Detection method of probiotics in fermented vegetable feed based on Internet of Things

Abstract

The present invention discloses a method for detecting probiotics in fermented tail vegetable feed based on the Internet of Things, which relates to the field of microbial detection technology and includes the following steps: obtaining complete echo signal time series data, and performing high-frequency sampling on the echo signal time series with a sliding time window of fixed length to construct a three-dimensional light signal transient feature vector. The present invention achieves accurate identification of abnormal disturbances in the light signal by constructing a three-dimensional light signal transient feature vector and introducing a disturbance residual and time-sensitive weight mechanism; combining the optical interference index with a dynamic sampling window adjustment strategy to automatically avoid refractive index mutation areas, avoid sensor misjudgment and data distortion, significantly improve the detection stability and accuracy of the probiotic activity state in a fermentation environment, and enhance the intelligent monitoring capability of feed quality.

Description

Technical Field

[0001] The present invention relates to the technical field of microbial detection, and in particular to a method for detecting probiotics in fermented vegetable feed based on the Internet of Things. Background Art

[0002] The detection of probiotics in fermented tail vegetable feed based on the Internet of Things refers to the use of Internet of Things technology to build an intelligent detection system that integrates perception, transmission and analysis, and conducts real-time monitoring and evaluation of the activity level or content of probiotics in feed made from tail vegetables through microbial fermentation. This method deploys multiple types of biosensors (such as pH, conductivity, gas composition, bioluminescence responders, etc.) to collect online indicators related to probiotic metabolism during the fermentation process, and uploads the collected data to the cloud platform or edge computing terminal with the help of a wireless transmission module. Combined with the built-in model, it makes intelligent judgments on the growth status of probiotics, fermentation maturity and its dynamic change trends, thereby achieving precise control of feed quality. This method can effectively improve the level of intelligent management of microbial activity in the process of agricultural waste resource utilization, and ensure the safe and efficient application of fermented tail vegetable feed in livestock and poultry farming.

[0003] The existing technology has the following deficiencies: In the existing technology, the use of optical fiber sensors to monitor the growth status of probiotics in fermented vegetable feed in real time has been gradually applied to the process of agricultural waste resource utilization. However, in the vegetable fermentation environment, probiotics decompose organic matter in large quantities under anaerobic or facultative conditions, and continuously release a variety of metabolic gases including carbon dioxide (CO2) and hydrogen sulfide (H2S). When these gases are unevenly distributed in the internal space of the fermentation container, they are prone to local accumulation in the end area of ​​the optical fiber sensor, causing a sudden change in the refractive index of the medium in the sensing area, and then causing nonlinear bending of the light signal propagation path, reflection angle deviation or multiple refraction anomalies, forming an "optical interference zone". Once the above anomaly occurs, the echo signal output by the optical fiber sensor will experience unstable fluctuations or systematic deviations, which will at least affect the short-term judgment of the probiotic activity data, and at worst cause distortion of the process control of the entire batch fermentation cycle. In addition, this type of refractive index disturbance phenomenon is hidden and unpredictable, and conventional calibration methods are difficult to identify in a timely manner. It is very easy to form a "false signal normal" data loop, which can lead to major quality and safety hazards such as uncontrolled reproduction of probiotics, misjudgment of fermentation termination, or serious decline in feed quality.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for detecting probiotics in fermented tail vegetable feed based on the Internet of Things. By constructing a three-dimensional optical signal transient feature vector and introducing a disturbance residual and time-sensitive weight mechanism, accurate identification of abnormal disturbances in the optical signal can be achieved; combining the optical interference index with the dynamic sampling window adjustment strategy, the refractive index mutation area is automatically avoided, sensor misjudgment and data distortion are avoided, and the detection stability and accuracy of the active state of probiotics in the fermentation environment are significantly improved, and the intelligent monitoring capability of feed quality is enhanced to solve the problems in the above-mentioned background technology.

[0006] In order to achieve the above object, the present invention provides the following technical solution: a method for detecting probiotics in fermented vegetable feed based on the Internet of Things, comprising the following steps:

[0007] Acquire the complete echo signal time series data, perform high-frequency sampling on the echo signal time series with a sliding time window of fixed length, and construct a three-dimensional optical signal transient feature vector;

[0008] The transient feature vector of the three-dimensional optical signal generated in each time window is calculated dimension by dimension with the feature vector of the standard optical signal established under a stable fermentation environment to generate a disturbance residual vector; the disturbance residual vectors are arranged in chronological order to form a disturbance residual vector sequence;

[0009] Normalize the disturbance residual vector sequence, calculate the ratio between the maximum disturbance residual and the minimum disturbance residual in each time window, and generate the disturbance response value;

[0010] Based on the dynamic trend of kurtosis change in the optical signal waveform, a disturbance time-sensitive weight factor is constructed;

[0011] The disturbance response value is multiplied point by point by the disturbance time-sensitive weight factor to generate a weighted disturbance value sequence; within the range of a continuous sliding window, the weighted disturbance value sequence is cumulatively summed to generate an optical interference index;

[0012] The optical interference index is compared with a preset optical interference reference threshold. When the optical interference index is greater than the reference threshold, the dynamic sampling window adjustment mechanism is activated to change the length and start and end time parameters of the original sampling time interval, and perform an offset operation on the sampling window to avoid the abnormal disturbance area in the optical signal and reduce the detection error caused by the sudden change of the refractive index.

[0013] Preferably, the specific steps of obtaining the optical fiber sensor echo signal time series and performing high-frequency sampling to construct a three-dimensional optical signal transient feature vector are as follows:

[0014] For the time series of echo signals output by the fiber optic sensor during the target monitoring period, the original signal is segmented using a sliding time window of a set length. Signal amplitude normalization and noise filtering operations are performed on the signal curve within each time window. The local standard deviation of the signal amplitude in the current time period is calculated to extract the signal intensity fluctuation characteristics.

[0015] Based on the normalized optical signal waveform within each time window, a time correlation analysis method is used to accurately match and locate the main peak time difference between the incident pulse signal and the echo signal. The time delay value of the signal from incident to return is calculated as the reflection delay characteristic in the current window.

