Method and system for automatic control of optical-electrical modulator bias point

Abstract

The application discloses a photoelectric modulator bias point automatic control method and system, and relates to the field of signal processing, which comprises the following steps: generating a composite disturbance voltage signal through a digital signal processor; injecting the bias electrode of a photoelectric modulator, collecting the optical signal output by the photoelectric modulator, and converting the optical signal into an electrical signal; extracting the amplitude of the fundamental wave component, the phase difference data of the second and third harmonic components and the fundamental wave; calculating dynamic characteristic parameters through spectrum analysis, and establishing the mapping relationship between the characteristic parameters and the bias point offset; outputting the offset direction and offset level of the bias point through a deep learning model; generating a bias voltage compensation amount through a fuzzy PID controller; and adjusting the bias electrode voltage, performing a smoothing filter, and eliminating high-frequency jitter. The application has the advantages that: through the composite disturbance signal, real-time spectrum analysis and deep learning optimization decision, combined with the fuzzy PID control and dynamic closed-loop mechanism, high-precision and stable bias point adjustment is realized, and the system performance and stability are significantly improved.

Description

Technical Field

[0001] This invention relates to the field of signal processing, and in particular to an automatic control method and system for the bias point of an optoelectronic modulator. Background Technology

[0002] Optoelectronic modulators are crucial components in optical communication systems, responsible for converting electrical signals into optical signals and modulating them. Their performance directly impacts the stability and transmission efficiency of the communication system. With advancements in communication technology and the diversification of application scenarios, the research and application of automatic bias point control methods for optoelectronic modulators provide a solid foundation and support for further innovation and application of optical communication technology.

[0003] Currently available automatic bias point control methods for optoelectronic modulators largely rely on single-frequency perturbation signals or simple control algorithms. This design often fails to provide sufficient dynamic response and precise adjustment capabilities when facing complex operating environments or frequently changing optical signal characteristics, leading to unstable system performance under high-frequency jitter or noise interference. Secondly, some methods employ relatively simple algorithms in spectrum analysis and feature parameter extraction, lacking targeted optimization and deep learning decision support. This makes them unable to effectively handle the dynamic adjustment requirements of complex data scenarios, affecting the system's real-time performance and adaptive capabilities. Furthermore, some commercially available methods are conservative in their control strategy selection, failing to fully utilize advanced technologies such as fuzzy PID controllers for compensation optimization. This limits the system's reliability and sustained performance under long-term stable operation and extreme working conditions. Summary of the Invention

[0004] To improve existing methods and systems, an automatic control method and system for the bias point of an optoelectronic modulator is provided. This method achieves high-precision and stable bias point adjustment by combining composite disturbance signals, real-time spectrum analysis, and deep learning optimization decision-making with fuzzy PID control and dynamic closed-loop mechanism, significantly improving system performance and stability.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] An automatic control method for the bias point of an optoelectronic modulator includes:

[0007] A composite perturbation voltage signal is generated by a digital signal processor. The signal is composed of a low-frequency sinusoidal signal and a broadband pseudo-random coded signal linearly superimposed.

[0008] The composite perturbation signal is injected into the bias electrode of the photoelectric modulator, and the optical signal output by the photoelectric modulator is collected in real time and converted into an electrical signal by a photodetector.

[0009] Real-time spectrum analysis of the electrical signal, synchronous extraction of the fundamental component amplitude, the second harmonic component amplitude, the third harmonic component and the phase difference between the fundamental component and the third harmonic component through band-pass filtering, analog-to-digital conversion and fast Fourier transform;

[0010] Based on the obtained spectrum analysis data, the dynamic characteristic parameter is calculated, and a mapping relationship between the characteristic parameter and the bias point offset is established.

[0011] The dynamic characteristic parameter is input into a pre-trained deep learning model, and the offset direction and offset level of the current bias point relative to the target working point are output, and bias point offset quantization data is obtained.

[0012] Based on the offset direction and offset level, a fuzzy PID controller is driven to generate a bias voltage compensation amount.

[0013] The bias voltage compensation amount is superimposed on the direct current bias end of the modulator, the bias electrode voltage is adjusted, high frequency jitter is eliminated through a voltage smoothing filter, and the above steps are repeated to perform dynamic closed-loop control on the bias point.

