CN118232155A

CN118232155A - Solid mode-locked laser control method and system based on twin convolutional neural network

Info

Publication number: CN118232155A
Application number: CN202410641747.9A
Authority: CN
Inventors: 关晨; 渠帅; 王晓飞
Original assignee: Laser Institute of Shandong Academy of Science
Current assignee: Laser Institute of Shandong Academy of Science
Priority date: 2024-05-23
Filing date: 2024-05-23
Publication date: 2024-06-21
Anticipated expiration: 2044-05-23
Also published as: CN118232155B

Abstract

The application relates to the technical field of laser pulse characteristic measurement, and provides a solid mode-locked laser control method and system based on a twin convolutional neural network, wherein the control method comprises the steps of training the twin convolutional neural network; acquiring one or more of a second time sequence signal, a second radio frequency spectrum signal or a second spectrum signal; converting the extracted features into a second transient feature sequence; inputting the second characteristic parameter into a second sub-network to obtain a second characteristic parameter; comparing to obtain Euclidean distance; determining that the real-time working state of the solid mode-locked laser does not comprise the stable mode-locked state in response to the Euclidean distance not being in the preset range; and adjusting the position of the X-cavity mirror assembly to enable the X-cavity mirror assembly to be in a stable mode locking state. The control method integrates the automatic analysis and adjustment functions of the convolutional neural network, can reduce manual intervention, improves the degree of automation and the stability, and reduces the maintenance cost and the labor cost. The output performance of the solid mode-locked laser can be continuously adjusted and optimized through real-time monitoring and adjustment.

Description

Solid mode-locked laser control method and system based on twin convolutional neural network

Technical Field

The application relates to the technical field of laser pulse characteristic measurement, in particular to a solid mode locking laser control method and system based on a twin convolutional neural network.

Background

The solid mode locking laser light source plays an important role in the fields of scientific research, medical diagnosis and treatment, material processing, communication technology, laser radar and the like. Its stable pulse output and high power characteristics make it an ideal choice for ultra-fast optical experiments and precision machining. In scientific research, solid-state mode-locked lasers are used to observe ultra-fast dynamic processes such as vibration and electronic behavior of atoms and molecules. In the medical field, the solid mode-locked laser can be used for ophthalmic surgery, skin treatment and cancer treatment, and has the characteristics of high efficiency and accuracy. In addition, the solid mode-locked laser plays an important role in material processing, and can be used in the fields of micromachining, laser marking, cutting and the like. In communication technology, solid-state mode-locked lasers are used in high-speed data transmission and fiber optic communication systems. Its advantages include stable pulse output, high power, excellent beam quality and compact design, resulting in wide applicability and reliability in various application scenarios.

However, in the process of designing and constructing the solid mode-locked laser light source, under the condition that the cavity type and the cavity length are preliminarily determined according to actual requirements or experimental conditions, the beam waist radius/diameter of the positions of the crystal and the saturated absorber is calculated by using the ABCD transmission matrix, and the cavity mirror with proper curvature is selected to be determined, so that the position and the size of the stable region are further determined to guide an experiment, and the position of a certain key cavity mirror can be manually adjusted and judged according to experimental phenomena in practice. For crystals doped with different ions, the mode locking stable region ranges are different, and particularly for crystals with smaller emission cross sections, the alignment condition is very harsh. If the position of the cavity mirror is not adjusted to reach the mode locking stable region, the whole mode locking cavity is difficult to form stable mode locking laser output and even can not output laser.

Therefore, how to adjust and judge the position of the key cavity mirror to form stable solid laser mode locking laser output is a technical problem which is difficult to solve at present.

Disclosure of Invention

The application provides a solid mode-locked laser control method and a solid mode-locked laser control system based on a twin convolutional neural network, which are used for solving the technical problem that the adjustment of a key cavity mirror to reach a mode-locked stable region range is difficult in the construction process of a solid mode-locked laser.

The solid mode-locked laser control method based on the twin convolutional neural network provided by the first aspect of the application comprises the following steps: establishing a database and a twin convolutional neural network; the system comprises a database, a twin convolutional neural network, a first transient state characteristic sequence and a second transient state characteristic parameter, wherein the database comprises the first transient state characteristic sequence and the first characteristic parameter corresponding to the first transient state sequence; training a twin convolutional neural network by adopting a database; the training modes of the first sub-network and the second sub-network are the same; the input of the first subnetwork is a first transient state characteristic sequence, the output of the first subnetwork is a first characteristic parameter, the first transient state characteristic sequence is obtained by extracting one or more of a first time sequence signal, a first radio frequency spectrum signal or a first spectrum signal of the solid mode-locked laser in different working states, and the working states comprise one of a non-light-emitting state, an output continuous light state, a Q-switched state, an incomplete mode-locked state and a stable mode-locked state; acquiring one or more of a second time sequence signal, a second radio frequency spectrum signal or a second spectrum signal of the solid mode-locked laser; the second time sequence signal, the second radio frequency spectrum signal or the second spectrum signal is a real-time sequence signal, a real-time radio frequency spectrum signal or a real-time spectrum signal of the solid mode-locked laser respectively; converting one or more of the second time sequence signal, the second radio frequency spectrum signal or the second spectrum signal into a second transient characteristic sequence after characteristic extraction; inputting the second transient characteristic sequence into a second sub-network to obtain a second characteristic parameter; comparing the second characteristic parameter with the first characteristic parameter to obtain Euclidean distance; determining that the real-time working state of the solid mode-locked laser does not comprise the stable mode-locked state in response to the Euclidean distance not being in the preset range; and adjusting the position of an X-cavity mirror assembly in the solid mode-locked laser according to the real-time working state, so that the solid mode-locked laser is in a stable mode-locked state.

