Laser oscillation method, laser oscillation system, generation method, generation device, and estimation program

Abstract

Provided is a laser oscillation method in which a SLM is disposed on a resonance path and a laser medium is irradiated with excitation light, whereby laser light is caused to oscillate in the resonance path, the laser oscillation method comprising: a data acquisition step of irradiating the laser medium with the excitation light and acquiring data related to the current oscillation; an estimation step of estimating, using an estimation program for outputting a parameter related to a phase pattern displayed on the SLM in response to the entry of the data related to the oscillation and information indicating a target oscillation state, a parameter related to the phase pattern according to the target oscillation state and the data related to the current oscillation acquired in the data acquisition step; and a control step of controlling the phase pattern displayed on the SLM on the basis of the parameter estimated in the estimation step.

Description

Laser oscillation method, laser oscillation system, generation method, generation device, and estimation program

[0001] The present disclosure relates to a laser oscillation method, a laser oscillation system, a generation method, a generation device, and an estimation program.

[0002] Patent Document 1 discloses a configuration in which a spatial light modulator is used as a reflecting mirror in a laser oscillation system.

[0003] JP 2009-289990 A

[0004] Here, in order to realize a desired oscillation state, it is difficult to manually create a phase pattern (modulation pattern) to be displayed on the spatial light modulator with high precision in a short period of time.

[0005] The present disclosure has been made in consideration of the above-described circumstances, and aims to easily and accurately create a phase pattern that realizes a desired oscillation state.

[0006] A laser oscillation method according to one aspect of the present disclosure is a laser oscillation method in which a spatial light modulator is placed on a resonance path and laser light is oscillated within the resonance path by irradiating a laser medium with excitation light, the method including: a data acquisition step of irradiating the laser medium with excitation light and acquiring data related to current oscillation; an estimation step of estimating parameters related to the phase pattern according to the data related to the current oscillation and a target oscillation state acquired in the data acquisition step using an estimation program that outputs parameters related to a phase pattern to be displayed on the spatial light modulator in response to input of the oscillation data and information indicating a target oscillation state; and a control step of controlling the phase pattern displayed on the spatial light modulator based on the parameters estimated in the estimation step.

[0007] In a laser oscillation method according to one aspect of the present disclosure, excitation light is irradiated onto a laser medium to acquire data related to current oscillation within a resonant path, and parameters related to a phase pattern are estimated by an estimation program based on the data related to the current oscillation and a target oscillation state. Then, a phase pattern displayed on a spatial light modulator is controlled based on the parameters. In this way, a phase pattern can be easily created by automatically controlling the phase pattern using a pre-prepared estimation program. Furthermore, because the estimation program receives inputs of data related to the current oscillation and a target (desired) oscillation state, a phase pattern that achieves a desired oscillation state can be created with high accuracy. As described above, the laser oscillation method according to one aspect of the present disclosure allows a phase pattern that achieves a desired oscillation state to be easily created with high accuracy.

[0008] In the data acquisition step, data relating to a non-oscillation state may be acquired as data relating to the current oscillation, and in the estimation step, parameters relating to the phase pattern may be estimated so as to transition from the non-oscillation state to the oscillation state. With this configuration, it is possible to easily and accurately create a phase pattern that appropriately transitions from a non-oscillation state to a desired oscillation state.

[0009] In the data acquisition step, data related to the current oscillation may be acquired as data related to the fluorescence emitted from the laser medium. With this configuration, even in a state before oscillation, data related to oscillation can be appropriately acquired from, for example, the intensity of amplified spontaneous emission (ASE). This makes it possible to create a phase pattern that achieves a desired oscillation state with higher accuracy.

[0010] In the data acquisition step, data related to an oscillation state may be acquired as data related to the current oscillation, and in the estimation step, parameters related to the phase pattern may be estimated so as to transition from the current oscillation state to another oscillation state. With this configuration, modulation parameters that appropriately transition from an oscillation state to a desired other oscillation state can be created easily and with high accuracy.

[0011] In the data acquisition step, data related to the laser oscillation light may be acquired as the data related to the current oscillation. With this configuration, data related to the oscillation can be appropriately acquired while the laser is oscillating. This makes it possible to create a phase pattern that realizes a desired oscillation state with higher accuracy.

[0012] The laser oscillation method may further include a learning step of generating an estimation program by learning a plurality of data sets in which first learning data related to a phase pattern displayed on the spatial light modulator and second learning data related to oscillation are associated with each other. In this manner, the accuracy of the estimation program can be improved by associating and learning a large amount of data related to oscillation (e.g., ASE intensity, a beam image of laser oscillation light, etc.) with the phase pattern. In other words, the provision of the learning step makes it possible to generate an estimation program with high accuracy for estimating parameters related to the phase pattern, thereby enabling the creation of a phase pattern that achieves a desired oscillation state with high accuracy.

[0013] The second learning data may include at least one of a beam image, intensity, spectrum, time waveform of the laser oscillation light, the type of laser medium, fluorescence emitted from the laser medium, information on the excitation light, the temperature in the data acquisition environment, and the position of the optical system. With this configuration, it is possible to generate an estimation program that can estimate parameters related to the phase pattern with high accuracy.

[0014] In the learning step, the data related to the current oscillation acquired in the data acquisition step may be used as second learning data. With this configuration, the learning step is incorporated into (performed simultaneously with) other steps related to oscillation control, making it possible to update the estimation program so as to improve its accuracy while performing oscillation control.

[0015] A laser oscillation system according to one aspect of the present disclosure is a laser oscillation system that oscillates laser light within a resonance path by irradiating a laser medium with excitation light, and includes: a spatial light modulator that is arranged on the resonance path and displays an arbitrary phase pattern; an acquisition unit that acquires data related to current oscillation; an estimation unit that estimates parameters related to the phase pattern according to the data related to current oscillation and a target oscillation state acquired by the acquisition unit using an estimation program that outputs parameters related to the phase pattern to be displayed on the spatial light modulator in response to input of the data related to oscillation and information indicating a target oscillation state; and a control unit that controls the phase pattern displayed on the spatial light modulator based on the parameters estimated by the estimation unit.

[0016] The acquisition unit may acquire data relating to a non-oscillation state as data relating to the current oscillation, and the estimation unit may estimate parameters relating to the phase pattern so as to transition from the non-oscillation state to the oscillation state.

