A method for controlling a light source for optical fiber communication

Abstract

The application discloses an optical fiber communication light source control method, and relates to the technical field of control analysis, which comprises the following steps: superimposing pilot perturbation data of a non-communication frequency band on a light source bias current; analyzing backscattering light power waveform data after the pilot perturbation data is superimposed, and extracting an intermodulation distortion component about an oscillation frequency and a pilot frequency in the waveform data; when the amplitude of the intermodulation distortion component is greater than a preset diagnosis threshold, injecting high-frequency modulation chirp data into the bias current, and using carrier dispersion effect to cause light source wavelength jitter data; synchronously updating oscillation amplitude data of a forward dynamic amplitude-frequency matrix; when the oscillation amplitude data decays below a preset safety threshold and is maintained for a preset time, and meanwhile dynamic phase difference data deviates from a fixed locking interval, gradually reducing and canceling the high-frequency modulation chirp data, jumping out of a parasitic resonant cavity diagnosis cycle, and outputting an oscillation suppression completion signal. The application has the effect of improving light source control efficiency.

Description

Technical Field

[0001] This application relates to the field of control and analysis technology, and in particular to a control method for an optical fiber communication light source. Background Technology

[0002] In fiber optic communication systems, backscattered light generated by microbending of the link can easily form a parasitic resonant cavity with the main optical path of the light source, triggering spontaneous oscillations within the light source. Existing light source control technologies generally suffer from drawbacks when dealing with such transient oscillations caused by sudden changes in the external environment, including unclear fault cause identification, lagging suppression methods, and highly destructive issues.

[0003] Specifically, most existing light source monitoring modules rely on a single static power threshold decision or basic steady-state frequency domain feature extraction. These conventional analysis methods struggle to capture the transient phase abrupt changes in micro-bending reflected light caused by external stress variations. Furthermore, they fail to effectively separate oscillations caused by internal drive power supply ripple coupling from those excited by external optical path parasitic feedback within complex signal sequences, leading to blind adjustments when faced with abnormal fluctuations. To suppress these spontaneous oscillations, traditional control logic typically involves drastically reducing the laser's bias current or forcibly blocking them using external micro-mechanical optical switches. However, in high-speed optical transmission scenarios, the inherent physical delay of mechanical components cannot match the evolution rate of the optical signal, while also introducing severe additional insertion loss fluctuations. Blindly applying current drops can instantly destroy the modulation depth and communication quality of the optical signal. These traditional passive defense methods essentially sacrifice normal data transmission services, failing to provide a lossless intervention mechanism that can precisely disrupt the phase matching conditions of micro-bending parasitic feedback without interrupting communication. This reduces light source control efficiency and warrants improvement. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this application provides a method for controlling optical fiber communication light sources.

[0005] This application provides a method for controlling an optical fiber communication light source, comprising the following steps: Forward optical power waveform data is acquired by a forward photodetector, backscattered optical power waveform data is acquired by a backward photodetector, and drive current ripple data is obtained at the laser driver sampling end. Short-time Fourier transform is performed on the forward optical power waveform data and the drive current ripple data to extract the corresponding forward dynamic amplitude-frequency matrix and drive dynamic amplitude-frequency matrix within the preset sliding time window, respectively. Calculate the coherence coefficient matrix between the forward dynamic amplitude-frequency matrix and the driving dynamic amplitude-frequency matrix; When the coherence value in the coherence coefficient matrix is ​​greater than the preset warning threshold and the amplitude envelope of the forward dynamic amplitude-frequency matrix exhibits an exponential growth characteristic exceeding the system damping coefficient, the parasitic resonant cavity diagnostic loop is triggered. The specific diagnostic loop for parasitic resonant cavities includes: Pilot perturbation data from non-communication frequency bands are superimposed on the light source bias current; The backscattered light power waveform data after superimposing pilot perturbation data is analyzed, and the intermodulation distortion component between the oscillation frequency and the pilot frequency is extracted from the waveform data. When the amplitude of the intermodulation distortion component is greater than the preset diagnostic threshold, high-frequency modulated chirped data is injected into the bias current to induce light source wavelength jitter data using the carrier dispersion effect. During the injection of high-frequency modulated chirped data, the dynamic phase difference data between the forward optical power waveform data and the backscattered optical power waveform data is continuously calculated, and the oscillation amplitude data of the forward dynamic amplitude-frequency matrix is ​​updated synchronously. When the oscillation amplitude data decays to below the preset safety threshold and remains below it for a preset time, and the dynamic phase difference data deviates from the fixed locking range, the high-frequency modulation chirp data is gradually reduced and canceled, the parasitic resonant cavity diagnostic loop is exited, and the oscillation suppression completion signal is output.

[0006] In summary, this application includes at least one of the following beneficial technical effects: 1. This application provides a method for controlling the light source of optical fiber communication. By superimposing high-frequency pilot perturbation data that does not affect the backbone communication into the optical path and continuously extracting the instantaneous phase characteristics of the optical power, the noise interference of the photodetector under extremely low light intensity is effectively filtered out. Furthermore, by extracting the intermodulation distortion component between the oscillation frequency and the pilot frequency, the single energy attenuation monitoring is extended to multi-dimensional nonlinear distortion tracking. This phase-frequency joint detection method can quickly and accurately lock the parasitic interference state induced by fiber microbending in complex alternating stress environments, significantly improving the online monitoring sensitivity and diagnostic accuracy of the optical network. 2. When the dynamic phase difference data is determined to deviate from the fixed locking range, high-frequency modulated chirped data with a specific ramp slope is injected into the laser bias current. The carrier dispersion effect is used to change the carrier concentration in the active region, thereby causing periodic changes in the refractive index and wavelength jitter of the light source. This controlled spectral frequency shift can be used as a physical perturbation to effectively break the interference phase lock between the environmental reflected echo and the original emitted light. This dynamic intervention process does not require interruption of high-speed backbone communication services, realizing lossless online self-healing of optical communication signals. 3. This application adopts a stepped descent envelope cancellation mechanism, combined with real-time calculation of the residual oscillation amplitude and comparison with a preset safety threshold. Only after confirming the elimination of external micro-bending stress is the control data gradually cleared, thus achieving a smooth transition of the intervention current. At the same time, for the intrinsic driving noise with coherent anomalies, the system can intelligently branch and trace the source, and suppress high-frequency ripple interference from the power supply level by adaptively adjusting the stopband attenuation parameter of the digital filter in the power supply regulator. This comprehensive control architecture, which covers active suppression and smooth cancellation of external sources and adaptive filtering and isolation of internal sources, gives optical communication equipment stronger environmental adaptability and all-weather operational reliability. Attached Figure Description

[0007] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 This is a flowchart of a method for controlling an optical fiber communication light source according to an embodiment of this application. Detailed Implementation

[0009] The following description, in conjunction with the implementation of this invention, is merely an example and illustration of the concept of this invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the inventive concept or exceed the scope defined in these claims, all of which should fall within the protection scope of this invention.

[0010] Application Overview: In existing technologies, fault diagnosis of long-distance fiber optic communication links often relies on static optical time-domain reflectometers or single receiver optical power threshold monitoring, which struggles to balance real-time performance and accurate source tracing. Traditional methods, when the fiber undergoes external mechanical micro-bending deformation, can easily create parasitic resonant cavity interference near the damage point, leading to phase locking and a surge in bit error rate. Existing equipment cannot simultaneously sense transient changes in the optical field phase difference without interrupting backbone services. Especially under high-speed modulation conditions where micro-bending intensifies, single energy threshold alarms exhibit severe lag and misjudgment, failing to meet the nanosecond-level low-latency fault mitigation and self-healing control requirements of modern optical communication networks.

[0011] To address the aforementioned issues, the inventors discovered a significant correlation between the intensity of the parasitic resonant cavity induced by fiber microbending and the dynamic phase difference and high-frequency intermodulation distortion components of the forward and backward optical fields. By actively injecting perturbations and constructing a phase-frequency coupled feedback model, precise diagnosis and disruption of the resonant state can be achieved. During the research, it was found that high-frequency modulated chirped current is extremely sensitive to breaking the phase lock of the parasitic cavity and has a significant destructive effect. However, instantaneously removing this perturbation easily triggers carrier density collapse and secondary optical thermal oscillations within the laser. Therefore, a step-wise dynamic removal of the chirped perturbation based on a comprehensive oscillation amplitude safety threshold was proposed. Further experimental verification incorporated the mapping relationship between the dynamic phase difference variation rate and the comprehensive oscillation amplitude into the adaptive adjustment mechanism of driver-end compensation and perturbation removal, forming a closed-loop feedback system from precise source tracing to smooth intervention.

