and related performance indicators (such as accuracy, recall rate, etc.). The processing and analysis results are fed back to the control system of the laser cutting machine for real-time monitoring and adjustment of cutting parameters to improve cutting quality and efficiency.

[0126] In a preferred embodiment of the present invention, the calculation formula of the pulse time difference is: ;

[0127] in, Indicates the pulse time difference; represents the average window size; , Respectively represent and The timestamp of each pulse; Represents a fixed offset for timestamps; , Respectively represent the and The timestamp jitter term of each pulse; Parameters representing nonlinear effects; Represents the index of the pulse train.

[0128] In an embodiment of the present invention, Indicates the pulse time difference, that is, the average time interval between adjacent pulses. This is an indicator for evaluating the stability and periodicity of the pulse signal. is the averaging window size, which determines the number of pulses considered when calculating the pulse time difference. A larger window size can improve the stability and accuracy of the calculation. , Respectively represent and The timestamp of each pulse is the basic data for calculating the pulse time difference. Indicates a fixed offset for timestamps, used to correct for timestamp deviations due to system delays or measurement errors. , Respectively represent the and The timestamp jitter item of each pulse is used to reflect the random fluctuation of the pulse timestamp. It is a parameter of nonlinear effect, which is used to describe the nonlinear time variation that may exist in the pulse sequence. This helps to simulate and calculate the time difference in the actual pulse signal more accurately. is the index of the pulse sequence, which is used to traverse and calculate the pulse time difference. By collecting the timestamp data of the pulse signal and substituting it into the above formula for calculation, the time difference between adjacent pulses can be obtained. By further analyzing and processing these time difference data, the stability, periodicity and possible nonlinear effects of the pulse signal can be evaluated.

[0129] By taking into account the fixed offset and jitter terms of the timestamp, as well as the parameters of the nonlinear effects, the formula is able to more accurately calculate the pulse time difference, thereby improving the accuracy of evaluating the stability and periodicity of the pulse signal.

[0130] By analyzing the changing trends and abnormal conditions of the pulse time difference, potential problems or faults in the pulse signal can be discovered in a timely manner, providing support for system maintenance and repair.

[0131] In a preferred embodiment of the present invention, the calculation formula of the repetition frequency is: ;

[0132] in, Indicates the repetition frequency; Indicates the number of pulse pairs measured; Indicates The weight of the pulse pair; Indicates and The time difference between the pulses; represents a correction term; Represents an index variable.

[0133] In an embodiment of the present invention, Represents the repetition frequency, which is the average number of pulse repetitions per unit time. This is an indicator for evaluating the characteristics of a pulse signal. is the number of pulse pairs measured and is used to determine the number of pulse pairs considered in calculating the repetition frequency. It is The weight of each pulse pair is used to reflect the importance of different pulse pairs when calculating the repetition frequency. It can be adjusted according to actual needs. It is and The time difference between the pulses is one of the basic data for calculating the repetition frequency. By accumulating these time differences, the total time length of the pulse sequence can be obtained. is a correction term used to correct the time difference deviation caused by system delay and measurement error. This is a key parameter that can improve the accuracy of the calculation results. The calculation formula of the correction term is: ;in, is the number of pulse pairs used to calculate the correction term; It is The actual time difference obtained by the measurement; It is the expected calculated The time difference between the two measurements.

[0134] During the pulse signal measurement process, the pulse timestamp data is collected. Through these timestamp data, the actual time difference between adjacent pulses is calculated Based on previous experimental data, the theoretical expected time difference between adjacent pulses was calculated. These theoretical expected values ​​represent the time difference we should measure if there is no systematic error. For each pair of adjacent pulses, calculate the deviation between the actual measured value and the theoretical expected value, that is, This deviation reflects the time difference deviation caused by systematic errors (such as delays, inaccuracies, environmental factors, etc.). Sum all deviation values ​​and divide by the number of pulse pairs measured , to obtain an average deviation, i.e., the correction term By adding the correction term, the time difference deviation caused by the system error can be corrected, thereby improving the accuracy of the calculation results.