[0014] Preferably, the composite disturbance voltage signal generated by the digital signal processor is linearly superimposed by a low-frequency sinusoidal signal and a wideband pseudo-random code signal, specifically including:

[0015] A low-frequency sinusoidal signal and a wideband pseudo-random code signal are generated by a digital signal processor based on the low-frequency working frequency range of the optoelectronic modulator;

[0016] The low-frequency sinusoidal signal and the wideband pseudo-random code signal are linearly superimposed to form a composite disturbance voltage signal, and the composite disturbance voltage signal contains low-frequency stationary disturbance and high-frequency random disturbance;

[0017] Based on the superimposed signal, the amplitude of the composite disturbance signal is adjusted within the working range of the optoelectronic modulator;

[0018] The generated composite disturbance signal is output by the digital signal processor.

[0019] Preferably, the composite disturbance signal is injected into the bias electrode of the optoelectronic modulator, and the optical signal output by the optoelectronic modulator is collected in real time, and the optical signal is converted into an electrical signal by an optoelectronic detector, specifically including:

[0020] The generated composite disturbance voltage signal is converted into an analog signal by a digital-to-analog converter, and the composite disturbance signal is injected into the bias electrode of the optoelectronic modulator;

[0021] The optoelectronic modulator adjusts the bias point based on the injected composite disturbance signal, thereby modulating the optical signal passing through it and outputting it;

[0022] The light signal is collected in real time by a photoelectric detector, and the light signal is converted into an electric signal based on the change of light intensity;

[0023] The converted electric signal is amplified by a low-noise amplifier, high-frequency noise is removed by a filter, and a clear electric signal is obtained.

[0024] Preferably, the real-time spectrum analysis of the electric signal includes synchronous extraction of the fundamental component amplitude, the second harmonic component amplitude, and the phase difference data of the third harmonic component and the fundamental wave through band-pass filtering, analog-to-digital conversion, and fast Fourier transform.

[0025] The clear electric signal obtained is converted into a digital signal by an analog-to-digital converter;

[0026] The digital signal is processed by fast Fourier transform to convert the time-domain signal into a frequency-domain signal, generating a frequency spectrum graph containing amplitude and phase data of the signal at different frequency points;

[0027] Based on the obtained frequency spectrum graph, the amplitude value corresponding to the frequency point is found to obtain the fundamental component amplitude;

[0028] The second harmonic frequency is twice the fundamental frequency, and the amplitude value of the second harmonic frequency point in the frequency spectrum is found to obtain the second harmonic component amplitude;

[0029] The third harmonic frequency is three times the fundamental frequency, and the amplitude value of the third harmonic frequency point in the frequency spectrum is found to obtain the third harmonic component amplitude;

[0030] The phase of the fundamental wave and the phase of the harmonic are obtained through the fundamental frequency point, and the phase difference between the fundamental wave and the harmonic is calculated.

[0031] Preferably, the mapping relationship between the characteristic parameters and the bias point offset is established based on the obtained spectrum analysis data, and the dynamic characteristic parameters are calculated, which specifically includes:

[0032] The dynamic characteristic parameters are calculated based on the obtained fundamental component amplitude, second harmonic component amplitude, and phase difference data of the third harmonic component and the fundamental wave;

[0033] The calculated dynamic characteristic parameters are converted into a linear representation of the bias point offset through a logarithmic mapping function.

[0034] Preferably, the dynamic characteristic parameters are input into a pre-trained deep learning model to output the offset direction and offset level of the current bias point relative to the target working point, and the bias point offset quantization data is obtained, which specifically includes:

[0035] The dynamic characteristic parameters are combined into a standardized input vector, and a one-dimensional convolution kernel is used to scan the input vector in the time domain, each convolution kernel extracts local features, and outputs a feature map.

[0036] The feature map is input into the LSTM unit in time sequence to generate a hidden state vector, and an original decision vector is generated through output;

[0037] The offset level is output through the classification model, and the continuous offset numerical value is output through the regression model, and the offset quantization data of the bias point is obtained by integration.

[0038] Preferably, the fuzzy PID controller is driven based on the offset direction and the offset level to generate the bias voltage compensation, which specifically includes:

[0039] Based on the obtained offset quantization data of the bias point, a set of fuzzy logic rules is set, including the control action corresponding to the offset direction and the level;

[0040] The actual bias point offset is converted into a fuzzy set through fuzzification, and the corresponding control output is generated according to the designed fuzzy control rule, and then the control amount of the fuzzy PID controller is obtained through defuzzification;

[0041] The bias voltage is adjusted based on the signal of the control amount of the fuzzy PID controller;

[0042] Based on the real-time offset direction and offset level, the parameters of the fuzzy logic rule and the fuzzy PID controller are updated to optimize the control performance and generate the bias voltage compensation.

