An intelligent adaptive control and self-feedback stabilization system for microcavity soliton optical frequency combs

Abstract

The present invention belongs to the field of laser technology, and specifically relates to an intelligent adaptive control and self-feedback stabilization system for a microcavity soliton optical frequency comb. The present invention adopts an auxiliary photothermal tuning scheme, combined with an FPGA chip to monitor and control the parameters of each device in the system. By analyzing the power curves of each state in the soliton optical frequency comb generation process, different adaptive tuning strategies are designed for it, and a complete closed-loop feedback tuning link is constructed, realizing the lumped excitation, parallel control, and collaborative feedback technology of the on-chip soliton optical frequency comb generation system. The overall system parameter indicators are excellent, the soliton excitation time is less than 60s, the soliton stable existence time is greater than 24h, and the overall system volume is only 30×20×15cm 3 , which greatly improves the ease of use, practicality and portability of frequency comb light sources, promotes the formation of new electro-optical hybrid integrated frequency comb signal source devices and systems, and serves new applications of ultra-stable optical clocks, ultra-fast optical communications, and ultra-sensitive optical sensors.

Description

Technical Field

[0001] The present invention belongs to the field of laser technology, and in particular relates to an intelligent adaptive control and self-feedback stabilization system of a microcavity soliton optical frequency comb. Background Art

[0002] As a special light source, the optical frequency comb appears in the frequency domain as a series of discrete frequency components with uniform intervals and a certain phase relationship. The optical frequency comb generated based on the microcavity Kerr effect has the advantages of high repetition rate, good coherence, and easy integration. It has greatly promoted the development of optical metrology, precision ranging, fiber-optic communication and other fields.

[0003] However, the current generation process of microcavity soliton optical frequency combs often requires complex control procedures. The generation and stability of the optical comb are extremely dependent on precise control of the microcavity state, especially the control of the thermal effects within the cavity. Currently, a variety of tuning schemes have been proposed, such as rapid frequency tuning, auxiliary photothermal balance, and sideband modulation. However, the implementation of these schemes currently requires the operator to manually control the state of the pump light or auxiliary light by monitoring the real-time spectrum or intracavity power. This leads to low tuning efficiency, long time consumption, and poor repeatability. In addition, the entire soliton excitation system is large and complex, making it difficult to integrate. To address these issues, there is an urgent need to invent a device that can achieve autonomous real-time monitoring, self-startup, and self-feedback stabilization of the microcavity soliton optical frequency comb generation system, and that can also achieve system miniaturization. Summary of the Invention

[0004] To address these issues, the present invention provides an intelligent adaptive control and self-feedback stabilization system for a microcavity soliton optical frequency comb. This system utilizes an auxiliary photothermal tuning scheme, monitoring and adjusting the state of the microcavity soliton optical frequency comb via an FPGA control board. Key adjustment parameters include the real-time power variations at the microcavity pump and auxiliary light transmission ends, as well as the wavelength, power, and tuning speed of the pump and auxiliary lasers. This ensures that the system can automatically identify and tune the optical frequency comb to produce a single soliton state, and maintains the comb's state through a feedback loop. This eliminates the need for manual intervention and offers exceptional practicality and ease of use.

[0005] An intelligent adaptive control and self-feedback stabilization system for a microcavity soliton optical frequency comb is divided into two parts: an optical system for an optical frequency comb generation link and an electrical system for a programmable feedback link.

[0006] The optical system of the optical frequency comb generation link includes: two narrow linewidth lasers (light source modules), two optical amplifiers, two filters, two circulators, an optical power meter and a microcavity.

[0007] Two narrow linewidth lasers are used as light source modules, providing pump light and auxiliary light respectively:

[0008] The pump light is connected to an optical amplifier through an optical fiber for signal amplification, and then connected to a circulator after passing through a filter and divided into two paths. One path enters the microcavity and adjusts the wavelength to the red detuning point of the microcavity resonance peak (adjusting the pump light power to reach the nonlinear threshold of the microcavity to generate four-wave mixing) to generate solitons. The other path is connected to an optical power meter to monitor the system power status.

