Lithium niobate coherent modulator control method, device and equipment based on gradient algorithm

Abstract

The present invention discloses a gradient algorithm-based lithium niobate coherent modulator control method, device, and equipment, relating to the technical field of electro-optical modulators. The method includes setting the phases of the I and Q paths of the lithium niobate IQ modulator to be the same based on a preprocessing algorithm; setting the bias voltages of the I, Q, and P paths of the lithium niobate IQ modulator to optimal bias points using a bias voltage locking algorithm; and determining the operating state of the lithium niobate IQ modulator under the current bias voltage based on a feedback algorithm, and performing error compensation adjustments to optimize the operating state of the lithium niobate IQ modulator. This application enables precise control of the modulator bias voltage.

Description

Technical Field

[0001] The present application relates to the technical field of electro-optical modulators, and in particular to a method, device and equipment for controlling a lithium niobate coherent modulator based on a gradient algorithm. Background Art

[0002] Lithium niobate optical modulators offer advantages such as wide bandwidth, fast response, low insertion loss, high signal-to-noise ratio, low half-wave voltage, and excellent stability. They are a mainstream electro-optical modulator and are widely used in optical communications, fiber optic sensors, and other fields. In high-speed optical communication systems, lithium niobate IQ (in phase-quadrature) modulators are key components due to their ability to implement complex modulation formats. However, to ensure efficient operation of the modulator, its bias voltage must be precisely controlled, a prerequisite for achieving high-quality optical signal transmission.

[0003] Traditional bias voltage control methods rely primarily on manual adjustments or simple automatic control strategies. These methods often struggle to adapt to rapidly changing environmental conditions, leading to unstable modulator performance and impacting communication quality. As optical communication systems continue to demand higher signal quality, the demands on modulator bias voltage control technology are also increasing. Therefore, achieving precise control of modulator bias voltage has become a pressing issue. Summary of the Invention

[0004] The present application provides a method, device and equipment for controlling a lithium niobate coherent modulator based on a gradient algorithm, which can achieve precise control of the modulator bias voltage.

[0005] In a first aspect, an embodiment of the present application provides a lithium niobate coherent modulator control method based on a gradient algorithm, which is used to modulate a lithium niobate IQ modulator with an IQ modulation structure in coherent modulation. The lithium niobate coherent modulator control method based on the gradient algorithm includes:

[0006] Based on the preprocessing algorithm, the phases of the I and Q paths of the lithium niobate IQ modulator are set to be the same;

[0007] The bias voltage locking algorithm is used to set the bias voltages of the I, Q, and P paths of the lithium niobate IQ modulator at the optimal bias point.

[0008] The working state of the lithium niobate IQ modulator under the current bias voltage is judged based on the feedback algorithm, and error compensation adjustment is performed to optimize the working state of the lithium niobate IQ modulator.

[0009] In combination with the first aspect, in one embodiment, setting the phases of the I path and the Q path of the lithium niobate IQ modulator to be the same based on the preprocessing algorithm specifically includes:

[0010] S101: Sweep the entire operating bias voltage range of the lithium niobate IQ modulator, output a curve of optical power versus bias voltage, obtain basic characteristic data of the lithium niobate IQ modulator, set the learning rate, convergence coefficient, convergence threshold, and the initial bias voltage of each channel, and record the initial optical power. Then go to S102;

[0011] S102: Calculate the gradient of the optical power of the I, Q, and P channels relative to the initial bias voltage, and move the bias voltages of the I, Q, and P channels in the direction of the maximum rising gradient of the optical power according to the set moving step size, obtain the bias voltages of the I, Q, and P channels after the shift, and record the optical power after the shift, and then go to S103;

[0012] S103: Compare the current optical power after movement with the latest initial optical power. If the optical power after movement is greater than the initial optical power, go to S105. If the optical power after movement is not greater than the initial optical power, and the latest learning rate is greater than the convergence threshold, go to S104. If the optical power after movement is not greater than the initial optical power, and the latest learning rate is not greater than the convergence threshold, then before the last bias voltage movement of the I-path, Q-path, and P-path, the bias voltages of the I-path, Q-path, and P-path obtained after movement are the bias voltages corresponding to the maximum optical power, indicating that the bias voltages of the I-path, Q-path, and P-path are set to the bias voltages corresponding to the maximum optical power, and end.

[0013] S104: Divide the current learning rate by the convergence coefficient to obtain the value as the latest learning rate, and go to S105;

[0014] S105: The current optical power value after the shift is used as the latest initial optical power, and the process goes to S102;

[0015] Among them, the set moving step size corresponding to the bias voltage of the I path is the latest learning rate multiplied by the gradient of the optical power of the I path compared to the initial bias voltage. The set moving step size corresponding to the bias voltage of the Q path is the latest learning rate multiplied by the gradient of the optical power of the Q path compared to the initial bias voltage. The set moving step size corresponding to the bias voltage of the P path is the latest learning rate multiplied by the gradient of the optical power of the P path compared to the initial bias voltage.

[0016] In combination with the first aspect, in one embodiment, the bias voltage of the I-path, Q-path, and P-path of the lithium niobate IQ modulator is set at the optimal bias point through the bias voltage locking algorithm, specifically including: locking the I-path and Q-path bias voltages based on a gradient descent algorithm, locking the P-path bias voltage based on a sweep spectrum algorithm, and locking the I-path, Q-path, and P-path bias voltages based on a pilot algorithm.

[0017] In combination with the first aspect, in one embodiment, locking the I-path and Q-path bias voltages based on the gradient descent algorithm specifically includes:

[0018] S201: When the bias voltage corresponding to the maximum optical power is obtained, the currently recorded shifted optical power is used as the first optical power, and the process proceeds to S202;

[0019] S202: Calculate the gradient of the optical power of the I and Q paths compared to the initial bias voltage, and move the bias voltages of the I and Q paths in the direction of the maximum decreasing gradient of the optical power according to a preset moving step size, obtain the bias voltages of the I and Q paths after the shift, and record the optical power after the shift, and go to S203;

[0020] S203: Compare the current optical power after movement with the latest first optical power. If the optical power after movement is less than the first optical power, go to S205. If the optical power after movement is not less than the first optical power and the latest learning rate is greater than the convergence threshold, go to S204. If the optical power after movement is not less than the first optical power and the latest learning rate is not greater than the convergence threshold, the bias voltages of the I-path, Q-path and P-path obtained before the last movement of the bias voltages of the I-path, Q-path and P-path are the bias voltages corresponding to the minimum optical power, indicating that the bias voltages of the I-path and Q-path are set to the bias voltages corresponding to the minimum optical power, and end.

[0021] S204: Divide the current learning rate by the convergence coefficient to obtain the value as the latest learning rate, and go to S205;

[0022] S205: The current optical power value after the shift is used as the latest first optical power, and the process goes to S202;

[0023] Among them, the preset moving step size corresponding to the bias voltage of the I path is the latest learning rate multiplied by the gradient of the optical power of the I path compared to the initial bias voltage, and the preset moving step size corresponding to the bias voltage of the Q path is the latest learning rate multiplied by the gradient of the optical power of the Q path compared to the initial bias voltage.

