A coupling device and a coupling method for coupling a laser to an optical fiber
By combining high-precision signal acquisition and intelligent control unit with active adjustment mechanism and machine learning model, the stability and accuracy problems of traditional laser coupling devices under high power are solved, realizing frictionless automatic coupling with nanometer-level precision, and improving laser coupling efficiency and reliability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANDONG XIEHE UNIV
- Filing Date
- 2026-03-27
- Publication Date
- 2026-06-09
AI Technical Summary
In high-power laser coupling processes, existing technologies suffer from problems such as the susceptibility of traditional optical power detectors to damage and high costs, low stability of mechanical adjustment mechanisms, and difficulty in achieving micron-level precise alignment, resulting in insufficient coupling efficiency and system reliability.
By combining a high-precision signal acquisition unit and an intelligent control unit with an active adjustment mechanism, and using a coupling state recognition model trained by machine learning algorithms, a combination of a voice coil motor and a flexible hinge is used to achieve frictionless, nanometer-level precision beam pointing control, indirectly sensing the coupling state and locking the optimal coupling point.
Achieving fully automated coupling in high-power and extreme environments eliminates reliance on traditional detectors, improves coupling efficiency and system reliability, and significantly enhances the stability and accuracy of laser coupling.
Smart Images
Figure CN122172390A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of laser equipment technology, specifically to a coupling device and coupling method for coupling laser optical fibers. Background Technology
[0002] In high-power laser processing and optical communication, efficiently coupling lasers to optical fibers with tiny cores is a crucial process. Current technologies generally rely on placing optical power detectors at the end of the optical path to provide feedback signals, guiding the alignment mechanism. However, when faced with kilowatt-level high-power lasers, special wavelengths, or extreme environments such as vacuum or radiation, traditional detectors suffer from easy damage, high cost, or malfunction, leading to automated coupling failure. Furthermore, traditional mechanical alignment mechanisms suffer from backlash and friction, making it difficult to achieve stable and reliable micron-level precision alignment, thus limiting coupling efficiency and system reliability.
[0003] For example, Chinese patent CN117270113A discloses a coupling device for coupling a laser to an optical fiber. Using this device, it is convenient to adjust the emission angle of multiple current beams, thereby facilitating the adjustment of the coupling position of multiple current beams and the adjustment of the area of the coupled current beams, making it convenient for cutting different materials. However, in actual use, this device uses a lead screw and other transmission methods, resulting in large errors and low stability, making it impossible for users to accurately control and align precisely.
[0004] To this end, we propose a coupling device and coupling method for coupling a laser to an optical fiber. Summary of the Invention
[0005] The purpose of this section is to outline some aspects of embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of these documents; however, such simplifications or omissions should not be construed as limiting the scope of the invention.
[0006] A coupling device for coupling a laser to an optical fiber includes a coupling cavity, a laser mounted on the coupling cavity, a focusing optical system, and an active adjustment mechanism for adjusting the optical path. The focusing optical system is disposed on the reflected optical path of a reflector and is used to converge the light beam reflected by the reflector to the end face of the coupling target. The active adjustment mechanism is driven by a driving component and further includes: A high-precision signal acquisition unit, which is electrically connected to the drive circuit of the drive component, is configured to acquire the drive current waveform of the drive component in real time during operation. An intelligent control unit is electrically connected to the drive circuit of the high-precision signal acquisition unit and the drive component; The intelligent control unit is configured as follows: a. Control the drive component to drive the active adjustment mechanism to move according to a preset scanning trajectory, so that the laser beam scans near the input end of the coupled target; b. During the scanning process, electrical signal waveform data from the high-precision signal acquisition unit are received synchronously; c. Extract features from the electrical signal waveform data and input the extracted features into a pre-trained coupled state recognition model; d. When the output of the coupling state recognition model indicates that the optimal coupling state has been reached, a locking signal is generated to stop the scanning motion of the drive component and maintain its current position.
