Method and device for rapidly detecting laser transmission mode in optical fiber without damage

Abstract

The invention belongs to the technical field of laser processing, and particularly discloses a nondestructive rapid detection method and device for a laser transmission mode in an optical fiber. Comprising the following steps: acquiring multiple frames of to-be-detected optical fiber vertical surface light field intensity distribution data pictures; judging whether the laser transmission mode in the to-be-detected optical fiber is stable or not based on the continuous to-be-detected optical fiber vertical surface light field intensity distribution data pictures, if so, entering the step 3, and if not, carrying out next judgment; and comparing the picture of which the judgment result is stable with the pre-stored images of different transmission modes to obtain the pre-stored image with the highest matching degree, and obtaining the transmission mode and the energy ratio of the optical fiber to be detected according to the optical fiber transmission mode and the energy ratio corresponding to the pre-stored image. By combining an advanced image processing technology, a deep learning model and a multi-task learning framework, rapid and accurate detection of the laser transmission mode in the optical fiber is realized, and an efficient and reliable solution is provided for performance monitoring and quality control of the optical fiber laser.

Description

Technical Field

[0001] The present invention belongs to the technical field of laser processing, and more specifically, relates to a method and device for quickly detecting a non-destructive laser transmission mode in an optical fiber. Background Art

[0002] As a new type of precision manufacturing technology, laser processing utilizes the interaction between high-energy photon beams and materials to achieve cutting, welding, surface treatment, drilling, and precision micro-machining of metals and non-metals, especially materials with high hardness, high brittleness, and high melting points. Due to its significant advantages such as high product quality, automation, low pollution, high efficiency, and easy large-scale production, laser processing has been widely applied in many important fields such as scientific research, biomedicine, industrial processing, and national defense security.

[0003] Among them, the high-power laser light source, as the core component, directly determines the performance indicators and processing effects of the laser processing system. According to the type of laser gain medium, laser light sources can be divided into solid laser sources, gas laser sources, semiconductor laser sources, fiber laser sources, etc. Among them, the continuous fiber laser source based on the waveguide structure has the characteristics of high average power, good beam quality, high energy conversion efficiency, good heat dissipation characteristics, compact structure, and high reliability, and is one of the most promising laser light sources in the current laser processing field.

[0004] However, with the continuous increase in the average power of fiber lasers, the thermally induced mode instability effect (TMI) has increasingly become a key factor affecting the stability performance of fiber lasers. How to quickly and non-destructively measure the mode stability of lasers is one of the key difficulties in achieving high-precision laser processing, improving processing consistency, and realizing high-quality laser processing. At the same time, in some special processing fields (such as lithium battery manufacturing, electronic component processing, and automobile manufacturing), it is often necessary to generate stable keyholes and molten pools in the processing area to achieve splash-free welding and reduce the generation of welding pores. Therefore, the beam mode adjustable fiber laser has emerged as the times require. Since it can effectively improve the welding speed and stability, and enhance the airtightness and aesthetics of welding, it has gradually become a potential new type of laser light source. However, how to identify the output mode and stability of the beam mode adjustable fiber laser in real time and non-destructively is one of the key technologies for it to truly move towards a broader application. Traditional laser mode testing methods require sampling using a spatial optical path. The testing instrument is bulky and difficult to effectively integrate, and the stability of the spatial testing optical path is poor, making it impossible to achieve real-time and rapid detection and identification. Summary of the Invention

[0005] In view of the above deficiencies or improvement requirements of the prior art, the present invention provides a non-destructive method and device for quickly detecting the laser transmission mode in an optical fiber. By using continuous pictures of the light field intensity distribution data, the stability of the laser transmission mode in the optical fiber is judged, and a convolutional neural network model is used to match the most similar pre-stored images, so as to quickly and accurately obtain the transmission mode and energy ratio of the optical fiber to be detected.

[0006] To achieve the above object, according to one aspect of the present invention, a non-destructive method for quickly detecting the laser transmission mode in an optical fiber is proposed, including the following steps:

[0007] Step 1: Obtain multiple frames of pictures of the light field intensity distribution data of the vertical plane of the optical fiber to be detected;

[0008] Step 2: Based on the continuous pictures of the light field intensity distribution data of the vertical plane of the optical fiber to be detected, judge whether the laser transmission mode in the optical fiber to be detected is stable. If it is, go to Step 3; if not, make the next judgment;

[0009] Step 3: Compare the pictures with a stable judgment result with the pre-stored images of different transmission modes to obtain the pre-stored image with the highest matching degree, and obtain the transmission mode and energy ratio of the optical fiber to be detected according to the optical fiber transmission mode and energy ratio corresponding to the pre-stored image.