[0012] Specifically, the diagnostic system first synchronously acquires forward and backscattered optical power waveforms after superimposed pilot perturbations. Through digital signal processing and frequency domain analysis, the intermodulation distortion component characterizing the nonlinear intensity is extracted, and a high-confidence dynamic phase difference data is calculated using an adaptive Gaussian weighted smoothing variant model. When the intermodulation distortion amplitude exceeds the limit and the dynamic phase difference data deviates from the fixed locking range, accompanied by a phase difference variation rate exceeding the abrupt change threshold, the system determines that external micro-bending resonance has occurred. It then automatically injects high-frequency modulated chirp data into the bias current, using carrier dispersion to induce wavelength jitter in the light source to forcibly break the parasitic cavity phase lock. During continuous suppression, the system calculates the effective energy in real time using a time-series damped smoothing variant model based on the attenuation characteristics of the comprehensive oscillation amplitude in the forward dynamic amplitude-frequency matrix. Using this as feedback, the chirp amplitude is gradually reduced in a stepwise manner according to a preset step size, forming a dynamic cancellation closed loop. For intrinsic drive noise interference where intermodulation distortion does not exceed the limit but coherence coefficient is abnormal, the system skips chirp injection and directly generates drive power supply filter compensation commands to adjust the stopband attenuation parameters of the digital filter.

[0013] Compared to existing technologies, traditional optical network monitoring methods rely on a single power drop characteristic and lack active intervention and physical-level phase compensation mechanisms. This makes them prone to large-scale service interruptions or systemic missed detections when encountering highly concealed parasitic resonances. This solution innovatively integrates transient phase difference tracking and high-frequency intermodulation frequency domain mapping techniques, achieving dynamic online suppression of malicious parasitic resonances by establishing a carrier dispersion-phase tearing coupling model. Unlike existing passive network outage detection or static, rigid filtering models, this solution intelligently triggers chirp injection modes based on real-time phase angle variation rates and continuously optimizes the disturbance withdrawal step size through a closed-loop feedback mechanism with damped attenuation calculations, significantly improving the reliability of optical path self-healing under complex alternating stress conditions.

[0014] Through the above technical solutions, this application effectively overcomes the problems of optical field interference distortion and laser frequency pull-locking caused by external fiber micro-bending deformation. While ensuring real-time lossless transmission of backbone communication services, it greatly improves the accuracy of parasitic resonant cavity identification and intervention. The dynamic chirped injection and stepped cancellation mechanism perfectly balance the high sensitivity of instantaneously breaking interference lock and the system stability advantage of preventing secondary thermal oscillations. The adaptive filtering and damping smoothing calculation functions ensure the accuracy of long-term online monitoring under complex high-frequency noise environments.

[0015] After introducing the basic concept of the present invention, the embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0016] Example This application discloses a method for controlling an optical fiber communication light source.

[0017] Reference Figure 1 A method for controlling a fiber optic communication light source includes the following steps: Forward optical power waveform data is acquired by a forward photodetector, backscattered optical power waveform data is acquired by a backward photodetector, and drive current ripple data is obtained at the laser driver sampling end. Short-time Fourier transform is performed on the forward optical power waveform data and the drive current ripple data to extract the corresponding forward dynamic amplitude-frequency matrix and drive dynamic amplitude-frequency matrix within the preset sliding time window, respectively. Calculate the coherence coefficient matrix between the forward dynamic amplitude-frequency matrix and the driving dynamic amplitude-frequency matrix; When the coherence value in the coherence coefficient matrix is ​​greater than the preset warning threshold and the amplitude envelope of the forward dynamic amplitude-frequency matrix exhibits an exponential growth characteristic exceeding the system damping coefficient, the parasitic resonant cavity diagnostic loop is triggered. The specific diagnostic loop for parasitic resonant cavities includes: Pilot perturbation data from non-communication frequency bands are superimposed on the light source bias current; The backscattered light power waveform data after superimposing pilot perturbation data is analyzed, and the intermodulation distortion component between the oscillation frequency and the pilot frequency is extracted from the waveform data. When the amplitude of the intermodulation distortion component is greater than the preset diagnostic threshold, high-frequency modulated chirped data is injected into the bias current to induce light source wavelength jitter data using the carrier dispersion effect. During the injection of high-frequency modulated chirped data, the dynamic phase difference data between the forward optical power waveform data and the backscattered optical power waveform data is continuously calculated, and the oscillation amplitude data of the forward dynamic amplitude-frequency matrix is ​​updated synchronously. When the oscillation amplitude data decays to below the preset safety threshold and remains below it for a preset time, and the dynamic phase difference data deviates from the fixed locking range, the high-frequency modulation chirp data is gradually reduced and canceled, the parasitic resonant cavity diagnostic loop is exited, and the oscillation suppression completion signal is output.

[0018] In this embodiment, the forward optical power waveform data refers to the continuous sampling sequence of the power of the light source output optical signal obtained by the forward photodetector as a function of time. Specifically, it can be implemented using a high-speed photodiode in conjunction with an analog-to-digital converter, and is used to monitor the basic light emission state at the main optical path in real time. The backscattered optical power waveform data refers to the discrete sequence of the power of the optical signal returned by Rayleigh scattering or microbending reflection in the optical fiber link collected by the backscattered photodetector as a function of time. Specifically, it can be implemented using a high-sensitivity avalanche photodiode, and is used to capture the back interference echo characteristics generated by external disturbances in the link. The drive current ripple data refers to the small AC fluctuation component contained in the power supply current obtained at the laser drive sampling end. Specifically, it can be implemented using a high-precision current sampling resistor combined with a differential amplifier circuit, and is used to investigate the crosstalk effect of intrinsic power supply noise on the emission frequency domain.

[0019] Among them, the forward dynamic amplitude-frequency matrix and the driving dynamic amplitude-frequency matrix refer to two-dimensional arrays containing three-dimensional information of time, frequency and amplitude obtained after performing a short-time Fourier transform on the time-domain waveform data. Specifically, they can be implemented by using a fast Fourier transform algorithm combined with a preset sliding window function. They are used to reveal the frequency distribution and dynamic evolution of optical power and driving current within a specific time period. The coherence coefficient matrix refers to a mathematical set that quantifies the degree of correlation between the forward dynamic amplitude-frequency matrix and the driving dynamic amplitude-frequency matrix at the same frequency point. Specifically, it can be implemented by using the calculation method of dividing the cross power spectral density by the self power spectral density. It is used to accurately distinguish whether the high-frequency oscillation of the optical path originates from internal driving noise or external parasitic resonance.

[0020] Among them, the system damping coefficient refers to the physical benchmark parameter characterizing the system's ability to dissipate and suppress oscillating energy. Specifically, it can be set by the natural decay constant under historical stable operating conditions, serving as a criterion for judging whether the forward optical power amplitude envelope has undergone malignant divergence and exponential growth. Pilot perturbation data refers to the tiny probe signal superimposed on the light source bias current, with a frequency strictly located in the non-communication frequency band. Specifically, it can be realized by using a specific frequency sine wave sequence generated by a digital frequency synthesizer, used to actively detect and excite hidden optical path nonlinear parasitic cavity effects. Intermodulation distortion component refers to the cross frequency product generated by the mixing of the original oscillation frequency and the newly added pilot frequency due to the nonlinear interference effect of the parasitic resonant cavity. Specifically, it can be realized by extracting the difference frequency or sum frequency peak in spectrum analysis, used as the core characteristic evidence for diagnosing external malignant microbending parasitic resonance. High-frequency modulation chirp data refers to the high-frequency modulation signal with a specific ramp slope in the injected bias current, specifically, it can be realized by using a high-frequency sawtooth wave digital sequence generator combined with digital-to-analog conversion, used to trigger the dispersion effect by changing the carrier concentration in the active region, thereby causing light source wavelength jitter data.

[0021] Among them, dynamic phase difference data refers to the instantaneous phase difference sequence between the forward light power waveform and the backscattered light power waveform on a microscopic time scale. Specifically, it can be achieved by extracting the analytical phase through Hilbert transform and combining it with an adaptive smoothing algorithm. It is used to reflect the evolution and disintegration process of the interference engagement state of the internal and external light fields in real time. The fixed locking interval refers to the steady-state phase difference tolerance range between the backlight and the forward light determined under steady-state conditions without external mechanical micro-bending interference. Specifically, it can be defined by adding a preset floating tolerance to the historical reference phase, and serves as a criterion for judging whether the parasitic resonance interference correlation has been completely destroyed by external physical perturbations.