[0007] Preferably, the coupling state recognition model is obtained by training a training dataset using a machine learning algorithm, and the method for constructing the training dataset includes: A calibration optical power detector is temporarily installed in the coupling cavity to measure the coupling efficiency during the training phase. The active adjustment mechanism is controlled to perform scanning, and the electrical signal waveform sequence acquired by the high-precision signal acquisition unit and its corresponding coupling efficiency value measured by the optical power detector are recorded simultaneously. Align the electrical signal waveform sequence with the coupling efficiency value to form a training sample set.
[0008] Preferably, feature extraction is performed on the electrical signal waveform data. The feature extraction includes: selecting at least a first time-domain feature from a time-domain feature group composed of root mean square (RMS), peak-to-peak value, skewness, and kurtosis; and selecting at least a first frequency-domain feature from a frequency-domain feature group composed of the amplitude of the main frequency component and the centroid of the spectrum, and combining the selected features into a feature vector.
[0009] Preferably, the coupling state recognition model is a convolutional neural network (CNN) model or a support vector machine (SVM) model; When using a CNN model, the input features are preprocessed one-dimensional time-series data of electrical signals or two-dimensional time-frequency graphs. When using the SVM model, the input features are a feature vector composed of the time-domain features and the frequency-domain features.
[0010] Preferably, the active adjustment mechanism is a two-axis fast-reflection mirror system, comprising: Reflecting mirrors are used to reflect laser beams; A frame for fixing the reflective lens; The inner frame is connected to the lens frame via a first flexible hinge, allowing the lens frame and the reflective lens to rotate about a first axis; The base frame is connected to the inner frame via a second flexible hinge, allowing the inner frame, the mirror frame, and the reflective lens to rotate about a second axis orthogonal to the first axis. The base frame is fixed within the coupling cavity.
[0011] Preferably, the driving assembly includes a first pair of voice coil motors and a second pair of voice coil motors. The first pair of voice coil motors acts differentially on opposite sides of the lens frame to drive the reflective lens to deflect around a first axis. The second pair of voice coil motors acts differentially on opposite sides of the inner frame to drive it to deflect around a second axis.
[0012] Preferably, the first-order natural frequencies of the first flexible hinge and the second flexible hinge are designed to be between 10Hz and 100Hz, and the current resolution of the high-precision signal acquisition unit is better than 0.1mA.
[0013] Preferably, the preset scanning trajectory is a spiral scanning trajectory.
[0014] Preferably, the laser is an fiber-coupled semiconductor laser with an output power greater than 500W, and the coupling target is an optical fiber with a core diameter of less than 100μm.
[0015] Coupling schemes are also provided, as follows: Step 1: Scan Startup Steps The intelligent control unit controls the drive components, which in turn drive the active adjustment mechanism to move the laser beam along a preset scanning trajectory near the input end face of the coupled target. Step 2, Signal Synchronization Acquisition Steps: During the scanning process, the high-precision signal acquisition unit synchronously acquires the drive current waveform data of the drive component in real time; Step 3: Feature Extraction Steps The intelligent control unit processes the collected drive current waveform data and extracts feature data containing time-domain and frequency-domain features; Step 4: Intelligent Status Recognition Steps The intelligent control unit inputs the feature data into a pre-trained coupling state recognition model to obtain the evaluation value output by the model, which characterizes the current coupling state; Step 5, Decision Locking Steps: The intelligent control unit continuously monitors the evaluation value. When it determines that the evaluation value has reached and stabilized at the preset optimal coupling threshold, it generates a locking command, controls the drive component to stop scanning and maintain the current position, and completes automatic coupling.