In some possible implementations, the solid mode-locked laser control method based on the twin convolutional neural network further includes: and determining that the real-time working state of the solid mode-locked laser is a stable mode-locked state in response to the Euclidean distance being in a preset range.

In some possible implementations, the first subnetwork and the second subnetwork each include a first convolutional layer, a first pooling layer, a second convolutional layer, and a second pooling layer that are sequentially connected; wherein the first convolution layer and the second convolution layer each employ Relu activation functions.

According to the solid mode-locked laser control method based on the twin convolutional neural network, which is provided by the first aspect of the application, the working state and the performance parameters of the solid mode-locked laser are determined by carrying out high-precision analysis and identification on the input time sequence signals, the radio frequency spectrum or the spectrum signals, so that the position state of the key cavity mirror is accurately adjusted. Due to the high efficiency and parallel processing capability of the convolutional neural network, the output of the solid mode-locked laser can be monitored and regulated in real time, so that a stable mode-locked state can be realized, and the method is suitable for environmental changes and fluctuation of working conditions. The control method integrates the automatic analysis and adjustment functions of the convolutional neural network, can reduce manual intervention, improves the automation degree and stability of the system, and reduces maintenance cost and labor cost. Through real-time monitoring and adjustment, the output performance of the solid mode-locked laser can be continuously adjusted and optimized, so that the solid mode-locked laser reaches the optimal working state, and the energy efficiency and the output quality are improved.

The solid mode-locked laser control system based on the twin convolutional neural network provided by the second aspect of the application adopts the solid mode-locked laser control method based on the twin convolutional neural network provided by the first aspect, and the solid mode-locked laser control system based on the twin convolutional neural network comprises: a pump source configured to generate light; wherein, the light carries energy; the X-cavity mirror assembly is arranged on the light path of the light; a laser crystal disposed in the X-cavity mirror assembly configured to absorb energy and generate a laser light; wherein the X-cavity component mirror is configured to reflect light to the laser crystal, receive and reflect laser light converted by the laser crystal; the detector is arranged on the output light path of the X-cavity mirror assembly and is configured to receive laser reflected by the X-cavity mirror assembly; the data acquisition card is connected with the detector and is configured to acquire data in the laser; the processor is connected with the data acquisition card and is configured to process data, determine and output the real-time working state of the solid mode-locked laser; and the state controller is connected with the processor and is configured to respond to the real-time working state without the stable mode locking state and adjust the position of the X-cavity mirror assembly according to the real-time working state so as to enable the solid mode locking laser to be in the stable mode locking state.

In some possible implementations, the X-ray cavity mirror assembly includes a first cavity mirror, a second cavity mirror, a third cavity mirror, a semiconductor saturation absorber mirror, and an output mirror; wherein the laser crystal is disposed between the first and second mirrors.

In some possible implementations, adjusting the position of the X-ray cavity mirror assembly according to the real-time operating state includes:

And adjusting the setting positions of the second cavity mirror and the output mirror according to the real-time working state.

In some possible implementations, the solid-mode-locked laser control system based on a twin convolutional neural network further includes: the first lens and the second lens are sequentially arranged between the pumping source and the first cavity mirror.

In some possible implementations, the data acquisition card includes a first acquisition module and a second acquisition module, where the first acquisition module is configured to acquire a first timing signal, a first radio frequency spectrum signal, or a first spectrum signal of the solid-mode-locked laser in different working states, and the working states include one of a non-light-emitting state, an output continuous light state, a Q-switched state, an incomplete mode-locked state, and a stable mode-locked state; the second acquisition module is configured to acquire one or more of a second timing signal, a second radio frequency spectrum signal or a second spectrum signal of the solid-state mode-locked laser; the second time sequence signal, the second radio frequency spectrum signal or the second spectrum signal is a real-time sequence signal, a real-time radio frequency spectrum signal or a real-time spectrum signal of the solid mode-locked laser respectively; the processor comprises: a building module configured to build a database and a twin convolutional neural network; the twin convolutional neural network comprises a first sub-network and a second sub-network; the database comprises first transient state characteristic parameters, wherein the first transient state characteristic parameters are obtained by extracting one or more of a first time sequence signal, a first radio frequency spectrum signal or a first spectrum signal; a training module configured to train the twin convolutional neural network using the database; the training modes of the first sub-network and the second sub-network are the same; the input of the first subnetwork is a first transient characteristic parameter, and the output of the first subnetwork is a first characteristic parameter; the input of the second subnetwork is a second transient characteristic parameter; the second transient characteristic parameter is obtained by extracting one or more of a second time sequence signal, a second radio frequency spectrum signal and a second spectrum signal, and the output of the second subnetwork is the second characteristic parameter; a comparison module configured to compare the euclidean distance of the second characteristic parameter with the first characteristic parameter; and the determining module is configured to determine that the real-time working state of the solid mode-locked laser does not comprise the stable mode-locked state in response to the Euclidean distance not being in the preset range.