[0022] The working process and principle of this application are as follows: First, forward optical power waveform data is acquired through a forward photodetector, backscattered optical power waveform data is acquired through a backward photodetector, and drive current ripple data is obtained at the laser driver sampling end. Then, a short-time Fourier transform is performed on the forward optical power waveform data and drive current ripple data to extract the corresponding forward dynamic amplitude-frequency matrix and drive dynamic amplitude-frequency matrix within a preset sliding time window, respectively. Next, the coherence coefficient matrix of the forward dynamic amplitude-frequency matrix and the drive dynamic amplitude-frequency matrix is ​​calculated. When the coherence value in the coherence coefficient matrix is ​​greater than a preset warning threshold and the amplitude envelope of the forward dynamic amplitude-frequency matrix exhibits an exponential growth characteristic exceeding the system damping coefficient, the parasitic resonant cavity diagnostic loop is triggered. In the diagnostic loop, pilot perturbation data of non-communication frequency band is first superimposed on the light source bias current, and then the backscattered optical power waveform data after superimposing the pilot perturbation data is analyzed. The intermodulation distortion component between the oscillation frequency and the pilot frequency is extracted from the waveform data. If the amplitude of the intermodulation distortion component is greater than a preset diagnostic threshold, high-frequency modulation chirp data is injected into the bias current to induce wavelength jitter data of the light source using the carrier dispersion effect. Then, during the injection of high-frequency modulation chirp data, the dynamic phase difference data between the forward optical power waveform data and the backscattered optical power waveform data is continuously calculated, and the oscillation amplitude data of the forward dynamic amplitude-frequency matrix is ​​updated synchronously. Finally, when the oscillation amplitude data decays to below the preset safety threshold and is maintained for a preset time, and the dynamic phase difference data deviates from the fixed locking range, the high-frequency modulation chirp data is gradually reduced and withdrawn, the parasitic resonator diagnostic loop is exited, and the oscillation suppression completion signal is output. In this way, accurate diagnosis and non-destructive dynamic elimination of parasitic resonators in optical fiber links are achieved, ensuring that optical signal transmission proceeds in the expected stable state.

[0023] Furthermore, a short-time Fourier transform is performed on the forward optical power waveform data and the drive current ripple data to extract the corresponding forward dynamic amplitude-frequency matrix and drive dynamic amplitude-frequency matrix within a preset sliding time window, including: The forward optical power waveform data and the drive current ripple data are segmented and extracted using a preset sliding time window to generate a forward optical power data frame sequence and a drive current ripple data frame sequence arranged in time order. Discrete Fourier transform calculations are performed on each frame of data in the forward optical power data frame sequence and the drive current ripple data frame sequence to obtain the amplitude component of each frame of data in the frequency domain. All amplitude components corresponding to the forward optical power data frame sequence are reassembled into a forward amplitude vector set, and all amplitude components corresponding to the drive current ripple data frame sequence are reassembled into a drive amplitude vector set. Following the time step sequence of the preset sliding time window, the vectors in the forward amplitude vector set are sequentially concatenated dimensionally to construct the forward dynamic amplitude-frequency matrix, and the vectors in the driving amplitude vector set are sequentially concatenated dimensionally to construct the driving dynamic amplitude-frequency matrix.

[0024] In this embodiment of the invention, the process of performing short-time Fourier transform on the forward optical power waveform data and the drive current ripple data to extract the corresponding forward dynamic amplitude-frequency matrix and drive dynamic amplitude-frequency matrix within a preset sliding time window involves first segmenting the discretized sampling data of the forward optical power waveform data and the drive current ripple data along the time axis using preset sliding time windows with both non-overlapping and partially overlapping time parameters. The window length of the preset sliding time window is set to a microsecond range of 100 to 200 microseconds, accompanied by a corresponding window sliding time step. The high-density sliding step interval, ranging from 10 to 20 microseconds, is set to be much smaller than the low-frequency disturbance period of conventional optical fiber communication. The underlying logic is to match the millisecond-level internal phase change process caused by external physical vibration of the micro-bent reflected light, effectively avoiding the failure to detect the short-wavelength characteristics of transient micro-bent disturbances with extremely strong suddenness. At the same time, the Hanning window function is introduced during the truncation process to smooth the data values ​​at the window boundary, reducing the spectral leakage error caused by the boundary truncation effect, thereby generating a forward optical power data frame sequence and a drive current ripple data frame sequence that are tightly arranged in absolute time order. Next, a Discrete Fourier Transform based on a standard short-time analytical variant is performed on each individual time slice of the forward optical power data frame sequence and the drive current ripple data frame sequence to obtain the physical true amplitude component of each frame in a specific frequency domain. The extraction formula for this amplitude component is designed as follows: In this calculation formula, The representation after transformation corresponds to the first Within the first time data frame The true physical amplitude components at discrete frequency points, when the fundamental input signal When the input is a forward optical power data frame sequence, the unit of measurement is milliwatts. The final output unit remains in physical milliwatts; when the basic input signal... When the input is a data frame sequence for driving current ripple, the unit of measurement is milliampere (mA). The output units are kept constant in physical milliamperes, and the constants in the formulas are also consistent. It represents the total number of sampled digital points contained within a single preset sliding time window data frame. It is a dimensionless integer constant of the counting type, determined by the hardware sampling rate. Representing the The first time data frame sequence after Hanning windowing processing Amplitude at a specific sampling point Represents the unit of imaginary arithmetic. This represents the discrete frequency index constant mapped onto the frequency axis, and is a dimensionless positional identifier; rigorous dimensional verification of the operations is performed here, due to the discrete complex exponential term. With the cumulative number of sampling points All are dimensionless, purely digital constants, and are input signals with actual physical units. After the integral accumulation over the window period, it is innovatively multiplied beforehand by a standardized coefficient that has undergone a variation. Normalization is performed to offset the inverted absolute energy accumulation difference caused by the different discretization sampling lengths, so that the amplified scalar value after accumulation is compressed again proportionally, resulting in... Physical dimensions and input The original dimensions are perfectly equivalent in the mathematical dimension, which effectively ensures that the amplitude components of the subsequent output can directly and accurately map and represent the actual physical power fluctuation range of the laser output optical path and driving circuit, bypassing the concept of abstract spectral energy. Subsequently, all amplitude components calculated from the forward optical power data frame sequence at all detection frequencies are recombined and compressed into a forward amplitude vector set based on the same time index. Simultaneously, all amplitude components independently calculated from the drive current ripple data frame sequence are also recombined into a drive amplitude vector set following the same recombination rule. Both sets contain a large number of independent column vectors with a single time frame as the dimension. Finally, strictly following the logical order of the time stepping within the preset sliding time window, each one-dimensional column vector in the forward amplitude vector set is sequentially matrix-concatenated along the horizontal time extension dimension, thereby constructing a three-dimensional vector set based on the sliding time frame sequence. A two-dimensional forward dynamic amplitude-frequency matrix with the discrete frequency index as the vertical axis is constructed. At the same time, the system uses the same spatial splicing dimension logic in parallel and synchronously to sequentially perform matrix-dimensional splicing on each independent vector in the set of driving amplitude vectors, thereby constructing a parallel driving dynamic amplitude-frequency matrix at the same spatiotemporal scale without delay. The successful construction of these two dynamic amplitude-frequency matrices enables the abstract system operating state to simultaneously possess the absolute frequency positioning attribute of high-frequency micro oscillations and the high-precision instantaneous time anchor point attribute of sudden environmental interference. They, as a complete multi-dimensional matrix data stream, will be directly input into the subsequent coherence coefficient matrix calculation architecture for cross-decoupling analysis.

[0025] Furthermore, the coherence coefficient matrix of the forward dynamic amplitude-frequency matrix and the driving dynamic amplitude-frequency matrix is ​​calculated, including: Extract the dominant frequency band forward vector and the dominant frequency band driving vector from the forward dynamic amplitude-frequency matrix; Calculate the cross power spectral density matrix of the dominant frequency band forward vector and the dominant frequency band driving vector, and calculate the auto power spectral density matrix of the dominant frequency band forward vector and the auto power spectral density matrix of the dominant frequency band driving vector respectively. The coherence coefficient matrix is ​​generated by dividing the square of the magnitude of the cross power spectral density matrix by the product of the two self power spectral density matrices.