[0009] The auxiliary light is connected to an optical amplifier through an optical fiber for signal amplification, and then connected to a circulator after passing through a filter and divided into two paths. One path enters the microcavity to compensate for the thermal effect in the cavity to achieve thermal stability, and the other path is connected to an optical power meter to monitor the system power status.

[0010] The electrical system of the program-controlled feedback link is formed by connecting the two narrow linewidth lasers of the optical system with an optical power meter and an FPGA control board.

[0011] The two narrow-linewidth lasers in the light source module are connected to the FPGA control board, which uses an optical power meter to monitor and control the pump light output wavelength in real time to meet the soliton generation conditions. After optical amplification and filtering, the pump light and auxiliary light are split through a circulator and connected to the optical power meter. The optical power meter converts the two optical signals it receives into electrical signal data and transmits them to the FPGA control board. The FPGA control board then adjusts the two narrow-linewidth lasers in the light source module after data processing based on the signals from the optical power meter, forming a closed-loop control.

[0012] The above-mentioned intelligent adaptive control and self-feedback stabilization system of the microcavity soliton optical frequency comb realizes intelligent adaptive control and self-feedback stabilization of the soliton optical frequency comb. The specific process is as follows:

[0013] Step 1: Initialize the system components, including setting the output wavelength and output power of the pump and auxiliary lasers, setting the output power of the optical amplifier, and turning on the pump and auxiliary lasers. The wavelengths of the pump and auxiliary lasers fall within the blue detuning of the resonance peak.

[0014] Step 2: The auxiliary light is scanned from the blue detuning point to near the resonance peak until the auxiliary light power drops sharply, indicating that the auxiliary light has entered the microcavity. At this point, observing the spectrum, several distinct bilaterally symmetrical sidebands can be seen around the auxiliary light, which are called Turing states.

[0015] Step 3: Start the first scan of the pump light. The pump light is scanned forward from the blue detuned side of the resonance peak to the red detuned side of the resonance peak to record the characteristic points. During this period, the system will experience a high-noise state, a multi-soliton state, a single soliton state, and soliton annihilation. The system records the pump light wavelength and power when entering the high-noise state and the pump light wavelength and power when soliton annihilation occurs.

[0016] Step 4: Adjust the pump light wavelength to the wavelength recorded in step 3 above when entering the high noise state.

[0017] Step 5: Start the second scan of the pump light: This scan uses the pump light wavelength and power information corresponding to the soliton annihilation state recorded in step 3 to identify and lock the single soliton state; when entering the single soliton state, record the wavelength and power of the pump light.

[0018] Step 6. Continue to monitor the power of the two narrow-linewidth lasers in real time using an optical power meter and take appropriate countermeasures based on the following situations.

[0019] In the first scenario, the system maintains a stable single soliton state. The pump light power may fluctuate slightly, but within a range of 0.04 dBm. The system then adjusts the wavelength of the auxiliary light based on this fluctuation to control the position of the resonance peak. Specifically, when the pump light power decreases slightly, indicating an increase in the power within the microcavity, the system's FPGA control board tunes the auxiliary light away from the resonance peak, causing the resonance peak to blueshift and the power within the microcavity to decrease. When the pump light power increases, the opposite operation is performed. This stabilizes the single soliton state.

[0020] Case 2: The pump light power is 0.04 dBm or more higher than the pump light power when the system enters the single soliton state in step 5 above. At this time, the system enters the soliton annihilation state. Return to step 4, re-tune to enter the single soliton state, and update the pump light wavelength and power for the system to enter the single soliton state.

[0021] At this point, the above process can realize the intelligent adaptive control and self-feedback stabilization of the soliton optical frequency comb.