In some possible implementations, the determining module is further configured to determine that the real-time operating state of the solid-mode-locked laser is a stable mode-locked state in response to the euclidean distance being within a preset range.

The solid mode-locked laser control system based on the twin convolutional neural network provided by the second aspect of the present application adopts the solid mode-locked laser control method based on the twin convolutional neural network provided by the first aspect, so that the beneficial technical effects that can be achieved by the solid mode-locked laser control system can be seen in the first aspect, and the description thereof is omitted.

Drawings

In order to more clearly illustrate the technical solution of the present application, the drawings that are needed in the embodiments will be briefly described below, and it will be obvious to those skilled in the art that other drawings can be obtained from these drawings without inventive effort.

Fig. 1 is a schematic flow chart of a solid mode-locked laser control method based on a twin convolutional neural network according to an embodiment of the present application;

FIG. 2 is a schematic diagram of a twin convolutional neural network according to an embodiment of the present application;

Fig. 3 is a schematic diagram of a first spectrum signal, a first timing signal, and a first radio frequency spectrum signal of a solid-state mode-locked laser in different states according to an embodiment of the present application;

FIG. 4 is a schematic diagram of a second spectrum signal, a second timing signal, and a second RF spectrum signal according to an embodiment of the present application;

FIG. 5 is a flow chart of data processing in a specific implementation provided by an embodiment of the present application;

Fig. 6 is a schematic structural diagram of a solid mode-locked laser control system based on a twin convolutional neural network according to an embodiment of the present application.

The graphic indicia:

100-a solid mode-locked laser control system based on a twin convolutional neural network; 10-a pump source; a 20-X endoscope assembly; 21-a first endoscope; 22-a second endoscope; 23-a third endoscope; 24-semiconductor saturation absorption mirror; 25-an output mirror; 30-laser crystals; 40-a detector; 50-a data acquisition card; a 60-processor; 70-state controller; 80-a first lens; 90-second lens.

Detailed Description

The technical solutions in the embodiments of the present application will be clearly described below with reference to the drawings in the embodiments of the present application. It will be apparent that the described embodiments are some, but not all, embodiments of the application. Based on the embodiments of the present application, other embodiments that may be obtained by those of ordinary skill in the art without making any inventive effort are within the scope of the present application.

Hereinafter, the terms "first," "second," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first", "a second", etc. may explicitly or implicitly include one or more such feature. In the description of the present application, unless otherwise indicated, the meaning of "a plurality" is two or more.

Furthermore, in the present application, the terms "upper," "lower," "inner," "outer," and the like are defined relative to the orientation in which the components are schematically depicted in the drawings, and it should be understood that these directional terms are relative concepts, which are used for descriptive and clarity relative thereto, and which may be varied accordingly with respect to the orientation in which the components are depicted in the drawings.

In order to solve the technical problems, the embodiment of the application provides a solid mode-locked laser control method based on a twin convolutional neural network. Referring to fig. 1, the solid mode-locked laser control method based on the twin convolutional neural network provided by the embodiment of the application is applied to a solid mode-locked laser control system based on the twin convolutional neural network, and the control method can be realized by the following steps S100-S800.

Step S100: and establishing a database and a twin convolutional neural network.

The step S100 may be implemented by establishing a database and establishing a twin convolutional neural network, and specifically may include the following steps S101 and S102.

Step S101: and establishing a database.

Step S102: and (5) establishing a twin convolutional neural network.

The step S101 and the step S102 are not in a fixed order, and may be performed in the order of the step S101 to the step S102, or may be performed in the order of the step S102 to the step S101.

In some possible implementations, step S101 may be implemented by the following steps S101a to S101 b.

Specifically, the twin convolutional neural network is a one-dimensional network, and the input data is one-dimensional data.

Step S101a: training data is acquired.

The training data is one or more of a first time sequence signal, a first radio frequency spectrum signal or a first spectrum signal of the solid mode-locked laser in different working states, and a first characteristic parameter corresponding to one or more of the first time sequence signal, the first radio frequency spectrum signal or the first spectrum signal. The working state comprises one of a non-light-emitting state, an output continuous light state, a Q-switched state, an incomplete mode locking state and a stable mode locking state.

Specifically, signals in a continuous light state, a Q-switched state, an incomplete mode locking state and a stable mode locking state are output and are collected when an X-cavity mirror assembly of the solid mode locking laser is in a stable region range, and signals in an uncorrupted state are collected when the X-cavity mirror assembly of the solid mode locking laser is in an unstable region range.

For example, 10000 sets may be recorded separately in different operating states.

At the beginning of acquiring training data, a solid mode-locked laser control system based on a twin convolutional neural network can be built, so that the acquired data can be conveniently processed and fed back.

Step S101b: and extracting features of the training data to obtain a database.

Specifically, one or more of the acquired first timing signal, the first radio frequency spectrum signal or the first spectrum signal is subjected to signal framing for feature extraction to obtain a first transient state feature sequence so as to establish a database. Wherein the first transient characteristic sequence corresponds to the first characteristic parameter.

The principle of feature extraction is not particularly limited in the present application.