[0043] Preferably, the bias voltage compensation is superimposed on the DC bias end of the modulator to adjust the bias electrode voltage, and high-frequency jitter is eliminated through a voltage smoothing filter, and the above steps are repeated to perform dynamic closed-loop control on the bias point, which specifically includes:

[0044] Based on the obtained bias voltage compensation, it is superimposed on the DC bias end of the modulator;

[0045] By adjusting the bias electrode voltage of the modulator, the bias current or voltage is changed to realize fine adjustment of the bias point;

[0046] During the adjustment of the compensation voltage and the bias electrode voltage, a smoothing filter is added in the circuit to eliminate high-frequency jitter and noise;

[0047] The bias voltage and the output signal of the system are monitored in real time, and the compensation voltage and the bias electrode voltage are adjusted according to the actual measurement results;

[0048] Through the feedback control mechanism, the bias voltage is adjusted according to the change of the output signal to maintain the stability of the bias point.

[0049] Preferably, the photoelectric modulator is a Mach-Zehnder lithium niobate modulator, the bias electrode includes: a main bias electrode for coarse adjustment of the operating point, the voltage range is -5V~+5V; a fine adjustment electrode for dynamic compensation, the voltage range is -0.5V~+0.5V; the compensation amount preferentially acts on the fine adjustment electrode, and when the cumulative compensation amount exceeds a threshold value, the main bias electrode is switched to.

[0050] Further, the photoelectric modulator bias point automatic control system is provided, comprising:

[0051] The composite disturbance signal generation module generates a composite disturbance signal of a low-frequency sinusoidal signal and a wideband pseudo-random coded signal through a digital signal processor, and provides a disturbance signal for the bias electrode;

[0052] The signal injection and optical signal acquisition module injects the composite disturbance signal into the bias electrode of the photoelectric modulator, acquires the optical signal output by the photoelectric modulator in real time and converts it into an electrical signal;

[0053] The spectrum analysis module performs fast Fourier transform analysis on the electrical signal, extracts fundamental wave, second harmonic, third harmonic components and phase difference data, and obtains spectrum analysis results;

[0054] The characteristic parameter calculation module calculates dynamic characteristic parameters based on the spectrum data and establishes a mapping relationship with the bias point offset;

[0055] The deep learning decision module inputs the dynamic characteristic parameters into a pre-trained deep learning model, outputs the offset direction and offset level of the bias point, and obtains bias point offset quantization data;

[0056] The fuzzy PID control module calculates the bias voltage compensation amount through a fuzzy PID controller according to the offset direction and offset level, and generates a control signal:

[0057] The closed-loop control module superimposes the bias voltage compensation amount to the direct current bias end of the modulator, adjusts the bias electrode voltage, eliminates high-frequency noise through a smoothing filter, and performs closed-loop control;

[0058] The processor is used for processing the calculation process of each formula and the construction calculation process of each model.

[0059] Compared with the prior art, the advantages of the present application are:

[0060] By introducing a composite disturbance voltage signal, combining low-frequency sinusoidal signals and wideband pseudo-random coded signals, the limitations of single-frequency disturbance are effectively avoided, thus providing more comprehensive bias point adjustment capabilities. Secondly, real-time spectral analysis and accurate decision-making process based on deep learning model make the dynamic adjustment of bias point more precise, and can monitor and adapt to changes in the working environment in real time. In addition, the fuzzy PID controller is used to optimize the compensation amount, combined with the dynamic closed-loop control mechanism, to ensure that the bias point of the modulator can be continuously maintained in the best working state, avoiding the interference of high-frequency jitter and noise, and further improving the stability and reliability of the system. Especially in the Mach-Zehnder lithium niobate modulator, through the joint adjustment of the main bias electrode and the fine adjustment electrode, the wide-range adjustment and fine adjustment requirements of the bias point can be efficiently met, ensuring the high-precision working effect. BRIEF DESCRIPTION OF DRAWINGS

[0061] Figure 1 Method flowchart for the present application;

[0062] Figure 2 Composite disturbance voltage signal generation schematic diagram for the present application;

[0063] Figure 3 Conversion electrical signal schematic diagram for the present application;

[0064] Figure 4 Spectral analysis data schematic diagram for the present application;

[0065] Figure 5 Dynamic characteristic parameter calculation schematic diagram for the present application;

[0066] Figure 6 Bias point offset quantization data acquisition schematic diagram for the present application;

[0067] Figure 7 Bias voltage compensation amount generation schematic diagram for the present application;

[0068] Figure 8 Dynamic closed-loop control schematic diagram for the present application. DETAILED DESCRIPTION

[0069] The following description is used to disclose the present application so that those skilled in the art can implement the present application. The preferred embodiments in the following description are only as examples, and other obvious modifications can be made by those skilled in the art.