[0026] In this embodiment of the invention, the process of calculating the coherence coefficient matrix of the forward dynamic amplitude-frequency matrix and the driving dynamic amplitude-frequency matrix first involves directionally extracting and extracting row and column data within a preset frequency band from the constructed forward dynamic amplitude-frequency matrix and the driving dynamic amplitude-frequency matrix along the frequency axis, forming the dominant frequency band forward vector and the dominant frequency band driving vector, respectively. The preset frequency band is set to a specific high-frequency range of 10 kHz to 500 kHz. The preset logic is that this frequency band can perfectly strip away the low-frequency noise of the ambient temperature drift of conventional optical communication below 1 kHz, while avoiding the thermal white noise of the system at the megahertz level, and accurately anchoring the typical oscillation fundamental frequency and its higher harmonic distribution area of ​​the optical parasitic resonator caused by link micro-bending and mechanical vibration. Next, based on the extracted dominant frequency band forward vector and dominant frequency band driving vector, a smoothing time window algorithm is introduced to calculate the cross-power spectral density matrix and their respective self-power spectral density matrices. To avoid the mathematical failure problem where the Fourier transform result under a single time slice always approaches 1 during coherent calculation, a time-dimensional weighted average variation calculation formula is designed as follows: In this calculation formula, The forward vector representing the dominant frequency band is in the th... The first time frame and the first The complex amplitude-frequency values ​​at each frequency point are derived from forward optical power detection, and the unit of measurement is objectively maintained as milliwatts. The complex amplitude-frequency value of the driving vector of the dominant frequency band under the same spatiotemporal coordinates is derived from the driving current ripple and its unit is milliampere. represent The conjugate of complex numbers; variables This represents the total number of observation frames included within the smoothing time window, typically chosen as a dimensionless integer constant between 5 and 10, used to smooth transient noise disturbances on the time scale; based on this formula, and Substituting the obtained principal power spectrum cross terms gives the element data of the cross power spectral density matrix. and Substituting the obtained system self-products and combining them forms the self-power spectral density matrix element data of the dominant frequency band forward vector; similarly, ... and Substituting the self-product combination generated into the self-power spectral density matrix element data is used to generate the dominant frequency band driving vector. Subsequently, the square of the magnitude of the cross-power spectral density matrix is ​​divided by the product of the two self-power spectral density matrices using normalized mapping logic, and an anti-escape noise floor basis is introduced to generate the final coherence coefficient matrix. The formula for extracting the coherence value is modified as follows: In the formula Representing time and frequency points The coherent value at the point, with a dynamic range that strictly converges to between 0 and 1, represents the linear coherence strength between the abnormal fluctuation of optical power and the abnormal ripple of the driving current. Represents the cross-power spectral density value. This represents the forward self-power spectral density value. The value representing the driving power spectral density is intentionally added to the denominator of the formula. The system's static noise floor threshold matrix parameters are pre-calibrated by collecting thermal noise floor data from the communication light source when the modulation signal is turned off. The purpose is to address the severe pseudo-coherence amplification and crash overflow effect caused by the division calculation due to the extremely small self-power spectrum at the denominator when the actual network link is in an absolutely silent or no obvious peak state. This improves the stability of the variant model's anti-disturbance judgment under extremely low signal-to-noise ratio conditions. Thus, the discrete and free time-frequency power and current parameters of each dimension are refined into a pure coherence coefficient matrix that can directly measure the weight induced by the parasitic resonant cavity. This coherence coefficient matrix is ​​then used as the decision basis and directly input to the next stage for source diagnosis of parasitic oscillations.

[0027] Furthermore, the steps for determining that the amplitude envelope of the forward dynamic amplitude-frequency matrix exhibits an exponential growth characteristic exceeding the system damping coefficient include: Extract the peak envelope data sequence from the forward dynamic amplitude-frequency matrix along the time dimension; Calculate the set of amplitude ratios between adjacent time points in the peak envelope data sequence; When all ratios in the amplitude ratio set are greater than the product of the natural constant and the system damping coefficient, it is determined that the amplitude envelope of the forward dynamic amplitude-frequency matrix exhibits an exponential growth characteristic exceeding the system damping coefficient.

[0028] In this embodiment of the invention, the process of determining that the amplitude envelope of the forward dynamic amplitude-frequency matrix exhibits an exponential growth characteristic exceeding the system damping coefficient involves, firstly, searching and extracting the maximum components in the amplitude plane frame by frame along the time axis within a preset oscillation frequency band representing the high-frequency parasitic resonance effect, targeting the output forward dynamic amplitude-frequency matrix. Using extreme value optimization logic, the dominant amplitude source intensity corresponding to each independent time slice is accurately identified. These discrete maximum components are then strictly concatenated and arranged according to their absolute time generation order to generate a peak envelope data sequence that can intuitively characterize the trajectory of the transient energy peak surge evolution of the system. The mathematical space mapping model for this core extraction action is defined as... In this specific extraction mechanism formula, For the forward dynamic amplitude-frequency matrix at the th The sliding time frame node and the first The absolute physical amplitude components at discrete monitoring frequency points have units inherited entirely from the detector electro-optic conversion mapping, and are therefore maintained in milliwatts. This is a preset high-frequency oscillation band detection set selected after filtering out the conventional communication baseband using a filter hardware. The extracted single envelope node element is then output through maximum / extreme value operations. It truly reflects the highest energy peak within the current short time frame. Since its intrinsic calculation process only involves the optimization of similar variables without changing the physical properties, its dimensions are strictly and legally preserved at the milliwatt level. Next, based on this processing, the system meticulously calculates the set of amplitude ratios between adjacent time nodes in the peak envelope data sequence. To avoid congestion caused by floating-point operations on complex non-exponential decomposition rules by the underlying digital signal microprocessor, and also to completely eliminate the division-by-zero crash error induced by the optical path energy approaching the system's dark current noise floor limit in the previous extremely short instant, this embodiment introduces and incorporates a feedforward differential ratio variation algorithm with a specific DC bias to rapidly obtain the dynamic growth rate. The specific calculation variation of this growth rate is set as follows: In this equation The calculated amplitude ratio data spanning adjacent single time frame periods. To preset a small DC bias correction constant, it is crucial that, in order to meet the requirements of... Apply the dimensional validity criteria for the additive fusion operation dimension. It is mandated to be declared and assigned the same milliwatt dimension, and its static value is recommended to be set at the factory based on the dark photocurrent characteristics to be extremely close to the lower limit of photoelectric thermal noise floor. When calculated using the first-order anti-overflow division formula, the core molecular group and the underlying denominator group both belong to the energy array carrying milliwatt physical units. During cross-dimensional division calculations, their physical units are absolutely offset and completely cancel each other out, resulting in an independent ratio output. All elements in the amplitude ratio set formed by superimposing and recombining multiple frames are automatically and completely transformed into a standard pure digital dimensionless potential scale. Subsequently, a hard diagnostic boundary threshold must be established to determine the outward divergence of the current optical system instability. In traditional control engineering feedback evolution models, detecting exponential divergence behavior typically requires a large number of cycles to perform power-law logarithmic convolution calculations. However, to meet the critical fault prevention requirements of low latency (nanosecond level) in fiber optic backbone communication and to maximize the release of main control computing power, this invention directly reduces the dimension of the deep nonlinear exponential envelope divergence solution to a primary linear multiplication comparison of the baseline intrinsic scalar. Specifically, the threshold for determining this defense line is set as the direct product of the natural constant and the system damping coefficient, i.e., the calculation defense line equation is constructed as follows: Within this formula deconstruction, As the mathematical natural fundamental constant representing the distribution characteristics of the base of the natural logarithm, it, as a purely irrational base point, is itself always in a dimensionless state without physical dimensions, along with the parameters of the product. This is a composite of the system damping coefficient, which is used to calibrate the inherent internal loss of the laser cavity and the suppression strength of the reverse isolation end face of the fiber optic connector within a specific temperature band. This characteristic coefficient can be directly obtained by applying a strong pulse impulse signal towards the main link output port and then deducing the slope of the drop envelope rate of the small echo ringing during the reference no-light modulation calibration phase of the device's power-on self-test. Its nominal value is usually constrained within the safe normal range of 0.42 to 0.78. Because this damping coefficient essentially characterizes the ratio of internal energy absorption to photon divergence attenuation in a homogeneous system, it must also belong to the category of purely dimensionless characteristic parameters. Therefore, the judgment criterion derived by cross-multiplying two dimensionless variables is... It must be indisputably confirmed as a constant scalar without any substantial physical units at the most stringent dimensional calculation level; then, proceeding to the final judgment verification process, each cross-frame ratio sequence feature element in the generated full set of dimensionless amplitude ratios is sequentially compared with the threshold boundary locked by the previous calculation in a one-way time-travel horizontal comparison. Only when all data slice elements in the amplitude ratio set within the preset continuous fault-tolerant verification window (to eliminate single-point jump false triggering behavior caused by random photon shot scattering, it is strongly recommended that the span of this moving verification window be preset to include 3 to 5 consecutive ratio judgment frame periods) all exhibit a strict and complete leapfrog greater than the specific numerical limit benchmark of the product of the natural constant and the system damping coefficient within their own continuous observation time span, can the mathematical logic divergent calculation law and the external photophysical cavity backlash be mapped. From a dual, rigorous perspective, the extremely critical warning can be instantly confirmed. It is certain that not only has the light energy broken out of the controllable wandering region of the background Gaussian noise and formed a pathological geometric progression of violent expansion, but more seriously, the speed of energy rise has substantially overwhelmed and unilaterally torn apart the maximum self-healing suppression limit that the original static damping coefficient of the system can provide. This completely triggers and outputs the final Boolean-type affirmative judgment signal, directly announcing that the amplitude envelope of the forward dynamic amplitude-frequency matrix has completely broken free from endogenous suppression and accurately generated an exponential growth characteristic exceeding the system damping coefficient. Then, it quickly utilizes the trigger attribute flag characteristic state with extremely high hardware interrupt response priority to hard-connect and directly wake up all the diagnostic and control intervention sub-links mounted in the background, preventing the physical occurrence of laser thermal burnout and large-scale business chain failure.