[0010] As a further preference, Step 2 includes the following steps:

[0011] (21) Extract the intensity of the pixel units of the picture of the light field intensity distribution data of the vertical plane;

[0012] (22) Construct a model stability calculation model, and input the intensity of the pixel units of two consecutive frames of pictures of the light field intensity distribution data of the vertical plane into the model stability calculation model to calculate the stability value;

[0013] (23) If the stability value meets the threshold, the laser transmission mode is stable, and go to Step 3; if not, make the next judgment.

[0014] As a further preference, the model stability calculation model includes:

[0015]

[0016] Wherein, X i is the intensity value of the i-th pixel unit of the previous frame of picture, Y i is the intensity value of the i-th pixel unit of the next frame of picture, and P is the stability value.

[0017] As a further preference, Step 3 further includes:

[0018] (31) Construct a convolutional neural network model, use the light intensity distribution image as the input of the convolutional neural network model, and the transmission mode category and power as the output. Optimize and train the convolutional neural network model to learn the light intensity distribution characteristics and the relationship between the leaked laser signal and power;

[0019] (32) Based on the optimized and trained convolutional neural network model, obtain the pre-stored image with the highest matching degree, and obtain the transmission mode and energy ratio of the fiber to be detected according to the fiber optic transmission mode and energy ratio corresponding to the pre-stored image.

[0020] As a further preference, the step (31) includes the following steps:

[0021] (311) Extract the light intensity distribution data, leaked laser signal data, and power data corresponding to the pre-stored image to construct a data set;

[0022] (312) Use a support vector machine model to learn the mapping relationship between the light intensity distribution characteristics, the leaked laser signal characteristics, and power;

[0023] (313) Construct a convolutional neural model. Among them, the input layer is used to take the light intensity distribution image as the input, the convolutional layer is used to extract the local features of the image, the pooling layer is used to reduce the number of parameters and improve the generalization ability, the fully connected layer is used to map the extracted features to the fully connected layer, and the output layer is used to output the probability of each category and the predicted power value;

[0024] (314) Based on the mapping relationship between the light intensity distribution characteristics, the leaked laser signal characteristics, and power, use the data set to train the convolutional neural model.

[0025] As a further preference, the construction of the convolutional neural model further includes: constructing a multi-task learning loss function:

[0026]

[0027] In the formula, N is the number of categories, yi is the actual category, is the predicted category, M is the number of samples, yj is the actual power value, is the predicted power value, and λ is the weight coefficient.

[0028] As a further preference, the step (32) further includes:

[0029] (321) Based on the optimized and trained convolutional neural network model, obtain the predicted transmission mode of the fiber to be detected and the pre-stored image corresponding to the predicted transmission mode;

[0030] (322) Calculate the similarity between the pre-stored image corresponding to the predicted transmission mode and the picture of the optical fiber to be detected. If the similarity meets the threshold, obtain the transmission mode and energy ratio of the optical fiber to be detected according to the optical fiber transmission mode and energy ratio corresponding to the pre-stored image. Otherwise, return to step (321).

[0031] According to another aspect of the present invention, there is also provided a non-destructive rapid detection system for the laser transmission mode in an optical fiber, including:

[0032] A data acquisition module for acquiring multiple pictures of the vertical light field intensity distribution data of the optical fiber to be detected;

[0033] A laser transmission mode detection module for judging whether the laser transmission mode in the optical fiber to be detected is stable based on consecutive pictures of the vertical light field intensity distribution data of the optical fiber to be detected, comparing the pictures with a stable judgment result with pre-stored images of different transmission modes to obtain the pre-stored image with the highest matching degree, and obtaining the transmission mode and energy ratio of the optical fiber to be detected according to the optical fiber transmission mode and energy ratio corresponding to the pre-stored image.