[0024] In combination with the first aspect, in one embodiment, locking the P-channel bias voltage based on the spectrum sweep algorithm specifically includes:

[0025] Set the bias voltage of the I and Q channels to the bias voltage corresponding to the minimum optical power, and set the bias voltage of the P channel to the bias voltage corresponding to the maximum optical power;

[0026] Sweep the bias voltage of the P channel and store the optical power. Obtain any pair of adjacent optical power maxima and minima from the stored optical power. Calculate the target optical power at the Quad point based on the optical power maxima and minima.

[0027] In the bias voltage range corresponding to the maximum optical power and the minimum optical power, the bias voltage point closest to the Quad point target optical power is obtained, determined as the Quad point of the P path, and the bias voltage of the P path is locked to the Quad point.

[0028] In conjunction with the first aspect, in one embodiment,

[0029] The I-channel, Q-channel, and P-channel bias voltages are locked based on the pilot algorithm, wherein the sine wave required for the pilot is simulated by the DAC of the bias control circuit and the sine wave of the specified frequency is simulated by the MCU timer;

[0030] The simulation of the pilot signal includes:

[0031] Set up the discrete sine signal array, add pilot duration, DC bias voltage, pilot sine amplitude, and set the counting threshold based on the pilot sine frequency;

[0032] Start the MCU timer interrupt, start timing from 0, and set the MCU timer count to 0. The internal clock of the MCU timer triggers the count to increase, and an interrupt is generated when the count reaches the count threshold;

[0033] Perform interrupt processing and generate pilot bias voltage, specifically:

[0034] VH=VDC+Sin[i]*A

[0035] Wherein, VH represents the pilot bias voltage, VDC represents the DC bias voltage, A represents the pilot sine amplitude, and Sin[i] represents the discrete sine signal array generated in real time. If the length i of the discrete sine signal array generated in real time exceeds the set length of the discrete sine signal array, i is reset to 0. If i exceeds the length of the discrete sine signal array, i is reset to 0. When the timing reaches the pilot duration, the MCU timer interrupt is disabled, and the simulation of pilot signal generation ends.

[0036] In combination with the first aspect, in one embodiment, locking the I-path, Q-path, and P-path bias voltages based on the pilot algorithm specifically includes:

[0037] S221: Load frequency f on path I I After collecting the optical power of the set time period, stop loading the sinusoidal pilot signal, perform FFT transformation on the optical power, and record the frequency f on the spectrum. I Normalized optical power at , go to S222;

[0038] S222: Determine frequency f IIs the normalized optical power at the input signal less than the set threshold? If so, go to S223; if not, adjust the bias voltage of the I path and go to S222.

[0039] S223: Load frequency f on Q path Q After collecting the optical power of the set time period, stop loading the sinusoidal pilot signal, perform FFT transformation on the optical power, and record the frequency f on the spectrum. Q Normalized optical power at , go to S224;

[0040] S224: Determine frequency f Q Is the normalized optical power at the position less than the set threshold? If so, go to S225; if not, adjust the bias voltage of the Q path and go to S223;

[0041] S225: Load frequency f on path I I A sinusoidal pilot signal with a frequency of f is loaded on the Q path. Q After collecting the optical power of the set time period, stop loading the sinusoidal pilot signal, perform FFT transformation on the optical power, and record the frequency f on the spectrum. I +f Q The normalized optical power and |f I -f Q Normalized optical power at |, go to S226;

[0042] S226: frequency f I +f Q The normalized optical power and |f I -f Q Whether the normalized optical power at | is less than the set threshold, if not, adjust the bias voltage of the P path and go to S225; if so, go to S227;

[0043] S227: Execute judgment: If the frequency loaded on path I is f I After the sinusoidal pilot signal is applied, the bias voltage of the I channel is not adjusted, and the frequency f is loaded on the Q channel. Q The bias voltage of the Q circuit is not adjusted after the sinusoidal pilot signal is sent, and the frequency f is loaded on the I circuit at the same time. I The sinusoidal pilot signal and the frequency f loaded on the Q path Q If the bias voltage of the P circuit is not adjusted after the sinusoidal pilot signal is received, the process ends; otherwise, the process goes to S221.

[0044] In conjunction with the first aspect, in one embodiment, judging the operating state of the lithium niobate IQ modulator under the current bias voltage based on the feedback algorithm and performing error compensation adjustment to optimize the operating state of the lithium niobate IQ modulator specifically includes:

[0045] S301: Setting the optical power change threshold and the normalized optical power threshold, and setting the bias voltages of the I, Q, and P channels at the optimal bias points. The optical power corresponding to the optimal optical power is then taken as the optimal optical power, and the process proceeds to S302.

[0046] S302: Read the current optical power regularly, calculate the change between the current optical power and the optimal optical power, and then go to S303;

[0047] S303: Determine whether the calculated change exceeds the optical power change threshold. If so, go to S304; if not, go to S302.

[0048] S304: Loading frequency f on path I I After collecting the optical power of the set time period, stop loading the sinusoidal pilot signal, perform FFT transformation on the optical power, and record the frequency f on the spectrum. I The normalized optical power at the Q path is loaded with a frequency of f Q After collecting the optical power of the set time period, stop loading the sinusoidal pilot signal, perform FFT transformation on the optical power, and record the frequency f on the spectrum. Q The normalized optical power at the location, while the frequency f is loaded on the I path I The sinusoidal pilot signal and the frequency f loaded on the Q path Q After collecting the optical power of the set time period, stop loading the sinusoidal pilot signal, perform FFT transformation on the optical power, and record the frequency f on the spectrum. I +f Q The normalized optical power and |f I -f Q The normalized optical power at |, according to the recorded frequency f I Normalized optical power at frequency f Q Normalized optical power at frequency f I +f Q Normalized optical power at frequency |f I -f Q Whether the normalized optical power at | exceeds the normalized optical power threshold to determine whether the operating point is offset. If so, go to S305; if not, go to S306;

[0049] S305: Lock the bias voltages of the I, Q, and P channels based on the pilot algorithm, perform offset correction, and then go to S306;

[0050] S306: Based on the change in the optimal optical power caused by the relocking of the working point or the change in the overall optical power, the optimal optical power is updated, and the process goes to S302.

[0051] In a second aspect, an embodiment of the present application provides a lithium niobate coherent modulator control device based on a gradient algorithm, wherein the lithium niobate coherent modulator control device based on a gradient algorithm includes:

[0052] A first setting module is configured to set the phases of the I path and the Q path of the lithium niobate IQ modulator to be the same based on a preprocessing algorithm;

[0053] A second setting module is used to set the bias voltages of the I-channel, Q-channel, and P-channel of the lithium niobate IQ modulator at an optimal bias point through a bias voltage locking algorithm;

[0054] The execution module is used to judge the working state of the lithium niobate IQ modulator under the current bias voltage based on the feedback algorithm, and perform error compensation adjustment to make the working state of the lithium niobate IQ modulator optimal.