For example, when the feature extraction is performed on the first timing signal, the principles of root mean square, kurtosis, skewness and the like of the time domain waveform can be adopted.

When the first radio frequency spectrum signal is subjected to feature extraction, the principles of spectrum features such as power spectrum density, frequency distribution and the like can be adopted.

When the characteristic extraction is performed on the first spectrum signal, the principles of peak wavelength, wavelength range, spectrum width and the like can be adopted.

The embodiment of the application does not limit the feature extraction, and any of the feature extraction principles described above may be adopted, or other principles other than those described above may be adopted.

In some possible implementations, the first timing signal, the first radio frequency spectrum signal, or the first spectral signal may be preprocessed prior to feature extraction of one or more of the first timing signal, the first radio frequency spectrum signal, or the first spectral signal, where the preprocessing may include denoising, normalization, or the like.

Step S200: training a twin convolutional neural network using a database.

Specifically, the twin convolutional neural network comprises a first sub-network and a second sub-network, and in the process of training the twin convolutional neural network, the training modes of the first sub-network and the second sub-network are the same. The training mode can be any training mode, and the training mode of the twin convolutional neural network is not particularly limited in the embodiment of the application.

After training is completed, the inputs of the first sub-network and the second sub-network can be the first transient characteristic sequence, and the outputs of the first sub-network and the second sub-network can be the first characteristic parameters.

In connection with fig. 2, 3, and 4, the first subnetwork and the second subnetwork may each include a first convolutional layer, a first pooling layer, a second convolutional layer, and a second pooling layer connected. Wherein, the first convolution layer and the second convolution layer can both adopt Relu activation functions, and the input of the first sub-network and the second sub-network is one-dimensional data.

Step S300: one or more of a second timing signal, a second radio frequency spectrum signal, or a second spectrum signal of the solid-state mode-locked laser is acquired.

After the twin convolutional neural network is trained, the real-time output performance of the solid mode-locked laser is further evaluated, and real-time signals are collected, wherein the real-time signals comprise a second time sequence signal, a second radio frequency spectrum signal or a second spectrum signal, and the real-time sequence signal, the real-time radio frequency spectrum signal and the real-time spectrum signal of the solid mode-locked laser are respectively.

Step S400: and converting one or more of the second time sequence signal, the second radio frequency spectrum signal or the second spectrum signal into a second transient characteristic sequence after characteristic extraction.

In step S400, a preprocessing operation may be performed on the real-time signal before performing signal framing feature extraction on the real-time signal, where the preprocessing may include denoising and normalization. Thus, the accuracy of the real-time signal can be ensured, and the measurement error is reduced while the feature extraction is convenient. And extracting the characteristics of the preprocessed signals, and converting the signals into a second transient characteristic sequence.

The feature extraction principle of the real-time signal is the same as that of the training data, so that the data uniformity is ensured.

For example, when the first timing signal uses root mean square for feature extraction, the second timing signal also uses root mean square for feature extraction.

Step S500: and inputting the second transient characteristic sequence into a second sub-network to obtain a second characteristic parameter.

Specifically, the second subnetwork is used for real-time signal service, and after the real-time signal is collected, the real-time signal is preprocessed and the characteristics are extracted to obtain a second transient characteristic sequence, the second transient characteristic sequence is input into the second subnetwork, and a second characteristic parameter is obtained. That is, the input of the second subnetwork is the second transient characteristic sequence and the output of the second subnetwork is the second characteristic parameter.

It can be understood that the first sub-network is used for outputting original data, and the second sub-network is used for outputting real-time data, so that the real-time working state of the solid mode-locked laser can be conveniently determined through the comparison result of the real-time data and the original data. The first subnetwork and the second subnetwork form a training and comparison model.

The real-time signal is the same as the first timing signal, the first radio frequency spectrum signal and the specific preprocessing method of the first spectrum signal.

Step S600: and comparing the second characteristic parameter with the first characteristic parameter to obtain the Euclidean distance.

And comparing the second characteristic parameter with the first characteristic parameter to obtain the Euclidean distance, so that the real-time working state of the solid mode-locked laser is determined through the Euclidean distance.

Specifically, the data in the database is compared with the real-time signals, the characteristic parameters output by the two sub-networks are compared, the Euclidean distance is adopted to evaluate the similarity between the real-time signals and the corresponding frames of the database, and the twin convolutional neural network is adjusted according to the result of the similarity evaluation, so that the performance and stability of the twin convolutional neural network are further improved. The laser state in which the current solid-state mode-locked laser operates can be determined according to the Euclidean distance.

It should be noted that, if the real-time signal includes multiple types, for example, the collected real-time signal includes the second timing signal and the second radio frequency spectrum signal, when the second characteristic parameters are output, two second characteristic parameters, that is, the timing second characteristic parameter and the radio frequency second characteristic parameter, are output. And in comparison, comparing the time sequence second characteristic parameter with the time sequence first characteristic parameter output by the first sub-network to obtain a time sequence Euclidean distance, and comparing the radio frequency second characteristic parameter with the radio frequency first characteristic parameter output by the first sub-network to obtain a radio frequency Euclidean distance. In other words, the real-time signal is compared with the data in the corresponding first sub-network, so that the output accuracy of the network is ensured.