[0022] Furthermore, since the power at the transmission end of the auxiliary light will show a gradual downward trend, due to the influence of environmental factors, it is impossible to directly judge whether the degree of auxiliary light entering the cavity is appropriate based on the power value. Therefore, in step 2, the judgment of whether the auxiliary light enters the Turing state from the initial state adopts a dual judgment standard of the threshold and the slope of the power curve to make the judgment more accurate:

[0023] Assume that the current auxiliary light transmission end power is P aux , set the threshold condition: P aux <-4dBm, the purpose is to exclude the influence of other power change inflection points. When the threshold condition is met, the state is judged according to the change trend of the power curve. Due to data jitter, the solution of directly using the first-order difference to calculate the slope is not effective, so the system of the present invention will record the minimum value P according to the changing power value. min To eliminate the impact of data jitter, when 10-15 times meet the conditions: P aux -P min >0, the system will determine that it has entered the Turing state.

[0024] Furthermore, in Steps 3 and 5, one scan was performed respectively. This is because due to the uncertainty of the tuning process, the power change in the cavity has a certain randomness. During each single-soliton evolution process, the number of initially formed solitons is uncertain, and the number of solitons reduced during the process of the pump light redshift is also uncertain. On the power curve, it is manifested that the number of steps and the jump variables appearing in each scan are uncertain. Therefore, it is impossible to determine which step represents the formation of a single soliton, which makes it extremely difficult to directly lock the single-soliton state during a single forward scan of the pump light. However, under the condition that the degree of the auxiliary light entering the cavity is the same, the single-soliton state, the soliton annihilation state, and the power values obtained by scanning the pump light multiple times in a short period are basically the same, and the error can be ignored. Thus, a key tuning strategy proposed by the present invention is: after the Turing state generated by the entry of the auxiliary light into the cavity, the pump light is frequency-scanned twice. The first time is used to record the key node data and the change trend of each state change, and the second time is to use the data of the first time for processing and judgment to achieve precise tuning into the single-soliton state. Specifically:

[0025] Denote the pump light power at the time of soliton annihilation in Step 3 above as P0. In Step 5, during the second forward scan of the pump light, as the soliton steps appear, the power gradually increases. When the threshold condition is satisfied: P0 - 0.15 < P pump < P0 - 0.04, the system determines that it has entered the single-soliton state.

[0026] Furthermore, in the specific tuning process, the PID control algorithm is added to further improve the precise control of the frequency-scanning speeds of the pump light and the auxiliary light. PID is the abbreviation of Proportional, Integral, and Differential. In an ideal continuous control system, the control law of PID is shown by the following formula.

[0027]

[0028] [[ID=1~5]] pump Among them, the proportional K p , the integral T i , and the differential control parameter T D are three parameters to be regulated; e(t) is the error between the current state value and the ideal state value; u(t) is the final output response. During the frequency-scanning process of the auxiliary light in the initial state, assume the expected power of the auxiliary light transmission end is P i , then the error e(t) is the difference between the current power and the expected power, that is, P aux - P i , and the output response u(t) is the wavelength tuning rate of the auxiliary light.

[0029] Furthermore, in step 5, the PID algorithm is also used to tune the sweep speed when the pump light is swept. The error e(t) is the difference between the current pump light transmission end power and the soliton annihilation state power obtained in step 3. The output response u(t) is the pump light wavelength tuning rate.

[0030] Compared with the existing technology, the present invention has the following beneficial effects:

[0031] 1. The information required for soliton control in the present invention can be obtained through the first scan of the pump light in step 3 above, including the pump light wavelength and power when the system enters a high-noise state and the pump wavelength and power when the soliton is annihilated. Therefore, there is no need for pre-testing to obtain the above information. The entire tuning process is carried out automatically without manual intervention.

[0032] 2. The present invention focuses on the power change trend rather than the absolute value of power. Traditional tuning schemes basically use the absolute value of power as the judgment threshold condition, which results in a set of parameters corresponding to only one system. However, since the judgment threshold conditions set in the present invention are all related to the power change trend, it is applicable to different systems.