[0070] The photoelectric modulator bias point automatic control system comprises:

[0071] Composite disturbance signal generation module: the module generates a composite disturbance signal of low-frequency sinusoidal signals and wideband pseudo-random coded signals through a digital signal processor, and provides a disturbance signal for the bias electrode;

[0072] Signal injection and optical signal acquisition module: the module injects the composite disturbance signal into the bias electrode of the optoelectronic modulator, acquires the optical signal output by the optoelectronic modulator in real time and converts it into an electrical signal;

[0073] Spectrum analysis module: the module performs fast Fourier transform analysis on the electrical signal, extracts fundamental, second harmonic, third harmonic components and phase difference data, and obtains spectrum analysis results;

[0074] Characteristic parameter calculation module: the module calculates dynamic characteristic parameters based on spectrum data and establishes a mapping relationship between the characteristic parameters and the bias point offset;

[0075] Deep learning decision module: the module inputs the dynamic characteristic parameters into a pre-trained deep learning model, outputs the offset direction and offset level of the bias point, and obtains bias point offset quantization data;

[0076] Fuzzy PID control module: the module calculates the bias voltage compensation amount through a fuzzy PID controller according to the offset direction and offset level, and generates a control signal:

[0077] Closed-loop control module: the module superimposes the bias voltage compensation amount to the DC bias end of the modulator, adjusts the bias electrode voltage, and eliminates high-frequency noise through a smoothing filter to perform closed-loop control;

[0078] Processor: the processor is used to process the calculation process of each formula and the construction calculation process of each model.

[0079] Referring to Figure 1 The optoelectronic modulator bias point automatic control method comprises the following steps:

[0080] Step one: generate a composite disturbance voltage signal through a digital signal processor, the signal being composed of a low-frequency sinusoidal signal and a wideband pseudo-random coded signal linearly superimposed;

[0081] Step two: inject the composite disturbance signal into the bias electrode of the optoelectronic modulator, acquire the optical signal output by the optoelectronic modulator in real time, and convert it into an electrical signal through a photodetector;

[0082] Step three: perform real-time spectrum analysis on the electrical signal, and synchronously extract the fundamental component amplitude, second harmonic component amplitude, third harmonic component and phase difference data of the fundamental component through bandpass filtering, analog-to-digital conversion and fast Fourier transform;

[0083] Step four: based on the obtained spectrum analysis data, calculate the dynamic characteristic parameters and establish a mapping relationship between the characteristic parameters and the bias point offset;

[0084] Step five: input the dynamic characteristic parameters into the pre-trained deep learning model, output the offset direction and offset level of the current bias point relative to the target working point, and obtain the bias point offset quantization data;

[0085] Step six: drive the fuzzy PID controller based on the offset direction and offset level to generate the bias voltage compensation;

[0086] Step seven: superimpose the bias voltage compensation to the direct current bias end of the modulator, adjust the bias electrode voltage, eliminate high frequency jitter through the application of voltage smoothing filter, and repeat the above steps to perform dynamic closed loop control on the bias point.

[0087] Referring to Figure 2 As shown, a composite disturbance voltage signal is generated by a digital signal processor, which is composed of linear superposition of a low-frequency sinusoidal signal and a wideband pseudo-random coded signal, specifically including:

[0088] A low-frequency sinusoidal signal and a wideband pseudo-random coded signal are generated by a digital signal processor based on the low-frequency working frequency range of the optoelectronic modulator;

[0089] The low-frequency sinusoidal signal and the wideband pseudo-random coded signal are linearly superimposed to form a composite disturbance voltage signal, which contains low-frequency stationary disturbance and high-frequency random disturbance;

[0090] Based on the superimposed signal, the amplitude of the composite disturbance signal is adjusted within the working range of the optoelectronic modulator;

[0091] The generated composite disturbance signal is output by the digital signal processor.

[0092] Specifically, the generation of the low-frequency sinusoidal signal can be realized by a digital signal processor, and the frequency of the low-frequency sinusoidal signal is set to The signal is expressed in time domain as:

[0093]

[0094] Wherein, is the voltage value of the low-frequency sinusoidal signal, is the amplitude of the low-frequency sinusoidal signal, is the frequency of the low-frequency sinusoidal signal, is the initial phase of the signal;

[0095] The wideband pseudo-random coded signal is usually generated by a pseudo-random sequence generation algorithm, such as pseudo-random binary sequence PRBS;

[0096] The low-frequency sinusoidal signal is linearly superimposed with the wideband pseudo-random coded signal to obtain a composite disturbance signal. In order to ensure that the amplitude of the composite disturbance signal is within the working range of the optoelectronic modulator, the amplitude of the signal needs to be adjusted. Finally, the adjusted composite disturbance signal is output by the digital signal processor.