[0016] After completing the reflection delay analysis, the fast Fourier transform algorithm is used to perform frequency domain transformation on the echo signal within each time window. The rate of change of the phase difference between adjacent main frequency components is calculated to quantify the phase offset amplitude generated during the propagation of different frequency components, forming a complete optical signal phase offset characteristic. In turn, a three-dimensional optical signal transient feature vector is constructed that reflects the instantaneous transmission state of the optical signal.

[0017] Preferably, the specific steps of constructing the disturbance residual vector and extracting the change trend are as follows:

[0018] Within each sliding time window, each component of the three-dimensional optical signal transient feature vector constructed in the current time window is subtracted from the corresponding component of the standard three-dimensional optical signal feature vector pre-calibrated in a stable fermentation environment, forming a disturbance residual vector that reflects the degree of deviation between the optical signal propagation state in the current time period and the standard state.

[0019] According to the continuity of the sliding time window, the disturbance residual vectors obtained in each time period are arranged in order according to the time sequence to construct the disturbance residual vector sequence;

[0020] Smoothing is performed on the disturbance residual vector sequence to eliminate local abnormal mutations caused by instantaneous environmental fluctuations while retaining the overall change trend.

[0021] Preferably, the specific steps of quantifying the disturbance intensity of the optical signal under the sudden change of the refractive index of the medium based on the disturbance response value are as follows:

[0022] Normalization is performed on the disturbance residual vector sequence constructed under the continuous time window. By subtracting the minimum value of the sequence from each component and then dividing it by the extreme value of the sequence, all disturbance residual information is standardized in a unified scale.

[0023] In each sliding time window, the maximum and minimum values ​​in the normalized disturbance residual sequence within the current window are extracted, and the numerical ratio between the two is calculated to characterize the range of fluctuation amplitude of the disturbance signal within the current time window, thereby forming a primary disturbance quantitative index reflecting the severity of the optical signal transmission state;

[0024] The disturbance signal fluctuation amplitude ratio is used as the disturbance response value corresponding to the current time window to specifically characterize the degree of instability of the propagation path of the optical signal caused by the sudden change of the medium refractive index in the current time period.

[0025] Preferably, the specific steps of constructing a disturbance time-sensitive weight factor based on the dynamic trend of the optical signal kurtosis change and performing weighted processing on the disturbance response value are as follows:

[0026] In each sliding time window, the optical signal waveform in the current time period is statistically analyzed, the fourth-order kurtosis value of the signal amplitude distribution is extracted, and the kurtosis change rate is calculated through continuous windows;

[0027] A time-sensitive weight factor is constructed based on the kurtosis change rate, and the exponentially weighted moving average algorithm is used to update the factor. By setting the time decay coefficient, the factor is dynamically focused on the latest signs of abnormal disturbances.

[0028] The time-sensitive weight factor is multiplied by the disturbance response value in the corresponding time window to obtain a weighted disturbance value sequence.

[0029] Preferably, the specific steps of generating the optical interference index reflecting the stability state of the optical signal are as follows:

[0030] In each sliding time window, the disturbance response value corresponding to the current time window is multiplied point by point by the disturbance time-sensitive weight factor updated by the exponentially weighted moving average algorithm to generate the weighted disturbance value under the current time window;

[0031] Arrange the weighted disturbance values ​​generated in consecutive time periods in the order of the sliding time window to construct a complete weighted disturbance value sequence;

[0032] The weighted disturbance value sequence is cumulatively summed within the current sliding analysis range to generate the optical interference index at the current sampling time point.

[0033] Preferably, when the optical interference index is greater than a reference threshold, the dynamic sampling window adjustment mechanism is activated to change the length and start and end time parameters of the original sampling time interval, and perform an offset operation on the sampling window so that the sampling segment avoids the abnormal disturbance area in the optical signal. The specific steps are as follows:

[0034] After obtaining the optical interference index at the current time point, in order to evaluate its deviation from the stable state, the optical interference offset rate is calculated using the optical interference index reference threshold as a comparison benchmark. The calculation expression is as follows:

[0035]

[0036] Where, I t is the optical interference index, I ref is the optical interference index reference threshold, δ t is the optical interference offset rate;

[0037] Based on the obtained optical interference offset rate δ t , enter the adaptive adjustment of the sampling window, expand the original sampling window length, and calculate the current time amplitude to be offset based on the offset rate to ensure that the sampling segment avoids the abnormal disturbance area in the signal. The calculation expression is as follows:

[0038]

[0039] Where ΔT is the original sampling window length, α is the window expansion sensitivity coefficient, tanh(·) is the hyperbolic tangent function, and ΔT ′ is the dynamically adjusted sampling window length, β is the offset sensitivity coefficient, ∈ t is the sampling window offset.

[0040] The technical effect of the present invention is as follows: by constructing a multi-dimensional optical signal transient characteristic vector, introducing a dynamic monitoring mechanism for disturbance residuals, and combining a time-sensitive weighting factor with a kurtosis change, high-precision identification of abnormal disturbances in optical signals is achieved. At the same time, through the continuous quantification of the optical interference index and the adaptive adjustment of the dynamic sampling window, the "optical interference zone" caused by the sudden change in the refractive index of the medium is effectively avoided, sensing misjudgment and control distortion are avoided, and the "false signal normal" problem that is prone to occur in traditional optical fiber sensors in fermentation scenarios is fundamentally solved. This method not only enhances the system's real-time perception of the active state of probiotics, but also ensures the stability and accuracy of feed quality monitoring, providing more intelligent and reliable detection support for the agricultural waste resource process. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction to the drawings required for use in the embodiments will be given below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0042] Figure 1 The present invention is a flow chart of the method for detecting probiotics in fermented vegetable feed based on the Internet of Things. DETAILED DESCRIPTION

[0043] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of this disclosure will be thorough and complete and will fully convey the concepts of the example embodiments to those skilled in the art.

[0044] The present invention provides Figure 1 The method for detecting probiotics in fermented tail vegetable feed based on the Internet of Things includes the following steps:

[0045] The complete echo signal time series data of the optical fiber sensor within the target monitoring period is acquired. High-frequency sampling is performed on the echo signal time series using a sliding time window of fixed length. Within each sliding time window, the intensity fluctuation characteristics, reflection delay characteristics, and phase offset characteristics of the optical signal are extracted to construct a three-dimensional optical signal transient feature vector to describe the instantaneous transmission state of the optical signal within the current time window.