[0033] 3. After the system enters the single soliton state, the optical power meter will continue to monitor the power of the two narrow-linewidth lasers. According to the two different situations, the system will automatically make different adjustment measures.

[0034] In summary, the present invention provides a programmable automatic microcavity soliton generation system that fully leverages the design flexibility, parallel processing, and miniaturization advantages of FPGAs, achieving lumped excitation, parallel control, and collaborative feedback technology for on-chip frequency comb light sources, thereby realizing the intelligence of optical frequency comb devices. Compared with traditional soliton excitation systems and tuning processes, the present invention has a shorter soliton excitation time. Traditional manual operations often require an excitation time of more than 20 minutes, while the present invention can achieve one-button start within 60 seconds. The generated soliton optical frequency comb is stable for more than 24 hours, and the overall system volume is only 30×20×15cm. 3 , which greatly improves the ease of use, practicality and portability of frequency comb light sources, promotes the formation of new electro-optical hybrid integrated frequency comb signal source devices and systems, and serves new applications of ultra-stable optical clocks, ultra-fast optical communications, and ultra-sensitive optical sensors. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a system structure block diagram of the present invention;

[0036] Figure 2 This is the automatic tuning state transition diagram of the present invention.

[0037] Figure 3 2 is a power variation curve diagram of the embodiment.

[0038] Figure 4 This is a single soliton spectrum diagram generated by the embodiment.

[0039] Figure numerals: 1-pump light laser, 2-auxiliary light laser, 3-first erbium-doped fiber amplifier, 4-second erbium-doped fiber amplifier, 5-first filter, 6-spectrometer, 7-optical power meter, 8-second filter, 9-first circulator, 10-first coupler (99:1), 11-second coupler (99:1), 12-second circulator, 13-microring cavity, 14-FPGA control board. DETAILED DESCRIPTION

[0040] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0041] This embodiment provides an intelligent adaptive control and self-feedback stabilization system for a microcavity soliton optical frequency comb. Figure 1 As shown, it includes 1-pump light laser, 2-auxiliary light laser, 3-first erbium-doped fiber amplifier, 4-second erbium-doped fiber amplifier, 5-first filter, 6-spectrometer, 7-optical power meter, 8-second filter, 9-first circulator, 10-first coupler (99:1), 11-second coupler (99:1), 12-second circulator, 13-microring cavity, and 14-FPGA control board.

[0042] Among them, the operating wavelength of the pump light laser is 1550nm, and the adjustable range is 1nm. The operating wavelength of the auxiliary light laser is 1535nm, and the adjustable range is 30nm. The operating wavelength of the two erbium-doped fiber amplifiers is 1550nm, the output power of the first erbium-doped fiber amplifier is 240mw, and the output power of the second erbium-doped fiber amplifier is 210mw. The optical power meter can be connected to two inputs, and the input power is required to be less than 0dbm. The spectrometer can measure the power range of -70dbm-+20dbm, the measurable spectrum range is 1200nm-2400nm, the wavelength accuracy is ±0.05nm, and the dynamic range is 55db. In this system, the spectrometer is only used to observe phenomena and does not participate in the automatic tuning process. This system uses a high resonance quality factor (Q ~ 5×10 6 The silicon nitride micro-ring cavity has a diameter of 376 μm and a waveguide cross-sectional size of 1650 × 800 nm. 2 , single-mode oscillation can be achieved in the anomalous dispersion region around 1550nm. The spectral width is ~150nm and the repetition frequency is 106.051GHz.