[0097] Referring to Figure 3 The composite disturbance signal is injected into the bias electrode of the optoelectronic modulator, and the optical signal output by the optoelectronic modulator is collected in real time. The optical signal is converted into an electrical signal by a photodetector, and the method specifically comprises the following steps:

[0098] The generated composite disturbance voltage signal is converted into an analog signal by a digital-to-analog converter, and the composite disturbance signal is injected into the bias electrode of the optoelectronic modulator;

[0099] The optoelectronic modulator adjusts the bias point based on the injected composite disturbance signal, thereby modulating the optical signal passing through it and outputting the optical signal;

[0100] The optical signal is collected in real time by a photodetector, and the optical signal is converted into an electrical signal based on the change in light intensity;

[0101] The converted electrical signal is amplified by a low-noise amplifier, and high-frequency noise is removed by a filter to obtain a clear electrical signal.

[0102] Specifically, the composite disturbance signal is converted from a digital signal into an analog voltage signal by a digital-to-analog converter, which is suitable for injection into the bias electrode of the optoelectronic modulator. The composite disturbance signal converted by the digital-to-analog converter is injected into the bias electrode of the optoelectronic modulator. The bias electrode of the optoelectronic modulator is used to adjust its operating point. Based on the input disturbance signal, the optoelectronic modulator will adjust the modulation depth of the optical signal, thereby affecting the intensity of the optical signal passing through the modulator.

[0103] The optoelectronic modulator adjusts its optical signal output according to the input composite disturbance signal. The output optical signal intensity formula of the modulator is:

[0104]

[0105] wherein, is the optical signal intensity output by the optoelectronic modulator, is the static light intensity of the optoelectronic modulator, is the modulation sensitivity of the modulator, is the voltage signal on the bias electrode;

[0106] The modulated light signal is collected by a photodetector. The photodetector converts the light signal into an electrical signal. In order to enhance the collected electrical signal and remove high-frequency noise, the electrical signal needs to be amplified by a low-noise amplifier, and then a filter is used to remove high-frequency noise to obtain a clear electrical signal.

[0107] Referring to Figure 4 The electrical signal is subjected to real-time spectral analysis, and the amplitude of the fundamental component, the amplitude of the second harmonic component, and the phase difference data of the third harmonic component and the fundamental component are extracted synchronously by band-pass filtering, analog-to-digital conversion, and fast Fourier transform, specifically including:

[0108] The clear electrical signal obtained is converted into a digital signal by an analog-to-digital converter;

[0109] The digital signal is processed by fast Fourier transform to convert the time-domain signal into a frequency-domain signal, generating a frequency spectrum graph containing the amplitude and phase data of the signal at different frequency points;

[0110] Based on the obtained frequency spectrum graph, the amplitude value corresponding to the frequency point is found to obtain the amplitude of the fundamental component;

[0111] The second harmonic frequency is twice the fundamental frequency, and the amplitude value of the second harmonic frequency point in the frequency spectrum is found to obtain the amplitude of the second harmonic component;

[0112] The third harmonic frequency is three times the fundamental frequency, and the amplitude value of the third harmonic frequency point in the frequency spectrum is found to obtain the amplitude of the third harmonic component;

[0113] The phase of the fundamental and the phase of the harmonic are obtained by the fundamental frequency point, and the phase difference between the fundamental and the harmonic is calculated.

[0114] Specifically, the time-domain signal is converted into a frequency-domain signal by fast Fourier transform, and the frequency spectrum graph is obtained by fast Fourier transform processing, containing the amplitude and phase information of each frequency point;

[0115] Based on the frequency spectrum graph, the amplitudes of the fundamental, second harmonic, and third harmonic components are obtained, and the phase difference is calculated according to the extracted phase information of the fundamental and the harmonic;

[0116] In order to avoid the influence of low-frequency noise or high-frequency interference in the frequency spectrum, a band-pass filter is used to extract the frequency band, and the design of the band-pass filter adjusts the passband of the filter according to the fundamental frequency and the harmonic frequency range.

[0117] Referring to Figure 5 Based on the obtained spectral analysis data, the dynamic characteristic parameters are calculated, and the mapping relationship between the characteristic parameters and the bias point offset is established, specifically including:

[0118] The dynamic characteristic parameter is calculated based on the obtained fundamental component amplitude, second harmonic component amplitude, third harmonic component and phase difference data of the fundamental wave;

[0119] The dynamic characteristic parameter is converted into a linear representation of the bias point offset through a logarithmic mapping function.