[0135] is an index variable used to traverse and calculate the time difference and weight of the pulse pair. By collecting the timestamp data of the pulse signal and calculating the time difference between adjacent pulses, the repetition frequency of the pulse signal can be obtained. The correction term in the formula is It is used to correct the time difference deviation caused by various factors, so as to improve the accuracy of the calculation results.

[0136] By introducing the correction , the formula can calculate the repetition frequency more accurately, thereby reducing the deviation caused by system delay, measurement error and other factors. By accurately calculating the repetition frequency, problems in the pulse signal can be discovered in a timely manner. This helps to optimize and adjust the system in a targeted manner and improve the performance and stability of the system.

[0137] In a preferred embodiment of the present invention, the total energy of the laser pulse is calculated as follows: ;

[0138] in, represents the total energy of the laser pulse; , represents the time range of the integration; Indicates the peak power of the pulse; represents the constant pi; represents the half-width parameter of the Lorentz distribution; represents the time variable; Indicates the center time of the pulse; represents the proportionality constant; represents the laser parameters; represents the reference laser parameter value; represents the power constant of noise and background light; represents the nonlinear effect correction coefficient; Represents the dispersion correction factor.

[0139] In an embodiment of the present invention, describes the basic shape of the laser pulse, which is a Centered on is the half-width parameter of the Lorentz distribution. It is the peak power of the pulse, which determines the maximum energy of the pulse. is the constant pi, used for mathematical normalization, Considering the laser parameters Effect of pulse energy. is a proportionality constant that adjusts the magnitude of the parameter's effect. It is the reference laser parameter value used to compare and calculate the effect of parameter changes on pulse energy. is the power constant of noise and background light, which represents the energy components other than the main pulse energy. It is the nonlinear effect correction factor, which is used to correct the pulse energy change caused by nonlinear effects (such as self-focusing, self-phase modulation, etc.). Nonlinear effects can distort the pulse shape and thus affect the total energy of the pulse. is the dispersion correction factor, which is used to correct the change in pulse energy caused by the dispersion effect. The dispersion effect will cause the pulse to be broadened or compressed during propagation, thereby changing the energy distribution of the pulse. By integral calculation, the total energy of the laser pulse within this time range can be obtained. .

[0140] The formula takes into account multiple key characteristics of laser pulses and various factors that affect pulse energy, so it can more accurately calculate the total energy of laser pulses. By adjusting the peak power, shape, width and other parameters of the pulse, as well as optimizing the working conditions of the laser, the desired pulse energy output can be obtained. It can provide preprocessing and data analysis support for the experiment, reduce blindness and repetitiveness in the experiment, and improve experimental efficiency. By calculating and analyzing pulse energy, the mechanism and law of the interaction between laser and matter can be understood.

[0141] In a preferred embodiment of the present invention, the calculation formula for recognition accuracy is: ;

[0142] in, Indicates recognition accuracy; Represents the total number of samples in the test data set; Indicates samples, the number of objects correctly identified and detected by the model in the entire test dataset; Indicates samples, the number of targets that the model missed in the entire test dataset; Indicates samples, the number of targets that the model misdetects in the entire test dataset.

[0143] In this embodiment of the present invention, there is a test data set, which includes Each sample was identified and tested by the model, and three results were obtained: correct identification , missed inspection and false detection For each sample, we count , and Then, the three types of results of all samples are accumulated to get the total , and Using the above formula, the total The number is divided by the total number of detected targets (i.e. ), and obtain the recognition accuracy .

[0144] The recognition accuracy rate comprehensively considers the correct recognition, missed detection and false detection of the model, and can fully reflect the recognition performance of the model on the test data set. The accuracy rate can be used to easily compare the performance of different models on the same test data set. By comparing the accuracy rates, you can intuitively determine which model is better. , and By analyzing the specific situation of the model, we can find out the shortcomings of the model in the recognition process and then optimize and improve it in a targeted manner. In practical applications, high-accuracy models can more accurately identify and detect targets, thereby improving the reliability and effectiveness of the application. For example, in the fields of autonomous driving, security monitoring, medical diagnosis, etc., high-accuracy recognition models are of great significance for ensuring safety, improving efficiency and improving user experience.