[0029] Furthermore, it also includes a branch for determining the driving coupled oscillation: When the amplitude of the intermodulation distortion component is not greater than the preset diagnostic threshold and the coherence value in the coherence coefficient matrix is ​​greater than the preset warning threshold, skip the injection step of high-frequency modulation chirped data. Generate drive power supply filter compensation instruction data; The drive power supply filter compensation command data is sent to the power supply low dropout linear regulator. The ripple amplitude corresponding to the drive current ripple data is reduced by adjusting the stopband attenuation parameter of the digital filter, and the parasitic resonant cavity diagnostic loop is exited.

[0030] In this embodiment of the invention, for a light source control system that identifies abnormal fluctuations, the process after entering the core source tracing diagnosis includes a specific processing procedure for the drive coupling oscillation determination branch. When the amplitude of the intermodulation distortion component between the oscillation frequency and the pilot frequency extracted by the system in the waveform analysis stage is not greater than a preset diagnosis threshold, and the corresponding coherence value in the coherence coefficient matrix generated by the parallel determination front end is greater than a preset warning threshold, the system determines from the underlying physical cause that no optical parasitic feedback resonant cavity caused by micro-bending has been formed outside the optical fiber link. Instead, it directly diagnoses the oscillation mechanism as an endogenous drive coupling induced by high-frequency noise from the drive side transmitted to the electro-optic conversion medium. Under this logical analysis framework, the preset diagnosis threshold is set to an extremely low optical power limit of 0.01 mW to 0.05 mW that can reflect the effective external optical nonlinear intermodulation noise floor to prevent false triggering by weak shot waves. The preset warning threshold is configured as a dimensionless value of 0.85 to 0.95 to ensure a rigorous locking and definition of the electro-optic high-intensity correlation.

[0031] After diagnosis, to avoid forcibly applying unnecessary optical interference perturbations to the optical path, which has no external parasitic microbending, and to eliminate the cost of unhelpful spectral broadening, the main control program immediately and proactively skips the step of injecting high-frequency modulation chirp data to the bias current side. Instead, the system then switches to internal level filtering correction logic to generate drive power supply filter compensation command data containing dynamically adjusted weights. The core basis of this command data lies in its ability to accurately calculate the stopband attenuation parameters of the digital filter that need to be updated specifically for the next intervention moment. To overcome the problem that rigid open-loop filter optimization caused by traditional preset fixed attenuation constants is prone to causing inner loop lock-up and control oscillation overshoot, this branch designs a stopband control variant formula based on adaptive normalized error feedback. The specific calculation equation is proposed as follows: In the calculation system of this stopband attenuation parameter, This represents the stopband attenuation parameters of the new digital filter that will be issued soon after calculation. It is used to quantitatively characterize the forced suppression level of high-frequency switching current noise in a specific high-risk frequency band. Its essence is the gain multiplication control instruction of the underlying register, so its dimensional characteristic is dimensionless pure digital. These are the current stopband attenuation parameters of the digital filter read and called from within the memory chip array, which also belong to the dimensionless system; The instantaneous ripple amplitude of the current drive current ripple data at the highest oscillation frequency point is mapped and extracted across frames in the system. Its dimension is strictly confirmed as physical milliampere because it is based on the transient current sampling conversion law of the acquisition side. The safe ripple baseline threshold for the drive current, calibrated and entered before the system leaves the factory, represents the physical safety limit of the residual white noise floor ripple under the optimal level state of the light-emitting component. Considering the characteristics of the broadband drive module, the reference range is set to 0.5 mA to 1.5 mA, and its dimension is also determined to be the same physical mA. To pre-set the adaptive adjustment of the convergence step size weight, the damping contraction constant, which serves as the internal anti-transition constant of the compensation control model, is directly given a dimensionless property. In order to seek an optimal balance between rapid suppression and steady-state surge protection, it is usually strictly and rigidly constrained by the underlying code of the system within a safety range of 0.15 to 0.35. Until the formula is analyzed to the final response, the system will directly package and load the integrated drive power supply filter compensation command data with the highly adaptable new attenuation parameters. Through the control of the main data bus, it will be sent instantaneously and accurately to the core leading device of the chassis communication light source board, namely the low dropout linear regulator of the power supply. By overwriting the corresponding digital register bits of the internal integrated signal processor array, the stopband attenuation parameters of the digital filter will be hard adjusted at the basic physical logic level. This will significantly improve and increase the physical anti-wave response depth of the high-frequency impedance on the power feeder in the predetermined frequency band. This will directly flatten and significantly reduce the deep ripple amplitude corresponding to the real drive current ripple data collected and transmitted from the source side. While completely cutting off and removing the source of the internal electromagnetic noise interference optical frequency oscillation energy transmission chain, the steady-state light emission will be successfully restored in the nanosecond-level action flow, and the system will safely and smoothly jump out of the current parasitic resonant cavity diagnostic loop system. The whole process will completely realize the continuous lossless instantaneous intervention and recovery from high-order self-traceability identification and judgment to intervention isolation and repair of complex link light source fluctuation signals.

[0032] Furthermore, the backscattered light power waveform data after superimposing pilot perturbation data is analyzed to extract the intermodulation distortion components between the oscillation frequency and the pilot frequency in the waveform data, including: The frequency value corresponding to the current highest frequency peak in the backscattered light power waveform data is obtained as the oscillation frequency data, and the pilot frequency data corresponding to the pilot perturbation data is obtained. Calculate the sum frequency and difference frequency values ​​of the oscillation frequency data and the pilot frequency data; In the frequency domain spectrum of the backscattered light power waveform data, the amplitude spectrum values ​​at the frequency points where the sum frequency value and the difference frequency value are located are extracted, and the extracted amplitude spectrum values ​​are used as intermodulation distortion components.

[0033] In this embodiment of the invention, the process of analyzing the backscattered light power waveform data after superimposed pilot perturbation data and extracting the intermodulation distortion components between the oscillation frequency and the pilot frequency in the waveform data firstly involves using a Fast Fourier Transform algorithm to perform full-band frequency domain mapping on the acquired backscattered light time-domain data sequence carrying pilot perturbations, generating backscattered radio-domain spectral data containing absolute amplitude physical quantities. Then, in this backscattered radio-domain spectral data, a global extremum optimization logic is used to locate and extract the frequency value corresponding to the current amplitude spectrum peak extremum point, which is directly used as a representative value. The oscillation frequency data with divergent characteristics generated in the resonant cavity is synchronously retrieved from the control register of the signal generator inside the system and the pilot frequency data corresponding to the pilot perturbation data on which this active perturbation action is based is obtained. The preset pilot frequency data is set to the non-communication baseband frequency domain range of 100 kHz to 200 kHz. This preset frequency selection logic is that this frequency band can perfectly avoid the extremely low frequency mechanical thermal drift noise and the main lobe of the high frequency service communication main signal above 100 MHz, ensuring that the injected pilot signal as a probe will not cause crosstalk to the normal optical communication data stream. Subsequently, based on the principle of nonlinear mixing effect of optical media, the underlying system calculates the sum and difference frequencies of the oscillation frequency data and pilot frequency data in the digital domain. Furthermore, to completely overcome the inherent mathematical drawbacks of traditional single-frequency amplitude extraction methods, which are easily affected by the Fourier fence effect and spectral leakage leading to energy measurement distortion, this step abandons the crude detection method targeting a single frequency point and introduces a variant model for intermodulation distortion feature extraction based on asymmetric compensation neighborhood energy integration. In the frequency domain spectrum of the backscattered light power waveform data, a windowed local energy convergence calculation is performed around the frequency points where the sum and difference frequencies are located. This extraction operation variant formula is set as follows: In this feature extraction formula, To calculate the integrated intermodulation distortion component of the output, its physical dimensions are precisely constrained to physical milliwatts, which characterize the optical power intensity. This represents the specific frequency point in the backscattered radio frequency domain spectral data. The independent amplitude spectrum values ​​at the location are directly inherited from the transimpedance amplifier output mapping of the photodetector, and their dimensions are objectively maintained as physical milliwatts. The sum and frequency values ​​generated by the pre-calculation. The generated difference frequency value; The preset spectral leakage compensation bandwidth parameter, in Hertz, is set to a small range of 500 Hertz to 1000 Hertz to capture the energy of adjacent side lobes that are defocused due to laser frequency jitter. The total number of discrete digital sampling frequency points falling within the compensation bandwidth is a dimensionless constant of the counting type. and These are the intermodulation asymmetry compensation factors set for the high-frequency and low-frequency regions of the system, mainly used to actively correct the physical phenomenon of amplitude asymmetry in high- and low-frequency intermodulation products caused by uneven carrier consumption rates in the active region of semiconductor lasers. They are both set as dimensionless weighting coefficients with values ​​between 0.8 and 1.2. The comprehensive amplitude spectrum value with precise milliwatt dimensions extracted accordingly will be directly used as high-quality intermodulation distortion component data and seamlessly transmitted along the bus to the subsequent diagnostic threshold comparison and judgment stage. This serves as the core quantitative criterion for judging whether the nonlinear intensity of the external parasitic resonator reaches the high-frequency chirp intervention standard.