[0120] Specifically, the dynamic characteristic parameter calculation formula is:

[0121]

[0122] wherein, is the dynamic characteristic parameter, is the fundamental component amplitude, is the second harmonic component amplitude, is the third harmonic phase difference, is the phase sensitive coefficient;

[0123] In order to make the relationship between the characteristic parameter and the bias point offset more consistent with the response characteristics of the actual system, the dynamic characteristic parameter is converted into the bias point offset using a logarithmic mapping function. The logarithmic mapping function can convert the nonlinear change of the dynamic characteristic parameter into the linear change of the bias point, which makes the adjustment of the bias voltage more smooth and avoids excessive fluctuations caused by nonlinear changes. The formula is:

[0124]

[0125] wherein, is the bias point offset, is the target characteristic parameter, is the modulator sensitivity coefficient.

[0126] Referring to FIG. 1, Figure 6 the dynamic characteristic parameter is input into a pre-trained deep learning model, and the output is the offset direction and offset level of the current bias point relative to the target working point. The bias point offset quantization data specifically includes:

[0127] The dynamic characteristic parameter is combined into a standardized input vector, and the input vector is scanned in the time domain by a one-dimensional convolution kernel. Each convolution kernel extracts local features and outputs a feature map;

[0128] The feature map is input into an LSTM unit in a time sequence to generate a hidden state vector, and an original decision vector is output through the output;

[0129] The classification model outputs the level of the offset, the regression model outputs continuous offset numerical values, and the bias point offset quantization data is obtained by integration.

[0130] Referring to FIG. 1, Figure 7As shown, the offset voltage compensation is generated based on the offset direction and offset level driving the fuzzy PID controller, and specifically includes:

[0131] Based on the acquired offset point offset quantization data, the fuzzy logic rule set is set, including the control action corresponding to the offset direction and level;

[0132] The actual offset point offset is converted into a fuzzy set by fuzzification, and the corresponding control output is generated according to the designed fuzzy control rule, and the control amount of the fuzzy PID controller is obtained by defuzzification;

[0133] The bias voltage is adjusted based on the signal of the fuzzy PID controller control amount;

[0134] Based on the real-time offset direction and offset level, the parameters of the fuzzy logic rule and the fuzzy PID controller are updated, the control performance is optimized, and the offset voltage compensation is generated.

[0135] Specifically, the offset point offset of the input signal is acquired and quantized into a discrete value. By measuring the offset of the system, the current offset and the offset direction and level are obtained, for example, large, medium, small, positive, and negative.

[0136] Based on the working characteristics and requirements of the system, the fuzzy control rule is designed. The role of the rule is to map the direction and level of the offset to the control action, to convert the actual offset data into a fuzzy set, to obtain the membership degree, and to generate the control output according to the fuzzy result and the designed fuzzy rule. For each fuzzy rule, the fuzzy result of each control action is obtained according to the input membership degree and the rule result;

[0137] The result of fuzzy reasoning is converted into a specific control output, and the bias voltage is adjusted according to the control amount obtained by defuzzification. In the control process, the system needs to monitor the changes of the offset direction and the offset level in real time, and adjust the parameters of the fuzzy rule and the PID controller according to the new data to optimize the control performance.

[0138] Referring to Figure 8 The offset voltage compensation is added to the DC bias end of the modulator to adjust the bias electrode voltage, and the high-frequency jitter is eliminated by applying a voltage smoothing filter. The above steps are repeated to perform dynamic closed-loop control on the offset point, and specifically include:

[0139] Based on the acquired offset voltage compensation, it is added to the DC bias end of the modulator;

[0140] By adjusting the bias electrode voltage of the modulator, the bias current or voltage is changed to achieve fine adjustment of the offset point;

[0141] In the adjustment process of the compensation voltage and the bias electrode voltage, a smoothing filter is added in the circuit to eliminate high-frequency jitter and noise;

[0142] The bias voltage and the output signal of the system are monitored in real time, and the compensation voltage and the bias electrode voltage are adjusted according to the actual measurement results;

[0143] Through a feedback control mechanism, the bias voltage is adjusted according to the change of the output signal to maintain the stability of the bias point.

[0144] Specifically, the bias voltage compensation is used to adjust the direct current bias point of the modulator, and the purpose is to adjust the working state of the modulator to an ideal working point. Before obtaining the compensation, the current bias voltage needs to be measured first and compared with the target value. The calculated compensation is added to the direct current bias end of the modulator to adjust the bias point of the modulator. The adjustment of the bias voltage is usually realized by the power supply module, that is, the voltage applied at the power supply end of the modulator adjusts the working state of the modulator;

[0145] The adjustment of the bias electrode voltage is used to finely adjust the bias current or voltage of the modulator, so as to further optimize the working point. The relationship between the bias electrode voltage and the bias current or voltage is defined by the input-output characteristics of the modulator, and is usually adjusted by a linear or nonlinear relationship;

[0146] In the real-time monitoring process, a closed-loop feedback control algorithm is used to adjust the bias voltage in real time to maintain its stability and accurately adjust the bias voltage so that the output signal of the modulator always remains near the target value.