[0145] In a preferred embodiment of the present invention, the calculation formula of the recall rate is: ;

[0146] in, represents the recall rate; Indicates The weight of the samples; Expressing the samples, the number of negative classes correctly predicted by the model; Represents the adjustment coefficient.

[0147] In an embodiment of the present invention, is the weighted true positive number of all samples This ensures that each sample is ranked according to its importance (determined by the weight ) contributes to the recall. is the weighted true positives and weighted false negatives for all samples Sum, which reflects the performance of the model on all positive samples. Indicates the adjustment of global accuracy. is the number of all correct predictions, is the total number of samples. The power is used to adjust the recall so that it not only considers the performance of positive samples, but also the accuracy of the overall classification. is an adjustment coefficient used to control the impact of global accuracy adjustment on recall. = 0, the formula is simplified to only consider the ratio of weighted true positives to weighted false negatives, which is the definition of traditional recall. When >0, the global accuracy has a positive impact on the recall rate; when When <0, the global precision has a negative impact on the recall.

[0148] By introducing weighted summation and global accuracy adjustment, a more comprehensive indicator for evaluating model performance is provided, which not only considers the prediction accuracy of positive samples, but also the accuracy of the overall classification. The introduction of makes the formula flexible, and the weight between global accuracy and traditional recall can be adjusted according to the needs of specific application scenarios. ,This formula can realize personalized evaluation of different samples, and is particularly suitable for the case where the importance of samples is uneven.

[0149] In a preferred embodiment of the present invention, according to the data processing and analysis results, the parameters of the laser emission module are automatically adjusted, including pump power, resonant cavity length, temperature, focal length of the camera, and exposure parameters, and the adjustment instructions are sent to the laser emission module and the camera, including:

[0150] The data processing and analysis module processes and analyzes the laser pulse data and image data, and outputs key parameters, including pulse width, repetition frequency, energy distribution laser characteristic parameters, and image clarity, contrast and image quality parameters;

[0151] The feedback control module receives the parameters output by the data processing and analysis module in real time, and formulates corresponding parameter adjustment schemes according to the received parameters, including increasing the pump power, shortening the resonant cavity length or lowering the temperature. If the image clarity is insufficient, the focus or exposure parameters of the camera are adjusted;

[0152] According to the adjustment scheme, the feedback control module generates corresponding adjustment instructions and sends the adjustment instructions to the laser emission module and the camera.

[0153] In an embodiment of the present invention, the laser emission module emits laser pulses, and the camera captures the target image at the same time. The data processing and analysis module receives the laser pulse data and image data, and performs detailed processing and analysis. Through analysis, the module outputs key parameters, including laser characteristic parameters (such as pulse width, repetition frequency, energy distribution) and image quality parameters (such as image clarity, contrast). The feedback control module receives the parameters output by the data processing and analysis module in real time. According to the received parameters, the feedback control module uses preset algorithms and logic to formulate corresponding parameter adjustment schemes. If the laser characteristic parameters do not meet expectations, possible measures may include increasing the pump power to increase the laser energy, shortening the resonant cavity length to optimize the laser mode, or reducing the temperature to stabilize the laser output. If the image quality parameters are not good, such as insufficient image clarity, the feedback control module will instruct the camera to adjust the focal length or exposure parameters to improve the image quality. According to the adjustment scheme, the feedback control module generates specific adjustment instructions and sends these instructions to the laser emission module and the camera. After receiving the adjustment instructions, the laser emission module and the camera adjust the corresponding parameters according to the instructions. After adjustment, the system performs data acquisition and processing again to form a closed-loop control, and continuously optimizes the effects of laser emission and image capture.

[0154] By adjusting the parameters of the laser emission module in real time, the stability and consistency of the laser output can be ensured, and the accuracy and efficiency of laser processing, measurement and other applications can be improved. Optimizing the focal length and exposure parameters of the camera can improve the image quality. The system can automatically adjust parameters according to environmental changes or application requirements, and has strong adaptive capabilities. This enables the system to work in different scenarios, improving the versatility and practicality of the system. Through the collaborative work of the data processing and analysis module and the feedback control module, the system realizes intelligent control of laser emission and image capture. This not only improves the automation of the system, but also reduces the need for manual intervention, reducing the complexity and cost of operation.