[0043] The baud rate of communication between the two narrow-linewidth lasers and the FPGA control board is 115200bps, ensuring high-speed real-time communication. The output data frame format of the narrow-linewidth laser used in this embodiment includes the following: Destination address (target device address code), Source address (host address code), Type (instruction type), Reg (destination register), Date bytes (data) and CRC checksum (Cyclical Redundancy Check). Among them, the CRC checksum uses the CRC-CCITT-16 format. Specifically, the checksum consists of two bytes. At the beginning, each bit of the CRC register is preset to 1, and then the CRC register is XORed with the 8-bit data. After that, the CRC register is shifted from high to low, and the most significant bit (MSB) position is padded with zero. If the least significant bit (LSB, which has been shifted out of the CRC register) is 1, the register is XORed with a predefined polynomial code. Otherwise, if the LSB is zero, no XOR is required. Repeat the above shift from high to low 8 times. After the first 8-bit data is processed, the value of the CRC register at this time is XORed with the next 8-bit data and shifted 8 times as before. After all characters are processed, the value in the CRC register is the final CRC value. The polynomial corresponding to the CRC-CCITT-16 format check code is: 16 +x 12 +x 5 +1, 0x1021 is 10001000000100001 in binary.

[0044] The optical power meter communicates with the FPGA control board at a baud rate of 9600bps to ensure stable data transmission. The returned data frame format consists of the following: a 4-bit hexadecimal frame header (0xEDFA), LEN (data length), ADDR (data address bit), DATA (12-bit hexadecimal data), and SUM (checksum). The data bits are divided into three groups based on the order of the three interfaces, containing power information for the three channels, denoted as DATA1-DATA6. The output power signals are then denoted as Power1-Power3 and calculated using the following formula:

[0045]

[0046] During the regulation process, the pump light power is recorded as P pump , the auxiliary optical power is recorded as P aux ,The specific automated control process includes the following steps, such as Figure 2 As shown:

[0047] Step 1: Initialize and start the pump light laser, tune its wavelength to 1550.3500nm, make it fall at the blue detuning point of the resonance peak, and set the initial output power to 10dBm; initialize and start the auxiliary light laser, tune its wavelength to 1533.712nm, also make it fall at the blue detuning point of the resonance peak, and set its initial output power to 10dBm.

[0048] Initialize and start the first erbium-doped fiber amplifier and the second erbium-doped fiber amplifier, wherein the output power of the first erbium-doped fiber amplifier is set to 240 mw, and the output power of the second erbium-doped fiber amplifier is set to 210 mw.

[0049] Step 2: The auxiliary light starts forward scanning and monitors the power decrease trend of the auxiliary light transmission end. At this time, the auxiliary light gradually approaches the resonance peak from the blue detuning point, and the power in the cavity gradually increases, so the auxiliary light power will gradually decrease. By setting the threshold condition P aux <-4dBm to eliminate the influence of other power inflection points, and at this time adjust the auxiliary light forward scanning rate through the PID algorithm. Afterwards, the system will record the minimum value P according to the changing power value min , when the condition is met 10 times: P aux -P min When the sampling point of the auxiliary light transmission end is greater than 0, the system will determine that the auxiliary light transmission end power curve has an upward inflection point. At this time, observing the spectrum, several obvious left-right symmetrical sidebands can be seen around the auxiliary light, which is called the Turing state. The current pump light transmission end power is P flag .

[0050] Step 3: Start tuning the pump light, slowly moving it from the blue detuning of the resonance peak to the red detuning of the resonance peak. At this point, you can ignore the power of the auxiliary light transmission end and only focus on the power of the pump light transmission end.

[0051] As the pump light gradually approaches the resonance peak from the blue detuned end, the power at the pump light transmission end will first show a slow decline trend. After the pump light enters the cavity, the pump light energy is transferred to the cavity, and the power at the pump light transmission end will show a steep decline trend. When the threshold condition P is met, pump <P flagWhen it is -0.5, the system determines that the pump light enters the cavity. At this time, observing the spectrum, it can be found that through the degenerate and non-degenerate four-wave mixing effects in the cavity, the sidebands begin to densely fill the entire spectrum, and this is called the high-noise state. The system will record the pump light wavelength λ0 in the high-noise state. After entering the high-noise state, the pump light is scanned forward by 75 pm to ensure that the system reaches the soliton annihilation state. During this period, as the pump light redshifts, it will gradually go through the multi-soliton state, single-soliton state, and soliton annihilation state. And as the number of solitons decreases, the power in the cavity also decreases. At this time, the power curve shows a stepped climb at the pump light transmission end. However, at this time, we cannot directly determine the existence of the single-soliton state accurately through the power. Record the power value in the soliton annihilation state as P0.