[0046] The core function of this step is to achieve real-time identification and quantitative characterization of optical transmission anomalies that may be caused by the metabolic process of probiotics in fermented tail vegetable feed, laying a data foundation for subsequent optical path perturbation detection and sampling control. In the complex and dynamically changing fermentation environment, probiotics release gases such as carbon dioxide and hydrogen sulfide during metabolism. The uneven spatial distribution of these gases within the fermentation vessel affects the refractive index of the medium surrounding the optical fiber, thereby changing the propagation path and characteristics of the optical signal. By acquiring the complete echo signal time series within the target monitoring period and dividing it into multiple sliding time windows of fixed length, local features of the optical signal at different time scales can be extracted. The optical signal intensity fluctuation feature extracted within each time window reflects the dynamic change trend of the optical signal energy over time and is a primary indicator for determining the stability of the sensing environment. The reflection delay feature reveals the temporal variation of the round-trip propagation path of the optical signal and is an important basis for identifying whether the optical path is extended or distorted due to gas perturbations or changes in the medium structure. The phase shift feature reflects the interference or microstructural perturbation effects of the light wave at different frequency components and is one of the most sensitive perturbation features. The three-dimensional transient characteristic vector of the optical signal, formed by the synthesis of these three, can comprehensively reflect the overall state of optical signal propagation during the current period. This step not only quantifies the complex optical change process into a processable numerical vector, but also provides high-quality raw data support for subsequent anomaly identification, disturbance modeling, and optical interference correction, thereby improving the overall probiotic monitoring system's perception accuracy and response efficiency to sudden disturbances.

[0047] The specific steps for obtaining the time series of the optical fiber sensor echo signal and performing high-frequency sampling to construct the three-dimensional optical signal transient feature vector are as follows:

[0048] For the time series of echo signals output by the fiber optic sensor during the target monitoring period, the original signal is segmented using a sliding time window of a set length. Signal amplitude normalization and noise filtering are performed on the signal curve within each time window to eliminate the interference of acquisition errors on signal strength evaluation. The local standard deviation of the signal amplitude in the current time period is calculated to extract the signal strength fluctuation characteristics.

[0049] Within each sliding time window, the raw echo signal collected by the fiber optic sensor is first amplitude normalized to eliminate the effects of sensor sensitivity variations and environmental noise on signal strength. A filtering algorithm (such as median filtering or Kalman filtering) is then applied to remove high-frequency interference from the signal, thereby retaining a representative amplitude variation trend. The local standard deviation of the normalized signal within this window is then calculated to measure the amplitude fluctuation of the signal intensity during that period. This intensity fluctuation characteristic reflects changes in the optical transmission medium caused by the metabolic activity of probiotics and is a key indicator for determining the initial onset of optical path disturbances.

[0050] Based on the normalized optical signal waveform within each time window, a time correlation analysis method is used to accurately match and locate the main peak time difference between the incident pulse signal and the echo signal. The time delay value of the signal from incident to return is calculated as the reflection delay characteristic in the current window.

[0051] After extracting the intensity features, a time correlation analysis is performed on the incident and echo signals within the current time window. By matching the main peak positions of the two, the propagation time difference between them is calculated, thereby obtaining the reflection delay time experienced by the current light signal in the fermentation medium. This delay time can reveal changes in the propagation path caused by microstructural changes or changes in medium density in the optical path, such as gas accumulation and liquid phase disturbances. The reflection delay feature helps determine whether nonlinear deformation has occurred in the optical transmission path and provides a quantitative basis in the time dimension for further identifying refractive index mutations.

[0052] After completing the reflection delay analysis, the fast Fourier transform algorithm is used to perform frequency domain transformation on the echo signal within each time window. The rate of change of the phase difference between adjacent main frequency components is calculated to quantify the phase offset amplitude generated during the propagation of different frequency components, forming a complete optical signal phase offset characteristic. In turn, a three-dimensional optical signal transient feature vector is constructed that reflects the instantaneous transmission state of the optical signal.

[0053] After acquiring the intensity and delay characteristics, a fast Fourier transform is performed on the echo signal within the current time window, converting it from the time domain to the frequency domain to analyze the phase changes of different frequency components. By calculating the rate of change of the phase difference between each main frequency component, the degree to which the optical signal is affected by medium perturbations during propagation is further evaluated. The phase shift feature has the advantage of being highly sensitive to small perturbations and can capture optical path interference signals that cannot be detected by conventional intensity or delay analysis. It provides supplementary information on microstructure dynamics to the three-dimensional feature vector, enhancing the accuracy and robustness of the entire perturbation identification model.

[0054] The transient feature vector of the three-dimensional optical signal generated in each time window is compared with the feature vector of the standard optical signal established under a stable fermentation environment, and the perturbation residual vector reflecting the optical path offset state is generated. The perturbation residual vector is then combined into a perturbation residual vector sequence in chronological order to dynamically represent the change trend between the current detection environment and the reference state.

[0055] The role of this step is to construct a disturbance residual data structure for dynamically monitoring the stability of the optical propagation path by performing a dimension-by-dimension difference analysis on the transient characteristics of the optical signal and the standard state, thereby achieving accurate identification and trend judgment of the perturbation changes in the fermentation environment. In the production process of fermented tail vegetable feed, the active metabolic behavior of probiotics can cause dynamic changes in the gas composition, density distribution and temperature field in the fermentation medium, thereby affecting the refractive index of the medium in which the optical fiber sensor is located, causing the optical signal to bend, delay or phase drift during transmission. In order to quantify this disturbance, this step compares the three-dimensional optical signal transient feature vector generated under each sliding time window with the standard feature vector established under ideal stable fermentation conditions, and calculates the deviation values ​​of intensity, time delay and phase respectively, thereby generating a disturbance residual vector. This vector not only accurately records the offset amplitude between the optical path at the current moment and the standard state, but also reflects which type of optical feature is disturbed, thereby providing a basis for interference classification. Further, all residual vectors are constructed as a disturbance residual vector sequence in chronological order, which can intuitively reflect the dynamic evolution trajectory of the optical propagation environment on the time axis. This sequence is crucial for identifying the persistence, periodicity, or suddenness of optical path disturbances, and serves as the fundamental data source for subsequently constructing disturbance response values ​​and triggering dynamic sampling mechanisms. In summary, this step, through residual quantization and serialization, establishes a mapping relationship between optical transmission characteristics and changes in the probiotics' active environment, significantly improving the detection system's ability to perceive and assess the timing of small disturbances.