[0155] When specifically applied, the above content specifically includes:

[0156] In a laser cutting machine, the laser emission module is responsible for generating high-energy laser pulses for cutting materials. The characteristics of these laser pulses (such as pulse width, repetition frequency, and energy distribution) affect the cutting effect. At the same time, the camera is used to capture the image of the material to be cut for accurate positioning and cutting path planning. The image quality parameters of the camera (such as image clarity and contrast) are crucial for accurately identifying the edges and features of the material. The data processing and analysis module receives the laser pulse data and the image data captured by the camera for detailed processing and analysis. For example, it analyzes whether the energy distribution of the laser pulse is uniform and whether the material thickness on the cutting path is consistent. At the same time, the image is processed by clarity enhancement, contrast adjustment, etc. to more accurately identify the edge of the material. The feedback control module receives the parameters output by the data processing and analysis module in real time and formulates an adjustment plan based on these parameters. If it is found that the energy distribution of the laser pulse is uneven, the feedback control module may instruct to increase the pump power to increase the laser energy, or shorten the resonant cavity length to optimize the laser mode to ensure the accuracy and efficiency of cutting. If the image clarity is insufficient and the edge of the material cannot be accurately identified, the feedback control module will instruct the camera to adjust the focal length or exposure parameters to improve the image quality and ensure the accuracy of the cutting path. According to the adjustment instructions generated by the feedback control module, the laser emission module and the camera adjust the corresponding parameters. After the adjustment, the system collects and processes data again to form a closed-loop control. This process is repeated until the final cutting effect and image quality are achieved.

[0157] Suppose that when cutting a piece of metal sheet, the initial laser pulse energy distribution is uneven, resulting in burrs on the cutting edge. Through the analysis of the data processing and analysis module, the problem of energy distribution is found. Based on the analysis results, the feedback control module instructs the laser emission module to increase the pump power and optimize the resonant cavity length to improve the energy distribution of the laser pulse. At the same time, the image captured by the camera shows that there is reflection on the surface of the metal sheet, resulting in insufficient image clarity. The feedback control module instructs the camera to adjust the exposure parameters to reduce the impact of reflections and improve image clarity. After adjustment, the laser cutting machine cuts again. At this time, the cutting edge is smooth and burr-free, and the cutting path is correct.

[0158] In a preferred embodiment of the present invention, the user is used to set and monitor the laser status, view and analyze the processing results, and export or print reports, including:

[0159] The system provides a human-computer interaction interface that allows users to enter and set system parameters through a graphical interface or command line interface. The parameters entered by the user will be verified by the system to ensure that they meet the operating specifications and safety requirements of the laser and camera. The verified parameters will be stored in the system configuration file.

[0160] The system collects the operating data of the laser in real time through sensors and interfaces, such as pump power, resonant cavity temperature, laser output energy, etc. The collected data will be processed and displayed on the user interface so that the user can understand the status of the laser in real time. If the status of the laser is abnormal, the system will trigger an alarm mechanism and notify the user through sound, light signal or email. The system processes the collected laser pulse data and image data to extract useful information. The processed results will be displayed on the user interface in the form of charts, curves or reports for users to view and analyze. According to user needs, the system can generate a detailed report containing laser pulse data, image data, processing results and laser status. The report can be exported to multiple formats, such as PDF, Excel, etc., so that users can view and share on different devices. The system also provides a printing function, allowing users to directly print reports for use in meetings, presentations or archiving.

[0161] When specifically applied, the above content specifically includes:

[0162] The system provides an intuitive graphical user interface (GUI) that allows personnel to enter and set the parameters of the laser and camera by clicking, dragging, and other operations. For example, users can set the percentage of pump power, the length of the resonant cavity, the repetition rate of the laser, etc. For more advanced users, the system also provides a command line interface (CLI) that allows them to quickly set parameters by entering specific commands. The parameters entered by the user will be strictly verified by the system to ensure that they meet the operating specifications and safety requirements of the laser and camera. For example, the system will check whether the pump power exceeds the maximum tolerance of the laser, whether the resonant cavity length is within the allowable range, etc. The verified parameters will be stored in the system's configuration file for use in subsequent experiments. Users can view and modify these parameters at any time.