[0034] Furthermore, high-frequency modulated chirped data is injected into the bias current to induce wavelength jitter data of the light source using the carrier dispersion effect, including: Generate a high-frequency sawtooth wave digital sequence with a frequency higher than the communication frequency band and an amplitude within the limit of the saturated output power of the light source; The high-frequency sawtooth wave digital sequence is converted into an analog current signal and superimposed on the bias current to form high-frequency modulated chirped data; Acquire data on the change in carrier concentration inside the laser after high-frequency modulation and chirping data injection; Based on the periodic changes in the refractive index of the active region caused by carrier concentration variation data, light source wavelength jitter data with nonlinear shift over time is generated.

[0035] In this embodiment of the invention, the specific process of injecting high-frequency modulated chirped data into the bias current and using the carrier dispersion effect to induce light source wavelength jitter data first requires configuring and generating a high-frequency sawtooth wave digital sequence by calling the direct digital frequency synthesizer inside the system microprocessor. In order to ensure that this active intervention means neither interrupting the ongoing high-speed backbone optical signal communication transmission nor excessively stimulating the heat loss inside the laser, the working fundamental frequency of the sawtooth wave sequence is preset to be within a specific frequency domain range of 200 kHz to 500 kHz. This preset frequency selection logic can perfectly avoid the service-carrying frequency band at the megahertz level and just fall into the optimal response activation band that is sufficient to induce the cavity plasma dispersion response. At the same time, the dynamic peak amplitude of the signal is strictly preset and locked within 5% of the basic static bias current fluctuation required for the saturated output power limit of the light-emitting device. The technical necessity of selecting a sawtooth wave instead of an ordinary sine wave is that its unique single-sided ramp linear impulse property can provide a continuous first-order frequency traction force in a single direction for the light-emitting optical path, which helps to break the parasitic locked state more efficiently. Subsequently, the system will convert the digital control instructions generated according to this sequence into continuous real-time analog current signals proportionally and without delay through a high-precision digital-to-analog converter hardware entity. This analog signal containing a specified ramp slope will be physically superimposed on the DC main bias current that is being powered through a bypass physical operation amplification adder network at the circuit baseband level, thereby mixing and interweaving to form high-frequency modulated chirped data with specific current forward ramp characteristics. Finally, it will be directly injected into the anode drive port of the semiconductor general-purpose laser chip along the independent drive bus. Next, it is essential to synchronously and accurately acquire data on the actual carrier concentration changes occurring inside the laser's active cavity after the full injection of high-frequency modulated chirp data. Given the physical concealment of electron-hole recombination motion within the microscopic semiconductor active region and the hardware limitations of not being able to perform slice measurements via wiring in mass production, this system pre-maps and calls a nonlinear reduced-order variant model based on feedforward observation and reconstruction of the physical equations of semiconductor heterojunction recombination rates in the core CNC base's code stack. This model is specifically used for real-time substitution estimation of the dynamic charge-discharge offset data reconstructing the microscopic carrier concentration. This carrier estimation and capture variant formula is precisely constructed as follows: Within this carrier reshaping equation, This data directly represents the change in internal carrier concentration at a specific time slice. Its physical properties are used to characterize the electron-hole pair wave flux within a unit volume microdomain, and its physical dimensions are precisely defined as the reciprocal of the cubic meter. ; It represents the real-time modulated analog current component obtained by separating and acquiring it from the output port of the pre-amplifier digital-to-analog converter. Its dimensions are strictly defined and are governed by the international electrical standard unit, the ampere. The intrinsic lifetime constant of spontaneous emission recombination of charge carriers in the active region of the laser, which is hard-programmed before leaving the factory, must be converted from the nanosecond level of the factory specification to the underlying internationally accepted unit of second in order to match the dimensional convection of the entire calculation framework. The physical constant representing the absolute constant fundamental charge of a single electron has a fixed absolute value approximately on the order of 10 to the power of -19. Its physical unit dimension belongs to the coulomb, which represents the quantity of charge. The equivalent transformation state is Ampere multiplied by a second. The geometric spatial volume constant representing the light-emitting active region of the central cavity inside the semiconductor optical communication light source chip of this model is directly read and calibrated from the original device mask layout, and the dimension is also cubic meters under the international standard system. The non-radiative dissipation capture efficiency coefficient represents the efficiency of electrons successfully injected into the luminescent active region after the external driving current crosses the structural contact isolation layer. It is constrained by the internal lattice leakage current effect. This parameter objectively reflects the proportion of charged particles that actually participate in subsequent photon radiation. Since it is essentially a percentage description of the material throughput, it is strictly locked by the underlying logic and defined as a pure, dimensionless numerical parameter.

[0036] As the injected carriers are repeatedly drawn in and violently accumulated within microsecond-level short wavelengths, the excited plasma free color extension dispersion effect forcibly drives and pulls the overall internal emission refractive index of the semiconductor micro-luminescent medium to undergo periodic torsional reshaping and violent deformation deflection. This refractive index shift ultimately directly leads to a lateral dynamic drift and tremor at the center peak of the output light, forming a chirp scouring effect. To free up hardware main memory and rapidly send offset reverse command data for subsequent disruption of external parasitic standing wave phase matching, this invention, based on a first-order low-power response dispersive wavelength cross-dimensional mapping relationship combined with a variant recombination equation, directly completes the solution through a first-order calculus approximation of order reduction mapping. This rapid solution, combined with the precisely arranged variant formula, is... Within this intervention mapping formula, The wavelength jitter data value of the newly generated light source, which represents the current output of the simulation and will soon be effective, is standardized in meters as the unit of measurement for microscopic waveform distance; to be substituted into the calculation. The newly generated payload for the previous processing node Dimensional data on changes in forward carrier concentration; The intrinsic differential mapping dispersion coefficient, which describes the sensitivity of plasma concentration to refractive index deflection, is essentially related to the gradient slope of the refractive index relative to the carrier density. Since the optical refractive index, which represents the strength of refraction in an optical system, inherently possesses typical dimensionless physical scalar properties, after the differential is derived by the gradient of the carrier concentration, the dimension of this differential mapping coefficient is objectively derived and constrained to cubic meters due to the inversion rule. The optical field confinement factor represents the optical field inside the Fabry-Perot resonant microcavity. It is mainly used to quantitatively evaluate the proportion of spatial energy overlap interception of the active region on the generated electromagnetic wave divergence mode. As an inherent pure fraction ratio constant of the system's spatial attributes, it is completely in the dimensionless world sequence. The center wavelength of the luminescent body at the factory, representing the luminous body's geometric calibration in a vacuum foundation before it is damaged by external human-induced current chirping disturbances, is also strictly defined in meters as the standard for foundation anchoring length. It represents the overall effective refractive index of the resonant cavity waveguide structure under the current ambient temperature gradient, and is a constant dimensionless solid dielectric refractive parameter defined by purely optical properties.

[0037] By continuously rewriting and re-evolving the data on beam wavelength jitter and frequency shift at the final output photon chain of the system through this extremely fast time-series update rate, the high-definition coherent relay length of the original monochromatic light wave transmitted in the optical fiber is forcibly stretched, squeezed, and even violently compressed and blocked by this system in the form of extremely strong electronic interference, to the point that it is impossible to piece together the limit span critical distance limit dimension of the external micro-curvature parasitic reverse standing wave. In this way, it directly and violently destroys, decomposes, and even completely dismantles the extremely stubborn and rigid dangerous fixed interference phase biting clamping and locking correlation relationship between the parasitic resonance wavefront caused by environmental scattering echoes at the external connection points of the entire network link and the original newly output photon wavefront inside the laser.

[0038] Furthermore, the dynamic phase difference data between the forward optical power waveform data and the backscattered optical power waveform data is continuously calculated, including: The instantaneous forward phase data of the forward optical power waveform data and the instantaneous back phase data of the backscattered optical power waveform data are extracted using Hilbert transform. Subtract the instantaneous backward phase data from the instantaneous forward phase data to obtain the instantaneous phase difference sequence; The instantaneous phase difference sequence is smoothed by time windowing to generate dynamic phase difference data.