[0016] The beneficial effects of this invention are: This invention indirectly senses the coupling state by detecting minute changes in the voice coil motor current, fundamentally eliminating reliance on traditional terminal optical power detectors and enabling fully automated coupling in high-power and extreme environments. By combining a low-stiffness flexible hinge with a voice coil motor, it achieves frictionless, nanometer-level precision beam pointing control and cleverly converts weak optical radiation pressure and photothermal stress into detectable electrical signals, significantly improving system sensitivity. Combined with intelligent identification of current characteristics using an artificial intelligence model, this method can quickly and accurately pinpoint the optimal coupling point, significantly improving coupling efficiency. Attached Figure Description
[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0018] in: Figure 1 This is a schematic diagram of the overall structure of the present invention; Figure 2 This is a schematic diagram of the coupling cavity Jeb in this invention; Figure 3 for Figure 2 Enlarged schematic diagram of the structure at point A; Figure 4 This is a schematic diagram of the laser structure in this invention; Figure 5 This is a schematic diagram of the active adjustment mechanism in this invention; Figure 6 This is a schematic diagram of the optical power detector structure in this invention; Figure 7 This is a schematic diagram of laser optical path transmission.
[0019] In the picture: 1. Coupled cavity; 2. Laser; 3. Focusing optical system; 4. Active adjustment mechanism; 41. Reflecting lens; 42. Lens frame; 43. First flexible hinge; 44. Inner frame; 45. Second flexible hinge; 46. Base frame; 5. Drive components; 51. First pair of voice coil motors; 52. Second pair of voice coil motors; 6. Intelligent control unit; 7. High-precision signal acquisition unit; 8. Coupled target; 9. Optical power detector. Detailed Implementation
[0020] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.
[0021] See appendix Figure 1-7 As shown, this invention provides a coupling device and method for coupling a laser to an optical fiber. By analyzing the current signal of the motor driving the active adjustment mechanism 4, an artificial intelligence model is used to indirectly sense and lock the optimal coupling state. This method is particularly suitable for high-power, special wavelength, or extreme environments where traditional detectors cannot be used.
[0022] Example 1: A coupling device for coupling a laser to an optical fiber includes a coupling cavity 1, a laser 2 mounted on the coupling cavity 1, a focusing optical system 3, and an active adjustment mechanism 4 for adjusting the optical path. The focusing optical system 3 is disposed on the reflected optical path of a reflecting mirror 41 and is used to focus the light beam reflected by the reflecting mirror 41 onto the end face of the coupling target 8. The active adjustment mechanism 4 is driven by a driving component 5 and further includes: The high-precision signal acquisition unit 7 is electrically connected to the drive circuit of the drive component 5 and is configured to acquire the drive current waveform of the drive component 5 in real time during operation. The intelligent control unit 6 is electrically connected to the high-precision signal acquisition unit 7 and the drive circuit of the drive component 5; The intelligent control unit 6 is configured as follows: a. Control the drive component 5 to drive the active adjustment mechanism 4 to move according to the preset scanning trajectory, so that the laser beam scans near the input end of the coupled target 8; b. During the scanning process, electrical signal waveform data from the high-precision signal acquisition unit 7 are received synchronously; c. Extract features from the electrical signal waveform data and input the extracted features into a pre-trained coupled state recognition model; d. When the output of the coupling state recognition model indicates that the optimal coupling state has been reached, a locking signal is generated to stop the scanning motion of the drive component 5 and maintain its current position.
[0023] Furthermore, the preset scanning trajectory is a spiral scanning trajectory.
[0024] Furthermore, the laser 2 is an optical fiber coupled semiconductor laser with an output power greater than 500W, and the coupling target 8 is an optical fiber with a core diameter of less than 100μm.
[0025] In this embodiment, the coupling device for the laser-coupled fiber includes a sealed coupling cavity 1, which integrates the core functional components. A laser 2 (e.g., a fiber-coupled semiconductor laser with an output power of 1000W) is fixed to one side of the cavity. A coupling target 8 (e.g., an optical fiber with a core diameter of 50μm) is fixed to the other side of the cavity. An active adjustment mechanism 4 and a focusing optical system 3 are sequentially arranged along the optical path between the laser 2 and the coupling target 8.