[0163] The system collects the operating data of the laser in real time through sensors and interfaces, such as pump power, resonant cavity temperature, laser output energy, etc. These data are crucial for understanding the performance and status of the laser. The collected data will be processed and displayed on the user interface in the form of charts, curves, etc. The user can clearly see the real-time status of the laser, such as fluctuations in pump power, changes in resonant cavity temperature, etc.

[0164] If the laser status is abnormal, such as too high pump power, too high resonant cavity temperature, or unstable laser output energy, the system will trigger an alarm mechanism. The alarm can be sound, light signal, or email, so that the user can take timely measures. The system processes the collected laser pulse data and image data to extract useful information. For example, the system can calculate characteristic parameters such as laser pulse width, repetition frequency, energy distribution, and quality parameters such as image clarity and contrast.

[0165] The processed results will be displayed on the user interface in the form of charts, curves or reports. Users can clearly see the characteristic curves of laser pulses, the quality distribution of images, etc., so as to conduct in-depth analysis and research. According to the needs of users, the system can generate detailed reports containing laser pulse data, image data, processing results and laser status. The report content is comprehensive and accurate, and it is an important basis for users to summarize experiments, write papers or make presentations. Reports can be exported in multiple formats, such as PDF, Excel, etc. Users can choose the appropriate format for export and sharing according to their needs. For example, users can export reports to PDF format for archiving or printing, or export them to Excel format for further data processing and analysis. The system also provides a printing function that allows users to print reports directly. Users can choose to print all or part of the report for use in meetings, presentations or archiving.