[0056] The specific steps of constructing the disturbance residual vector and extracting the change trend are as follows:

[0057] Within each sliding time window, each component of the three-dimensional optical signal transient feature vector constructed in the current time window is subtracted from the corresponding component of the standard three-dimensional optical signal feature vector pre-calibrated in a stable fermentation environment, forming a disturbance residual vector that reflects the degree of deviation between the optical signal propagation state in the current time period and the standard state.

[0058] In each sliding time window, the three characteristic components of the three-dimensional optical signal transient feature vector extracted in the current window, namely the signal intensity fluctuation value, the reflection delay time value and the phase offset amplitude value, are first subjected to a one-to-one difference calculation with the corresponding components in the standard feature vector established through long-term sampling and statistical analysis under stable fermentation conditions. The difference result constitutes a disturbance residual vector, which is used to quantitatively express the degree of deviation of the optical signal propagation state in the current window relative to the ideal reference state. This operation not only provides basic data for identifying abnormal fluctuations caused by external disturbances, but also, through dimensional decomposition, can determine which type of optical signal characteristics have undergone mutations, thereby providing pre-emptive support for the classification of interference sources.

[0059] According to the continuity of the sliding time window, the disturbance residual vectors obtained in each time period are arranged in order according to the time sequence to construct a disturbance residual vector sequence, which is used to preserve the change trend of the disturbance in the time domain and reflect the dynamic evolution process of the probiotic metabolic environment.

[0060] After obtaining the perturbation residual vectors for each time window, these vectors are then concatenated sequentially according to the chronological order of the sliding window to construct a complete perturbation residual vector sequence. This sequence preserves the continuous evolution of the perturbation in the temporal dimension and can reveal the deviation trajectory between the optical propagation environment and the standard state throughout the fermentation cycle. By constructing this time series, not only can the time period and duration of the perturbation be observed, but also the presence of periodic interference, sudden disturbances, or cumulative effects can be preliminarily identified, providing a dynamic basis for determining the trend of optical perturbations.

[0061] Smoothing is performed on the disturbance residual vector sequence to eliminate local abnormal mutations caused by instantaneous environmental fluctuations while retaining the overall change trend, making the residual data sequence more suitable for subsequent optical path stability modeling and interference identification, and improving the accuracy and timeliness of fermentation environment change judgments.

[0062] To reduce the impact of external transient noise and random system fluctuations on disturbance identification, a smoothing algorithm, such as a moving average or exponential smoothing method, is applied to the constructed disturbance residual vector sequence to filter and reduce noise on the disturbance values ​​in each time window. This smoothed residual sequence more accurately reflects the macroscopic changes in the optical signal propagation state over time, helping to eliminate misleading judgment results caused by occasional outliers. This improves the accuracy and stability of subsequent determinations of optical interference levels, adjustments to sampling strategies, and optimization of fermentation process control.

[0063] Normalize the perturbation residual vector sequence, calculate the ratio between the maximum perturbation residual and the minimum perturbation residual in each time window, and generate a perturbation response value to represent the perturbation intensity caused by the sudden change in the refractive index of the medium in the current time window.

[0064] The role of this step is to standardize the original disturbance residual data and calculate the degree of disturbance amplitude change within each time window through extreme value analysis, thereby achieving dynamic quantification and sensitive expression of the stability state of the optical signal. In the complex fermented tail vegetable feed environment, the metabolic activity of probiotics and the medium changes caused by them (such as gas accumulation, temperature rise or density fluctuations) will affect the refractive index of the area where the optical fiber sensor is located, thereby interfering with the propagation path of the optical signal, manifesting as drastic changes in signal intensity, delay or phase. Since these changes occur with random, local and nonlinear characteristics, if the data is not normalized, it will lead to the inability to compare the disturbance values ​​under different time windows horizontally, seriously affecting the accuracy of subsequent anomaly identification. Therefore, the first step of the normalization process is to unify all disturbance residual values ​​into a standard scale interval, eliminating the masking effect of dimensional differences and occasional noise on the overall fluctuation pattern. On this basis, by extracting the maximum and minimum values ​​of the normalized residual sequence within each sliding time window, the ratio of the two is further calculated. This ratio effectively reflects the fluctuation amplitude affected by the stability of optical signal propagation in the current time period. This disturbance response value not only quantitatively characterizes the degree of disturbance in a single window but also serves as a core input parameter in subsequent steps, participating in multiple key decision-making processes such as dynamic weighting, optical interference index construction, and sampling window adjustment. In short, this step achieves highly sensitive extraction of optical signal disturbance intensity through mathematical normalization and extreme value ratio analysis, providing the system with rapid and accurate response to transient disturbances and forming a fundamental building block in the entire disturbance identification system.

[0065] The specific steps for quantifying the disturbance intensity of the optical signal under the sudden change of the medium refractive index based on the disturbance response value are as follows:

[0066] Normalization is performed on the disturbance residual vector sequence constructed under the continuous time window. By subtracting the minimum value of the sequence from each component and then dividing it by the range value of the sequence, all disturbance residual information is standardized on a unified scale, which facilitates the comparability of disturbance data in different time periods and the mathematical stability of subsequent processing.

[0067] After the construction of the disturbance residual vector sequence is completed, in order to improve the comparability and processing stability of disturbance data in different time periods, the sequence is first normalized. The specific method is to uniformly perform the "minimum value shift + range scaling" operation on the disturbance residual values ​​in each time window, that is, subtract the minimum value in the window from each component, and then divide it by the difference between the maximum and minimum values ​​in the window. This normalization method compresses all disturbance data to the [0,1] interval, effectively avoiding the calculation deviation caused by the difference in numerical scales between different windows, while enhancing the numerical stability of subsequent ratio calculations. This step plays a basic standardization role in the entire disturbance identification process, so that the disturbance response value has a unified dimension and dynamic sensitivity.