[0026] Initialization and Scan Start-up: The user initiates the automatic coupling program via the intelligent control unit 6. The intelligent control unit 6 then sends a command to the drive component 5, driving the active adjustment mechanism 4 to move along a spiral scanning trajectory. This trajectory ensures that the laser beam systematically covers a circular area on the input end face of the coupling target 8. Its advantages include a continuous and thorough search path, allowing for rapid traversal of the possible optimal coupling point area from the center outwards.
[0027] Synchronous signal acquisition: During the scanning process, the high-precision signal acquisition unit 7 synchronously acquires the drive current waveform of the drive component 5 in real time at a sampling rate of no less than 500 kS / s. This ensures a strict correspondence between the mechanical scanning motion and the electrical signals reflecting the system state, providing a true and synchronized data foundation for subsequent analysis.
[0028] Intelligent recognition and decision-making: The intelligent control unit 6 extracts features from the collected current waveform data in real time and inputs them into a pre-trained coupling state recognition model. This model continuously outputs an evaluation value representing the current predicted coupling efficiency.
[0029] Locking and Completion: When the model output value reaches the preset optimal threshold and remains stable, the intelligent control unit 6 immediately generates a locking signal, commanding the drive component 5 to stop scanning and maintain its current position. The entire process is fully automatic, requiring no manual intervention or reliance on terminal optical power feedback, fundamentally solving the problem of automatic coupling in detector-free scenarios.
[0030] Example 2: Furthermore, the coupling state recognition model is obtained by training a training dataset using a machine learning algorithm, and the method for constructing the training dataset includes: A calibration optical power detector 9 is temporarily installed in the coupling cavity 1 to measure the coupling efficiency during the training phase. The active adjustment mechanism 4 is controlled to perform scanning, and the electrical signal waveform sequence acquired by the high-precision signal acquisition unit 7 and its corresponding coupling efficiency value measured by the optical power detector 9 are recorded simultaneously. Align the electrical signal waveform sequence with the coupling efficiency value to form a training sample set.
[0031] Model training is required before the device leaves the factory or during maintenance and calibration. The specific steps are as follows: Introducing a calibration detector: In the coupling cavity 1, between the focusing optical system 3 and the coupling target 8, a calibration optical power detector 9 (such as a water-cooled thermopile power meter) is temporarily installed.
[0032] Data acquisition: The active adjustment mechanism 4 is controlled to perform a wide-range scan. During this process, two key data are recorded synchronously and at high frequency: one is the drive current waveform sequence acquired by the high-precision signal acquisition unit 7, and the other is the actual coupling efficiency value measured by the optical power detector 9.
[0033] Dataset Construction: A large number of collected current waveform data segments are aligned with the coupling efficiency values at the same time to form a massive training sample set. The direct effect of this method is that it enables the AI model to learn the complex, non-linear mapping relationship between the "specific changing pattern of the driving current" and the "optimal coupling state".
[0034] Model Training: The selected machine learning algorithm is trained using this dataset to obtain a coupling state identification model that can accurately predict coupling efficiency based on current characteristics. After training is complete, the calibration optical power detector 9 can be removed.
[0035] Example 3: Furthermore, feature extraction is performed on the electrical signal waveform data. The feature extraction includes: selecting at least a first time-domain feature from a time-domain feature group composed of root mean square (RMS), peak-to-peak value, skewness, and kurtosis; and selecting at least a first frequency-domain feature from a frequency-domain feature group composed of the amplitude of the main frequency component and the centroid of the spectrum, and combining the selected features into a feature vector.
[0036] Furthermore, the coupling state recognition model is a convolutional neural network (CNN) model or a support vector machine (SVM) model; When using a CNN model, the input features are preprocessed one-dimensional time-series data of electrical signals or two-dimensional time-frequency graphs. When using the SVM model, the input features are a feature vector composed of the time-domain features and the frequency-domain features.