[0052] Step 4: Tune the pump laser wavelength to the wavelength λ0 recorded when entering the high-noise state in Step 3.

[0053] Step 5: Scan the pump laser wavelength forward through the PID algorithm and control the tuning rate. The error is set as the current power and the power in the soliton annihilation state. As the soliton steps appear, the power gradually increases. When the threshold condition is satisfied: P0 - 0.15 < P pump < P0 - 0.04, the system determines that it has entered the single-soliton state. At this time, stop scanning the pump laser and record the current power value and the pump laser wavelength. Figure 4 For the single-soliton spectrogram generated in this embodiment.

[0054] Step 6: Continuously monitor the power value at the pump transmission end. If the condition is satisfied: P pump > P0 - 0.04, it is determined that the system enters the soliton annihilation state. At this time, it is necessary to return to Step 4 again and tune the system to enter the single-soliton state again. Figure 3 On the left is the pump light power change curve graph of the embodiment, and on the right is the auxiliary light power change curve graph.

[0055] If the system maintains a stable single-soliton state, at this time, the power fluctuation range of the pump light is within 0.04 dbm. The system adjusts the wavelength of the auxiliary light according to the fluctuation to control the position of the resonance peak; when the pump light power drops slightly, it means that the power in the microcavity rises. The FPGA control board of the system will tune the auxiliary light away from the resonance peak, so that the resonance peak blueshifts and the power in the microcavity drops; when the pump light power rises, the opposite operation is performed to achieve the stability of the single-soliton state. Thus, the entire system forms a closed-loop feedback control link and can automatically generate and stably maintain the single-soliton state.