[0068] In each sliding time window, the maximum and minimum values ​​in the normalized disturbance residual sequence within the current window are extracted, and the numerical ratio between the two is calculated to characterize the range of fluctuation amplitude of the disturbance signal within the current time window, thereby forming a primary disturbance quantitative index reflecting the severity of the optical signal transmission state;

[0069] After normalization, the maximum and minimum values ​​of the normalized perturbation residual are extracted from each sliding time window, and the ratio between the two is calculated to form a primary amplitude indicator representing the strength of the perturbation in that window. A larger ratio indicates greater instability in the optical signal propagation state during that time period, potentially due to drastic fluctuations caused by factors such as gas accumulation and refractive index mutations. Conversely, a smaller ratio indicates a relatively stable optical path. This step, by extracting the range of numerical fluctuations, provides core data support for quantifying perturbation intensity, and is characterized by its simple structure, efficient computation, and high sensitivity.

[0070] The disturbance signal fluctuation amplitude ratio is used as the disturbance response value corresponding to the current time window to specifically characterize the degree of instability of the propagation path of the optical signal caused by the sudden change of the medium refractive index in the current time period, providing a dynamic judgment basis for subsequent burst interference identification and sampling window adjustment mechanism.

[0071] The maximum-to-minimum disturbance ratio calculated within each time window is used as the disturbance response value corresponding to that window, serving as a direct quantitative expression of the optical path stability during that time period. This disturbance response value can not only independently determine the degree of optical signal anomaly within a single time segment, but also serve as an input parameter for subsequent time-weighted processing, interference index calculation, and dynamic sampling control. It serves as a "disturbance degree quantifier" within the entire method system, a key intermediary in the transition from feature changes to disturbance identification, directly affecting the ultimate monitoring accuracy and response speed.

[0072] The dynamic trend of kurtosis changes in the optical signal waveform is used to construct a disturbance time-sensitive weight factor. This weight factor uses an exponentially weighted moving average algorithm to weight the disturbance response value, emphasizing the instantaneous abnormal fluctuation area in the optical signal and suppressing the proportion of small noise in the cumulative weight.

[0073] This step achieves highly sensitive identification of unusual disturbances in the optical signal by constructing a time-sensitive weighting factor that reflects the dynamic characteristics of the optical signal waveform and applying a differentiated weighting to the disturbance response values. This process effectively suppresses background noise from interfering with the overall disturbance determination. During probiotic detection in fermented vegetable feed, the echo signals collected by the fiber optic sensor are often affected by various factors within the fermentation medium, such as transient accumulation of gas release, dramatic pH fluctuations, or sudden changes in medium density. These factors often manifest as sudden changes in the signal waveform morphology, rather than simply abnormal numerical intensity. Therefore, relying solely on the disturbance response value is insufficient to accurately identify sudden disturbances. To capture this "morphological level" of sudden changes, this step first analyzes the kurtosis rate of change of the optical signal waveform over a continuous time window. Kurtosis is a statistical characteristic of signal sharpness; a high kurtosis rate indicates the presence of sudden abnormal structures in the signal. Subsequently, a time-sensitive disturbance weighting factor is constructed based on the kurtosis rate of change and dynamically updated using an exponentially weighted moving average algorithm. This gives greater weight to recent unusual fluctuations and less weight to stable sections in the weighting calculation. This amplifies true abnormal disturbance responses while attenuating non-substantial offsets caused by minor system noise or accidental fluctuations. Ultimately, using this weighted mechanism to process disturbance response values ​​significantly improves the accuracy, real-time performance, and reliability of optical path disturbance identification, providing a precise and efficient input basis for the subsequent construction of the optical interference index and the intelligent sampling and adjustment mechanism. This step serves as a "signal amplifier for abnormality identification" within the entire monitoring system and is a key technical link in enhancing the system's transient perception and anti-interference capabilities.

[0074] The specific steps of constructing the disturbance time-sensitive weight factor based on the dynamic trend of the optical signal kurtosis change and weighting the disturbance response value are as follows:

[0075] Within each sliding time window, the optical signal waveform within the current time period is statistically analyzed to extract the fourth-order kurtosis value of the signal amplitude distribution. The kurtosis change rate is calculated through continuous windows to characterize the dynamic evolution trend of instantaneous spikes or distortion areas in the optical signal, serving as a preliminary characteristic indicator for judging the sensitivity to abnormal disturbances.

[0076] In each sliding time window, the waveform distribution of the optical signal collected during this time period is first analyzed to extract its fourth-order statistical feature, the kurtosis value. Kurtosis is an indicator that measures the sharpness of the signal waveform. A larger value indicates that the signal may contain sudden spikes or distortions, and is highly sensitive to disturbances. Subsequently, the rate of change of kurtosis over time is obtained by differential calculation between consecutive time windows, which is used to quantify the waveform stability of the optical signal in the time dimension. This rate of change serves as an important precursor to determining the probability of abnormal disturbances. It can detect in advance those periods that have not yet caused significant changes in signal strength but have already shown abnormal waveform trends, providing a dynamic perception basis for the subsequent construction of a sensitive weighting mechanism.

[0077] A time-sensitive weighting factor is constructed based on the rate of change of kurtosis, and is updated using an exponentially weighted moving average algorithm. By setting a time decay coefficient, sudden changes in kurtosis within the recent window are given a higher weight, while stable fluctuations within the long-term window are given a lower weight, dynamically focusing on the latest signs of abnormal disturbances.

[0078] The Exponential Weighted Moving Average (EWMA) algorithm is an algorithm that dynamically smooths time series data. Its core idea is to assign higher weights to recent data and gradually decaying lower weights to older data, making the overall average more sensitive to current trends.

[0079] The purpose of introducing an exponentially weighted moving average (EWMA) algorithm into the time-sensitive weighting factor constructed based on the kurtosis of the optical signal is to update the value of the weighting factor in real time and make it more responsive to abnormal fluctuations that may occur in the "latest moment." Compared with the traditional simple moving average (SMA), the EWMA does not average the fluctuation trends of all past windows. Instead, it uses a decay coefficient to maximize the impact of kurtosis changes in the latest time window on the overall factor, thereby achieving stronger responsiveness to sudden disturbances. Furthermore, interference information in older windows automatically decays over time, preventing "historical interference" from influencing current judgments. Therefore, the EWMA can highlight the disturbance characteristics of the current window while retaining a certain historical background trend, improving the system's sensitivity, robustness, and timeliness. This is particularly suitable for complex scenarios such as probiotic fermentation processes, where optical signals fluctuate frequently and abnormal changes are unpredictable. In short, this algorithm, which updates information with a "recent priority" approach, is a key tool for achieving highly timely identification and real-time control feedback.