[0017] The acquisition unit may acquire data related to fluorescence emitted from the laser medium as data related to the current oscillation.

[0018] The acquisition unit may acquire data relating to an oscillation state as data relating to the current oscillation, and the estimation unit may estimate parameters relating to the phase pattern so as to transition from the current oscillation state to another oscillation state.

[0019] The acquisition unit may acquire data related to laser oscillation light as data related to the current oscillation.

[0020] The laser oscillation system may further include a learning unit that generates an estimation program by learning a plurality of data sets that associate first learning data related to a phase pattern displayed on the spatial light modulator with second learning data related to oscillation.

[0021] The second learning data may include at least one of a beam image, intensity, spectrum, time waveform of the laser oscillation light, type of laser medium, fluorescence emitted from the laser medium, information on the excitation light, temperature in the data acquisition environment, and the position of the optical system.

[0022] The learning unit may set the data relating to the current oscillation acquired by the acquisition unit as the second learning data.

[0023] A generation method according to one aspect of the present disclosure is a method for generating an estimation program that outputs parameters related to a phase pattern displayed on a spatial light modulator on a resonant path, and includes a learning data collection step of collecting multiple pieces of first learning data related to the phase pattern displayed on the spatial light modulator and second learning data related to oscillation within the resonant path, and a learning step of generating the estimation program by learning multiple data sets in which the first learning data and the second learning data collected in the learning data collection step correspond to each other.

[0024] A generation device according to one aspect of the present disclosure is a generation device for an estimation program that outputs parameters related to a phase pattern displayed on a spatial light modulator on a resonant path, and includes a collection unit that collects multiple pieces of first learning data related to the phase pattern displayed on the spatial light modulator and second learning data related to oscillation within the resonant path, and a learning unit that generates the estimation program by learning multiple data sets that associate the first learning data with the second learning data collected by the collection unit.

[0025] An estimation program according to one aspect of the present disclosure is an estimation program that outputs parameters related to a phase pattern displayed on a spatial light modulator arranged on a resonator, and causes a computer to execute an acquisition process that acquires data related to current oscillation detected by irradiating a laser medium with excitation light, and an estimation process that outputs parameters related to the phase pattern displayed on the spatial light modulator based on the data related to the current oscillation acquired in the acquisition process and information indicating a target oscillation state.

[0026] According to one aspect of the present disclosure, a phase pattern that realizes a desired oscillation state can be created easily and with high accuracy.

[0027] FIG. 1 is a diagram for explaining an overview of a laser oscillation system according to an embodiment. FIG. 2 is a diagram schematically illustrating the configuration of a laser oscillation system. FIG. 3 is a diagram illustrating an example of a phase pattern to be displayed on an SLM. FIG. 4 is a flowchart illustrating the procedure of an oscillation characteristic phase. FIG. 5 is a flowchart illustrating the procedure of an oscillation phase. FIG. 6 is a flowchart illustrating the procedure of a learning phase. FIG. 7 is a flowchart illustrating the procedure when a learning phase is performed during an oscillation characteristic phase. FIG. 8 is a flowchart illustrating the procedure when transitioning from an oscillation phase to an oscillation characteristic phase. FIG. 9 is a diagram schematically illustrating the configuration of a laser oscillation system according to a modified example.

[0028] Hereinafter, this embodiment will be described in detail with reference to the drawings. In each drawing, the same or corresponding parts are designated by the same reference numerals, and duplicated explanations will be omitted.

[0029] First, an overview of a laser oscillation system according to this embodiment will be described with reference to FIG. 1 . FIG. 1 is a diagram illustrating an overview of the laser oscillation system according to this embodiment. The laser oscillation system according to this embodiment uses a laser oscillator configuration in which a spatial light modulator (SLM) is arranged on the resonance path, and achieves a desired oscillation state by controlling the phase pattern displayed on the SLM using machine learning technology. While laser oscillators using general optical elements have limitations on resonance conditions, a laser oscillator using an SLM can achieve various resonance conditions that are difficult to achieve using general optical elements by adjusting the phase pattern displayed on the SLM. There are a wide variety of phase patterns that can be displayed on an SLM. By performing machine learning using a large amount of data as training data, such a wide variety of phase patterns can be easily created with high accuracy. This laser oscillation system can achieve new oscillation characteristics that were previously unachievable.

[0030] Specifically, as shown in FIG. 1 , an estimation program is prepared in advance by machine learning, with various data related to the oscillation of a laser oscillator as input parameters and parameters related to a phase pattern displayed on an SLM as output parameters. The oscillation-related data here refers to information such as the intensity (output power) of the laser light, the beam intensity distribution, the ASE intensity, the spectrum, and the fluorescence waveform. Then, by inputting the data related to the oscillation of the laser oscillator as input parameters into the estimation program, parameters related to a phase pattern corresponding to the oscillation-related data are output as output parameters, and a phase pattern corresponding to the output parameters is displayed on the SLM. By repeating this process, the phase pattern displayed on the SLM can be appropriately adjusted using actual measured values ​​of the oscillation-related data, thereby obtaining desired oscillation characteristics.

[0031] Fig. 2 is a diagram schematically illustrating the configuration of the laser oscillation system 1. As shown in Fig. 2, the laser oscillation system 1 includes a laser oscillator 10, a detection unit 20, a control unit 30, a computer 40 (acquisition unit, estimation unit, learning unit), and a monitor 50.

[0032] The laser oscillator 10 includes a laser diode 11, a mirror 12, a laser medium 13, a quarter-wave plate 14, a polarizing beam splitter 15, an SLM 16, and a mirror 17. As described above, the laser oscillator 10 has an optical configuration in which one of the two resonator mirrors included in the configuration is replaced with the SLM 16. The laser oscillator 10 oscillates laser light within a resonant path by irradiating the laser medium 13 with excitation light.

[0033] The laser diode 11 is a light source that irradiates the laser medium 13 with excitation light. The laser diode 11 irradiates the laser medium 13 with excitation light of an oscillation wavelength that corresponds to the laser medium 13, for example, excitation light of 940 nm. Note that other light sources may be used instead of the laser diode 11. The excitation light emitted from the laser diode 11 is irradiated onto the laser medium 13 via the mirror 12.