[0055] In a third aspect, an embodiment of the present application provides a lithium niobate coherent modulator control device based on a gradient algorithm, wherein the lithium niobate coherent modulator control device based on a gradient algorithm includes a processor, a memory, and a lithium niobate coherent modulator control program based on a gradient algorithm stored on the memory and executable by the processor, wherein when the lithium niobate coherent modulator control program based on a gradient algorithm is executed by the processor, the steps of the lithium niobate coherent modulator control method based on a gradient algorithm are implemented.

[0056] The beneficial effects of the technical solutions provided in the embodiments of the present application include:

[0057] (1) Using the gradient ascent algorithm for preprocessing and the gradient descent algorithm to find the optimal bias voltage point of the modulator, the bias voltages of different paths can be adjusted synchronously, and the influence between paths can be compensated in real time. Compared with the ordinary thermal spectrum sweep algorithm, the time and complexity of iterative adjustment are greatly reduced. In addition, the dynamic nature of the gradient algorithm enables it to adapt to IQ modulators with similar bias voltage ranges, and has good compatibility.

[0058] (2) Using DAC to simulate the sine wave required by the pilot method, and using digital FFT filtering instead of hardware filtering, no additional chip devices are required, and the detection quality meets the detection requirements of the pilot method; compared with the method of using a function generator to generate a sine signal and hardware filtering detection, it can save material costs, reduce the size of the control circuit board, and improve the locking accuracy without complicating the circuit design, which is conducive to integrated applications;

[0059] (3) By designing an automatic optimization algorithm, it is effective in reducing the impact of external interference and heat accumulation effects on the modulator during long-term operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1Design schematic diagram for bias voltage system of lithium niobate IQ modulator;

[0061] Figure 2 This is a flow chart of a control method for a lithium niobate coherent modulator based on a gradient algorithm of the present application;

[0062] Figure 3 This is a functional module diagram of the lithium niobate coherent modulator control device based on the gradient algorithm of this application;

[0063] Figure 4 This is a schematic diagram of the hardware structure of the lithium niobate coherent modulator control device based on the gradient algorithm of this application. DETAILED DESCRIPTION

[0064] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0065] In order to make the objectives, technical solutions and advantages of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0066] In the first aspect, an embodiment of the present application provides a lithium niobate coherent modulator control method based on a gradient algorithm, which is used to modulate the lithium niobate IQ modulator of the IQ modulation structure in coherent modulation, monitor the optical power by measuring and controlling the photodiode at the output end of the lithium niobate IQ modulator, find the optimal operating point through the bias voltage control algorithm, and lock the bias voltages of each channel to the optimal operating point to achieve high-speed signal modulation.

[0067] The lithium niobate coherent modulator control method of the present application can reduce iteration time and complexity, quickly and accurately lock the optimal bias point, and combine the pilot method in the control process to help achieve accurate locking and real-time optimization of the optimal working point. In the traditional pilot method, a function generator chip, an operational amplifier and other devices are required to form a sine wave generation circuit. The present application uses a DAC (digital-to-analog converter) to simulate the sine wave required by the pilot method and uses a digital FFT (Fast Fourier Transform) for filtering. This method reuses multiple DACs and multiple ADCs (analog-to-digital converters), and there is no need to design a sine wave generation circuit and a pilot detection circuit. Compared with the method of using a function generator to generate a sine signal and hardware filtering, it simplifies the circuit design while ensuring the detection quality, saves material costs, reduces the size of the control circuit board, and is conducive to integrated applications.

[0068] First of all, it should be noted that the lithium niobate IQ modulator in this application adopts a typical IQ modulation structure in coherent modulation, see Figure 1 As shown in Figure 1, the lithium niobate IQ modulator consists of two sub-MZI (multimode interferometer) structures and a parent MZI structure. The upper and lower sub-MZI structures are called the I path and Q path, respectively, and the parent MZI structure is called the P path. When the modulator is operating, the phase difference between the I path and the Q path is 90 degrees.

[0069] Common operating points for Mach-Zehnder modulators include the slope quadrature point (Quad point), the minimum optical power point (Null point), and the maximum optical power point (Max point). For common modulation formats such as QPSK and PAM-4, the bias voltages for the I and Q channels should be set at the Null point, and the bias voltage for the P channel should be set at the Quad point.

[0070] The bias voltage control system of lithium niobate IQ modulator includes lithium niobate IQ modulator chip, bias voltage control circuit and bias voltage control algorithm. The design diagram is as follows: Figure 1 shown. Figure 1 In the figure, 1 is a lithium niobate IQ modulator; 2 is an optical waveguide; 3 is a multimode interferometer (MMI), which functions as a beam splitter and combiner in the modulator; 4 is a thermal phase shifter for the I path, used to control the phase of the I path; 5 is a thermal phase shifter for the Q path, used to control the thermal phase shift of the Q path; 6 is a thermal phase shifter for the P path, used to control the phase difference between the I and Q paths; and 7 is a monitoring photodiode (MPD), used to detect optical power and convert it into photocurrent. It should be noted that the MPD can be either an on-chip integrated MPD or an external MPD device. In practical applications, at least one of the two forms is sufficient. Figure 1 Regarding the MPD settings, the MPD on the left corresponds to the MPD form integrated on the chip, and the MPD on the right corresponds to the external MPD device form.

[0071] For the bias voltage control circuit: the photocurrent is converted into an analog voltage signal through a transimpedance amplifier (TIA) circuit; the analog voltage signal is converted into a digital signal through an analog-to-digital converter (ADC) and sent to the main microcontroller unit (MCU); after the main control runs the control algorithm, the digital bias voltage signal is sent to a multi-channel digital-to-analog converter (DAC); the multi-channel DAC converts the digital bias voltage signal into an analog bias voltage and loads it onto the I-channel, Q-channel, and P-channel thermal phase shifters.

[0072] In one embodiment, referring to Figure 2 , Figure 2 This is a flow chart of the control method of lithium niobate coherent modulator based on gradient algorithm in this application. Figure 2As shown, the control method of the lithium niobate coherent modulator based on the gradient algorithm includes:

[0073] S1: Based on the preprocessing algorithm, the phases of the I and Q paths of the lithium niobate IQ modulator are set to be the same;

[0074] The lithium niobate coherent modulator control method of the present application specifically includes three parts: preprocessing, bias voltage locking and automatic optimization.

[0075] In step S1, the preprocessing algorithm aims to align the phases of the I and Q paths of the lithium niobate IQ modulator to facilitate subsequent locking. First, the entire operating bias voltage range of the lithium niobate IQ modulator is scanned, and the output optical power versus bias voltage curve is stored to obtain basic characteristic data of the lithium niobate IQ modulator. The initial bias voltage is then set using a gradient rise method to lock the I, Q, and P path bias voltages at the maximum optical power point. At this point, the I and Q paths of the lithium niobate IQ modulator are in phase.