[0080] After obtaining the kurtosis change rate, a time-sensitive weighting factor is constructed to reflect the importance of each time window to the final disturbance judgment. To ensure that the weighting factor is more sensitive to recent window changes while suppressing interference from old data, the factor is dynamically updated using an exponentially weighted moving average algorithm. By reasonably setting the attenuation coefficient, the sudden change in kurtosis that appears in the latest window is given a higher weight, while the changes that have stabilized in the historical window are given a lower weight, thereby achieving information focusing and signal priority control. This weighting factor enhances the disturbance response system's ability to respond to "sudden" and "discontinuous" events, making it not only dependent on the amplitude, but also able to judge the changes in the waveform structure itself.

[0081] The time-sensitive weight factor is multiplied by the disturbance response value in the corresponding time window to obtain a weighted disturbance value sequence. This sequence is more sensitive to sharp disturbance signals and can effectively amplify the recognition results of instantaneous optical path anomaly signals. At the same time, it weakens the impact of small disturbances caused by background noise on the system decision results, significantly improving the entire system's real-time recognition capability for sudden disturbance states.

[0082] The constructed time-sensitive weight factor is multiplied point by point with the disturbance response value in the current window to generate a weighted disturbance value sequence. This sequence retains the numerical characteristics of the original disturbance intensity while introducing the ability to dynamically control changes in abnormal waveform structure, giving transient mutation events a greater weight in the overall index while effectively weakening small random disturbances. This weighting method significantly improves the system's recognition accuracy and anti-interference ability in complex signal backgrounds, providing a more reliable data foundation for subsequent optical interference index calculations, real-time alarm mechanisms, and sampling strategy adjustments, ensuring the stability and intelligence level of the probiotic detection system during the fermentation process.

[0083] The disturbance response value is multiplied point by point by the disturbance time-sensitive weight factor to generate a weighted disturbance value sequence; within the continuous sliding window range, the weighted disturbance value sequence is cumulatively summed to generate an optical interference index, which represents the stability of the optical signal at the current sampling time point;

[0084] This step integrates local disturbance intensity with the ability to identify sudden changes. Through multiplication and accumulation, it generates an optical interference index (OI) that dynamically and continuously reflects the stability of optical signal propagation, providing a key basis for subsequent intelligent judgment of whether the system has entered an abnormal state. During the probiotic detection process in fermented tail vegetable feed, the uncertainty and dynamic nature of the fermentation environment make the signals collected by the fiber optic sensor susceptible to sudden changes in the medium's refractive index, resulting in unstable propagation. While a single disturbance response value can characterize the degree of local disturbance within the current window, it lacks sensitivity to sudden changes in the signal and the ability to identify continuous trends. Therefore, this step first performs a point-by-point multiplication of the disturbance response value within each time window with its corresponding time-sensitive disturbance weighting factor to generate a weighted disturbance value sequence. This sequence retains the original numerical information of the disturbance intensity while enhancing the response to abnormal fluctuations through a weighting mechanism, thereby improving the system's ability to distinguish between real disturbances and noise disturbances. The system then accumulates and sums this weighted disturbance value sequence over a continuous sliding time range to generate an OI that can be dynamically updated over time. This index can be considered a score of the optical environment stability at the current point in time, reflecting whether the current optical path propagation state deviates from normal physiological fermentation conditions. It not only serves as a criterion for real-time monitoring but also triggers the system to automatically adjust the sampling window, perform data filtering, or perform alarm mechanisms. By introducing this index, the entire detection system is able to integrate multiple local disturbance signals into a unified judgment, achieving a transition from point-by-point data to global decision-making, significantly enhancing the intelligence and adaptability of probiotic fermentation status monitoring.

[0085] The specific steps for generating an optical interference index reflecting the stability state of an optical signal are as follows:

[0086] In each sliding time window, the disturbance response value corresponding to the current time window is multiplied point by point with the disturbance time-sensitive weight factor updated by the exponentially weighted moving average algorithm to generate the weighted disturbance value under the current time window, which is used to simultaneously reflect the information of the two dimensions of disturbance intensity and sudden disturbance sensitivity;

[0087] In each sliding time window, the system first obtains two key input parameters: one is the disturbance response value calculated within the time window, which is used to represent the disturbance intensity of the optical signal due to the sudden change in the refractive index during this time period; the other is the disturbance time-sensitive weight factor generated by the exponentially weighted moving average algorithm, which is used to highlight the abnormal fluctuation window with sudden burst characteristics. By performing a point-by-point product operation on these two parameters, the weighted disturbance value under the current time window is generated. This product not only retains the intensity characteristics of the original disturbance, but also introduces a differentiated evaluation mechanism for abnormal sensitivity, realizing the amplified response to sudden disturbances and the weakening of the stable window, thereby improving the recognition accuracy and response agility of the entire system.

[0088] The weighted disturbance values ​​generated in consecutive time periods are arranged in sequence according to the sliding time window order to construct a complete weighted disturbance value sequence. This sequence dynamically records the comprehensive disturbance performance of the optical signal in each time window and provides the input basis for the subsequent overall trend quantification.

[0089] After sequentially generating weighted disturbance values ​​in each sliding time window, the system arranges and combines these values ​​in chronological order to construct a complete sequence of weighted disturbance values. This sequence is a dynamic data set with a timeline structure that can reflect in real time the evolution of the comprehensive disturbance over multiple consecutive time periods during the optical signal propagation process. Its value lies in that it not only records the severity of the disturbance at each moment, but also preserves the persistence and cumulative trends of the disturbance. This provides the basic data structure for subsequent operations such as trend identification, segmentation, and stability assessment, and serves as a bridge from local disturbance to global state judgment.

[0090] The weighted disturbance value sequence is cumulatively summed within the current sliding analysis range to generate the optical interference index at the current sampling time point. This index serves as an important indicator reflecting the stability of the optical signal propagation state under the current fermentation environment, and provides a decision-making basis for the system to perform real-time interference identification and sampling window adjustment.