[0034] As a further preference, the data acquisition module includes:

[0035] A sealed dark box;

[0036] An optical fiber fixing component provided on the sealed dark box. The optical fiber fixing component includes an optical fiber clamp, a clamp fixing frame, and an optical fiber pressing block. The optical fiber clamp is installed in the clamp fixing frame and fixed to the clamp fixing frame through a threaded through hole. The optical fiber pressing block is connected to the clamp fixing frame through a hinge. The optical fiber to be detected is placed in the optical fiber clamp and pressed tightly by the optical fiber pressing block; and

[0037] A microscopic imaging component provided in the sealed dark box. The optical fiber to be measured is placed inside the microscopic imaging component. The microscopic imaging component measures the leakage light signal intensity and intensity distribution of the optical fiber to be detected in real time through the integrated photosensitive CCD array and microscopic lens group inside it, that is, the vertical light field intensity distribution data of the optical fiber to be detected.

[0038] As a further preference, in the laser transmission mode detection module, the mode stability calculation model includes:

[0039]

[0040] Among them, X i is the intensity value of the i-th pixel unit of the previous frame of picture, Y i is the intensity value of the i-th pixel unit of the next frame of picture, and P is the stability value.

[0041] Generally speaking, compared with the prior art by the above technical solutions conceived by the present invention, the main technical advantages are as follows:

[0042] 1. The present invention can quickly analyze and identify the laser transmission mode in the transmission optical fiber and detect the mode stability in real time by detecting the leakage laser signals generated by phenomena such as the intrinsic scattering of the optical fiber (Rayleigh scattering, imperfect waveguide structure, etc.) and thermally induced mode instability.

[0043] 2. Through the automated image processing and pattern recognition technology, this solution can quickly determine the stability of the laser transmission mode in the optical fiber from multiple frames of light field intensity distribution data, reducing manual intervention and processing time and improving the detection efficiency. At the same time, by using the convolutional neural network model to deeply learn the relationship between the light intensity distribution characteristics, leakage laser signals and power, the recognition accuracy of the transmission mode is enhanced, ensuring the reliability of the detection results.

[0044] 3. By constructing a multi-task learning loss function, the convolutional neural network model of the present invention can not only identify the optical fiber transmission mode but also predict the power value, improving the adaptability and generalization ability of the model. This multi-task learning framework allows the system to maintain a high recognition accuracy and stability evaluation ability when dealing with new or different optical fiber transmission modes.

[0045] 4. The non-destructive detection technology provided by the present invention enables real-time monitoring and quality control of the transmission mode of the fiber laser without affecting its normal operation. This is crucial for the maintenance and performance optimization of fiber lasers, as it can help detect and solve potential problems in a timely manner, reduce equipment failure rates, extend equipment life, thereby reducing maintenance costs and increasing production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 is a top view of the data acquisition module of a non-destructive in-fiber laser transmission mode rapid detection system according to an embodiment of the present invention;

[0047] Figure 2 is Figure 1 the front view of the data acquisition module in

[0048] Figure 3 is Figure 1 the left view of the data acquisition module in

[0049] Figure 4 is a schematic diagram of laser signal leakage caused by transmission optical fiber scattering;

[0050] Figure 5 is a schematic diagram of the light intensity distribution of the transmissible modes of a large mode field optical fiber at different normalized frequencies;

[0051] Figure 6 In Figure 6In (b) is the contour distribution diagram of the single-mode field, Figure 6 In (c) is the height map of the two-mode distribution, Figure 6 In (d) is the contour distribution diagram of the two-mode field, Figure 6 In (e) is the height map of the four-mode distribution, Figure 6 In (f) is the contour distribution diagram of the four-mode field;

[0052] Figure 7 are the light intensity distribution diagrams of different laser transmission modes.

[0053] In all the drawings, the same reference numerals denote the same technical features, specifically: 1-bottom lining housing; 2-sealing cover plate; 3-hinge; 4-microscopic imaging assembly; 5-optical fiber clamp; 6-clamp fixing bracket; 7-threaded through hole; 8-optical fiber pressing block; 9-hinge; 10-threaded through hole; 11-through hole. Detailed implementation manners

[0054] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0055] A non-destructive and rapid detection method for the laser transmission mode in an optical fiber provided by an embodiment of the present invention has the following implementation principle:

[0056] As Figure 5 、 Figure 6 、 Figure 7 shown, based on the principle of total internal reflection of the medium, the laser signal in the optical fiber waveguide is theoretically a leak-free transmission. Without destroying the waveguide structure, the external detection device cannot obtain information such as the laser transmission mode, frequency, intensity, and phase in the optical fiber from the fiber surface. However, due to the existence of intrinsic scattering (Rayleigh scattering, etc.), internal stress in the optical fiber, imperfect waveguide structure, waveguide microbending, thermally induced mode instability, etc., it is often difficult for the optical fiber waveguide to achieve perfect leak-free transmission. Especially for high-energy laser light sources used in industrial processing, their average power can often reach several kilowatts or even tens of kilowatts per single fiber (or pulse energy of several millijoules or even hundreds of millijoules), which makes the leakage light signals caused by the above defects can be collected by ordinary commercial optical microscopic detectors.