[0037] This embodiment details two preferred technical paths for feature extraction and model selection.
[0038] Path A (based on handcrafted features and SVM model): Feature extraction: The intelligent control unit extracts the root mean square (RMS) (reflecting the average power of the signal) and kurtosis (reflecting the sharpness of the signal distribution, sensitive to impulsive features) from the current waveform data in the time domain, and extracts the spectral centroid (reflecting the frequency region where the signal energy is concentrated) from the frequency domain using Fast Fourier Transform (FFT). These three features are combined into a feature vector.
[0039] Model Application: Input this feature vector into a Support Vector Machine (SVM) regression model. SVM models perform well when processing small to medium-sized, structured feature data, offering high computational efficiency. The advantage of this approach is its relatively strong model interpretability and lower processor resource requirements, making it suitable for embedded system deployment.
[0040] Path B (based on raw data and CNN model): Feature extraction: After standardization, filtering and other preprocessing of the current waveform data, a segment of one-dimensional time series data is directly used as input.
[0041] Model Application: A one-dimensional convolutional neural network (CNN) model is used. CNNs can automatically learn and extract deep, manually designed features from raw data. The advantage of this approach is that it avoids the subjectivity and information loss of manual feature design, and usually achieves higher accuracy and robustness, making it particularly suitable for processing complex signal patterns.
[0042] Furthermore, the active adjustment mechanism 4 is a two-axis fast-reflection mirror system, comprising: Reflecting mirror 41 is used to reflect laser beams; Frame 42 is used to fix the reflective lens 41; The inner frame 44 is connected to the lens frame 42 via a first flexible hinge 43, allowing the lens frame 42 and the reflective lens 41 to rotate about a first axis; The base frame 46 is connected to the inner frame 44 via a second flexible hinge 45, allowing the inner frame 44, the mirror frame 42, and the reflective lens 41 to rotate about a second axis orthogonal to the first axis; The base frame 46 is fixed inside the coupling cavity 1.
[0043] Furthermore, the driving assembly 5 includes a first pair of voice coil motors 51 and a second pair of voice coil motors 52. The first pair of voice coil motors 51 act differentially on opposite sides of the lens frame 42 to drive the reflective lens 41 to deflect around a first axis. The second pair of voice coil motors 52 act differentially on opposite sides of the inner frame 44 to drive it to deflect around a second axis.
[0044] Furthermore, the first natural frequencies of the first flexible hinge 43 and the second flexible hinge 45 are designed to be between 10Hz and 100Hz, and the current resolution of the high-precision signal acquisition unit 7 is better than 0.1mA.
[0045] This embodiment details the core mechanical structure that enables high-precision beam pointing control and high-sensitivity signal sensing.
[0046] The active adjustment mechanism 4 is a two-axis fast-reflection mirror system, and its mechanical structure is as follows: Kinematic chain: The reflecting lens 41 is fixed by the lens frame 42, and the lens frame 42 is connected to the inner frame 44 by the first flexible hinge 43, forming a degree of freedom of motion about the first axis. The inner frame 44 is then connected to the base frame 46 fixed to the coupling cavity 1 by the second flexible hinge 45, forming a degree of freedom of motion about the second axis.
[0047] Drive method: The drive assembly 5 includes a first pair of voice coil motors 51 and a second pair of voice coil motors 52. They are all installed in a differential manner, that is, each pair of motors acts on opposite sides of the moving part (frame 42 or inner frame 44). The effect of differential drive is to realize push-pull bidirectional drive, significantly improve the control bandwidth, stiffness and linearity of the system, and actively suppress external disturbances.