Claims

1. An intelligent adaptive control and self-feedback stabilization system for a microcavity soliton optical frequency comb, characterized by: It is divided into two parts: the optical system of the optical frequency comb generation link and the electrical system of the program-controlled feedback link; The optical system of the optical frequency comb generation link includes: two narrow linewidth lasers, two optical amplifiers, two filters, two circulators, an optical power meter and a microcavity; Two narrow linewidth lasers provide pump light and auxiliary light for the light source respectively: The pump light is connected to an optical amplifier through an optical fiber for signal amplification. It then passes through a filter and is connected to a circulator to be split into two paths. One path enters the microcavity and adjusts the wavelength to the red detuning point of the microcavity resonance peak to generate solitons. The other path is connected to an optical power meter to monitor the system power status. The auxiliary light is connected to an optical amplifier through an optical fiber for signal amplification. Then, it passes through a filter and is connected to a circulator to be divided into two paths. One path enters the microcavity to compensate for the thermal effect in the cavity to achieve thermal stability, and the other path is connected to an optical power meter to monitor the power status of the system. The electrical system of the program-controlled feedback link is formed by connecting the two narrow linewidth lasers of the above optical system with an optical power meter and an FPGA control board; Two narrow-linewidth lasers are connected to an FPGA control board, which monitors and controls the pump light output wavelength in real time to meet soliton generation requirements. After optical amplification and filtering, the pump light and auxiliary light are split by a circulator and connected to an optical power meter. The optical power meter converts the two optical signals it receives into electrical data and transmits them to the FPGA control board. The FPGA control board then controls the two narrow-linewidth lasers based on the data processed by the optical power meter, forming a closed-loop control system. The specific process for achieving intelligent adaptive control and self-feedback stabilization of the soliton optical frequency comb is as follows: Step 1: Initialize the system components, including setting the output wavelength and output power of the pump light and auxiliary light, setting the output power of the optical amplifier, and turning on the pump light and auxiliary light; the wavelengths of the pump light laser and auxiliary light laser fall at the blue detuning point of the resonance peak; Step 2: The auxiliary light is scanned from the blue detuning position to the vicinity of the resonance peak until the auxiliary light power drops sharply and enters the Turing state; The judgment of the auxiliary light entering the Turing state from the initial state adopts the dual judgment criteria of threshold and power curve slope: Assume that the current auxiliary light transmission end power is P aux , set the threshold condition: P aux <-4dBm, when the threshold condition is met, the state is judged according to the change trend of the power curve; the minimum value P is recorded according to the changing power value min To eliminate the impact of data jitter, when 10-15 times meet the conditions: P aux -P min When the sampling point is greater than 0, the system determines that it has entered the Turing state; Step 3: Start the first sweep of the pump light. The pump light is scanned forward from the blue-detuned side of the resonance peak to the red-detuned side of the resonance peak to record the characteristic points. During this period, the system gradually experiences the high-noise state, the multi-soliton state, the single-soliton state, and the soliton annihilation. The system records the pump light wavelength and power when entering the high-noise state, and the pump light wavelength and power when soliton annihilation occurs. Step 4: Adjust the pump light wavelength to the wavelength recorded in step 3 above when entering the high noise state; Step 5: Start the second scan of the pump light: This scan uses the pump light wavelength and power information corresponding to the soliton annihilation state recorded in step 3 to identify and lock the single soliton state; when entering the single soliton state, record the pump light wavelength and power; Step 6. Continue to monitor the power of the two narrow-linewidth lasers in real time using an optical power meter and take appropriate countermeasures based on the following situations: Case 1: The system maintains a stable single soliton state. At this time, the power fluctuation range of the pump light is within 0.04dBm. The system adjusts the wavelength of the auxiliary light according to the fluctuation to control the position of the resonance peak. Specifically, when the pump light power drops slightly, indicating that the power in the microcavity increases, the system's FPGA control board will tune the auxiliary light away from the resonance peak, thereby causing the resonance peak to blueshift and the power in the microcavity to decrease. When the pump light power increases, the opposite operation is performed to achieve the stability of the single soliton state. Case 2: The pump light power is 0.04 dBm or more higher than the pump light power when the system enters the single soliton state in step 5 above. At this time, the system enters the soliton annihilation state. Return to step 4, re-tune to enter the single soliton state, and update the pump light wavelength and power for the system to enter the single soliton state.

2. The intelligent adaptive control and self-feedback stabilization system for a microcavity soliton optical frequency comb according to claim 1, characterized in that: The judgment basis for entering the single soliton state in step 5 is: Record the pump light power at the time of soliton annihilation in step 3 as P0. During the red sweep of the pump light, as the soliton step appears, the power gradually increases. When the threshold condition: P0 - 0.15 < P pump < P0 - 0.04 is satisfied, the system determines that it enters the single soliton state.

3. The intelligent adaptive control and self-feedback stabilization system for a microcavity soliton optical frequency comb according to claim 1, characterized in that: The tuning of the auxiliary light wavelength is controlled by a PID control algorithm: Among them, the ratio K p , integral T i , differential control parameter T D are the three parameters that need to be controlled; e(t) is the error between the current state value and the ideal state value; u(t) is the final output response; in the auxiliary light sweep process of the initial state, the expected power of the auxiliary light transmission end is set to P i , the error e(t) is the difference between the current power and the expected power P aux -P i , the output response u(t) is the auxiliary light wavelength tuning rate.

4. The intelligent adaptive control and self-feedback stabilization system for a microcavity soliton optical frequency comb according to claim 1, characterized in that: In step 5, the PID algorithm is also used to tune the sweep speed when the pump light is swept. The error e(t) is the difference between the current pump light transmission end power and the soliton annihilation state power obtained in step 3. The output response u(t) is the auxiliary light wavelength tuning rate.

Images