[0057] Furthermore, the present invention realizes transmission mode recognition by using the intrinsic scattering (Rayleigh scattering) of optical fibers. Rayleigh scattering is caused by the inhomogeneity of material atoms or molecules and the material structure, which makes the refractive index of the material microscopically inhomogeneous and causes the scattering of transmitted light waves. This kind of scattering is inherent in the material, cannot be eliminated, and the scattering direction is random, and is linearly related to the light intensity of the transmitted laser mode.

[0058] At the same time, in order to suppress nonlinear effects such as stimulated Brillouin scattering and stimulated Raman scattering in high-power fiber lasers, large-mode-field few-mode fibers with a core diameter of more than 20 μm are often used. For lasers with adjustable output modes, multimode fibers are required. There are obvious differences in the light intensity distribution of different transmission modes. For large-mode fibers with a core diameter of more than 20 μm, there are micron-level differences in the light intensity distribution of different high-order transmission modes.

[0059] The leakage light caused by Rayleigh scattering is in a random direction and is linearly related to the power. At the same time, there are micron-level differences in the light intensity distribution of different transmission modes in large-mode-field fibers. According to the Abbe diffraction limit formula, the minimum scale that can be observed by an optical microscope is 1 / 2 of the wavelength of the incident light. Using an optical microscope with a light source in the 1-μm band, in principle, a resolution of not less than 500 nm can be achieved, and the light intensity distribution inside the fiber core can be recognized.

[0060] Based on the above principle, the present invention can quickly analyze and identify the laser transmission mode in the transmission fiber and detect the mode stability in real time by detecting the leakage laser signal generated by the intrinsic scattering (such as Rayleigh scattering) of the optical fiber, using a micron-level microscopic system, and combining the light intensity distribution characteristics of the optical fiber transmission mode.

[0061] Specifically, a non-destructive method for quickly detecting the laser transmission mode in an optical fiber according to the present invention includes the following steps:

[0062] Step 1: Obtain multiple frames of data pictures of the vertical light field intensity distribution of the optical fiber to be detected;

[0063] Step 2: Based on the continuous data pictures of the vertical light field intensity distribution of the optical fiber to be detected, judge whether the laser transmission mode in the optical fiber to be detected is stable. If so, go to Step 3; if not, make the next judgment;

[0064] Step 3: Compare the pictures with a stable judgment result with the pre-stored images of different transmission modes to obtain the pre-stored image with the highest matching degree, and obtain the transmission mode and energy ratio of the optical fiber to be detected according to the optical fiber transmission mode and energy ratio corresponding to the pre-stored image.

[0065] Based on any of the above embodiments or a combination of multiple embodiments, Step 2 includes the following steps:

[0066] (21) Extract the intensity of pixel units in the image of the vertical plane light field intensity distribution data;

[0067] (22) Construct a mode stability calculation model, and input the intensity of pixel units in the images of the vertical plane light field intensity distribution data for two consecutive frames into the mode stability calculation model to calculate the stability value;

[0068] (23) If the stability value meets the threshold, the laser transmission mode is stable, and proceed to step three; if not, make the next judgment.

[0069] Based on any one of the above embodiments or a combination of multiple embodiments, the mode stability calculation model includes:

[0070]

[0071] where X i is the intensity value of the i-th pixel unit in the previous frame image, Y i is the intensity value of the i-th pixel unit in the next frame image, and P is the stability value.

[0072] Based on any one of the above embodiments or a combination of multiple embodiments, step three further includes:

[0073] (31) Construct a convolutional neural network model, use the light intensity distribution image as the input of the convolutional neural network model, and the transmission mode category and power as the output, and optimize and train the convolutional neural network model to learn the relationship between the light intensity distribution characteristics, the leaked laser signal and the power;

[0074] (32) Based on the optimized and trained convolutional neural network model, obtain the pre-stored image with the highest matching degree, and obtain the transmission mode and energy ratio of the fiber to be detected according to the fiber transmission mode and energy ratio corresponding to the pre-stored image.