[0034] The laser medium 13 is a material that amplifies light by causing stimulated emission at a rate exceeding the light absorption rate during laser oscillation. The laser medium 13 may be made of, for example, a rod-shaped Yb:YaG ceramic. The laser medium 13 may also be made of other solids such as Nd:YAG or YVO4, or gases such as He—Ne gas, Ar gas, or CO2 gas, or an organic solvent-based liquid containing dye molecules, or a semiconductor. The light amplified by the laser medium 13 reaches the SLM 16 via polarization optical elements such as a quarter-wave plate 14 and a polarizing beam splitter 15.

[0035] The SLM 16 is disposed on the resonant path, has an input unit for a control signal, and displays a phase pattern based on the control signal. The SLM 16 may modulate light passing through the resonant path by displaying the phase pattern. The SLM 16 is, for example, a reflective liquid crystal on silicon (LCOS) spatial light modulator. Parameters for the phase pattern are generated, for example, by a computer 40, and the SLM 16 is displayed in response to a control signal from the control unit 30. The phase pattern may be, for example, a hologram pattern. Note that "disposed on the resonant path" includes not only cases where the SLM 16 is disposed inside the resonant path (for example, when a transmissive SLM 16 is used), but also cases where the SLM 16 is disposed as a resonant mirror at the end of the resonant path.

[0036] The SLM 16 displays, for example, a lens pattern as the phase pattern. For example, a Fresnel lens pattern may be used as the lens pattern. When a Fresnel lens is used, the "parameters related to the phase pattern," which are the output parameters described above, may be three parameters: a focal length (f) indicating the radius of curvature, a horizontal axis (x) indicating the display coordinate on the SLM 16, and a vertical axis (y) indicating the display coordinate on the SLM 16. These three parameters are training data used in training to generate the estimation program described below. Specifically, oscillation-related data (e.g., ASE intensity distribution) detected by the detection unit 20 in response to changes in the three parameters is recorded, and the f, x, and y parameters are mapped to the ASE intensity and prepared as a training data set for machine learning (details will be described later). Then, a training model is prepared for determining the focal length (f) and coordinates (x, y) of the Fresnel lens so as to achieve a target oscillation state based on the data related to the oscillation state, and training is performed.

[0037] In the following description, the phase pattern is assumed to be a Fresnel lens pattern, but the present invention is not limited to this, and the lens pattern may be another lens pattern such as an aspherical lens pattern. Furthermore, the phase pattern may be a pattern created by Zernike terms, a pattern that can be drawn with sin and cos, or the like. Fig. 3 shows, as an example of a pattern to be displayed on the SLM 16, a phase pattern in which a coma aberration correction component (pattern) created using a Zernike term is superimposed on a focusing component created using a Zernike term. Note that, when the phase pattern is other than a Fresnel lens pattern, the above-described "parameters related to the phase pattern" are not limited to f, x, and y.

[0038] In the following description, the SLM 16 is a reflective liquid crystal spatial light modulator and is arranged on the resonance path, but a transmissive SLM may also be used. The location of the SLM is not limited to the resonance path, and it may be arranged on the laser diode 11 side (for excitation), or it may be arranged both on the resonance path and for excitation.

[0039] The light reflected by the SLM 16 is reflected by the polarizing beam splitter 15 and reaches the detection unit 20 via the mirror 17. In FIG. 2, the dotted line represents the excitation light, and the solid line represents the laser oscillation light. The resonance path in FIG. 2 runs from the mirror 12 to the SLM 16. Light amplification by stimulated emission is achieved by oscillating the laser between the mirror 12 and the SLM 16. A portion of the amplified laser oscillation light and the excitation light is extracted from the polarizing beam splitter 15 and directed toward the mirror 17. The mirror 12 is a flat mirror that transmits the excitation light and reflects the oscillation light. The quarter-wave plate 14 determines the polarization of the light incident on the SLM 16. The mirror 17 reflects only the laser oscillation light and transmits the excitation light. In this way, the laser oscillation light reflected by the mirror 17 is detected by the detection unit 20. The transmitted excitation light is blocked by a beam damper (not shown) located behind the mirror 17. A bandpass filter such as a high-pass filter that passes only the laser oscillation light (e.g., 1030 nm) and blocks the excitation light (e.g., 940 nm) may be used instead of the mirror 17. However, since the filter is likely to generate heat when the power of the excitation light is high, it is preferable to use the mirror 17 to transmit the excitation light.

[0040] The detection unit 20 detects laser oscillation light or fluorescence. According to this optical system, the detection unit 20 can monitor both laser oscillation light and fluorescence (spontaneous emission light) emitted from the laser medium. This is because the fluorescence has a peak wavelength at the same wavelength as the laser oscillation light (1030 nm in this embodiment). A second detection unit may be provided in addition to the detection unit 20, located close to the laser medium, to detect fluorescence emitted from the laser medium. This allows fluorescence monitoring with a relatively large signal intensity. The detection unit 20 may be any device capable of detecting changes in light before and after oscillation. For example, the detection unit 20 may be a photodiode (PD), a power meter that monitors output intensity, a spectrum analyzer, a spectroscope, a CMOS camera, a CCD camera, or the like. Information detected (monitored) by the detection unit 20 may be, for example, ASE intensity. When detecting a pre-oscillation state (non-oscillation state), the detection unit 20 may detect, for example, fluorescence intensity, a fluorescence spectrum, or a fluorescence time waveform. When detecting the state after oscillation (oscillation state), the detection unit 20 may detect, for example, the intensity and spectrum of the output laser light, the shape and intensity of the oscillation beam, the time waveform of the oscillation light, etc. The detection unit 20 outputs the detection result to the computer 40 as "data related to the current oscillation."

[0041] The computer 40 functions as an acquisition unit, an estimation unit, a learning unit, and a generation device. The computer 40 acquires information from the detection unit 20, displays the information on the monitor 50, and outputs the information to the control unit 30.

[0042] The computer 40 performs two main phases: a learning phase and an estimation phase. The estimation phase includes an oscillation phase in which a transition from a non-oscillation state to an oscillation state occurs, and an oscillation characteristics phase in which a transition from an oscillation state occurs. The processing of each phase will be described below in order.