[0091] Finally, the system performs a cumulative summation operation on the weighted disturbance value sequence constructed above, integrating the disturbance information at each time point within the current sliding analysis window to form a single value - the optical interference index. This index value is a one-time comprehensive judgment result on the stability of the entire optical signal propagation state at the current sampling time point. The larger the value, the more severe the disturbance to the system and the higher the instability of the optical path; the smaller the value, the more stable the current optical signal propagation environment. Through this index, the system can quickly determine whether there is an optical anomaly at present, and accordingly drive the real-time sampling adjustment mechanism, control feedback mechanism or alarm module, thereby improving the autonomous adaptability and real-time control capabilities of the overall detection system in a dynamic environment.

[0092] The optical interference index is compared with a preset optical interference reference threshold. When the optical interference index is greater than the reference threshold, a dynamic sampling window adjustment mechanism is activated to change the length and start and end time parameters of the original sampling time interval. The sampling window is offset to avoid abnormal disturbance areas in the optical signal, reduce detection errors caused by sudden changes in the refractive index, and improve the stability and accuracy of optical signal data acquisition.

[0093] When the optical interference index is greater than the reference threshold, the dynamic sampling window adjustment mechanism is activated to change the length and start and end time parameters of the original sampling time interval, and perform an offset operation on the sampling window to make the sampling segment avoid the abnormal disturbance area in the optical signal. The specific steps are as follows:

[0094] After obtaining the optical interference index at the current time point, in order to evaluate its deviation from the stable state, the optical interference offset rate is calculated using the optical interference index reference threshold as a comparison benchmark. The calculation expression is as follows:

[0095]

[0096] Where, I t It is the optical interference index, which indicates the optical interference index value at the corresponding time point of the current sliding time window, reflecting the overall propagation stability of the optical signal at the current moment. The higher the value, the more serious the refractive index disturbance in the current environment and the more unstable the optical fiber signal. It is the key dynamic input for judging whether the current optical path state deviates from the stable value of the fermentation condition and directly participates in the calculation of the offset rate. ref is the optical interference index reference threshold, δ t is the optical interference deviation rate, which represents the dimensionless deviation index of the current optical signal disturbance degree based on the reference threshold. t >1 indicates a significant deviation from the stable operating condition, δ t ≈1 means it is close to the boundary, δ t <1 indicates weak interference;

[0097] By optical interference the offset rate δ t It provides a continuous and differentiable quantitative indicator of disturbance intensity, which provides a basic response basis for dynamically adjusting the sampling window and avoids the sampling rigidity and response lag caused by binary judgment logic.

[0098] Based on the obtained optical interference offset rate δ t , enter the adaptive adjustment of the sampling window, expand the original sampling window length to adapt to the change of interference intensity, and calculate the current time amplitude to be offset according to the offset rate to ensure that the sampling segment avoids the abnormal disturbance area in the signal. The calculation expression is as follows:

[0099]

[0100] Where ΔT is the original sampling window length, which represents the length of the standard sliding sampling time interval set in the undisturbed state. It is the basic time unit for continuous sampling during the fermentation process and determines how much length of data is intercepted from the light signal sequence for feature extraction and analysis each time. α is the window expansion sensitivity coefficient, which controls δ t The intensity of the influence on the change of the sampling window length, the value range is 0-1, tanh(·) is the hyperbolic tangent function, ΔT ′ is the length of the dynamically adjusted sampling window, when the current optical disturbance offset rate is δ t Afterwards, ΔT is corrected in real time according to its strength, and the actual sampling time length is obtained, so that the system can automatically lengthen the sampling window according to the degree of disturbance, thereby spanning the entire interference area and preventing a single window from being seriously polluted by the disturbance. β is the offset sensitivity coefficient, which is used to adjust The amplification effect of the offset distance controls the influence of the disturbance intensity on the offset degree of the sampling window. The value range is 0.5-2.5, ∈ t The sampling window offset is the distance by which the current sampling window is offset on the time axis, so that the sampling window of the system is away from the center of abnormal disturbance in the optical signal, thereby avoiding the interference peak and obtaining more representative and stable detection data.

[0101] The main function of this step is to convert the perception of the degree of optical disturbance into a dynamic adjustment mechanism for the window size and position, thereby enhancing the system's ability to avoid sudden interference, improving the stability and accuracy of the sampling data, and ensuring the high reliability of the entire probiotic detection method in complex fermentation environments.

[0102] The above-mentioned IoT-based method for detecting probiotics in fermented tail vegetable feed can significantly improve the anti-interference ability and data credibility of the fiber optic sensing system in a complex biological fermentation environment. This method achieves high-precision identification of abnormal disturbances in the optical signal by constructing a multi-dimensional transient feature vector of the optical signal, introducing a dynamic monitoring mechanism for the disturbance residual, and combining a time-sensitive weighting factor with the kurtosis change. At the same time, through the continuous quantification of the optical interference index and the adaptive adjustment of the dynamic sampling window, the "optical interference zone" caused by the sudden change in the refractive index of the medium is effectively avoided, thus avoiding sensor misjudgment and control distortion, and fundamentally solving the "false signal normal" problem that is prone to occur in traditional fiber optic sensors in fermentation scenarios. This method not only enhances the system's real-time perception of the active state of probiotics, but also ensures the stability and accuracy of feed quality monitoring, providing more intelligent and reliable detection support for the resource utilization process of agricultural waste.

[0103] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0104] The above description is merely illustrative of certain exemplary embodiments of the present invention. It goes without saying that those skilled in the art will be able to modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims.

[0105] It should be noted that, in this document, if there are relational terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprises", "comprising" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device that includes the element.