Step S700: and determining that the real-time working state of the solid mode-locked laser does not comprise the stable mode-locked state in response to the Euclidean distance not being in the preset range.

And if the obtained Euclidean distance is not in the preset range, indicating that the real-time working state of the solid mode locking laser is not the stable mode locking state. The preset range represents a reference range of euclidean distance of the solid mode-locked laser in a stable mode-locked state, and specific numerical values of the preset range can be set according to the use requirement of the actual solid mode-locked laser.

Step S800: and adjusting the position of an X-cavity mirror assembly in the solid mode-locked laser according to the real-time working state, so that the solid mode-locked laser is in a stable mode-locked state.

And when the solid mode-locked laser is not in the stable mode-locked state, the solid mode-locked laser is in the stable mode-locked state by adjusting the position of the X-cavity mirror assembly.

That is, when the twin convolutional neural network evaluates that the solid mode-locked laser is in a state of not emitting light, outputting continuous light, adjusting Q, and not locking the mode completely, the position of one or two key cavity mirrors in the X-cavity mirror assembly needs to be adjusted, so that the solid mode-locked laser is located within the stable mode-locked region. Thus, the solid-state mode-locked laser can achieve stable output.

In addition, when the solid mode-locked laser is in disorder due to external disturbance and other factors, the self-feedback adjustment can be realized through the control method.

In some possible implementations, the control method further includes step S900.

Step S900: and determining that the real-time working state of the solid mode-locked laser is a stable mode-locked state in response to the Euclidean distance being in a preset range.

At this time, when the solid mode-locked laser is judged to be in the stable mode-locked state, the adjustment of the X-cavity mirror assembly is not required.

Specifically, according to the control method provided by the embodiment of the application, the working state and the performance parameters of the solid mode-locked laser are determined by analyzing and identifying the input time sequence signal, the radio frequency spectrum signal or the spectrum signal with high precision, so that the position state of the key cavity mirror is accurately adjusted. Due to the high efficiency and parallel processing capability of the convolutional neural network, the control system can monitor and regulate the output of the solid mode-locked laser in real time, so that the solid mode-locked laser can realize a stable mode-locked state and adapt to environmental changes and fluctuation of working conditions. The control method integrates the automatic analysis and adjustment functions of the convolutional neural network, can reduce manual intervention, improves the automation degree and stability of a control system, and reduces maintenance cost and labor cost. Through real-time monitoring and adjustment, the output performance of the solid mode-locked laser can be continuously adjusted and optimized, so that the solid mode-locked laser reaches the optimal working state, and the energy efficiency and the output quality are improved.

In a specific implementation manner, referring to fig. 5, the solid mode-locked laser control method based on the twin convolutional neural network provided by the embodiment of the present application may be implemented by the following steps S1 to S7.

Step S1: spectrum, time sequence and radio frequency spectrum signals are acquired.

This step may correspond to step S300 described above.

Step S2: and (5) data acquisition.

Step S3: input data and processing, denoising, normalization and the like.

This step may be a preprocessing process for the real-time signal.

Step S4: and constructing and training a twin convolutional neural network.

This step may be a part of the steps in step S100 and step S200 described above.

It should be noted that, in this implementation, the step S300 is performed first, and then the step S100 and the step S200 are performed, which can be understood as a process of continuously acquiring the real-time signal, and the real-time signal is continuously acquired, so as to implement real-time adjustment of the unstable mode locking state.

Step S5: and (5) feature comparison.

Step S6: and (5) similarity analysis.

Step S5 and step S6 may correspond to step S600 described above.

Step S7: and (5) generating a control signal and regulating in real time.

This step may correspond to step S800 described above.

The present application also provides an embodiment of a solid-state mode-locked laser control system 100 based on a twin convolutional neural network, corresponding to the embodiment of the control method described above. The solid-state mode-locked laser control system 100 based on the twin convolutional neural network adopts the solid-state mode-locked laser control method based on the twin convolutional neural network provided by the embodiment.

Referring to fig. 6, the solid-state mode-locked laser control system 100 based on a twin convolutional neural network includes a pump source 10, an X-cavity mirror assembly 20, a laser crystal 30, a detector 40, a data acquisition card 50, a processor 60, and a state controller 70.

The pump source 10 is used to generate light, which carries energy. Specifically, the pump source 10 is used to provide energy to excite atoms or ions of the laser crystal 30 so that it is in a higher energy state in preparation for emission of stimulated radiation.

The X-ray cavity mirror assembly 20 is disposed on an optical path of light, and the X-ray cavity mirror assembly 20 includes a first cavity mirror 21, a second cavity mirror 22, a third cavity mirror 23, a semiconductor saturation absorption mirror 24, and an output mirror 25.