[0043] (Learning Phase) The computer 40 generates an estimation program by learning a plurality of data sets in which first learning data related to the phase pattern displayed on the SLM 16 is associated with second learning data related to oscillation. The estimation program is a learning model that outputs parameters related to the phase pattern displayed on the SLM 16 in response to input of data related to oscillation and information indicating a target oscillation state. The estimation program is a program that causes the computer 40 to execute an acquisition process that acquires data related to current oscillation detected by irradiating the laser medium 13 with excitation light, and an estimation process that outputs parameters related to the phase pattern displayed on the SLM 16 based on the data related to the current oscillation acquired in the acquisition process and information indicating a target oscillation state.

[0044] The first learning data includes, for example, three parameters: a focal length (f) indicating the radius of curvature of the Fresnel lens pattern, a horizontal axis (x), and a vertical axis (y). The second learning data includes, for example, at least one of a beam image, intensity, spectrum, and time waveform of the laser oscillation light, the type of laser medium 13, information on the excitation light, the temperature in the data acquisition environment, and the position of the optical system. The type of the laser medium 13 may be distinguished, for example, by the discharge volume or the enclosed gas pressure if the laser medium 13 is a gas, by the molecular design if the laser medium 13 is a liquid, or by the size, concentration, and shape if the laser medium 13 is a solid. The information on the excitation light may include the current, the angle of incidence on the laser medium 13, the excitation region, etc. The temperature may be the room temperature, the cooling temperature of the excitation light source, the cooling temperature of the laser medium, etc. The position of the optical system may be the positions, angles, and arrangements of the optical elements, the SLM 16, the laser medium 13, and various measuring instruments, etc. The learning data may be a data set that associates the phase pattern displayed on the SLM 16 with data relating to oscillation, and may be selected appropriately according to the target oscillation state.

[0045] The computer 40 may execute the learning phase during an oscillation phase or an oscillation characteristics phase, which will be described later. That is, the computer 40 may generate an estimation program using the "data related to the current oscillation" acquired during the oscillation phase or the oscillation characteristics phase as the second learning data described above.

[0046] (Oscillation Phase) As described above, the oscillation phase is a phase in which the laser transitions from a pre-oscillation state (non-oscillation state) to an oscillation state. The computer 40 functions as an acquisition unit that acquires "data related to the current oscillation" from the detection unit 20. The computer 40 stores the acquired information. In the oscillation phase, the computer 40 acquires at least data related to the non-oscillation state as data related to the current oscillation. The data related to the non-oscillation state is data that enables estimation of the state of light even before oscillation, such as ASE intensity, fluorescence intensity, fluorescence spectrum, and fluorescence time waveform. In this way, the computer 40 may acquire data related to the fluorescence emitted from the laser medium 13 as data related to the non-oscillation state.

[0047] The computer 40 functions as an estimation unit that estimates parameters related to the phase pattern displayed on the SLM 16 by inputting data related to the current oscillation (data related to the non-oscillation state) and information indicating a target oscillation state into an estimation program. Here, the computer 40 estimates parameters related to the phase pattern so as to transition from a non-oscillation state to an oscillation state. The information indicating the target oscillation state is data confirming that oscillation has occurred, such as the oscillation wavelength. The parameters related to the phase pattern are, for example, three parameters: focal length (f), horizontal axis (x), and vertical axis (y), which are coefficients for forming a Fresnel lens. As an optimization method for determining the parameters related to the phase pattern, for example, the Nermidad method, grid search, gradient descent, Newton's method, Levenberg-Marquardt method, etc. may be used. Estimating the parameters related to the phase pattern may be synonymous with creating a phase pattern (e.g., a Fresnel lens pattern). The computer 40 transmits the parameters (output parameters) related to the estimated phase pattern to the control unit 30.

[0048] The control unit 30 controls the phase pattern displayed on the SLM 16 based on the output parameters estimated by the computer 40. The control unit 30 transmits a control signal to the SLM 16 so that a phase pattern according to the output parameters is displayed on the SLM 16. The SLM 16 displays the phase pattern according to the control signal, i.e., the phase pattern that transitions from a non-oscillating state to an oscillating state.

[0049] (Oscillation characteristic phase) As described above, the oscillation characteristic phase is a phase in which the laser transitions from an already oscillating state (oscillation state) to another oscillation state. The computer 40 functions as an acquisition unit that acquires "data related to the current oscillation" from the detection unit 20. The computer 40 stores the acquired information. In the oscillation characteristic phase, the computer 40 acquires at least data related to the oscillation state as data related to the current oscillation. The data related to the oscillation state is data that enables estimation of the state of light after oscillation, such as the intensity and spectrum of the output laser light, the oscillation beam shape, intensity, and the time waveform of the oscillation light. In this way, the computer 40 may acquire data related to the laser oscillation light as data related to the oscillation state.

[0050] The computer 40 functions as an estimation unit that estimates parameters related to the phase pattern to be displayed on the SLM 16 by inputting data related to the current oscillation (data related to the oscillation state) and information indicating a target oscillation state into an estimation program. Here, the computer 40 estimates parameters related to the phase pattern so as to transition from the current oscillation state to another oscillation state. The information indicating the target oscillation state here is data related to the target oscillation characteristics, such as intensity. The parameters related to the phase pattern are, for example, three parameters: focal length (f), horizontal axis (x), and vertical axis (y), which are coefficients for forming a Fresnel lens. The computer 40 transmits the estimated parameters (output parameters) related to the phase pattern to the control unit 30.

[0051] The control unit 30 controls the phase pattern displayed on the SLM 16 based on the output parameters estimated by the computer 40. The control unit 30 transmits a control signal to the SLM 16 so that a phase pattern according to the output parameters is displayed on the SLM 16. The SLM 16 displays a phase pattern according to the control signal, i.e., a phase pattern that transitions from one oscillation state to another oscillation state.

[0052] Next, the processing of each of the above-mentioned phases will be described with reference to FIGS.

[0053] 4 is a flowchart showing the procedure of the oscillation characteristic phase. As shown in FIG. 4, in the oscillation characteristic phase, first, a target oscillation state is determined in the computer 40 (step S1). The target oscillation state (e.g., target oscillation characteristics) may be obtained and determined, for example, by input from a user. The target oscillation characteristics may be, for example, maximization of intensity.