[0075] Based on any one of the above embodiments or a combination of multiple embodiments, step (31) includes the following steps:

[0076] (311) Extract the light intensity distribution data, leaked laser signal data and power data corresponding to the pre-stored image, and construct a data set;

[0077] (312) Use the support vector machine model to learn the mapping relationship between the light intensity distribution characteristics, the leaked laser signal characteristics and the power;

[0078] (313) Construct a convolutional neural model, where the input layer is used to input the light intensity distribution image, the convolutional layer is used to extract the local features of the image, the pooling layer is used to reduce the number of parameters and improve the generalization ability, the fully connected layer is used to map the extracted features to the fully connected layer, and the output layer is used to output the probability of each category and the predicted power value;

[0079] (314) Based on the mapping relationship between the light intensity distribution characteristics, the characteristics of the leaked laser signal, and the power, the convolutional neural model is trained using the data set.

[0080] Based on any of the above embodiments or a combination of multiple embodiments, the building of the convolutional neural model further includes: building a multi-task learning loss function:

[0081]

[0082] In the formula, N is the number of categories, yi is the actual category, is the predicted category, M is the number of samples, yj is the actual power value, is the predicted power value, and λ is the weight coefficient.

[0083] Based on any of the above embodiments or a combination of multiple embodiments, step (32) further includes:

[0084] (321) Based on the optimized and trained convolutional neural network model, obtain the predicted transmission mode of the optical fiber to be detected and the pre-stored image corresponding to the predicted transmission mode;

[0085] (322) Calculate the similarity between the pre-stored image corresponding to the predicted transmission mode and the picture of the optical fiber to be detected. If the similarity meets the threshold, obtain the transmission mode and energy ratio of the optical fiber to be detected according to the optical fiber transmission mode and energy ratio corresponding to the pre-stored image. Otherwise, return to step (321).

[0086] Based on any of the above embodiments or a combination of multiple embodiments, in this embodiment, a non-destructive rapid detection system for the laser transmission mode in an optical fiber is used to obtain multiple pictures of the vertical light field intensity distribution data of the optical fiber to be detected.

[0087] Perform mode stability judgment: Extract the intensity of the pixel unit. Build a mode stability calculation model to calculate the stability value. If the stability value meets the threshold, the laser transmission mode is stable. Otherwise, perform the next judgment.

[0088] Match the pre-stored image: Compare the stable picture with the pre-stored images of different transmission modes. Obtain the pre-stored image with the highest matching degree. Obtain the transmission mode and energy ratio of the optical fiber to be detected according to the optical fiber transmission mode and energy ratio corresponding to the pre-stored image.

[0089] Among them, in the laser transmission mode detection module, the mode stability calculation model includes:

[0090]

[0091] Among them, X i is the intensity value of the i-th pixel unit of the previous frame of picture, Y iis the intensity value of the i-th pixel unit of the subsequent frame image, and P is the stable value.

[0092] The construction of the convolutional neural network model includes:

[0093] (1) Dataset construction: Extract the light intensity distribution data, leaked laser signal data, and power data corresponding to the pre-stored images. Use the support vector machine model to learn the mapping relationship between the light intensity distribution characteristics, leaked laser signal characteristics, and power.

[0094] (2) CNN model design: Input layer: Light intensity distribution image. Convolutional layer: Extract local features of the image. Pooling layer: Reduce the number of parameters and improve the generalization ability. Fully connected layer: Map the extracted features to the fully connected layer. Output layer: Output the probability of each category and the predicted power value.

[0095] (3) Construct a multi-task learning loss function:

[0096]

[0097] In the formula, N is the number of categories, yi is the actual category, is the predicted category, M is the number of samples, yj is the actual power value, is the predicted power value, and λ is the weight coefficient.

[0098] (4) Best matching image retrieval

[0099] Predicted transmission mode acquisition: Use the optimized and trained CNN model to obtain the predicted transmission mode of the optical fiber to be detected and the corresponding pre-stored image.

[0100] Similarity calculation: Calculate the similarity between the pre-stored image corresponding to the predicted transmission mode and the image of the optical fiber to be detected. If the similarity meets the threshold, obtain the transmission mode and energy ratio of the optical fiber to be detected according to the optical fiber transmission mode and energy ratio corresponding to the pre-stored image, otherwise return to the predicted transmission mode acquisition step.