[0076] S2: Using the bias voltage locking algorithm, the bias voltages of the I, Q, and P paths of the lithium niobate IQ modulator are set at the optimal bias points.

[0077] In step S2, the bias voltage locking algorithm is used to achieve optimal bias point locking for I-path, Q-path, and P-path. First, a preliminary locking is performed, using a gradient algorithm to lock the bias voltages of I-path and Q-path at the Null point, then scanning the spectrum analysis data to lock the bias voltage of P-path at the Quad point, and then performing precise locking, adding pilot signals to I-path and Q-path, sampling the optical power, and performing FFT transformation. By analyzing the spectrum, it is possible to accurately determine whether the bias voltages of I-path, Q-path, and P-path are at the optimal working point. If not, fine-tune them, and repeat this step until I-path, Q-path, and P-path are all at the optimal working point. The present application also provides a method for using DAC to simulate the generation of the sine wave required by the pilot method. This method does not require additional equipment. Compared with the method of using a function generator to generate a sine signal, it can save material costs, reduce the size of the control circuit board, and improve the locking accuracy without complicating the circuit design, which is conducive to integrated applications.

[0078] S3: Based on the feedback algorithm, the operating state of the lithium niobate IQ modulator under the current bias voltage is determined, and error compensation adjustments are made to optimize the operating state of the lithium niobate IQ modulator. This automatic optimization algorithm, based on the principle of feedback control, ensures that the lithium niobate IQ modulator maintains optimal operating conditions despite long-term operation and environmental factors.

[0079] Step S3 uses a feedback algorithm to determine whether the working state of the lithium niobate IQ modulator is optimal under the current bias voltage, and makes adjustments to compensate for errors caused by long-term operation of the modulator and environmental influences.

[0080] Furthermore, in one embodiment, based on a preprocessing algorithm, the phases of the I and Q paths of the lithium niobate IQ modulator are set to be the same, specifically including:

[0081] S101: Sweep the entire operating bias voltage range of the lithium niobate IQ modulator, output a curve of optical power versus bias voltage, obtain basic characteristic data of the lithium niobate IQ modulator, set the learning rate, convergence coefficient, convergence threshold, and the initial bias voltage of each channel, and record the initial optical power. Then go to S102;

[0082] It should be noted that, since the value of the learning rate may change later, the learning rate set in step S101 is the initial learning rate;

[0083] S102: Calculate the gradient of the optical power of the I, Q, and P channels relative to the initial bias voltage, and move the bias voltages of the I, Q, and P channels in the direction of the maximum rising gradient of the optical power according to the set moving step size, obtain the bias voltages of the I, Q, and P channels after the shift, and record the optical power after the shift, and then go to S103;

[0084] S103: Compare the current optical power after movement with the latest initial optical power. If the optical power after movement is greater than the initial optical power, it means that the bias voltage is moving in the direction of increasing optical power, then go to S105. If the optical power after movement is not greater than the initial optical power, and the latest learning rate is greater than the convergence threshold, then it means that the optical power has exceeded the maximum point and the gradient algorithm has not converged. Go to S104. If the optical power after movement is not greater than the initial optical power, and the latest learning rate is not greater than the convergence threshold, then it means that the gradient algorithm has converged. At this time, revert to the last bias voltage, that is, before the last bias voltage movement of I-way, Q-way, and P-way, the bias voltages of I-way, Q-way, and P-way after movement are obtained, which are the bias voltages corresponding to the maximum optical power, indicating that the bias voltages of I-way, Q-way, and P-way are set to the bias voltages corresponding to the maximum optical power. End;

[0085] It should be noted that, when the optical power after the first movement is compared with the latest initial optical power, the optical power after the movement refers to the optical power after the movement recorded after the first movement, and the initial optical power refers to the initial optical power recorded for the first time. For subsequent comparisons of the optical power after the movement with the latest initial optical power, if the value of the initial optical power has changed, the latest initial optical power refers to the initial optical power after the change; the current optical power after the movement refers to the optical power after the movement recorded after the latest movement;

[0086] S104: Divide the current learning rate by the convergence coefficient to obtain a value as the latest learning rate, and go to S105; that is, divide the current learning rate by the convergence coefficient to obtain a value as the latest learning rate value;

[0087] S105: The current optical power value after the movement is used as the latest initial optical power, and the process goes to S102; that is, the initial optical power value is replaced by the current optical power value after the movement;

[0088] Among them, the set moving step size corresponding to the bias voltage of the I path is the latest learning rate multiplied by the gradient of the optical power of the I path compared to the initial bias voltage. The set moving step size corresponding to the bias voltage of the Q path is the latest learning rate multiplied by the gradient of the optical power of the Q path compared to the initial bias voltage. The set moving step size corresponding to the bias voltage of the P path is the latest learning rate multiplied by the gradient of the optical power of the P path compared to the initial bias voltage.

[0089] Furthermore, in one embodiment, the bias voltages of the I-path, Q-path, and P-path of the lithium niobate IQ modulator are set at the optimal bias point through a bias voltage locking algorithm, specifically including: locking the I-path and Q-path bias voltages based on a gradient descent algorithm, locking the P-path bias voltage based on a sweep spectrum algorithm, and locking the I-path, Q-path, and P-path bias voltages based on a pilot algorithm.

[0090] Furthermore, in one embodiment, the I-path and Q-path bias voltages are locked based on a gradient descent algorithm, specifically including:

[0091] S201: When the bias voltage corresponding to the maximum optical power is obtained, the currently recorded shifted optical power is used as the first optical power, and the process proceeds to S202;

[0092] In step S1, the bias voltages of the I, Q, and P channels have been set to the bias voltages corresponding to the maximum optical power;

[0093] S202: Calculate the gradient of the optical power of the I and Q paths compared to the initial bias voltage, and move the bias voltages of the I and Q paths in the direction of the maximum decreasing gradient of the optical power according to a preset moving step size, obtain the bias voltages of the I and Q paths after the shift, and record the optical power after the shift, and go to S203;

[0094] S203: Compare the current optical power after movement with the latest first optical power. If the optical power after movement is less than the first optical power, it means that the bias voltage is moving in the direction of decreasing optical power. Go to S205. If the optical power after movement is not less than the first optical power and the latest learning rate is greater than the convergence threshold, it means that the optical power has crossed the minimum point and the gradient algorithm has not converged. Go to S204. If the optical power after movement is not less than the first optical power and the latest learning rate is not greater than the convergence threshold, it means that the gradient algorithm has converged. At this time, rewind the last bias voltage, that is, before the last bias voltage movement of the I-way, Q-way, and P-way, the bias voltages of the I-way and Q-way after movement are obtained, which are the bias voltages corresponding to the minimum optical power, indicating that the bias voltages of the I-way and Q-way are set to the bias voltages corresponding to the minimum optical power. End;

[0095] S204: Divide the current learning rate by the convergence coefficient to obtain the value as the latest learning rate, and go to S205;

[0096] S205: The current optical power value after the shift is used as the latest first optical power, and the process goes to S202;

[0097] Among them, the preset moving step size corresponding to the bias voltage of the I path is the latest learning rate multiplied by the gradient of the optical power of the I path compared to the initial bias voltage, and the preset moving step size corresponding to the bias voltage of the Q path is the latest learning rate multiplied by the gradient of the optical power of the Q path compared to the initial bias voltage.