[0054] Next, the oscillation characteristics for the current state (oscillator state) are acquired in the computer 40 (step S2). The oscillation characteristics for the current state may be data relating to the current oscillation acquired from the detection unit 20, for example.

[0055] Next, in the computer 40, data relating to the current oscillation (data relating to the oscillation state) and information indicating the target oscillation state are input into an estimation program, which is a learning model, and output parameters relating to the phase pattern are estimated (step S3).

[0056] Next, the control unit 30 causes a phase pattern based on the output parameters estimated by the computer 40 to be displayed (reflected) on the SLM 16 of the oscillator (step S4).

[0057] Then, in this state, the computer 40 acquires the oscillation characteristics, specifically data relating to the current oscillation (step S5), and determines whether the target oscillation state has been achieved (for example, whether the intensity has been maximized) (step S6). If the determination result shows that the target oscillation state has not been achieved, the output parameters are predicted again using the learning model (step S3), and if the target oscillation state has been achieved, the oscillation characteristics phase is terminated.

[0058] 5 is a flowchart showing the procedure of the oscillation phase. As shown in FIG. 5, in the oscillation phase, first, data supporting oscillation (data that can confirm that oscillation has occurred) is selected in the computer 40 (step S11). The data supporting oscillation may be, for example, "that the oscillation wavelength is 1030 nm."

[0059] Next, information on the current state (oscillator state) is acquired in the computer 40 (step S12). The information on the current state is, for example, data on the current oscillation acquired from the detection unit 20, or may be data on a non-oscillation state, such as data on the fluorescence emitted from the laser medium 13 (spectrum, etc.).

[0060] Next, in the computer 40, data relating to the current oscillation (data relating to the non-oscillation state) and information indicating the target oscillation state are input into an estimation program, which is a learning model, and output parameters relating to the phase pattern are estimated (step S13).

[0061] Next, the control unit 30 causes a phase pattern based on the output parameters estimated by the computer 40 to be displayed (reflected) on the SLM 16 of the oscillator (step S14).

[0062] Then, in this state, the computer 40 acquires data relating to the current oscillation (step S15), and determines whether the oscillation state is established (for example, whether the oscillation wavelength is 1030 nm) (step S16). If the determination result shows that the oscillation state is not established, the output parameters are predicted again using the learning model (step S13). If the oscillation state is established, the oscillation phase is terminated.

[0063] 6 is a flowchart showing the procedure of the learning phase. As shown in FIG. 6, in the learning phase, first, the type of data to be used as learning data is considered (data preparation) (step S101). For example, the beam intensity distribution for the phase pattern may be used as the data in the oscillation state, and for example, the ASE intensity for the mirror angle may be used as the data in the non-oscillation state.

[0064] Next, the data determined in step S101 is collected (step S102). Here, a large amount of data necessary for learning is acquired. The large amount of data may be image data acquired by the detection unit 20, which is a camera, or intensity acquired by the detection unit 20, which is a PD.

[0065] Next, the acquired large amount of data is subjected to data preprocessing in the computer 40 (step S103). In the data preprocessing, for example, necessary data is selected, organized, and expanded. Then, the preprocessed data is divided into a training set and a validation set (step S104).

[0066] Next, a model to be used for learning is selected (step S105), where a machine learning method, learning method, and algorithm are selected.

[0067] Then, in the computer 40, multiple data sets that associate first learning data related to the phase pattern displayed on the SLM 16 with second learning data related to the oscillation are learned, and model training is performed (step S106).

[0068] Thereafter, the performance of the learning model is evaluated (step S107), and the parameters are tuned. The processes of steps S103 to S107 described above are repeated until an optimal model is obtained. Through these processes, an optimal model is completed (step S108).

[0069] 7 is a flowchart showing the procedure for implementing the learning phase during the oscillation characteristic phase. As shown in FIG. 7, in the oscillation characteristic phase, first, a target oscillation state is determined in the computer 40 (step S201). Next, the computer 40 acquires oscillation characteristics for the current state (oscillator state) (step S202).

[0070] Next, in the computer 40, data related to the current oscillation (data related to the oscillation state) and information indicating the target oscillation state are input to an estimation program, which is a learning model, and output parameters related to the phase pattern are estimated (step S203). Next, the control unit 30 displays (reflects) the phase pattern based on the output parameters estimated by the computer 40 on the SLM 16 of the oscillator (step S204).

[0071] Then, in this state, the computer 40 acquires the oscillation characteristics, specifically data relating to the current oscillation (step S205), and determines whether the target oscillation state has been achieved (for example, whether the intensity has been maximized) (step S206).If the determination result shows that the target oscillation state has not been achieved, the output parameters are predicted again using the learning model (step S203).If the target oscillation state has been achieved, the learning model is updated and the oscillation characteristics phase is terminated.

[0072] The data relating to the current oscillation acquired in step S205 is treated as second learning data and preprocessed (step S301). Then, the above-mentioned data division (step S302), model selection (step S303), model training (step S304), and model evaluation (step S305) are performed, and the optimal model is updated as needed to achieve the target oscillation state (step S306). The learning model updated in this way is used as an estimation program to perform the processes of step S203 and the like.

[0073] 8 is a flowchart showing the procedure for transitioning from the oscillation phase to the oscillation characteristics phase. As shown in FIG. 8, in the oscillation phase, first, data supporting oscillation (data confirming that oscillation has occurred) is selected in the computer 40 (step S401). The data supporting oscillation may be, for example, "ASE intensity being equal to or greater than XX."

[0074] Next, information on the current state (oscillator state) is acquired in the computer 40 (step S402). The information on the current state is, for example, data on the current oscillation acquired from the detection unit 20, or may be data on the non-oscillating state such as ASE intensity.

[0075] Next, in the computer 40, data relating to the current oscillation (data relating to the non-oscillation state) and information indicating the target oscillation state are input to an estimation program, which is a learning model, and output parameters relating to the phase pattern are estimated (step S403). Note that the number of output parameters may be, for example, two or more, and may be a phase pattern generation coefficient, an excitation current, etc.

[0076] Next, the control unit 30 reflects the output parameters estimated by the computer 40 in the oscillator (step S404). Specifically, a phase pattern may be displayed on the SLM 16, and an excitation current may be set.