[0147] It should be noted that the above-mentioned embodiments of the present application are only for description, and do not represent the advantages and disadvantages of the embodiments. The above describes a specific embodiment of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are possible or may be advantageous.

[0148] Each embodiment in the specification is described in a progressive manner, and the same or similar parts between each embodiment can be referred to each other. Each embodiment focuses on the difference from other embodiments.

[0149] The above is only the preferred embodiment of the present application, and does not limit the present application. Any modification, equivalent replacement, improvement, etc. made within the principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for automatic control of the bias point of an electro-optic modulator, characterized in that, The application relates to a method for dynamically adjusting the bias point of an optical modulator. The method comprises the following steps: generating a composite disturbance voltage signal by a digital signal processor, the signal being composed of a low-frequency sinusoidal signal and a wideband pseudo-random coded signal linearly superimposed; injecting the composite disturbance signal into the bias electrode of the optoelectronic modulator, collecting the optical signal output by the optoelectronic modulator in real time, and converting the optical signal into an electrical signal by an optoelectronic detector; performing real-time spectral analysis on the electrical signal, and synchronously extracting the amplitude of the fundamental component, the amplitude of the second harmonic component, and the phase difference between the third harmonic component and the fundamental component by band-pass filtering, analog-to-digital conversion, and fast Fourier transform; calculating dynamic characteristic parameters based on the obtained spectral analysis data, and establishing a mapping relationship between the characteristic parameters and the bias point offset; inputting the dynamic characteristic parameters into a pre-trained deep learning model to output the offset direction and offset level of the current bias point relative to the target working point, and obtaining bias point offset quantization data; driving a fuzzy PID controller based on the offset direction and offset level to generate a bias voltage compensation amount; 2. The method of claim 1, wherein, superimposing the bias voltage compensation amount on the direct current bias end of the modulator, adjusting the bias electrode voltage, eliminating high-frequency jitter through a voltage smoothing filter, and repeating the above steps to perform dynamic closed-loop control on the bias point. The method for dynamically adjusting the bias point of an optical modulator comprises the following steps: generating a low-frequency sinusoidal signal and a wideband pseudo-random coded signal by a digital signal processor based on the low-frequency working frequency range of the optoelectronic modulator; linearly superimposing the low-frequency sinusoidal signal and the wideband pseudo-random coded signal to form a composite disturbance voltage signal, wherein the composite disturbance voltage signal contains low-frequency stationary disturbance and high-frequency random disturbance; adjusting the amplitude of the superimposed signal to make the amplitude of the composite disturbance signal within the working range of the optoelectronic modulator; 3. The method of claim 1, wherein, outputting the generated composite disturbance signal by the digital signal processor. The method for dynamically adjusting the bias point of an optical modulator comprises the following steps: converting the generated composite disturbance voltage signal into an analog signal by a digital-to-analog converter, and injecting the composite disturbance signal into the bias electrode of the optoelectronic modulator; adjusting the bias point of the optoelectronic modulator based on the injected composite disturbance signal, and then modulating the optical signal passing through the optoelectronic modulator and outputting the optical signal; collecting the optical signal in real time by an optoelectronic detector, and converting the optical signal into an electrical signal based on the change in light intensity; 4. The method of claim 1, wherein, amplifying the converted electrical signal by a low-noise amplifier, removing high-frequency noise by a filter, and obtaining a clear electrical signal. The method for dynamically adjusting the bias point of an optical modulator comprises the following steps: converting the clear electrical signal into a digital signal by an analog-to-digital converter; processing the digital signal by fast Fourier transform to convert the time-domain signal into a frequency-domain signal, and generating a frequency spectrum diagram containing the amplitude and phase data of the signal at different frequency points; Based on the obtained frequency spectrum, the fundamental component amplitude value is obtained by searching the amplitude value corresponding to the frequency point; The second harmonic frequency is twice the fundamental frequency, and the second harmonic component amplitude value is obtained by finding the amplitude value of the second harmonic frequency point in the frequency spectrum; The third harmonic frequency is three times the fundamental frequency, and the third harmonic component amplitude value is obtained by finding the amplitude value of the third harmonic frequency point in the frequency spectrum; The phase of the fundamental wave and the phase of the harmonic are obtained through the fundamental frequency point, and the phase difference between the fundamental wave and the harmonic is calculated.