[0101] In the above embodiments, convolutional layer parameters: filter size k, stride s, padding p. Pooling layer parameters: pooling layer size 2×2, stride 2. Fully connected layer parameters: number of neurons in the fully connected layer. Learning rate: learning rate of the Adam optimizer. Batch size: batch size during training. Number of iterations: number of iterations during training. Regression loss weight λ: weight for balancing classification and regression tasks. Similarity threshold θ: threshold for screening similar images.

[0102] More specifically, in the above embodiments, data acquisition and preprocessing include: Light intensity distribution data: Use a microscopic imaging system to capture the light intensity distribution image of the fiber optic transmission mode. Leaked laser signal data: Use an optical detector to collect the leaked laser signal. Power measurement data: Use a power meter to measure the output power of the fiber laser.

[0103] Feature extraction includes: Light intensity distribution features: Extract features from the light intensity distribution image, such as the light intensity center, the shape and size of the light intensity distribution. Leaked signal features: Extract features from the leaked laser signal, such as signal intensity, frequency, and duration. Power features: Directly use the measured power data as features.

[0104] Convolution layer calculation:

[0105]

[0106] Pooling layer calculation:

[0107]

[0108] In one embodiment, cosine similarity is used for similarity calculation:

[0109]

[0110] where A and B are the feature vectors of two images respectively, · represents the dot product, and ||A|| and ||B|| are the norms of the vectors respectively.

[0111] Mode superposition:

[0112]

[0113] where αi is the superposition ratio determined according to the similarity magnitude, and mode i are the three main transmission modes with the highest similarity.

[0114] Through the above solution, a convolutional neural network model can be constructed that first performs image similarity judgment before learning the mapping relationship and combines mode superposition. This model will be able to screen out images with higher similarity to the fiber optic transmission mode to be detected, and then learn the complex mapping relationship between the light intensity distribution characteristics, the leaked laser signal, and the power, providing more accurate technical support for the performance monitoring and quality control of fiber lasers.

[0115] Such as Figure 1 、 Figure 2 and Figure 3 shown, according to another aspect of the present invention, there is also provided a non-destructive fast detection system for the laser transmission mode inside the optical fiber, including:

[0116] A data acquisition module for obtaining multiple pictures of the light field intensity distribution data of the vertical plane of the optical fiber to be detected;

[0117] A laser transmission mode detection module, which is used to judge whether the laser transmission mode in the fiber to be detected is stable based on continuous pictures of the intensity distribution data of the vertical plane light field of the fiber to be detected, and compare the pictures with stable judgment results with pre-stored images of different transmission modes to obtain the pre-stored image with the highest matching degree, and obtain the transmission mode and energy ratio of the fiber to be detected according to the fiber transmission mode and energy ratio corresponding to the pre-stored image.

[0118] Based on any of the above embodiments or a combination of multiple embodiments, the data acquisition module includes: a sealed dark box; an optical fiber fixing component arranged on the sealed dark box, the optical fiber fixing component includes an optical fiber clamp 5, a clamp fixing frame 6, and an optical fiber pressing block 8. The optical fiber clamp 5 is installed in the clamp fixing frame 6 and fixed to the clamp fixing frame 6 through a threaded through hole 7. The optical fiber pressing block 8 is connected to the clamp fixing frame 6 through a hinge 9. The fiber to be detected is placed in the optical fiber clamp 5 and pressed by the optical fiber pressing block 8; and a microscopic imaging component 4, arranged in the sealed dark box, the fiber to be measured is placed inside the microscopic imaging component 4, and the microscopic imaging component 4 measures the intensity and intensity distribution of the leakage light signal of the fiber to be detected in real time through the integrated photosensitive CCD array and microscopic lens group inside it, that is, the intensity distribution data of the vertical plane light field of the fiber to be detected.

[0119] Based on any of the above embodiments or a combination of multiple embodiments, in the laser transmission mode detection module, the mode stability calculation model includes:

[0120]

[0121] Among them, X i is the intensity value of the i-th pixel unit of the previous frame of picture, Y i is the intensity value of the i-th pixel unit of the next frame of picture, and P is the stability value.