[0077] Then, in this state, the computer 40 acquires data relating to the current oscillation (step S405), and determines whether the oscillation state is in effect (for example, whether the desired ASE intensity is reached) (step S406). If the determination result shows that the oscillation state is not in effect, the output parameters are predicted again using the learning model (step S403).

[0078] If the oscillation state is achieved, the process proceeds to the oscillation characteristic phase. Specifically, in the oscillation characteristic phase, the computer 40 determines a target oscillation state (step S407). The target oscillation state (e.g., the target oscillation characteristic) may be, for example, a flat-top pattern.

[0079] Next, the oscillation characteristics for the current state (oscillator state) are acquired in the computer 40 (step S408). The oscillation characteristics for the current state are, for example, data related to the current oscillation acquired from the detection unit 20, and may be an oscillation beam image or the like.

[0080] Next, in the computer 40, data relating to the current oscillation (data relating to the oscillation state) and information indicating the target oscillation state are input into an estimation program, which is a learning model, and output parameters relating to the phase pattern are estimated (step S409).

[0081] Next, the control unit 30 displays (reflects) a phase pattern based on the output parameters estimated by the computer 40 on the SLM 16 of the oscillator (step S410). In this state, the computer 40 acquires oscillation characteristics, specifically data related to the current oscillation (step S411), and determines whether the target oscillation state has been achieved (step S412). If the determination result shows that the target oscillation state has not been achieved, the output parameters are predicted again using the learning model (step S409). If the target oscillation state has been achieved, the oscillation characteristics phase ends. This determination may be made, for example, by analyzing the intensity distribution or confirming the flat-top pattern.

[0082] Next, the effects of the laser oscillation method according to this embodiment will be described.

[0083] The laser oscillation method according to this embodiment is a laser oscillation method in which an SLM 16 is placed on a resonance path, and laser light is oscillated within the resonance path by irradiating the laser medium 13 with excitation light, and includes the following steps: a data acquisition step of irradiating the laser medium 13 with excitation light and acquiring data related to the current oscillation; an estimation step of estimating parameters related to the phase pattern according to the data related to the current oscillation and the target oscillation state acquired in the data acquisition step, using an estimation program that outputs parameters related to the phase pattern to be displayed on the SLM 16 in response to input of the oscillation data and information indicating the target oscillation state; and a control step of controlling the phase pattern displayed on the SLM 16 based on the parameters estimated in the estimation step.

[0084] In the laser oscillation method according to this embodiment, excitation light is irradiated onto the laser medium 13 to acquire data related to the current oscillation within the resonant path, and parameters related to the phase pattern are estimated by an estimation program based on the data related to the current oscillation and a target oscillation state. Then, the phase pattern displayed on the SLM 16 is controlled based on the parameters. In this way, the phase pattern is automatically controlled using a pre-prepared estimation program, making it possible to easily create a phase pattern. Furthermore, because the estimation program receives inputs of data related to the current oscillation and a target (desired) oscillation state, it is possible to create with high accuracy a phase pattern that achieves a desired oscillation state. As described above, the laser oscillation method according to this embodiment makes it possible to easily and accurately create a phase pattern that achieves a desired oscillation state.

[0085] In the data acquisition step, data relating to a non-oscillation state may be acquired as data relating to the current oscillation, and in the estimation step, parameters relating to the phase pattern may be estimated so as to transition from the non-oscillation state to the oscillation state. With this configuration, it is possible to easily and accurately create a phase pattern that appropriately transitions from a non-oscillation state to a desired oscillation state.

[0086] In the data acquisition step, data related to the fluorescence emitted from the laser medium 13 may be acquired as data related to the current oscillation. With this configuration, even in a state before oscillation, data related to oscillation can be appropriately acquired from, for example, ASE (Amplified Spontaneous Emission) intensity. This makes it possible to create a phase pattern that achieves a desired oscillation state with higher accuracy. Note that, while conventionally, this was limited to control of the transverse mode, by focusing on the ASE intensity, it is now possible to automatically find oscillation conditions in a state before oscillation.

[0087] In the data acquisition step, data related to an oscillation state may be acquired as data related to the current oscillation, and in the estimation step, parameters related to a phase pattern may be estimated so as to transition from the current oscillation state to another oscillation state. With this configuration, phase parameters that appropriately transition from an oscillation state to a desired other oscillation state can be created easily and with high accuracy.

[0088] In the data acquisition step, data related to the laser oscillation light may be acquired as the data related to the current oscillation. With this configuration, data related to the oscillation can be appropriately acquired while the laser is oscillating. This makes it possible to create a phase pattern that realizes a desired oscillation state with higher accuracy.

[0089] The above laser oscillation method may further include a learning step of generating an estimation program by learning a plurality of data sets in which first learning data related to the phase pattern displayed on the SLM 16 is associated with second learning data related to the oscillation. In this manner, the accuracy of the estimation program can be improved by learning a large amount of data related to the oscillation (e.g., ASE intensity, a beam image of the laser oscillation light, etc.) in association with the phase pattern. In other words, the provision of the learning step makes it possible to generate an estimation program with high accuracy in estimating parameters related to the phase pattern, and to create a phase pattern that achieves a desired oscillation state with high accuracy.

[0090] The second learning data may include at least one of a beam image, intensity, spectrum, and time waveform of the laser oscillation light, the type of the laser medium 13, information on the excitation light, the temperature in the data acquisition environment, and the position of the optical system. With this configuration, it is possible to generate an estimation program that can estimate parameters related to the phase pattern with high accuracy.

[0091] In the learning step, the data related to the current oscillation acquired in the data acquisition step may be used as second learning data. With this configuration, the learning step is incorporated into (performed simultaneously with) other steps related to oscillation control, making it possible to update the estimation program so as to improve its accuracy while performing oscillation control.

[0092] Although the embodiments of the present disclosure have been described above, the present disclosure is not limited to the above embodiments. For example, although a configuration using the SLM 16 has been described, a deformable mirror or the like may be used instead depending on the desired laser oscillation characteristics. Specifically, when the desired laser oscillation characteristics are control of the intensity distribution of the output beam, a deformable mirror can be used.