Specifically, the first cavity mirror 21, the second cavity mirror 22, the third cavity mirror 23, the semiconductor saturation absorption mirror 24 and the output mirror 25 form a typical X-shaped locking cavity, a solid mode locking laser is designed based on an ABCD transmission matrix, and a resonant cavity with two arms of equal length is built, so that no interval or split exists between two stable regions, and the stable region range is relatively large. Among other things, the semi-conductor saturation absorbing mirror 24 is a critical component for achieving and stabilizing mode locking operation, and the semi-conductor saturation absorbing mirror 24 is capable of enabling mode locking by its nonlinear absorption characteristics. When the laser intensity is low, the semiconductor saturation absorber 24 exhibits high absorption, so that noise and unstable components in the laser cavity are absorbed. As the laser intensity increases to a certain threshold, the absorption of the semiconductor saturation absorption mirror 24 decreases and the transmittance increases, thereby promoting the formation of pulses of higher intensity and initiating the mode locking process. The semiconductor saturation absorber 24 can also help maintain stability of the mode-locked state. Due to its saturated absorption characteristics, it is possible to reduce absorption when the laser pulse reaches a certain intensity, allowing a high intensity pulse to propagate stably within the cavity. This ability to dynamically adjust absorption helps to suppress pulse fluctuations within the cavity, maintaining pulse stability and consistency.

A laser crystal 30 is disposed in the X-ray cavity mirror assembly 20, and the laser crystal 30 is configured to receive energy from the light and convert it into laser light. Specifically, the positions of the first and second mirrors 21 and 22 and the laser crystal 30 are adjusted, and the laser crystal 30 is placed near the center of the calculated stable region range. Thus, the laser light generated by the excitation of the laser crystal 30 can be directly transmitted to the first and second mirrors 21 and 22, and reflected in the X-lock cavity by the first and second mirrors 21 and 22.

The laser crystal 30 is used as a laser medium, and can absorb the energy of the pump source 10, and electrons are excited to a high energy state. Then, photons having a specific wavelength are emitted by the stimulated radiation returning to the ground state, forming a laser. The X-ray cavity mirror assembly 20 functions to form an optical resonant cavity that passes through the lasing medium multiple times by reflecting photons, increasing the probability of stimulated radiation and thus amplifying the light intensity.

The detector 40 is disposed on the output optical path of the X-ray cavity mirror assembly 20, and is configured to receive the laser light reflected by the X-ray cavity mirror assembly 20. Specifically, the detector 40 is disposed on the output optical path of the output mirror 25 so as to receive the laser light emitted from the output mirror 25. Among the lasers received by detector 40 are timing signals, radio frequency spectrum signals, and spectral signals that may be indicative of the operating state of the solid-mode locked laser.

The data acquisition card 50 is connected with the detector 40 and is used for acquiring data in the laser, wherein the data are a time sequence signal, a radio frequency spectrum signal and a spectrum signal.

In some possible implementations, the data acquisition card 50 includes a first acquisition module and a second acquisition module. The first acquisition module is used for acquiring a first time sequence signal, a first radio frequency spectrum signal or a first spectrum signal of the solid mode-locked laser in different working states and a first characteristic parameter corresponding to the first time sequence signal, the first radio frequency spectrum signal or the first spectrum signal; the working state comprises one of a non-light-emitting state, an output continuous light state, a Q-switched state, an incomplete mode locking state and a stable mode locking state.

The signals in the continuous light state, the Q-switched state, the incomplete mode locking state and the stable mode locking state are output and are collected when the X-cavity mirror assembly of the solid mode locking laser is in the stable region range, and the signals in the non-light-emitting state are collected when the X-cavity mirror assembly of the solid mode locking laser is in the unstable region range.

The second acquisition module is configured to acquire a second time sequence signal, a second radio frequency spectrum signal or a second spectrum signal of the solid-state mode-locked laser; the second time sequence signal, the second radio frequency spectrum signal or the second spectrum signal is a real-time sequence signal, a real-time radio frequency spectrum signal or a real-time spectrum signal of the solid mode-locked laser respectively.

That is, the first acquisition module is configured to perform step S101a in the above-described control method embodiment. The second acquisition module is configured to execute step S300 in the above-described control method embodiment.

The processor 60 is connected to the data acquisition card 50 for processing the data, determining the real-time operating state of the solid-state mode-locked laser according to the processed data, and transmitting the real-time operating state to the state controller 70.

Specifically, the processor 60 includes a setup module, a training module, a comparison module, and a determination module.

The building module is used for building the twin convolutional neural network. The twin convolutional neural network comprises a first sub-network and a second sub-network; that is, the setup module is configured to perform step S102 in the above-described control method embodiment.

The training module is used for training the twin convolutional neural network by using the training library; the training modes of the first sub-network and the second sub-network are the same; the input of the first subnetwork is a first transient characteristic parameter, and the first transient characteristic parameter is obtained by extracting one or more of a first timing signal, a first radio frequency spectrum signal and a first spectrum signal through characteristics; the output of the first subnetwork is a first characteristic parameter; the input of the second subnetwork is a second transient characteristic parameter; the second transient characteristic parameter is obtained by extracting one or more of a second time sequence signal, a second radio frequency spectrum signal and a second spectrum signal, and the output of the second subnetwork is the second characteristic parameter; that is, the training module is configured to perform step S200 in the above-described control method embodiment.

Specifically, the first subnetwork and the second subnetwork may each include a first convolution layer, a first pooling layer, a second convolution layer, and a second pooling layer that are sequentially connected; wherein the first convolution layer and the second convolution layer each employ Relu activation functions.

The comparison module is used for comparing Euclidean distance between the second characteristic parameter and the first characteristic parameter; that is, the comparison module is configured to perform step S600 in the above-described control method embodiment.