[0098] Furthermore, in one embodiment, locking the P-channel bias voltage based on a spectrum sweep algorithm specifically includes:

[0099] S211: Setting the bias voltages of the I and Q paths to the bias voltages corresponding to the minimum optical power, and setting the bias voltage of the P path to the bias voltage corresponding to the maximum optical power;

[0100] In steps S201 to S205, the bias voltages of the I and Q paths have been set to the bias voltages corresponding to the minimum optical power, i.e., the Null point; in step S1, the bias voltage of the P path has been set to the Max point;

[0101] S212: Sweep the bias voltage of the P channel and store the optical power, obtain any pair of adjacent optical power maximum and optical power minimum values ​​from the stored optical power, and calculate the Quad point target optical power based on the optical power maximum and optical power minimum values;

[0102] Half of the sum of the maximum and minimum optical power is the target optical power at the Quad point.

[0103] S213: In the bias voltage range corresponding to the maximum optical power and the minimum optical power, obtain the bias voltage point closest to the Quad point target optical power, determine it as the Quad point of the P path, and lock the bias voltage of the P path to the Quad point. At this time, the I path and the Q path are orthogonal.

[0104] Furthermore, in one embodiment, the I-channel, Q-channel, and P-channel bias voltages are locked based on a pilot algorithm, wherein the sine wave required for the pilot is simulated by a DAC of the bias control circuit, and a sine wave of a specified frequency is simulated by an MCU timer.

[0105] The simulation of the pilot signal includes:

[0106] S21: Set the discrete sine signal array, add the pilot duration, DC bias voltage, pilot sine amplitude, and set the counting threshold according to the pilot sine frequency;

[0107] By modifying the amplitude and counting threshold, sine wave signals with different amplitudes and frequencies can be generated;

[0108] S22: Start MCU timer interrupt, start timing from 0, and set the MCU timer count to 0. The internal clock of the MCU timer triggers the count to increase, and an interrupt is generated when the count reaches the count threshold;

[0109] S23: Perform interrupt processing and generate a pilot bias voltage. Specifically:

[0110] VH=VDC+Sin[i]*A

[0111] Where VH represents the pilot bias voltage, VDC represents the DC bias voltage, A represents the pilot sine amplitude, and Sin[i] represents the real-time generated discrete sine signal array. If the length i of the real-time generated discrete sine signal array exceeds the set discrete sine signal array length, i is reset to 0. When the timing reaches the pilot duration, the MCU timer interrupt is disabled, and the simulation of pilot signal generation ends. Using the MCU timer interrupt signal DAC to simulate the pilot signal eliminates the need for additional components and power supply circuits, simplifying circuit design and saving space.

[0112] Furthermore, in one embodiment, the I-path, Q-path, and P-path bias voltages are locked based on a pilot algorithm, specifically including:

[0113] S221: Load frequency f on path I I After collecting the optical power of the set time period (i.e. collecting the optical power of a short period of time), stop loading the sinusoidal pilot signal, perform FFT transformation on the optical power, and record the frequency f on the spectrum. I Normalized optical power at , go to S222;

[0114] S222: Determine frequency f I Is the normalized optical power at less than the set threshold? If so, go to S223; if not, adjust the bias voltage of the I path (specifically, fine-tune the bias voltage of the I path) and go to S222;

[0115] S223: Load frequency f on Q path Q After collecting the optical power of the set time period, stop loading the sinusoidal pilot signal, perform FFT transformation on the optical power, and record the frequency f on the spectrum. Q Normalized optical power at , go to S224;

[0116] S224: Determine frequency f Q Is the normalized optical power at less than the set threshold? If so, go to S225; if not, adjust the bias voltage of the Q path (specifically, fine-tune the bias voltage of the Q path) and go to S223;

[0117] S225: Load frequency f on path I I A sinusoidal pilot signal with a frequency of f is loaded on the Q path. Q After collecting the optical power of the set time period, stop loading the sinusoidal pilot signal, perform FFT transformation on the optical power, and record the frequency f on the spectrum. I +f Q The normalized optical power and |f I -f Q Normalized optical power at |, go to S226;

[0118] S226: frequency f I +f Q The normalized optical power and |f I -f Q | whether the normalized optical power at each of the optical paths is less than the set threshold value; if not, adjust the bias voltage of the P path (specifically, fine-tune the bias voltage of the P path), and go to S225; if so, go to S227;

[0119] S227: Execute judgment: If the frequency loaded on path I is f I After the sinusoidal pilot signal is applied, the bias voltage of the I channel is not adjusted, and the frequency f is loaded on the Q channel. Q The bias voltage of the Q circuit is not adjusted after the sinusoidal pilot signal is sent, and the frequency f is loaded on the I circuit at the same time. I The sinusoidal pilot signal and the frequency f loaded on the Q path Q If the bias voltage of the P circuit is not adjusted after the sinusoidal pilot signal is received, the process ends; otherwise, the process goes to S221.

[0120] That is, after loading the pilot signal three times, no fine-tuning operation is performed, which means that the I, Q, and P channels are all at the optimal working point and the algorithm ends; if fine-tuning is performed, in order to eliminate inter-channel interference, it is necessary to jump to step S221 for further detection and verification.

[0121] The collected optical power is fed into the MCU timer for FFT, which then searches for the corresponding pilot frequency. Compared to hardware circuit filtering for collecting optical power at the pilot frequency, this method uses a simpler circuit and offers more flexible detection. At this point, the bias voltage lock algorithm is complete, the I and Q paths are at the Null point, the P path is at the Quad point, the I and Q paths are orthogonal, and the optical power is at its minimum. This is when the signal loading effect is optimal.

[0122] Furthermore, in one embodiment, the operating state of the lithium niobate IQ modulator under the current bias voltage is determined based on a feedback algorithm, and error compensation adjustment is performed to optimize the operating state of the lithium niobate IQ modulator, specifically including:

[0123] S301: Setting the optical power change threshold and the normalized optical power threshold, and setting the bias voltages of the I, Q, and P channels at the optimal bias points. The optical power corresponding to the optimal optical power is then taken as the optimal optical power, and the process proceeds to S302.

[0124] S302: Read the current optical power regularly, calculate the change between the current optical power and the optimal optical power, and then go to S303;

[0125] S303: Determine whether the calculated change exceeds the optical power change threshold. If so, go to S304; if not, go to S302.