[0093] A modified laser oscillation system configuration shown in FIG. 9 may also be employed. The resonance path in FIG. 9 extends from the output mirror 82 to the SLM 16. In this configuration, even when the intensity of the excitation light is high, the excitation light does not enter the SLM 16, preventing damage to the SLM 16 due to the excitation light. The reflecting mirror 81 shown in FIG. 9 is tilted at 45° to transmit the excitation light and reflect the oscillation light. The oscillation light reflected by the reflecting mirror 81 enters the Brewster window 84, which determines the polarization of the light incident on the SLM 16. In this manner, the Brewster window 84 serves a purpose similar to that of the quarter-wave plate 14 shown in FIG. 2. The oscillation light generated within the resonator is extracted by the output mirror 82. The output mirror 82 may be, for example, a partial reflection mirror that reflects 95% of the excitation light and oscillation light (transmitting 5%). The light transmitted through the output mirror 82 is reflected by the oscillation light reflecting mirror 83 and detected by the detection unit 20. The oscillation light reflecting mirror 83 is a mirror that has the property of reflecting the wavelength of the oscillation light (for example, 1030 nm) and transmitting the wavelength of the excitation light (for example, 940 nm).

[0094] Finally, various exemplary embodiments included in the present disclosure are described below in [E1] to [E19].

[0095] [E1] A laser oscillation method in which a spatial light modulator is placed on a resonance path and laser light is oscillated within the resonance path by irradiating a laser medium with excitation light, the laser oscillation method comprising: a data acquisition step of irradiating the laser medium with the excitation light and acquiring data related to current oscillation; an estimation step of estimating parameters related to a phase pattern according to the data related to the current oscillation acquired in the data acquisition step and a target oscillation state using an estimation program that outputs parameters related to a phase pattern to be displayed on the spatial light modulator in response to input of data related to oscillation and information indicating a target oscillation state; and a control step of controlling the phase pattern displayed on the spatial light modulator based on the parameters estimated in the estimation step.

[0096] [E2] The laser oscillation method according to E1, wherein in the data acquisition step, data relating to a non-oscillation state is acquired as data relating to the current oscillation, and in the estimation step, parameters relating to the phase pattern are estimated so as to transition from the non-oscillation state to the oscillation state.

[0097] [E3] The laser oscillation method according to [E2], wherein in the data acquisition step, data relating to fluorescence emitted from the laser medium is acquired as the data relating to the current oscillation.

[0098] [E4] A laser oscillation method according to any one of [E1] to [E3], wherein in the data acquisition step, data relating to an oscillation state is acquired as data relating to the current oscillation, and in the estimation step, parameters relating to the phase pattern are estimated so as to transition from the current oscillation state to another oscillation state.

[0099] [E5] The laser oscillation method according to [E4], wherein in the data acquisition step, data relating to laser oscillation light is acquired as the data relating to the current oscillation.

[0100] [E6] A laser oscillation method according to any one of [E1] to [E5], further comprising a learning step of generating the estimation program by learning a plurality of data sets in which first learning data relating to a phase pattern displayed on the spatial light modulator and second learning data relating to oscillation are associated with each other.

[0101] [E7] The laser oscillation method according to [E6], wherein the second learning data includes at least one of a beam image, intensity, spectrum, time waveform of the laser oscillation light, type of laser medium, fluorescence emitted from the laser medium, information on the excitation light, temperature in the data acquisition environment, and the position of the optical system.

[0102] [E8] The laser oscillation method according to [E6] or [E7], wherein in the learning step, the data relating to the current oscillation acquired in the data acquisition step is used as the second learning data.

[0103] [E9] A laser oscillation system that oscillates laser light within a resonance path by irradiating a laser medium with excitation light, comprising: a spatial light modulator that is arranged on the resonance path and modulates light passing through the resonance path by displaying an arbitrary phase pattern; an acquisition unit that acquires data related to current oscillation; an estimation unit that estimates parameters related to the phase pattern according to the data related to the current oscillation acquired by the acquisition unit and a target oscillation state using an estimation program that outputs parameters related to the phase pattern to be displayed on the spatial light modulator in response to input of data related to oscillation and information indicating a target oscillation state; and a control unit that controls the phase pattern displayed on the spatial light modulator based on the parameters estimated by the estimation unit.

[0104] [E10] The laser oscillation system according to [E9], wherein the acquisition unit acquires data relating to a non-oscillation state as data relating to the current oscillation, and the estimation unit estimates parameters relating to the phase pattern so as to transition from the non-oscillation state to an oscillation state.

[0105] [E11] The laser oscillation system according to [E10], wherein the acquisition unit acquires data relating to fluorescence emitted from the laser medium as the data relating to the current oscillation.

[0106] [E12] A laser oscillation system according to any one of [E9] to [E11], wherein the acquisition unit acquires data relating to an oscillation state as the data relating to the current oscillation, and the estimation unit estimates parameters relating to the phase pattern so as to transition from the current oscillation state to another oscillation state.

[0107] [E13] The laser oscillation system according to [E12], wherein the acquisition unit acquires data relating to laser oscillation light as the data relating to the current oscillation.

[0108] [E14] The laser oscillation system according to any one of [E9] to [E13], further comprising a learning unit that generates the estimation program by learning a plurality of data sets in which first learning data relating to a phase pattern displayed on the spatial light modulator and second learning data relating to oscillation are associated with each other.

[0109] [E15] The laser oscillation system described in [E14], wherein the second learning data includes at least one of a beam image, intensity, spectrum, time waveform of the laser oscillation light, type of laser medium, fluorescence emitted from the laser medium, information on the excitation light, temperature in the data acquisition environment, and the position of the optical system.

[0110] [E16] The laser oscillation system according to [E14] or [E15], wherein the learning unit sets the data relating to the current oscillation acquired by the acquisition unit as the second learning data.

[0111] [E17] A method for generating an estimation program that outputs parameters related to a phase pattern displayed on a spatial light modulator on a resonance path, the method comprising: a learning data collection step of collecting multiple pieces of first learning data related to the phase pattern displayed on the spatial light modulator and second learning data related to oscillation in the resonance path; and a learning step of generating the estimation program by learning multiple data sets in which the first learning data and the second learning data collected in the learning data collection step correspond to each other.

[0112] [E18] A generating device for an estimation program that outputs parameters related to a phase pattern displayed on a spatial light modulator on a resonance path, the generating device comprising: a collection unit that collects a plurality of first learning data related to the phase pattern displayed on the spatial light modulator and a plurality of second learning data related to oscillation in the resonance path; and a learning unit that generates the estimation program by learning a plurality of data sets in which the first learning data and the second learning data collected by the collection unit correspond to each other.