And the determining module is used for determining that the real-time working state of the solid mode-locked laser does not comprise the stable mode-locked state in response to the Euclidean distance not being in the preset range. That is, the determining module is configured to perform step S700 in the above-described control method embodiment.

The determining module is further used for determining that the real-time working state of the solid mode-locked laser is a stable mode-locked state in response to the Euclidean distance being in a preset range. That is, the determining module is further configured to perform step S900 in the above-described control method embodiment.

One end of the state controller 70 is connected to the processor 60 for adjusting the setting positions of the second cavity mirror 22 and the output mirror 25 in the X-cavity mirror assembly 20 according to the real-time operation state when the real-time operation state does not include the stable mode-locking state, thereby placing the solid mode-locked laser in the stable mode-locking state.

Specifically, the location may include parameters such as coordinates, angles, and the like.

The other end of the state controller 70 is connected with the second cavity mirror 22 and the output mirror 25, when the solid mode-locked laser control system 100 based on the twin convolutional neural network judges that the solid mode-locked laser is not in a stable mode-locked state, the second cavity mirror 22 and the output mirror 25 which are driven by the electric drive can be adjusted in a self-adaptive mode, and the setting positions of the second cavity mirror 22 and the output mirror 25 are changed to enable the X-shaped mode-locked cavity to be in a stable mode-locked area range, so that the solid mode-locked laser can realize stable output at the moment, and the control of the solid mode-locked laser is completed.

With continued reference to fig. 6, the solid-state mode-locked laser control system 100 based on a twin convolutional neural network provided in an embodiment of the present application further includes a first lens 80 and a second lens 90, where the first lens 80 and the second lens 90 are sequentially disposed between the pump source 10 and the first cavity mirror 21. By providing the first lens 80 and the second lens 90, light can be transmitted to the first cavity mirror 21 better.

In the solid mode-locked laser control system 100 based on the twin convolutional neural network provided by the embodiment of the application, the twin convolutional neural network can perform high-precision analysis and identification on the input time sequence signal, the radio frequency spectrum signal or the spectrum signal, so that the working state and the performance parameters of the solid mode-locked laser are determined, and the position states of the second cavity mirror 22 and the output mirror 25 are accurately adjusted. Due to the high efficiency and parallel processing capability of the twin convolutional neural network, the solid mode-locked laser control system 100 based on the twin convolutional neural network can monitor and regulate the output of the solid mode-locked laser in real time, so that the solid mode-locked laser can realize a stable mode-locked state, and adapt to environmental changes and fluctuation of working conditions. The twin convolutional neural network can dynamically adjust and feed back to the solid mode-locked laser control system 100 based on the twin convolutional neural network according to the data collected in real time, so that the solid mode-locked laser control system 100 based on the twin convolutional neural network has self-adaptability, and the output stability and the excellent performance are ensured. The application integrates the automatic analysis and adjustment functions of the twin convolutional neural network, can reduce manual intervention, improves the automation degree and stability of the solid mode-locked laser control system 100 based on the twin convolutional neural network, and reduces maintenance cost and labor cost. Through real-time monitoring and adjustment, the solid mode-locked laser control system 100 based on the twin convolutional neural network can continuously adjust and optimize the output performance of the solid mode-locked laser, so that the solid mode-locked laser reaches the optimal working state, and the energy efficiency and the output quality are improved. Meanwhile, the solid mode-locked laser control system 100 based on the twin convolutional neural network adaptively adjusts the mode-locked pulse output of the solid mode-locked laser in real time by utilizing a multi-dimension and multi-angle laser database, so that the mode-locked stability and robustness are improved, and the redundant operation of the solid mode-locked laser in the implementation is improved.

It is noted that other embodiments of the application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. This application is intended to cover any variations, uses, or adaptations of the application following, in general, the principles of the application and including such departures from the present disclosure as come within known or customary practice within the art to which the application pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope of the application being indicated by the following claims.

It is to be understood that the application is not limited to the precise arrangements and instrumentalities shown in the drawings, which have been described above, and that various modifications and changes may be effected without departing from the scope thereof. The scope of the application is limited only by the appended claims.

Claims

1. The solid mode-locked laser control method based on the twin convolutional neural network is characterized by comprising the following steps of:

Establishing a database and a twin convolutional neural network; the system comprises a database, a twin convolutional neural network and a storage unit, wherein the database comprises a first transient characteristic sequence and a first characteristic parameter corresponding to the first transient sequence, and the twin convolutional neural network comprises a first sub-network and a second sub-network;

Training the twin convolutional neural network using the database; the training modes of the first sub-network and the second sub-network are the same; the input of the first subnetwork is the first transient state characteristic sequence, the output of the first subnetwork is the first characteristic parameter, the first transient state characteristic sequence is obtained by extracting one or more of a first time sequence signal, a first radio frequency spectrum signal or a first spectrum signal of the solid mode-locked laser in different working states, and the working states comprise one of a non-light-emitting state, an output continuous light state, a Q-switched state, an incomplete mode-locked state and a stable mode-locked state;

Acquiring one or more of a second time sequence signal, a second radio frequency spectrum signal or a second spectrum signal of the solid mode-locked laser; the second time sequence signal, the second radio frequency spectrum signal or the second spectrum signal is a real-time sequence signal, a real-time radio frequency spectrum signal or a real-time spectrum signal of the solid mode-locked laser respectively;

Converting one or more of the second time sequence signal, the second radio frequency spectrum signal or the second spectrum signal into a second transient characteristic sequence after characteristic extraction;

inputting the second transient characteristic sequence into the second sub-network to obtain a second characteristic parameter;

Comparing the second characteristic parameter with the first characteristic parameter to obtain Euclidean distance;

Determining that the real-time working state of the solid mode-locked laser does not comprise the stable mode-locked state in response to the euclidean distance not being within a preset range;

and adjusting the position of an X-cavity mirror assembly in the solid mode-locked laser according to the real-time working state, so that the solid mode-locked laser is in the stable mode-locked state.