[0126] S304: Loading frequency f on path I I After collecting the optical power of the set time period, stop loading the sinusoidal pilot signal, perform FFT transformation on the optical power, and record the frequency f on the spectrum. I The normalized optical power at the Q path is loaded with a frequency of f Q After collecting the optical power of the set time period, stop loading the sinusoidal pilot signal, perform FFT transformation on the optical power, and record the frequency f on the spectrum. Q The normalized optical power at the location, while the frequency f is loaded on the I path I The sinusoidal pilot signal and the frequency f loaded on the Q path Q After collecting the optical power of the set time period, stop loading the sinusoidal pilot signal, perform FFT transformation on the optical power, and record the frequency f on the spectrum. I +f Q The normalized optical power and |f I -f Q The normalized optical power at |, according to the recorded frequency f INormalized optical power at frequency f Q Normalized optical power at frequency f I +f Q Normalized optical power at frequency |f I -f Q Whether the normalized optical power at | exceeds the normalized optical power threshold to determine whether the operating point has shifted. If so, it indicates that the overall optical power has changed, and the process goes to S305. If not, it indicates that the working state of the lithium niobate IQ modulator is still optimal, and the process goes to S306.

[0127] S305: Lock the bias voltages of the I, Q, and P channels based on the pilot algorithm, perform offset correction, and then go to S306;

[0128] S306: Based on the change in the optimal optical power caused by the relocking of the working point or the change in the overall optical power, the optimal optical power is updated, and the process goes to S302.

[0129] It should be noted that the gradient algorithm is an ideal choice for bias voltage control in lithium niobate IQ modulators due to its efficient locking speed. By calculating the gradient of the modulator's output signal, the gradient algorithm quickly locates the bias voltage with optimal signal quality, achieving efficient automatic adjustment. This efficiency is reflected in its ability to synchronously adjust the bias voltages of different paths, compensating for interpath influences in real time, significantly reducing the time and complexity of iterative adjustments. Furthermore, the dynamic nature of the gradient algorithm enables it to adapt to IQ modulators with different bias voltage ranges.

[0130] In order to further improve the accuracy of locking, this application adopts the pilot method for precise locking. By adding pilot signals to the I and Q paths, performing FFT transformation after sampling the optical power, and analyzing the spectrum, it can accurately determine whether the bias voltages of the I, Q, and P paths are at the optimal working point. After the gradient algorithm quickly locks the bias voltage, it can further refine the adjustment to ensure that the modulator continues to operate in the optimal working state, thereby improving the working stability of the modulator and the overall performance of the optical communication system.

[0131] This application utilizes an efficient gradient algorithm, combined with the precise locking of the pilot method, to achieve dynamic and precise control of the bias voltage. Compared with traditional control methods, the control algorithm proposed in this application has better adaptability and stability, and can effectively cope with problems such as environmental changes and equipment aging, ensuring that the modulator can maintain the best working state under various conditions, thereby improving the reliability and transmission efficiency of the optical communication system.

[0132] In a second aspect, an embodiment of the present application further provides a lithium niobate coherent modulator control device based on a gradient algorithm.

[0133] In one embodiment, referring to Figure 3 , Figure 3 This is a functional module diagram of the lithium niobate coherent modulator control device based on the gradient algorithm of this application. Figure 3 As shown, the lithium niobate coherent modulator control device based on the gradient algorithm includes: a first setting module, a second setting module, and an execution module.

[0134] The first setting module is used to set the phases of the I and Q paths of the lithium niobate IQ modulator to be the same based on a preprocessing algorithm; the second setting module is used to set the bias voltages of the I, Q, and P paths of the lithium niobate IQ modulator to the optimal bias point through a bias voltage locking algorithm; the execution module is used to judge the working state of the lithium niobate IQ modulator under the current bias voltage based on a feedback algorithm, and to perform error compensation adjustment to optimize the working state of the lithium niobate IQ modulator.

[0135] In a third aspect, an embodiment of the present application provides a lithium niobate coherent modulator control device based on a gradient algorithm. The lithium niobate coherent modulator control device based on a gradient algorithm can be a personal computer (PC), a laptop computer, a server, or other device with data processing capabilities.

[0136] Reference Figure 4 , Figure 4 This is a hardware structure diagram of a lithium niobate coherent modulator control device based on a gradient algorithm involved in an embodiment of the present application. In the embodiment of the present application, the lithium niobate coherent modulator control device based on a gradient algorithm may include a processor, a memory, a communication interface, and a communication bus.

[0137] The communication bus may be of any type and is used to interconnect the processor, memory, and communication interface.

[0138] Communication interfaces include input / output (I / O) interfaces, physical interfaces, and logical interfaces, used to interconnect components within the gradient-based lithium niobate coherent modulator control device, as well as interfaces used to interconnect the gradient-based lithium niobate coherent modulator control device with other devices (e.g., other computing devices or user equipment). Physical interfaces can include Ethernet interfaces, fiber optic interfaces, ATM interfaces, etc.; user equipment can include displays, keyboards, etc.

[0139] The memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0140] The processor may be a general-purpose processor that can call a gradient-algorithm-based lithium niobate coherent modulator control program stored in a memory and execute the gradient-algorithm-based lithium niobate coherent modulator control method provided in the embodiments of the present application. For example, the general-purpose processor may be a central processing unit (CPU). The method executed when the gradient-algorithm-based lithium niobate coherent modulator control program is called may refer to the various embodiments of the gradient-algorithm-based lithium niobate coherent modulator control method of the present application, and will not be further described here.

[0141] Those skilled in the art will understand that Figure 4 The hardware structure shown in the figure does not constitute a limitation to the present application and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0142] The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices. The terms "first", "second" and "third" are used to distinguish different objects, etc., and do not represent a sequence, nor do they limit the "first", "second" and "third" to different types.

[0143] In the description of the embodiments of this application, the words "exemplary," "for example," or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary," "for example," or "for example" in the embodiments of this application should not be construed as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary," "for example," or "for example" is intended to present the relevant concepts in a concrete manner.

[0144] In the description of the embodiments of the present application, unless otherwise specified, “ / ” means or, for example, A / B can mean A or B; “and / or” in the text is merely a description of the association relationship of associated objects, indicating that three relationships may exist, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, “multiple” refers to two or more than two.

[0145] In some processes described in the embodiments of the present application, multiple operations or steps are included that appear in a specific order. However, it should be understood that these operations or steps may not be performed in the order in which they appear in the embodiments of the present application or may be performed in parallel. The sequence numbers of the operations are only used to distinguish between different operations, and the sequence numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations or steps may be performed in sequence or in parallel, and these operations or steps may be combined.

[0146] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, of course, it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for enabling a terminal device to execute the methods described in each embodiment of the present application.