[0122] In a preferred embodiment, the bottom lining housing 1 and the sealing cover plate 2 are connected by a hinge 3 to form a sealed dark box to avoid interference from external noise and light. The inner sides of the bottom lining housing 1 and the sealing cover plate 2 are coated with a light-absorbing material to absorb internal diffuse reflected stray light. An opening is provided at the top of the microscopic imaging assembly 4 to connect the inside and outside. When the device is in use, the optical fiber to be measured is placed inside the microscopic imaging assembly 4 through the opening. Among them, the photosensitive CCD array measures the intensity and intensity distribution of the optical fiber leakage light signal in real time through a microscopic lens group. The optical fiber fixing assembly consists of an optical fiber fixture (V-groove) 5, a fixture fixing frame 6, and an optical fiber pressing block 8. The optical fiber fixture (V-groove) 5 is installed inside the fixture fixing frame 6 and fixed through a threaded through hole 7. The optical fiber pressing block 8 is connected to the fixture fixing frame 6 by a hinge 9. When the device is in use, the optical fiber is placed in the optical fiber fixture (V-groove) 5 and pressed tightly by the optical fiber pressing block 8. At the same time, the optical fiber fixing assembly is symmetrically placed on both sides of the microscopic imaging assembly 4 to fix the optical fiber, so as to realize the stable operation of the device. The electrical signal interface 12 is used to upload the optical signal intensity and intensity distribution data measured by the device to the upper computer for analysis. The transmission mode recognition and stability analysis functions are realized by calculating and analyzing the collected data through the upper computer control software.

[0123] Based on an alternative embodiment of the present invention, by detecting the leakage laser signal generated by the intrinsic scattering (such as Rayleigh scattering) of the optical fiber, using a micron-level microscopic system, and combining the optical intensity distribution characteristics of the optical fiber transmission mode, it is possible to quickly analyze and identify the laser transmission mode in the transmission optical fiber and detect the mode stability in real time. Specifically,

[0124] The transmission mode recognition function is to obtain the optical field intensity distribution data of the optical fiber vertical plane through the microscopic imaging assembly, and then compare and identify the obtained single-frame image data with the pre-stored images of different transmission modes in the control software database to obtain the three main transmission modes with the highest similarity. The three main transmission modes are superimposed in proportion in sequence according to the similarity, and the best superimposed transmission mode is obtained by comparing with the single-frame image data, and finally the optical fiber transmission mode and energy ratio are obtained. The transmission mode stability analysis function is to obtain the optical field intensity distribution data of the optical fiber vertical plane through the CCD array probe, and then calculate and analyze the intensity data of each pixel unit of the two consecutive frame image data obtained:

[0125]

[0126] Among them, X i is the intensity value of the i-th pixel unit of the previous frame of the picture, Y i is the intensity value of the i-th pixel unit of the subsequent frame of the picture, and P is the stability value. When the mode instability value is "1", it means that the two consecutive frame modes are completely different, and when the value is "0", it means that the two consecutive frame modes are completely the same.

[0127] Those skilled in the art can easily understand that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A non-destructive method for rapid detection of laser transmission modes in optical fibers, comprising the following steps: Step 1: Obtain multiple frames of light field intensity distribution data images of the vertical plane of the optical fiber to be detected; Step 2: Based on the continuous vertical light field intensity distribution data images of the optical fiber to be tested, determine whether the laser transmission mode in the optical fiber to be tested is stable. If so, proceed to step 3; if not, proceed to the next judgment; Step three, compare the image judged to be stable with pre-stored images of different transmission modes to obtain the pre-stored image with the highest matching degree, and obtain the transmission mode and energy proportion of the optical fiber to be tested based on the optical fiber transmission mode and energy proportion corresponding to the pre-stored image.

2. A non-destructive method for rapid detection of laser transmission modes in optical fibers according to claim 1, characterized in that: Step 2 includes the following steps: (21) extracting the pixel unit intensity of the vertical light field intensity distribution data image; (22) constructing a mode stability calculation model, and inputting the pixel unit intensity of two consecutive frames of vertical light field intensity distribution data images into the mode stability calculation model to calculate the stability value; (23) If the stability value meets the threshold, the laser transmission mode is stable and the process goes to step 3; if not, the process proceeds to the next judgment.

3. A non-destructive method for rapid detection of laser transmission modes in optical fibers according to claim 2, characterized in that: The mode stability calculation model includes: Among them, X i is the intensity value of the i-th pixel unit in the previous frame, Y i is the intensity value of the i-th pixel unit in the next frame, and P is the stable value.