[0113] [E19] An estimation program that outputs parameters related to a phase pattern displayed on a spatial light modulator arranged on a resonator, the program causing a computer to execute: an acquisition process that acquires data related to current oscillation detected by irradiating a laser medium with excitation light; and an estimation process that outputs parameters related to the phase pattern displayed on the spatial light modulator based on the data related to the current oscillation acquired in the acquisition process and information indicating a target oscillation state.

[0114] 1...laser oscillation system, 10...laser oscillator, 13...laser medium, 16...SLM (spatial light modulator), 20...detection unit, 30...control unit, 40...computer (acquisition unit, estimation unit, learning unit).

Claims

1. A laser oscillation method in which a spatial light modulator is disposed on a resonance path and laser light is oscillated in the resonance path by irradiating a laser medium with excitation light, the method including: a data acquisition step of irradiating the laser medium with the excitation light and acquiring data related to the current oscillation; an estimation step of estimating parameters related to a phase pattern displayed on the spatial light modulator according to the data related to the current oscillation acquired in the data acquisition step and a target oscillation state, using an estimation program that outputs parameters related to the phase pattern displayed on the spatial light modulator in response to input of data related to the oscillation and information indicating the target oscillation state; and a control step of controlling the phase pattern displayed on the spatial light modulator based on the parameters estimated in the estimation step.

2. The laser oscillation method according to claim 1, wherein in the data acquisition step, data related to a non-oscillation state is acquired as the data related to the current oscillation, and in the estimation step, the parameters related to the phase pattern are estimated so as to transition from the non-oscillation state to the oscillation state.

3. The laser oscillation method according to claim 2, wherein in the data acquisition step, data related to fluorescence emitted from the laser medium is acquired as the data related to the current oscillation.

4. The laser oscillation method according to any one of claims 1 to 3, wherein in the data acquisition step, data related to an oscillation state is acquired as the data related to the current oscillation, and in the estimation step, the parameters related to the phase pattern are estimated so as to transition from the current oscillation state to another oscillation state.

5. The laser oscillation method according to claim 4, wherein in the data acquisition step, data related to laser oscillation light is acquired as the data related to the current oscillation.

6. The laser oscillation method according to any one of claims 1 to 5, further comprising a learning step of generating the estimation program by performing multiple learning on a data set in which first learning data related to a phase pattern displayed on the spatial light modulator is associated with second learning data related to the oscillation.

7. The laser oscillation method according to claim 6, wherein the second learning data includes at least one of a beam image, intensity, spectrum, time waveform of the laser oscillation light, a type of the laser medium, fluorescence emitted from the laser medium, information on the excitation light, temperature in a data acquisition environment, and a position of an optical system.

8. The laser oscillation method according to claim 6 or 7, wherein in the learning step, data related to the current oscillation acquired in the data acquisition step is used as the second learning data.

9. A laser oscillation system that oscillates laser light in a resonance path by irradiating a laser medium with excitation light, comprising: a spatial light modulator disposed on the resonance path and displaying an arbitrary phase pattern; an acquisition unit that acquires data related to the current oscillation; an estimation unit that estimates parameters related to the phase pattern according to the data related to the current oscillation acquired by the acquisition unit and a target oscillation state, using an estimation program that outputs parameters related to the phase pattern displayed on the spatial light modulator in response to an input of data related to the oscillation and information indicating the target oscillation state; and a control unit that controls the phase pattern displayed on the spatial light modulator based on the parameters estimated by the estimation unit.

10. The laser oscillation system according to claim 9, wherein the acquisition unit acquires data related to a non-oscillation state as data related to the current oscillation, and the estimation unit estimates parameters related to the phase pattern so as to transition from the non-oscillation state to the oscillation state.

11. The laser oscillation system according to claim 10, wherein the acquisition unit acquires data related to fluorescence emitted from the laser medium as data related to the current oscillation.

12. The laser oscillation system according to any one of claims 9 to 11, wherein the acquisition unit acquires data related to an oscillation state as data related to the current oscillation, and the estimation unit estimates parameters related to the phase pattern so as to transition from the current oscillation state to another oscillation state.

13. The laser oscillation system according to claim 12, wherein the acquisition unit acquires data related to laser oscillation light as data related to the current oscillation.

14. The laser oscillation system according to any one of claims 9 to 13, further comprising a learning unit that generates the estimation program by performing multiple learning on a data set associating first learning data related to the phase pattern displayed on the spatial light modulator with second learning data related to the oscillation.

15. The laser oscillation system according to claim 14, wherein the second learning data includes at least one of a beam image of laser oscillation light, intensity, spectrum, time waveform, type of laser medium, fluorescence emitted from the laser medium, information on excitation light, temperature in the data acquisition environment, and position of the optical system.

16. The laser oscillation system according to claim 14 or 15, wherein the learning unit uses the data related to the current oscillation acquired by the acquisition unit as the second learning data.

17. A method for generating an estimation program that outputs parameters related to a phase pattern displayed on a spatial light modulator on a resonance path, the method including: a learning data collection step of collecting a plurality of first learning data related to the phase pattern displayed on the spatial light modulator and second learning data related to oscillation in the resonance path; and a learning step of generating the estimation program by performing multiple learning on a data set in which the first learning data and the second learning data collected in the learning data collection step are associated with each other.

18. A generation device for generating an estimation program that outputs parameters related to a phase pattern displayed on a spatial light modulator on a resonance path, the device including: a collection unit that collects a plurality of first learning data related to the phase pattern displayed on the spatial light modulator and second learning data related to oscillation in the resonance path; and a learning unit that generates the estimation program by performing multiple learning on a data set in which the first learning data and the second learning data collected by the collection unit are associated with each other.

19. An estimation program that outputs parameters related to a phase pattern displayed on a spatial light modulator disposed on a resonator, the program causing a computer to execute: an acquisition process of acquiring data related to current oscillation detected by irradiating a laser medium with excitation light; and an estimation process of outputting parameters related to the phase pattern displayed on the spatial light modulator based on the data related to the current oscillation acquired in the acquisition process and information indicating a target oscillation state.

Images