[0166] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A black phosphorus-based mid-infrared band laser data processing system, characterized in that: include: A laser emission module, used to generate and output laser pulses in the mid-infrared band; An optical path control module is used to control the propagation path, focusing state and polarization characteristics of the laser pulse so that the laser beam irradiates the target area; The signal detection and acquisition module is used to capture laser pulse data and image data of the target area using a camera, and to record the image data in real time through a data acquisition card. The laser pulse data includes the intensity, frequency, and phase parameter data of the laser pulse; Data processing and analysis module, used for preprocessing, feature extraction, pattern recognition and performance evaluation of laser pulse data and image data, including pulse width, repetition frequency, energy distribution, nonlinear effect analysis, data identification and comparison, to obtain data processing and analysis results; A feedback control module is used to automatically adjust the parameters of the laser emission module according to the data processing and analysis results, including pump power, resonant cavity length, temperature, focal length of the camera, and exposure parameters, and send adjustment instructions to the laser emission module and the camera; Human-computer interaction interface, used by users to set and monitor laser status, view and analyze processing results, and export or print reports; Control the propagation path, focusing state and polarization characteristics of the laser pulse so that the laser beam illuminates the target area, including: Measure the parameters of the laser pulse output by the laser emission module, and construct the optical path layout according to the location of the target area, including the propagation path of the laser pulse, the location of the focus point, and the polarization direction; Determine the lenses, mirrors, and grating optical elements, and install the optical elements at predetermined positions according to the constructed optical path; Use a reflector to change the propagation direction of the laser pulse, and adjust the angle and position of the reflector to make the laser pulse propagate along the optical path; By adjusting the position and focal length of the lens, the laser pulse forms a focusing state in the target area, including point focusing and line focusing; The polarization direction of the laser pulse is regulated by a polarization controller, and the polarization direction of the laser pulse is controlled by adjusting the parameters of the polarization controller; By adjusting and setting the optical elements of lenses, mirrors, gratings and polarization controllers, the laser pulses propagate along the constructed optical path so that the laser beam illuminates the target area; Capture laser pulse data and use a camera to capture image data of the target area, and record the image data in real time through a data acquisition card, including: Use a photoelectric detector to detect the laser pulse output by the laser emission module in real time and measure the intensity, frequency and phase parameters of the laser pulse; The photoelectric detector converts the detected laser pulse parameters into electrical signals and transmits them to the data acquisition card; The data acquisition card collects the electrical signal transmitted by the photoelectric detector in real time at a set sampling rate; Use the camera to aim at the target area, convert the captured image data into video signals, and transmit them to the data acquisition card; The data acquisition card acquires the video signal transmitted by the camera in real time at a set frame rate and records the image data in real time; Preprocessing, feature extraction, pattern recognition and performance evaluation of laser pulse data and image data, including pulse width, repetition frequency, energy distribution, data identification and comparison, to obtain data processing and analysis results, including: The laser pulse data is smoothed by Gaussian filtering, and the filtering parameters are set to perform denoising and contrast enhancement operations on the image data to obtain pre-processed laser pulse data and image data; According to the preprocessed laser pulse data, the pulse time difference is calculated to obtain the pulse width, and the number of pulses is counted within a given time window to calculate the repetition frequency; Analyze the pulse waveform in the laser pulse data, calculate the total energy of the laser pulse, and detect the edge information in the image based on the preprocessed image data, and extract the texture features and shape features of the image; According to the pulse time difference and repetition frequency, the parameters of the decision tree model are set, and the decision tree model is trained using a training data set containing different types of laser pulses and corresponding labels to obtain a classifier for the laser pulse type; according to the extracted image features, the parameters of the convolutional neural network model are set, including the number of convolutional layers and the filter size, and the convolutional neural network model is trained using a training data set containing the target area and the corresponding labels to obtain a data identifier for detecting and identifying targets in the image; Use the trained decision tree model to classify and identify the laser pulse data to be identified, output the pulse type label, and calculate the classification accuracy; use the trained convolutional neural network model to detect and identify the target image to be identified, output the target position and type label, and calculate the recognition accuracy and recall rate indicators; According to the classification of laser pulse data features and the detection and recognition of image data targets, the processing and analysis results of laser pulse data and image data are sorted and output; The calculation formula of pulse time difference is: ; in, Indicates the pulse time difference; represents the average window size; , Respectively represent and The timestamp of each pulse; Represents a fixed offset for timestamps; , Respectively represent the and The timestamp jitter term of each pulse; Parameters representing nonlinear effects; represents the index of the pulse train; The formula for calculating the repetition frequency is: ; in, Indicates the repetition frequency; Indicates the number of pulse pairs measured; Indicates The weight of the pulse pair; Indicates and The time difference between the pulses; represents a correction term; Represents an index variable; The total energy of the laser pulse is calculated as: ; in, represents the total energy of the laser pulse; , Indicates the time range of the integration; Indicates the peak power of the pulse; represents the constant pi; represents the half-width parameter of the Lorentz distribution; represents the time variable; Indicates the center time of the pulse; represents the proportionality constant; represents the laser parameters; represents the reference laser parameter value; represents the power constant of noise and background light; represents the nonlinear effect correction coefficient; represents the dispersion correction factor; The calculation formula for recognition accuracy is: ; in, Indicates recognition accuracy; Represents the total number of samples in the test data set; Indicates samples, the number of objects correctly identified and detected by the model in the entire test dataset; Indicates samples, the number of targets that the model missed in the entire test dataset; Indicates samples, the number of targets that the model misdetected in the entire test dataset; The calculation formula for recall is: ; in, represents the recall rate; Indicates The weight of the samples; Expressing the samples, the number of negative classes correctly predicted by the model; Represents the adjustment coefficient.

2. The black phosphorus-based mid-infrared band laser data processing system according to claim 1, characterized in that: According to the data processing and analysis results, the parameters of the laser emission module are automatically adjusted, including pump power, resonant cavity length, temperature, camera focal length, exposure parameters, and the adjustment instructions are sent to the laser emission module and camera, including: The data processing and analysis module processes and analyzes the laser pulse data and image data, and outputs key parameters, including pulse width, repetition frequency, energy distribution laser characteristic parameters, and image clarity, contrast and image quality parameters; The feedback control module receives the parameters output by the data processing and analysis module in real time, and formulates corresponding parameter adjustment schemes according to the received parameters, including increasing the pump power, shortening the resonant cavity length or lowering the temperature. If the image clarity is insufficient, the focus or exposure parameters of the camera are adjusted; According to the adjustment scheme, the feedback control module generates corresponding adjustment instructions and sends the adjustment instructions to the laser emission module and the camera.

Images