[0048] Sensitivity Design: The first natural frequency of the first flexible hinge 43 and the second flexible hinge 45 is designed to be around 25Hz. This design achieves a balance between "high stiffness" and "high sensitivity": frequencies above 10Hz can effectively resist low-frequency vibrations in the environment; frequencies below 100Hz ensure that the system has sufficient sensitivity to weak light radiation pressure and photothermal stress (i.e., "photo-induced stress"), and can produce detectable mechanical deformation.
[0049] Foundation of Sensing Accuracy: The high-precision signal acquisition unit 7 achieves a current resolution of 0.05mA. This specification is the hardware guarantee for realizing weak signal detection. When "photo-enhancing" causes deformation of the flexible hinge, it leads to changes in the load of the voice coil motor, resulting in fluctuations in the drive current from the nanoampere level to the microampere level. High-resolution sampling ensures that these weak features containing coupling state information can be captured.
[0050] The effect of this embodiment is that the high-sensitivity mechanical design effectively converts the "photodynamic force" in the optical domain into "deformation" in the mechanical domain, and then converts it into "current signal change" through high-precision sampling in the electrical domain, thereby providing a physical basis for the realization of the "detectorless" scheme.
[0051] The core principle of this invention is "optical-mechanical-electrical coupling effect and AI intelligent recognition." It utilizes a physical phenomenon overlooked by traditional technologies: during high-power laser coupling, the light radiation pressure and photothermal effect acting on the optical element (collectively referred to as "photo-force") fine-tune the stress state of the mechanical structure. This minute change sensitively modulates the current of the drive motor. By detecting and analyzing this modulation, the current coupling state can be indirectly and accurately deduced.
[0052] The overall workflow of the device of the present invention begins with a user start command and ends with the system automatically locking into the optimal coupling state, as follows: The user initiates the automatic coupling program via the intelligent control unit 6. Subsequently, the intelligent control unit 6 sends specific drive signals to the first pair of voice coil motors 51 and the second pair of voice coil motors 52 of the drive assembly 5, commanding them to drive the active adjustment mechanism 4 to move along a preset spiral scanning trajectory. This trajectory causes the reflecting mirror 41, fixed on the frame 42, to begin a combined motion around the first and second axes, thereby driving the reflected laser beam to systematically scan outwards from a central point along a spiral path on the focal plane in front of the fiber end face of the coupling target 8.
[0053] During this scanning process, the high-precision signal acquisition unit 7 continuously and synchronously acquires the drive current waveforms of the first pair of voice coil motors 51 and the second pair of voice coil motors 52 in real time with a sampling rate of no less than 500 kS / s and a current resolution better than 0.1 mA. This current signal completely encodes all load change information experienced by the drive component 5 during its movement.
[0054] As the laser beam gradually approaches and achieves efficient coupling with the fiber core of the coupling target 8 during scanning, a qualitative change occurs in the interaction between the laser beam and the reflective mirror 41. On the one hand, the change in momentum carried by the laser photons exerts a microNewton-level optical radiation pressure on the reflective mirror 41; on the other hand, the thermal expansion caused by the slight absorption of laser energy by the mirror coating generates photothermal stress. These two effects are collectively referred to as photo-coercivity, which is applied to the reflective mirror 41 and attempts to change its orientation.
[0055] The optical force is transmitted through the frame 42 to the first flexible hinge 43 and the second flexible hinge 45. Because the first-order natural frequencies of these two flexible hinges are intentionally designed within a low-stiffness range of 10Hz to 100Hz, they are extremely sensitive to weak force signals. Therefore, the optical force overcomes the angular stiffness of the flexible hinges, causing a nearly invisible mechanical deformation at the nanometer or microradian level. This deformation manifests as a slight tendency for the frame 42 and inner frame 44 to deviate from their original trajectories of motion.