[0106] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

Claims

1. A method for detecting probiotics in fermented vegetable feed based on the Internet of Things, characterized in that: The following steps are involved: Acquire the complete echo signal time series data, perform high-frequency sampling on the echo signal time series with a sliding time window of fixed length, and construct a three-dimensional optical signal transient feature vector; The transient feature vector of the three-dimensional optical signal generated in each time window is calculated dimension by dimension with the feature vector of the standard optical signal established under a stable fermentation environment to generate a disturbance residual vector; the disturbance residual vectors are arranged in chronological order to form a disturbance residual vector sequence; Normalize the disturbance residual vector sequence, calculate the ratio between the maximum disturbance residual and the minimum disturbance residual in each time window, and generate the disturbance response value; Based on the dynamic trend of kurtosis change in the optical signal waveform, a disturbance time-sensitive weight factor is constructed; Performing a point-by-point product operation on the disturbance response value and the disturbance time-sensitive weight factor to generate a weighted disturbance value sequence; Perform cumulative summation on the weighted disturbance value sequence within the continuous sliding window to generate an optical interference index; comparing the optical interference index with a preset optical interference reference threshold; When the optical interference index is greater than the reference threshold, the dynamic sampling window adjustment mechanism is activated to change the length and start and end time parameters of the original sampling time interval, and perform an offset operation on the sampling window to avoid the abnormal disturbance area in the optical signal, thereby reducing the detection error caused by the sudden change of the refractive index. The specific steps of constructing the disturbance time-sensitive weight factor based on the dynamic trend of the optical signal kurtosis change and weighting the disturbance response value are as follows: In each sliding time window, the optical signal waveform in the current time period is statistically analyzed, the fourth-order kurtosis value of the signal amplitude distribution is extracted, and the kurtosis change rate is calculated through continuous windows; A time-sensitive weight factor is constructed based on the kurtosis change rate, and the exponentially weighted moving average algorithm is used to update the factor. By setting the time decay coefficient, the factor is dynamically focused on the latest signs of abnormal disturbances. The time-sensitive weight factor is multiplied by the disturbance response value in the corresponding time window to obtain a weighted disturbance value sequence.

2. The method for detecting probiotics in fermented vegetable feed based on the Internet of Things according to claim 1, wherein The specific steps for obtaining the time series of the optical fiber sensor echo signal and performing high-frequency sampling to construct the three-dimensional optical signal transient feature vector are as follows: For the time series of echo signals output by the fiber optic sensor during the target monitoring period, the original signal is segmented using a sliding time window of a set length. Signal amplitude normalization and noise filtering operations are performed on the signal curve within each time window. The local standard deviation of the signal amplitude in the current time period is calculated to extract the signal intensity fluctuation characteristics. Based on the normalized optical signal waveform within each time window, a time correlation analysis method is used to accurately match and locate the main peak time difference between the incident pulse signal and the echo signal. The time delay value of the signal from incident to return is calculated as the reflection delay characteristic in the current window. After completing the reflection delay analysis, the fast Fourier transform algorithm is used to perform frequency domain transformation on the echo signal within each time window. The rate of change of the phase difference between adjacent main frequency components is calculated to quantify the phase offset amplitude generated during the propagation of different frequency components, forming a complete optical signal phase offset characteristic. In turn, a three-dimensional optical signal transient feature vector is constructed that reflects the instantaneous transmission state of the optical signal.

3. The method for detecting probiotics in fermented vegetable feed based on the Internet of Things according to claim 1, wherein The specific steps of constructing the disturbance residual vector and extracting the change trend are as follows: Within each sliding time window, each component of the three-dimensional optical signal transient feature vector constructed in the current time window is subtracted from the corresponding component of the standard three-dimensional optical signal feature vector pre-calibrated in a stable fermentation environment, forming a disturbance residual vector that reflects the degree of deviation between the optical signal propagation state in the current time period and the standard state. According to the continuity of the sliding time window, the disturbance residual vectors obtained in each time period are arranged in order according to the time sequence to construct the disturbance residual vector sequence; Smoothing is performed on the disturbance residual vector sequence to eliminate local abnormal mutations caused by instantaneous environmental fluctuations while retaining the overall change trend.

4. The method for detecting probiotics in fermented vegetable feed based on Internet of Things according to claim 1, wherein The specific steps for quantifying the disturbance intensity of the optical signal under the sudden change of the medium refractive index based on the disturbance response value are as follows: Normalization is performed on the disturbance residual vector sequence constructed under the continuous time window. By subtracting the minimum value of the sequence from each component and then dividing it by the extreme value of the sequence, all disturbance residual information is standardized in a unified scale. In each sliding time window, the maximum and minimum values ​​in the normalized disturbance residual sequence within the current window are extracted, and the numerical ratio between the two is calculated to characterize the range of fluctuation amplitude of the disturbance signal within the current time window, thereby forming a primary disturbance quantitative index reflecting the severity of the optical signal transmission state; The disturbance signal fluctuation amplitude ratio is used as the disturbance response value corresponding to the current time window to specifically characterize the degree of instability of the propagation path of the optical signal caused by the sudden change of the medium refractive index in the current time period.

5. The method for detecting probiotics in fermented vegetable feed based on the Internet of Things according to claim 1, wherein The specific steps for generating an optical interference index reflecting the stability state of an optical signal are as follows: In each sliding time window, the disturbance response value corresponding to the current time window is multiplied point by point by the disturbance time-sensitive weight factor updated by the exponentially weighted moving average algorithm to generate the weighted disturbance value under the current time window; Arrange the weighted disturbance values ​​generated in consecutive time periods in the order of the sliding time window to construct a complete weighted disturbance value sequence; The weighted disturbance value sequence is cumulatively summed within the current sliding analysis range to generate the optical interference index at the current sampling time point.

6. The method for detecting probiotics in fermented vegetable feed based on the Internet of Things according to claim 1, wherein When the optical interference index is greater than the reference threshold, the dynamic sampling window adjustment mechanism is activated to change the length and start and end time parameters of the original sampling time interval, and perform an offset operation on the sampling window to make the sampling segment avoid the abnormal disturbance area in the optical signal. The specific steps are as follows: After obtaining the optical interference index at the current time point, in order to evaluate its deviation from the stable state, the optical interference offset rate is calculated using the optical interference index reference threshold as a comparison benchmark. The calculation expression is as follows: Where, I t is the optical interference index, I ref is the optical interference index reference threshold, δ t is the optical interference offset rate; Based on the obtained optical interference offset rate δ t , enter the adaptive adjustment of the sampling window, expand the original sampling window length, and calculate the current time amplitude to be offset based on the offset rate to ensure that the sampling segment avoids the abnormal disturbance area in the signal. The calculation expression is as follows: Where ΔT is the original sampling window length, α is the window expansion sensitivity coefficient, tanh(·) is the hyperbolic tangent function, and ΔT ′ is the dynamically adjusted sampling window length, β is the offset sensitivity coefficient, ∈ t is the sampling window offset.

Images