[0039] In this embodiment of the invention, the specific process of continuously calculating the dynamic phase difference data of the forward optical power waveform data and the backscattered optical power waveform data firstly requires calling the Hilbert transform algorithm inside the digital signal processor to map the time-domain discretized forward optical power waveform data and the backscattered optical power waveform data to the orthogonal phase space to construct a complex analytic signal. Then, by solving the arctangent function of the imaginary part and the real part of the complex signal, the instantaneous forward phase data of the forward optical power waveform data and the instantaneous back phase data corresponding to the backscattered optical power waveform data on the running trajectory are accurately calculated and extracted. This process completely gets rid of the disadvantage of the extremely limited waveform resolution caused by the traditional hard-triggered zero-crossing detection, and realizes the absolute tracking of the phase fluctuation of the optical field at every microsecond-level sampling point. The physical dimensions of the two extracted instantaneous phase data are all standardly defined as the internationally used unit radian, which reflects the circumferential angular displacement. Next, since directly linearly subtracting the two phase signals inevitably suffers from the underlying mathematical limitation of the periodic truncation of the arctangent function, causing severe spurious phase jumps when the phase crosses the positive and negative circumference boundaries, the system, after obtaining a preliminary difference array by directly subtracting the instantaneous backward phase data from the instantaneous forward phase data, forcibly introduces a continuous phase unwrap algorithm to perform absolute continuity correction and compensation on the array, eliminating all unreasonable periodic intervals and thus obtaining a pure instantaneous phase difference sequence that can truly characterize the cumulative evolution trajectory of the microscopic spatial optical path of the two light waves throughout the round-trip cycle of the long-distance fiber optic link. Subsequently, to completely eliminate the random phase wander jitter caused by the shot noise of the photodetector under extremely low light intensity, this embodiment abandons the conventional rectangular smoothing window with equal weights and proposes and applies an adaptive Gaussian weighted smoothing variant model based on the backward energy confidence to perform high-order time windowing smoothing on the instantaneous phase difference sequence. The adaptive variant smoothing formula is constructed as follows: In this variant computing architecture, This represents the current generation after confidence level smoothing. The dynamic phase difference data of each discrete time node is maintained in radians as the final output physical representation dimension. The first after unwinding and straightening Numerical values ​​are extracted from the instantaneous phase difference sequence of each historical moment, with the same dimension in radians; The first one representing the front channel synchronization cache The absolute value of the true amplitude of the backscattered light power waveform data at a historical moment, whose actual dimension is inherited from the photoelectric receiver as a physical milliwatt, is cleverly modified and borrowed here as a dynamic weighting coefficient to characterize the noise resistance confidence of the current phase data. The necessity of this modification is that when the backscattered light falls to the bottom noise region due to the Rayleigh scattering characteristics of the link, the signal-to-noise ratio of the corresponding phase is extremely poor or even phase failure occurs. Through the automatic clamping and scaling of this milliwatt-level power amplitude, the unreliable phase difference under low light intensity can be dynamically deweighted, which greatly improves the strong robustness of the overall dynamic phase difference smooth tracking. The Gaussian time decay window function is constructed to assign lower memory weights to more distant historical data, where... The hysteresis index for traversing the smooth window is an integer constant, without physical units. The preset Gaussian smoothing standard deviation factor is preferably set in the dimensionless range of 2 to 4 to ensure that the smooth transition does not lose its noise filtering effect due to excessively rapid attenuation, nor does it mask the true phase abrupt inflection point due to excessively slow attenuation. The total length of the preset sliding smoothing time window is set to a pure digital dimensionless counting interval containing 15 to 30 sampling points to perfectly cover the half-cycle span of a single high-frequency oscillation induced by micro-bending deformation.

[0040] Furthermore, the determination steps for dynamic phase difference data deviating from the fixed locking interval include: Obtain reference phase difference data under historical conditions without micro-bending; A fixed locking interval with a preset floating tolerance is constructed, centered on an integer multiple of pi value of the reference phase difference data; The dynamic phase difference data is compared with the boundary values ​​of the fixed locking interval; When the dynamic phase difference data falls outside the boundary value of the fixed locking interval and the phase difference change rate data is greater than the phase change threshold, it is determined that the dynamic phase difference data deviates from the fixed locking interval.

[0041] In this embodiment of the invention, the process of determining whether the dynamic phase difference data deviates from the fixed locking interval first involves obtaining reference phase difference data recorded during continuous operation of the system without external mechanical micro-bending from a historical steady-state database. This data accurately represents the inherent steady-state interference fundamental frequency origin of the long-span fiber optic link under a static temperature field. Next, using this reference phase difference data as the anchor point, an integer multiple of the constant pi is introduced to induce a phase shift, and a preset floating tolerance is superimposed around it, thereby constructing a fixed locking interval with clearly defined boundaries. The mathematical set of this locking interval is set as follows: In the formula This is the reference phase difference data with the dimension of radians; It is an integer constant representing the interference period and is a dimensionless parameter; It is a natural constant; The preset floating tolerance parameter, with its dimension in radians, is preset within the range of 0.15 to 0.35 radians. This preset logic is designed to fully accommodate the slight optical path creep caused by the optical fiber under a slow diurnal temperature gradient, and avoid misinterpreting it as high-frequency mechanical vibration. Subsequently, the input dynamic phase difference data needs to be compared with the upper and lower boundary values ​​of the fixed locking interval in real time. Simultaneously, to avoid misjudgments caused by shot noise jumps in the underlying photoelectric detector, the distortion rate of the transient phase must be verified. This embodiment abandons the conventional first-order difference method that amplifies high-frequency noise, and uses a time-weighted gradient variant formula based on an exponential forgetting factor to calculate the phase difference variation rate data. The specific variant is as follows: ; in the formula The extracted phase difference change rate data at the current time point is used to quantify the phase tearing speed of the optical field, with the dimension being radians per second; and Representing the present moment and historical retrospect respectively. The dynamic phase difference data input at each step size is all in radians. For dimensionless step index; The preset sliding backtracking depth is set to a dimensionless integer between 5 and 15 to ensure that the observation window can accommodate the complete mutation front. This represents the underlying physical discrete sampling period, measured in seconds. A preset smooth convergence parameter, used to control the geometric decay rate of historical weights, is strictly defined as a dimensionless constant between 0.2 and 0.6. Finally, when the input dynamic phase difference data falls outside the boundary value of the fixed locking interval, and the absolute value of the calculated phase difference change rate is greater than the preset phase abrupt change threshold, the system determines that the dynamic phase difference data has completely deviated from the fixed locking interval, indicating that external distortion has torn apart the parasitic interference-locked steady state. The preset phase abrupt change threshold is set in radians per second, between 500 and 1500 radians per second, using this high threshold to achieve the engineering goal of 100% filtering out false temperature drift alarms.

[0042] Furthermore, gradually reduce and remove high-frequency modulation chirped data, including: The stepped descent envelope data is generated by decreasing the amplitude control word data of the high-frequency modulated chirp data by a preset step size. Within each decreasing period of the stepped descent envelope data, the oscillation amplitude data of the forward dynamic amplitude-frequency matrix is ​​reread; If the oscillation amplitude data rebounds above the preset safety threshold, the decrease will stop and the current amplitude control word data will be maintained; If the oscillation amplitude data remains below the preset safety threshold, the decrementing operation continues until the amplitude control word data returns to zero, thus completing the cancellation of the high-frequency modulation chirp data.

[0043] In this embodiment of the invention, the process of gradually reducing and canceling the high-frequency modulation chirped data first involves the main control chip linearly decreasing the amplitude control word data sent to the bias circuit according to a predetermined program and a preset step size, thereby generating a stepped descent envelope data containing multiple buffer steps in the time domain. The preset step size is set to a dimensionless scaling factor of 5% to 10% of the peak amplitude command value, and the holding period for each step is set to 10 to 20 microseconds, aiming to avoid secondary thermal oscillations caused by instantaneous collapse of the optical field through a controlled current soft landing mode. Next, within each decreasing cycle of the stepped descent envelope data, the system needs to reread the transient spectral amplitude of the forward dynamic amplitude-frequency matrix in the target parasitic frequency band. To overcome the misjudgment caused by the dark current noise spike of the photodiode, an energy-weighted smoothing variant model based on time-damped decay is introduced to calculate the comprehensive oscillation amplitude data within the current decreasing cycle. The variant formula is: In the formula, For the first The comprehensive oscillation amplitude data within each step cycle is measured in milliwatts. For monitoring the first window The forward transient amplitude at a dimensionless, purely digital index frequency offset point, with the dimension in milliwatts; The preset observation window point constant is a dimensionless integer between 50 and 100 to cover the complete aftereffect fading transition zone. A preset digital damping convergence parameter, used to apply attenuation penalty to stale sampled data, is set to a dimensionless constant between 0.05 and 0.15. Subsequently, the output composite oscillation amplitude data... The system compares the data with a preset safety threshold, which represents the baseline residual power of the system under conditions of no microbending interference. This threshold is measured in milliwatts and is set to be between 0.02 and 0.08 milliwatts. If the oscillation amplitude data is determined to rebound and surge above the preset safety threshold, it indicates that the external microbending stress coupling has not been completely decoupled. The system immediately terminates the decreasing process and maintains the current amplitude control word data as the dynamic suppression baseline. Conversely, if continuous comparisons confirm that the oscillation amplitude data remains stably below the preset safety threshold, the system continues to perform decreasing fine-tuning operations until the amplitude control word data gradually returns to zero in steps, completing the smooth cancellation and reset clearing of the entire high-frequency modulation chirp data in the fiber optic link.