4. The non-destructive method for rapid detection of laser transmission modes in optical fibers according to claim 1, characterized in that: Step three also includes: (31) constructing a convolutional neural network model, using the light intensity distribution image as the input of the convolutional neural network model, and the transmission mode category and power as the output, and optimizing and training the convolutional neural network model to learn the relationship between the light intensity distribution characteristics, the leakage laser signal and the power; (32) Based on the optimized and trained convolutional neural network model, the pre-stored image with the highest matching degree is obtained, and the transmission mode and energy proportion of the optical fiber to be detected are obtained according to the optical fiber transmission mode and energy proportion corresponding to the pre-stored image.

5. The non-destructive method for rapid detection of laser transmission modes in optical fibers according to claim 4, characterized in that: The step (31) comprises the following steps: (311) extracting light intensity distribution data, leaked laser signal data, and power data corresponding to the pre-stored image, and constructing a data set; (312) A support vector machine model is used to learn the mapping relationship between light intensity distribution characteristics, leakage laser signal characteristics and power; (313) Construct a convolutional neural model, wherein the input layer is used to take the light intensity distribution image as input, the convolution layer is used to extract local features of the image, the pooling layer is used to reduce the number of parameters and improve the generalization ability, the fully connected layer is used to map the extracted features to the fully connected layer, and the output layer is used to output the probability of each category and the predicted power value; (314) Based on the mapping relationship between light intensity distribution characteristics, leakage laser signal characteristics and power, the convolutional neural model is trained using a data set.

6. A non-destructive method for rapid detection of laser transmission modes in optical fibers according to claim 4, characterized in that: The constructing of the convolutional neural model further includes: constructing a multi-task learning loss function: Where N is the number of categories, yi is the actual category, is the predicted category, M is the number of samples, yj is the actual power value, is the predicted power value, and λ is the weight coefficient.

7. The non-destructive method for rapid detection of laser transmission modes in optical fibers according to claim 4, characterized in that: The step (32) further comprises: (321) obtaining a predicted transmission mode of the optical fiber to be detected and a pre-stored image corresponding to the predicted transmission mode based on the optimized trained convolutional neural network model; (322) Calculate the similarity between the pre-stored image corresponding to the predicted transmission mode and the image of the optical fiber to be detected. If the similarity meets the threshold, obtain the transmission mode and energy proportion of the optical fiber to be detected based on the optical fiber transmission mode and energy proportion corresponding to the pre-stored image. Otherwise, return to step (321).

8. A non-destructive rapid detection system for laser transmission modes in optical fibers, characterized in that: include: A data acquisition module is used to obtain multiple frames of light field intensity distribution data images of the vertical surface of the optical fiber to be detected; The laser transmission mode detection module is used to determine whether the laser transmission mode in the optical fiber to be detected is stable based on continuous images of the vertical light field intensity distribution data of the optical fiber to be detected, and compare the image with the stable judgment result with the pre-stored images of different transmission modes to obtain the pre-stored image with the highest matching degree, and obtain the transmission mode and energy proportion of the optical fiber to be detected according to the optical fiber transmission mode and energy proportion corresponding to the pre-stored image.

9. The non-destructive rapid detection system for laser transmission mode in optical fiber according to claim 8, characterized in that: The data acquisition module comprises: Sealed dark box; An optical fiber fixing assembly is arranged on the sealed dark box, the optical fiber fixing assembly comprises an optical fiber clamp (5), a clamp fixing frame (6), and an optical fiber pressing block 8, the optical fiber clamp (5) is installed in the clamp fixing frame (6) and fixed to the clamp fixing frame (6) via a threaded through hole (7), the optical fiber pressing block (8) is connected to the clamp fixing frame (6) via a hinge (9), and the optical fiber to be detected is placed in the optical fiber clamp (5) and pressed by the optical fiber pressing block (8); and The microscopic imaging component (4) is arranged in the sealed dark box, and the optical fiber to be tested is placed inside the microscopic imaging component (4). The microscopic imaging component (4) measures the leakage light signal intensity and intensity distribution of the optical fiber to be tested in real time through the photosensitive CCD array and microlens group integrated inside the microscopic imaging component (4), that is, the vertical light field intensity distribution data of the optical fiber to be tested.

10. A non-destructive rapid detection system for laser transmission modes in optical fibers according to claim 9, characterized in that: In the laser transmission mode detection module, the mode stability calculation model includes: Among them, X i is the intensity value of the i-th pixel unit in the previous frame, Y i is the intensity value of the i-th pixel unit in the next frame, and P is the stable value.

Images