[0056] This minute deformation tendency of the active adjustment mechanism 4 immediately changes the mechanical load on the first pair of voice coil motors 51 and the second pair of voice coil motors 52. In order to strictly follow the position command issued by the intelligent control unit 6 and maintain the preset scanning trajectory, the drive circuit of the voice coil motor must instantly adjust its output force to counteract this deformation. According to the force-current proportionality law of the voice coil motor, this adjustment of the output force is linearly and without delay modulated into a minute change in the drive current. Therefore, the information of the optical force acting on the reflective mirror 41 is converted into a characteristic modulation signal in the drive current waveform with high fidelity.
[0057] The intelligent control unit 6 continuously acquires these modulated drive current waveform data from the high-precision signal acquisition unit 7 and performs real-time signal processing on them. The processing includes extracting characteristic quantities from the current waveform that can characterize the signal strength and reflect its statistical distribution and frequency components, such as root mean square value, peak-to-peak value, skewness, kurtosis, and the amplitude of the dominant frequency component and the spectral centroid obtained through fast Fourier transform.
[0058] Subsequently, the intelligent control unit 6 combines these extracted feature data into a feature vector and inputs it into its internal pre-trained coupling state recognition model. This model is a regression model trained using machine learning algorithms. In the supervised learning phase involving the calibration optical power detector 9, it has deeply learned the complex nonlinear mapping relationship between different current characteristic modes and the actual coupling efficiency. At this point, based on the input real-time features, the model quickly infers and outputs an evaluation value representing the current predicted coupling state.
[0059] The intelligent control unit 6 continuously monitors this evaluation value. When the system determines that this value has reached its historical maximum and stabilized at the preset optimal threshold, it indicates that the current beam position is the optimal coupling point. The intelligent control unit 6 immediately generates a locking signal, which commands the drive assembly 5 to stop its helical scanning motion and maintain the current drive current of the first pair of voice coil motors 51 and the second pair of voice coil motors 52, thereby locking the attitude of the reflector 41 and the direction of the laser beam at this optimal state. At this point, the entire automatic coupling process of the high-power laser is complete.
[0060] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of protection claimed by the present invention. The scope of protection of the present invention is defined by the appended claims and their equivalents.
Claims
1. A coupling device for coupling a laser to an optical fiber, comprising a coupling cavity (1), a laser (2) mounted on the coupling cavity (1), a focusing optical system (3), and an active adjustment mechanism (4) for adjusting the optical path, wherein the focusing optical system (3) is disposed on the reflected optical path of a reflecting mirror (41) for converging the light beam reflected by the reflecting mirror (41) to the end face of a coupling target (8), and the active adjustment mechanism (4) is driven by a driving component (5), characterized in that, Also includes: A high-precision signal acquisition unit (7) is electrically connected to the drive circuit of the drive component (5) and is configured to acquire the drive current waveform of the drive component (5) in real time during operation. The intelligent control unit (6) is electrically connected to the drive circuit of the high-precision signal acquisition unit (7) and the drive component (5); The intelligent control unit (6) is configured as follows: a. Control the drive component (5) to drive the active adjustment mechanism (4) to move according to the preset scanning trajectory, so that the laser beam scans near the input end of the coupled target (8); b. During the scanning process, electrical signal waveform data from the high-precision signal acquisition unit (7) are received synchronously; c. Extract features from the electrical signal waveform data and input the extracted features into a pre-trained coupled state recognition model; d. When the output of the coupling state recognition model indicates that the optimal coupling state has been reached, a locking signal is generated to stop the scanning motion of the drive component (5) and maintain its current position.
2. The coupling device for coupling an optical fiber to a laser according to claim 1, characterized in that, The coupling state recognition model is obtained by training a training dataset using a machine learning algorithm. The method for constructing the training dataset includes: A calibration optical power detector (9) is temporarily installed in the coupling cavity (1) to measure the coupling efficiency during the training phase; The active adjustment mechanism (4) is controlled to perform scanning, and the electrical signal waveform sequence acquired by the high-precision signal acquisition unit (7) and the corresponding coupling efficiency value measured by the optical power detector (9) are recorded. Align the electrical signal waveform sequence with the coupling efficiency value to form a training sample set.