[0147] The above are only preferred embodiments of the present application and do not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A lithium niobate coherent modulator control method based on a gradient algorithm, used to modulate a lithium niobate IQ modulator with an IQ modulation structure in coherent modulation, characterized in that: The lithium niobate coherent modulator control method based on the gradient algorithm includes: Based on the preprocessing algorithm, the phases of the I and Q paths of the lithium niobate IQ modulator are set to be the same; The bias voltage locking algorithm is used to set the bias voltages of the I, Q, and P paths of the lithium niobate IQ modulator at the optimal bias point. Based on the feedback algorithm, the working state of the lithium niobate IQ modulator under the current bias voltage is judged, and error compensation adjustment is performed to optimize the working state of the lithium niobate IQ modulator; The method of setting the phases of the I and Q paths of the lithium niobate IQ modulator to be the same based on the preprocessing algorithm specifically includes: S101: Sweep the entire operating bias voltage range of the lithium niobate IQ modulator, output a curve of optical power versus bias voltage, obtain basic characteristic data of the lithium niobate IQ modulator, set the learning rate, convergence coefficient, convergence threshold, and the initial bias voltage of each channel, and record the initial optical power. Then go to S102; S102: Calculate the gradient of the optical power of the I, Q, and P channels relative to the initial bias voltage, and move the bias voltages of the I, Q, and P channels in the direction of the maximum rising gradient of the optical power according to the set moving step size, obtain the bias voltages of the I, Q, and P channels after the shift, and record the optical power after the shift, and then go to S103; S103: Compare the current optical power after movement with the latest initial optical power. If the optical power after movement is greater than the initial optical power, go to S105. If the optical power after movement is not greater than the initial optical power, and the latest learning rate is greater than the convergence threshold, go to S104. If the optical power after movement is not greater than the initial optical power, and the latest learning rate is not greater than the convergence threshold, then before the last bias voltage movement of the I-path, Q-path, and P-path, the bias voltages of the I-path, Q-path, and P-path obtained after movement are the bias voltages corresponding to the maximum optical power, indicating that the bias voltages of the I-path, Q-path, and P-path are set to the bias voltages corresponding to the maximum optical power, and end. S104: Divide the current learning rate by the convergence coefficient to obtain the value as the latest learning rate, and go to S105; S105: The current optical power value after the shift is used as the latest initial optical power, and the process goes to S102; Among them, the set moving step size corresponding to the bias voltage of the I path is the latest learning rate multiplied by the gradient of the optical power of the I path compared to the initial bias voltage. The set moving step size corresponding to the bias voltage of the Q path is the latest learning rate multiplied by the gradient of the optical power of the Q path compared to the initial bias voltage. The set moving step size corresponding to the bias voltage of the P path is the latest learning rate multiplied by the gradient of the optical power of the P path compared to the initial bias voltage.

2. The method for controlling a lithium niobate coherent modulator based on a gradient algorithm according to claim 1, wherein: The bias voltage locking algorithm is used to set the bias voltages of the I, Q, and P paths of the lithium niobate IQ modulator at the optimal bias point, specifically including: locking the I and Q path bias voltages based on a gradient descent algorithm, locking the P path bias voltage based on a sweep spectrum algorithm, and locking the I, Q, and P path bias voltages based on a pilot algorithm.

3. The control method of a lithium niobate coherent modulator based on a gradient algorithm according to claim 2, wherein: The locking of the I-path and Q-path bias voltages based on the gradient descent algorithm specifically includes: S201: When the bias voltage corresponding to the maximum optical power is obtained, the currently recorded shifted optical power is used as the first optical power, and the process proceeds to S202; S202: Calculate the gradient of the optical power of the I and Q paths compared to the initial bias voltage, and move the bias voltages of the I and Q paths in the direction of the maximum decreasing gradient of the optical power according to a preset moving step size, obtain the bias voltages of the I and Q paths after the shift, and record the optical power after the shift, and go to S203; S203: Compare the current optical power after movement with the latest first optical power. If the optical power after movement is less than the first optical power, go to S205. If the optical power after movement is not less than the first optical power and the latest learning rate is greater than the convergence threshold, go to S204. If the optical power after movement is not less than the first optical power and the latest learning rate is not greater than the convergence threshold, the bias voltages of the I-path, Q-path and P-path obtained before the last movement of the bias voltages of the I-path, Q-path and P-path are the bias voltages corresponding to the minimum optical power, indicating that the bias voltages of the I-path and Q-path are set to the bias voltages corresponding to the minimum optical power, and end. S204: Divide the current learning rate by the convergence coefficient to obtain the value as the latest learning rate, and go to S205; S205: The current optical power value after the shift is used as the latest first optical power, and the process goes to S202; Among them, the preset moving step size corresponding to the bias voltage of the I path is the latest learning rate multiplied by the gradient of the optical power of the I path compared to the initial bias voltage, and the preset moving step size corresponding to the bias voltage of the Q path is the latest learning rate multiplied by the gradient of the optical power of the Q path compared to the initial bias voltage.

4. The method for controlling a lithium niobate coherent modulator based on a gradient algorithm according to claim 3, wherein: The locking of the P-channel bias voltage based on the spectrum sweep algorithm specifically includes: Set the bias voltage of the I and Q channels to the bias voltage corresponding to the minimum optical power, and set the bias voltage of the P channel to the bias voltage corresponding to the maximum optical power; Sweep the bias voltage of the P channel and store the optical power. Obtain any pair of adjacent optical power maxima and minima from the stored optical power. Calculate the target optical power at the Quad point based on the optical power maxima and minima. In the bias voltage range corresponding to the maximum optical power and the minimum optical power, the bias voltage point closest to the Quad point target optical power is obtained, determined as the Quad point of the P path, and the bias voltage of the P path is locked to the Quad point.

5. The method for controlling a lithium niobate coherent modulator based on a gradient algorithm according to claim 2, wherein: The I-channel, Q-channel, and P-channel bias voltages are locked based on the pilot algorithm, wherein the sine wave required for the pilot is simulated by the DAC of the bias control circuit and the sine wave of the specified frequency is simulated by the MCU timer; The simulation of the pilot signal includes: Set up the discrete sine signal array, add pilot duration, DC bias voltage, pilot sine amplitude, and set the counting threshold based on the pilot sine frequency; Start the MCU timer interrupt, start timing from 0, and set the MCU timer count to 0. The internal clock of the MCU timer triggers the count to increase, and an interrupt is generated when the count reaches the count threshold; Perform interrupt processing and generate pilot bias voltage, specifically: Wherein, VH represents the pilot bias voltage, VDC represents the DC bias voltage, A represents the pilot sine amplitude, and Sin[i] represents the discrete sine signal array generated in real time. If the length i of the discrete sine signal array generated in real time exceeds the set length of the discrete sine signal array, i is reset to 0. If i exceeds the length of the discrete sine signal array, i is reset to 0. When the timing reaches the pilot duration, the MCU timer interrupt is disabled, and the simulation of pilot signal generation ends.