[0044] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described or use similar methods to replace them, as long as they do not deviate from the concept of the invention, they should all fall within the protection scope of the present invention.

[0045] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0046] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A method for controlling a fiber optic communication light source, characterized in that, Includes the following steps: Forward optical power waveform data is acquired by a forward photodetector, backscattered optical power waveform data is acquired by a backward photodetector, and drive current ripple data is obtained at the laser driver sampling end. Short-time Fourier transform is performed on the forward optical power waveform data and the drive current ripple data to extract the corresponding forward dynamic amplitude-frequency matrix and drive dynamic amplitude-frequency matrix within the preset sliding time window, respectively. Calculate the coherence coefficient matrix between the forward dynamic amplitude-frequency matrix and the driving dynamic amplitude-frequency matrix; When the coherence value in the coherence coefficient matrix is ​​greater than the preset warning threshold and the amplitude envelope of the forward dynamic amplitude-frequency matrix exhibits an exponential growth characteristic exceeding the system damping coefficient, the parasitic resonant cavity diagnostic loop is triggered. The diagnostic loop for parasitic resonant cavities specifically includes: Pilot perturbation data from non-communication frequency bands are superimposed on the light source bias current; The backscattered light power waveform data after superimposing pilot perturbation data is analyzed, and the intermodulation distortion component between the oscillation frequency and the pilot frequency is extracted from the waveform data. When the amplitude of the intermodulation distortion component is greater than the preset diagnostic threshold, high-frequency modulated chirped data is injected into the bias current to induce light source wavelength jitter data using the carrier dispersion effect. During the injection of high-frequency modulated chirped data, the dynamic phase difference data between the forward optical power waveform data and the backscattered optical power waveform data is continuously calculated, and the oscillation amplitude data of the forward dynamic amplitude-frequency matrix is ​​updated synchronously. When the oscillation amplitude data decays to below the preset safety threshold and remains below it for a preset time, and the dynamic phase difference data deviates from the fixed locking range, the high-frequency modulation chirp data is gradually reduced and canceled, the parasitic resonant cavity diagnostic loop is exited, and the oscillation suppression completion signal is output.

2. The optical fiber communication light source control method according to claim 1, characterized in that, Short-time Fourier transforms are performed on the forward optical power waveform data and the drive current ripple data to extract the corresponding forward dynamic amplitude-frequency matrix and drive dynamic amplitude-frequency matrix within a preset sliding time window, including: The forward optical power waveform data and the drive current ripple data are segmented and extracted using a preset sliding time window to generate a forward optical power data frame sequence and a drive current ripple data frame sequence arranged in time order. Discrete Fourier transform calculations are performed on each frame of data in the forward optical power data frame sequence and the drive current ripple data frame sequence to obtain the amplitude component of each frame of data in the frequency domain. All amplitude components corresponding to the forward optical power data frame sequence are reassembled into a forward amplitude vector set, and all amplitude components corresponding to the drive current ripple data frame sequence are reassembled into a drive amplitude vector set. Following the time step sequence of the preset sliding time window, the vectors in the forward amplitude vector set are sequentially concatenated dimensionally to construct the forward dynamic amplitude-frequency matrix, and the vectors in the driving amplitude vector set are sequentially concatenated dimensionally to construct the driving dynamic amplitude-frequency matrix.

3. The optical fiber communication light source control method according to claim 2, characterized in that, Calculate the coherence coefficient matrix between the forward dynamic amplitude-frequency matrix and the driving dynamic amplitude-frequency matrix, including: Extract the dominant frequency band forward vector and the dominant frequency band driving vector from the forward dynamic amplitude-frequency matrix; Calculate the cross power spectral density matrix of the dominant frequency band forward vector and the dominant frequency band driving vector, and calculate the auto power spectral density matrix of the dominant frequency band forward vector and the auto power spectral density matrix of the dominant frequency band driving vector respectively. The coherence coefficient matrix is ​​generated by dividing the square of the magnitude of the cross power spectral density matrix by the product of the two self power spectral density matrices.

4. The optical fiber communication light source control method according to claim 1, characterized in that, The steps for determining whether the amplitude envelope of the forward dynamic amplitude-frequency matrix exhibits an exponential growth characteristic exceeding the system damping coefficient include: Extract the peak envelope data sequence from the forward dynamic amplitude-frequency matrix along the time dimension; Calculate the set of amplitude ratios between adjacent time points in the peak envelope data sequence; When all ratios in the amplitude ratio set are greater than the product of the natural constant and the system damping coefficient, it is determined that the amplitude envelope of the forward dynamic amplitude-frequency matrix exhibits an exponential growth characteristic exceeding the system damping coefficient.

5. The optical fiber communication light source control method according to claim 1, characterized in that, It also includes a branch for determining the driving coupled oscillation: When the amplitude of the intermodulation distortion component is not greater than the preset diagnostic threshold and the coherence value in the coherence coefficient matrix is ​​greater than the preset warning threshold, skip the injection step of high-frequency modulation chirped data. Generate drive power supply filter compensation instruction data; The drive power supply filter compensation command data is sent to the power supply low dropout linear regulator. The ripple amplitude corresponding to the drive current ripple data is reduced by adjusting the stopband attenuation parameter of the digital filter, and the parasitic resonant cavity diagnostic loop is exited.

6. The optical fiber communication light source control method according to claim 1, characterized in that, The backscattered light power waveform data after superimposing pilot perturbation data is analyzed to extract the intermodulation distortion components between the oscillation frequency and the pilot frequency, including: The frequency value corresponding to the current highest frequency peak in the backscattered light power waveform data is obtained as the oscillation frequency data, and the pilot frequency data corresponding to the pilot perturbation data is obtained. Calculate the sum frequency and difference frequency values ​​of the oscillation frequency data and the pilot frequency data; In the frequency domain spectrum of the backscattered light power waveform data, the amplitude spectrum values ​​at the frequency points where the sum frequency value and the difference frequency value are located are extracted, and the extracted amplitude spectrum values ​​are used as intermodulation distortion components.

7. The optical fiber communication light source control method according to claim 1, characterized in that, Injecting high-frequency modulated chirped data into the bias current, and using the carrier dispersion effect to induce wavelength jitter data in the light source, including: Generate a high-frequency sawtooth wave digital sequence with a frequency higher than the communication frequency band and an amplitude within the limit of the saturated output power of the light source; The high-frequency sawtooth wave digital sequence is converted into an analog current signal and superimposed on the bias current to form high-frequency modulated chirped data; Acquire data on the change in carrier concentration inside the laser after high-frequency modulation and chirping data injection; Based on the periodic changes in the refractive index of the active region caused by carrier concentration variation data, light source wavelength jitter data with nonlinear shift over time is generated.

8. The optical fiber communication light source control method according to claim 1, characterized in that, Continuously calculate the dynamic phase difference data between the forward optical power waveform data and the backscattered optical power waveform data, including: The instantaneous forward phase data of the forward optical power waveform data and the instantaneous back phase data of the backscattered optical power waveform data were extracted using Hilbert transform. Subtract the instantaneous backward phase data from the instantaneous forward phase data to obtain the instantaneous phase difference sequence; The instantaneous phase difference sequence is smoothed by time windowing to generate dynamic phase difference data.

9. The optical fiber communication light source control method according to claim 1, characterized in that, The steps for determining whether dynamic phase difference data deviates from the fixed locking range include: Obtain reference phase difference data under historical conditions without micro-bending; A fixed locking interval with a preset floating tolerance is constructed, centered on an integer multiple of pi value of the reference phase difference data; The dynamic phase difference data is compared with the boundary values ​​of the fixed locking interval; When the dynamic phase difference data falls outside the boundary value of the fixed locking interval and the phase difference change rate data is greater than the phase change threshold, it is determined that the dynamic phase difference data deviates from the fixed locking interval.

10. The optical fiber communication light source control method according to claim 1, characterized in that, Gradually reduce and remove high-frequency modulation chirp data, including: The stepped descent envelope data is generated by decreasing the amplitude control word data of the high-frequency modulated chirp data by a preset step size. Within each decreasing period of the stepped descent envelope data, the oscillation amplitude data of the forward dynamic amplitude-frequency matrix is ​​reread; If the oscillation amplitude data rebounds above the preset safety threshold, the decrease will stop and the current amplitude control word data will be maintained; If the oscillation amplitude data remains below the preset safety threshold, the decrementing operation continues until the amplitude control word data returns to zero, thus completing the cancellation of the high-frequency modulation chirp data.

Images