2. The solid-state mode-locked laser control method based on a twin convolutional neural network according to claim 1, further comprising:

and determining that the real-time working state of the solid mode-locked laser is the stable mode-locked state in response to the Euclidean distance being in the preset range.

3. The method for controlling a solid-state mode-locked laser based on a twin convolutional neural network as claimed in claim 1,

The first sub-network and the second sub-network comprise a first convolution layer, a first pooling layer, a second convolution layer and a second pooling layer which are sequentially connected; wherein the first convolution layer and the second convolution layer each employ Relu activation functions.

4. A solid-state mode-locked laser control system based on a twin convolutional neural network, characterized in that a solid-state mode-locked laser control method based on a twin convolutional neural network as claimed in any one of claims 1 to 3 is adopted, the solid-state mode-locked laser control system based on a twin convolutional neural network comprising:

a pump source configured to generate light; wherein the light carries energy;

The X-cavity mirror assembly is arranged on the light path of the light;

A laser crystal disposed in the X-cavity mirror assembly configured to absorb the energy and generate a laser light; wherein the X-cavity assembly mirror is configured to reflect the light rays to the laser crystal, receive and reflect the laser light converted by the laser crystal;

the detector is arranged on the output light path of the X-cavity mirror assembly and is configured to receive the laser reflected by the X-cavity mirror assembly;

the data acquisition card is connected with the detector and is configured to acquire data in the laser;

the processor is connected with the data acquisition card and is configured to process the data, determine and output the real-time working state of the solid mode-locked laser;

And the state controller is connected with the processor and is configured to respond to the real-time working state without the stable mode locking state, and adjust the position of the X-cavity mirror assembly according to the real-time working state so that the solid mode locking laser is in the stable mode locking state.

5. The solid state mode locked laser control system based on a twin convolutional neural network of claim 4,

The X-cavity mirror assembly comprises a first cavity mirror, a second cavity mirror, a third cavity mirror, a semiconductor saturation absorption mirror and an output mirror; wherein the laser crystal is disposed between the first and second mirrors.

6. The solid state mode locked laser control system based on a twin convolutional neural network of claim 5,

Adjusting the position of the X-ray endoscope assembly according to the real-time operating state includes:

7. The solid state mode locked laser control system based on a twin convolutional neural network of claim 5,

The solid mode-locked laser control system based on the twin convolutional neural network further comprises: the first lens and the second lens are sequentially arranged between the pumping source and the first cavity mirror.

8. The solid state mode locked laser control system based on a twin convolutional neural network of claim 4,

The data acquisition card comprises a first acquisition module and a second acquisition module, wherein the first acquisition module is configured to acquire a first timing signal, a first radio frequency spectrum signal or a first spectrum signal of the solid mode-locked laser in different working states, and the working states comprise one of a non-light emitting state, an output continuous light state, a Q-switched state, a non-complete mode-locked state and a stable mode-locked state; the second acquisition module is configured to acquire one or more of a second timing signal, a second radio frequency spectrum signal or a second spectrum signal of the solid-state mode-locked laser; the second time sequence signal, the second radio frequency spectrum signal or the second spectrum signal is a real-time sequence signal, a real-time radio frequency spectrum signal or a real-time spectrum signal of the solid mode-locked laser respectively;

The processor includes:

A building module configured to build a database and a twin convolutional neural network; wherein the twin convolutional neural network comprises a first sub-network and a second sub-network; the database comprises a first transient state characteristic parameter, wherein the first transient state characteristic parameter is obtained by extracting one or more of the first timing signal, the first radio frequency spectrum signal or the first spectrum signal;

A training module configured to train the twin convolutional neural network using the database; the training modes of the first sub-network and the second sub-network are the same; the input of the first sub-network is the first transient characteristic parameter, and the output of the first sub-network is the first characteristic parameter; the input of the second sub-network is a second transient characteristic parameter; the second transient characteristic parameter is obtained by extracting one or more of the second time sequence signal, the second radio frequency spectrum signal or the second spectrum signal, and the output of the second sub-network is a second characteristic parameter;

a comparison module configured to compare euclidean distances of the second characteristic parameter and the first characteristic parameter;

and the determining module is configured to determine that the real-time working state of the solid mode-locked laser does not comprise the stable mode-locked state in response to the Euclidean distance not being in a preset range.

9. The solid state mode locked laser control system based on the twin convolutional neural network of claim 8,

And the determining module is further configured to determine that the real-time working state of the solid mode-locked laser is the stable mode-locked state in response to the euclidean distance being in the preset range.

10. The solid state mode locked laser control system based on the twin convolutional neural network of claim 8,