5. The method of claim 1, wherein, The mapping relationship between the characteristic parameters and the bias point offset is established based on the obtained frequency spectrum analysis data, and the dynamic characteristic parameters are calculated. Based on the obtained fundamental component amplitude value, second harmonic component amplitude value, third harmonic component and phase difference data of the fundamental wave, the dynamic characteristic parameters are calculated. The calculated dynamic characteristic parameters are converted into a linear representation of the bias point offset through a logarithmic mapping function.

6. The method of claim 1, wherein, The dynamic characteristic parameters are input into the pre-trained deep learning model, and the output is the offset direction and offset level of the current bias point relative to the target working point, and the bias point offset quantization data is obtained. The dynamic characteristic parameters are combined into a standardized input vector, and a one-dimensional convolution kernel is used to slide and scan the input vector in the time domain. Each convolution kernel extracts local features and outputs a feature map. The feature map is input into the LSTM unit in time sequence to generate a hidden state vector, and the output is the original decision vector. The classification model outputs the level of the offset, and the regression model outputs the continuous offset value. The bias point offset quantization data is obtained by integrating the two.

7. The method of claim 1, wherein, Based on the offset direction and offset level, the fuzzy PID controller is driven to generate the bias voltage compensation. Based on the obtained bias point offset quantization data, the fuzzy logic rule set is set, including the control action corresponding to the offset direction and level. The actual bias point offset is converted into a fuzzy set through fuzzification, and the corresponding control output is generated according to the designed fuzzy control rule. The control amount of the fuzzy PID controller is obtained through defuzzification. The bias voltage is adjusted based on the signal of the fuzzy PID controller control amount. Based on the real-time offset direction and offset level, the fuzzy logic rules and the parameters of the fuzzy PID controller are updated to optimize the control performance and generate the bias voltage compensation.

8. The method of claim 1, wherein, The bias voltage compensation is superimposed on the DC bias end of the modulator to adjust the bias electrode voltage. The high-frequency jitter is eliminated through the application of the voltage smoothing filter. The above steps are repeated to perform dynamic closed-loop control on the bias point. Based on the obtained bias voltage compensation, it is superimposed on the DC bias end of the modulator. By adjusting the bias electrode voltage of the modulator, the bias current or voltage is changed to achieve fine adjustment of the bias point. During the adjustment of the compensation voltage and the bias electrode voltage, a smoothing filter is added to the circuit to eliminate high-frequency jitter and noise. The bias voltage and output signal of the system are monitored in real time, and the compensation voltage and bias electrode voltage are adjusted according to the actual measurement results. Through the feedback control mechanism, the bias voltage is adjusted according to the change of the output signal to maintain the stability of the bias point.

9. The method of claim 1, wherein, The photoelectric modulator is a Mach-Zehnder lithium niobate modulator, the bias electrode includes: a main bias electrode for coarse adjustment of the operating point, the voltage range is-5V~+5V; a fine adjustment electrode for dynamic compensation, the voltage range is-0.5V~+0.5V; the compensation amount preferentially acts on the fine adjustment electrode, and when the cumulative compensation amount exceeds the threshold, the main bias electrode is switched to.

10. An optical modulator bias point automatic control system for implementing the optical modulator bias point automatic control method according to any one of claims 1 to 9, characterized by, Comprise: Composite disturbance signal generation module: the module generates a composite disturbance signal of a low-frequency sinusoidal signal and a wideband pseudo-random coded signal through a digital signal processor, and provides a disturbance signal for the bias electrode; Signal injection and optical signal acquisition module: the module injects the composite disturbance signal into the bias electrode of the photoelectric modulator, acquires the optical signal output by the photoelectric modulator in real time and converts it into an electrical signal; Spectrum analysis module: the module performs fast Fourier transform analysis on the electrical signal, extracts fundamental, second harmonic, third harmonic components and phase difference data, and obtains spectrum analysis results; Characteristic parameter calculation module: the module calculates dynamic characteristic parameters based on spectrum data and establishes a mapping relationship with bias point offset; Deep learning decision module: the module inputs the dynamic characteristic parameters into a pre-trained deep learning model, outputs the offset direction and offset level of the bias point, and obtains bias point offset quantization data; Fuzzy PID control module: the module calculates the bias voltage compensation amount according to the offset direction and offset level through a fuzzy PID controller, and generates a control signal: Closed-loop control module: the module superimposes the bias voltage compensation amount to the DC bias end of the modulator, adjusts the bias electrode voltage, and eliminates high-frequency noise through a smoothing filter to perform closed-loop control; Processor: the processor is used to process the calculation process of each formula and the construction calculation process of each model.

Images