3. The coupling device for laser-coupled optical fiber according to claim 2, characterized in that, Feature extraction is performed on electrical signal waveform data. The feature extraction includes: selecting at least a first time-domain feature from a time-domain feature group consisting of root mean square (RMS), peak-to-peak value, skewness, and kurtosis; and selecting at least a first frequency-domain feature from a frequency-domain feature group consisting of the amplitude of the main frequency component and the centroid of the spectrum; and combining the selected features into a feature vector.
4. The coupling device for laser-coupled optical fiber according to claim 3, characterized in that, The coupling state recognition model is a convolutional neural network (CNN) model or a support vector machine (SVM) model. When using a CNN model, the input features are preprocessed one-dimensional time-series data of electrical signals or two-dimensional time-frequency graphs. When using the SVM model, the input features are a feature vector composed of the time-domain features and the frequency-domain features.
5. The coupling device for laser-coupled optical fiber according to claim 1, characterized in that, The active adjustment mechanism (4) is a two-axis fast-reflection mirror system, including: Reflecting mirror (41) is used to reflect laser beams; A frame (42) is used to fix the reflective lens (41). The inner frame (44) is connected to the lens frame (42) via a first flexible hinge (43), allowing the lens frame (42) and the reflective lens (41) to rotate about a first axis; The base frame (46) is connected to the inner frame (44) via a second flexible hinge (45), allowing the inner frame (44), the mirror frame (42) and the reflective lens (41) to rotate about a second axis orthogonal to the first axis; The base frame (46) is fixed inside the coupling cavity (1).
6. The coupling device for laser-coupled optical fiber according to claim 5, characterized in that, The drive assembly (5) includes a first pair of voice coil motors (51) and a second pair of voice coil motors (52). The first pair of voice coil motors (51) act differentially on opposite sides of the lens frame (42) to drive the reflective lens (41) to deflect around a first axis. The second pair of voice coil motors (52) act differentially on opposite sides of the inner frame (44) to drive it to deflect around a second axis.
7. The coupling device for coupling an optical fiber to a laser according to claim 6, characterized in that, The first natural frequency of the first flexible hinge (43) and the second flexible hinge (45) is designed to be between 10Hz and 100Hz, and the current resolution of the high-precision signal acquisition unit (7) is better than 0.1mA.
8. The coupling device for laser-coupled optical fiber according to claim 1, characterized in that, The preset scanning trajectory is a spiral scanning trajectory.
9. The coupling device for coupling an optical fiber to a laser according to claim 1, characterized in that, The laser (2) is an optical fiber coupled semiconductor laser with an output power greater than 500W, and the coupling target (8) is an optical fiber with a core diameter of less than 100μm.
10. A coupling method for a laser-coupled optical fiber coupling device as described in claim 1, characterized in that, Includes the following steps: Step 1: Scan Startup Steps The intelligent control unit (6) controls the drive component (5) and drives the active adjustment mechanism (4) to drive the laser beam to perform a preset scanning trajectory movement near the input end face of the coupled target (8); Step 2, Signal Synchronization Acquisition Steps: During the scanning process, the high-precision signal acquisition unit (7) synchronously acquires the drive current waveform data of the drive component (5) in real time; Step 3: Feature Extraction Steps The intelligent control unit (6) processes the collected drive current waveform data and extracts feature data containing time-domain and frequency-domain features; Step 4: Intelligent Status Recognition Steps The intelligent control unit (6) inputs the feature data into the pre-trained coupling state recognition model and obtains the evaluation value output by the model that represents the current coupling state; Step 5, Decision Locking Steps: The intelligent control unit (6) continuously monitors the evaluation value. When it determines that the evaluation value has reached and stabilized at the preset optimal coupling threshold, it generates a locking command, controls the drive component (5) to stop scanning and maintain the current position, and completes automatic coupling.