6. The method for controlling a lithium niobate coherent modulator based on a gradient algorithm according to claim 5, wherein: The locking of the I-path, Q-path, and P-path bias voltages based on the pilot algorithm specifically includes: S221: Load frequency f on path I I After collecting the optical power of the set time period, stop loading the sinusoidal pilot signal, perform FFT transformation on the optical power, and record the frequency f on the spectrum. I Normalized optical power at , go to S222; S222: Determine frequency f I Is the normalized optical power at the input signal less than the set threshold? If so, go to S223; if not, adjust the bias voltage of the I path and go to S222. S223: Load frequency f on Q path Q After collecting the optical power of the set time period, stop loading the sinusoidal pilot signal, perform FFT transformation on the optical power, and record the frequency f on the spectrum. Q Normalized optical power at , go to S224; S224: Determine frequency f Q Is the normalized optical power at the position less than the set threshold? If so, go to S225; if not, adjust the bias voltage of the Q path and go to S223; S225: Load frequency f on I path I A sinusoidal pilot signal with a frequency of f is loaded on the Q path. Q After collecting the optical power of the set time period, stop loading the sinusoidal pilot signal, perform FFT transformation on the optical power, and record the frequency f on the spectrum. I +f Q The normalized optical power and |f I -f Q Normalized optical power at |, go to S226; S226: frequency f I +f Q The normalized optical power and |f I -f Q Whether the normalized optical power at | is less than the set threshold, if not, adjust the bias voltage of the P path and go to S225; if so, go to S227; S227: Execute judgment: If the frequency loaded on path I is f I After the sinusoidal pilot signal is applied, the bias voltage of the I channel is not adjusted, and the frequency f is loaded on the Q channel. Q The bias voltage of the Q circuit is not adjusted after the sinusoidal pilot signal is sent, and the frequency f is loaded on the I circuit at the same time. I The sinusoidal pilot signal and the frequency f loaded on the Q path Q If the bias voltage of the P circuit is not adjusted after the sinusoidal pilot signal is received, the process ends; otherwise, the process goes to S221.

7. The method for controlling a lithium niobate coherent modulator based on a gradient algorithm according to claim 6, wherein: The method of judging the working state of the lithium niobate IQ modulator under the current bias voltage based on the feedback algorithm and performing error compensation adjustment to optimize the working state of the lithium niobate IQ modulator specifically includes: S301: Setting the optical power change threshold and the normalized optical power threshold, and setting the bias voltages of the I, Q, and P channels at the optimal bias points. The optical power corresponding to the optimal optical power is then taken as the optimal optical power, and the process proceeds to S302. S302: Read the current optical power regularly, calculate the change between the current optical power and the optimal optical power, and then go to S303; S303: Determine whether the calculated change exceeds the optical power change threshold. If so, go to S304; if not, go to S302. S304: Loading frequency f on path I I After collecting the optical power of the set time period, stop loading the sinusoidal pilot signal, perform FFT transformation on the optical power, and record the frequency f on the spectrum. I The normalized optical power at the Q path is loaded with a frequency of f Q After collecting the optical power of the set time period, stop loading the sinusoidal pilot signal, perform FFT transformation on the optical power, and record the frequency f on the spectrum. Q The normalized optical power at the location, while the frequency f is loaded on the I path I The sinusoidal pilot signal and the frequency f loaded on the Q path Q After collecting the optical power of the set time period, stop loading the sinusoidal pilot signal, perform FFT transformation on the optical power, and record the frequency f on the spectrum. I +f Q The normalized optical power and |f I -f Q The normalized optical power at |, according to the recorded frequency f I Normalized optical power at frequency f Q Normalized optical power at frequency f I +f Q Normalized optical power at frequency |f I -f Q Whether the normalized optical power at | exceeds the normalized optical power threshold to determine whether the operating point is offset. If so, go to S305; if not, go to S306; S305: Lock the bias voltages of the I, Q, and P channels based on the pilot algorithm, perform offset correction, and then go to S306; S306: Based on the change in the optimal optical power caused by the relocking of the working point or the change in the overall optical power, the optimal optical power is updated, and the process goes to S302.

8. A lithium niobate coherent modulator control device based on a gradient algorithm, characterized in that: The lithium niobate coherent modulator control device based on the gradient algorithm includes: A first setting module is configured to set the phases of the I path and the Q path of the lithium niobate IQ modulator to be the same based on a preprocessing algorithm; A second setting module is used to set the bias voltages of the I-channel, Q-channel, and P-channel of the lithium niobate IQ modulator at an optimal bias point through a bias voltage locking algorithm; An execution module, configured to determine the operating state of the lithium niobate IQ modulator under the current bias voltage based on a feedback algorithm, and perform error compensation adjustment to optimize the operating state of the lithium niobate IQ modulator; The method of setting the phases of the I and Q paths of the lithium niobate IQ modulator to be the same based on the preprocessing algorithm specifically includes: S101: Sweep the entire operating bias voltage range of the lithium niobate IQ modulator, output a curve of optical power versus bias voltage, obtain basic characteristic data of the lithium niobate IQ modulator, set the learning rate, convergence coefficient, convergence threshold, and the initial bias voltage of each channel, and record the initial optical power. Then go to S102; S102: Calculate the gradient of the optical power of the I, Q, and P channels relative to the initial bias voltage, and move the bias voltages of the I, Q, and P channels in the direction of the maximum rising gradient of the optical power according to the set moving step size, obtain the bias voltages of the I, Q, and P channels after the shift, and record the optical power after the shift, and then go to S103; S103: Compare the current optical power after movement with the latest initial optical power. If the optical power after movement is greater than the initial optical power, go to S105. If the optical power after movement is not greater than the initial optical power, and the latest learning rate is greater than the convergence threshold, go to S104. If the optical power after movement is not greater than the initial optical power, and the latest learning rate is not greater than the convergence threshold, then before the last bias voltage movement of the I-path, Q-path, and P-path, the bias voltages of the I-path, Q-path, and P-path obtained after movement are the bias voltages corresponding to the maximum optical power, indicating that the bias voltages of the I-path, Q-path, and P-path are set to the bias voltages corresponding to the maximum optical power, and end. S104: Divide the current learning rate by the convergence coefficient to obtain the value as the latest learning rate, and go to S105; S105: The current optical power value after the shift is used as the latest initial optical power, and the process goes to S102; Among them, the set moving step size corresponding to the bias voltage of the I path is the latest learning rate multiplied by the gradient of the optical power of the I path compared to the initial bias voltage. The set moving step size corresponding to the bias voltage of the Q path is the latest learning rate multiplied by the gradient of the optical power of the Q path compared to the initial bias voltage. The set moving step size corresponding to the bias voltage of the P path is the latest learning rate multiplied by the gradient of the optical power of the P path compared to the initial bias voltage.

9. A lithium niobate coherent modulator control device based on a gradient algorithm, characterized in that: The lithium niobate coherent modulator control device based on the gradient algorithm includes a processor, a memory, and a lithium niobate coherent modulator control program based on the gradient algorithm stored in the memory and executable by the processor, wherein when the lithium niobate coherent modulator control program based on the gradient algorithm is executed by the processor, the steps of the lithium niobate coherent modulator control method based on the gradient algorithm as described in any one of claims 1 to 7 are implemented.

Images