A frequency offset correction method in HPLC+HRF dual-mode communication

Abstract

The present application relates to the technical field of ecosystem service function evaluation, and relates to a frequency offset correction method in HPLC+HRF dual-mode communication.The method comprises the following steps: S1, collecting phase sequence data from an HPLC communication link, collecting frequency offset estimation data and corresponding confidence data from an HRF communication link, and adaptively adjusting a preset detection threshold of the phase sequence data according to the confidence data; S2, performing pulse abnormal sample detection and elimination on the phase sequence data according to the adjusted preset detection threshold; the present application dynamically adjusts the detection threshold through the confidence data, so that the abnormal detection process can be matched with the current interference environment in real time, thereby maintaining the integrity of the phase sequence data, solving the problem of insufficient adaptability of the fixed detection threshold in the dynamic change scene of electromagnetic interference, providing a more reliable phase data basis for subsequent interpolation compensation and frequency offset estimation, and further improving the frequency offset correction precision of the dual-mode communication system.

Description

Technical Field

[0001] This invention relates to the field of communication system technology, and more specifically, to a frequency offset correction method in HPLC+HRF dual-mode communication. Background Technology

[0002] With the rapid development of modern communication technology, multimode communication systems have gradually become a research hotspot due to their diversity and robustness. Among them, the HPLC (high performance carrier link) and HRF (high precision radio frequency link) dual-mode communication system combines the advantages of both links and can achieve highly reliable and high-precision data transmission in complex environments. Especially in environments with strong electromagnetic pulse interference, such as underground substations, the dual-mode communication system can effectively improve the anti-interference capability and system stability of communication, and ensure the real-time monitoring and safe operation of important power systems.

[0003] However, in the process of implementing the technical solution of the invention in the embodiments of this application, it was found that the above-mentioned technology has at least the following technical problems:

[0004] In existing technologies, phase sequence data is first acquired from the HPLC link, and abnormal samples are identified and removed using a fixed detection threshold. Then, HRF frequency offset estimation is used to interpolate and compensate for the missing samples. Finally, the frequency offset estimate is calculated by combining the two sets of data and filtered. This fully utilizes the complementarity of the dual-link data and improves the accuracy and stability of frequency offset estimation. However, the fixed detection threshold cannot adapt to the instantaneous changes in the electromagnetic pulse intensity of underground substations, resulting in over- or under-removal of abnormal samples, which affects the accuracy of subsequent interpolation and compensation. Moreover, interpolation and compensation often ignore the phase gradient constraint of the neighborhood of abnormal samples, making it difficult for the interpolation curve to accurately reflect the real signal characteristics, reducing the reliability of frequency offset estimation, and making it difficult to fully utilize the synergistic advantages of dual-mode communication data. Therefore, a frequency offset correction method in HPLC+HRF dual-mode communication is proposed. Summary of the Invention

[0005] The purpose of this invention is to provide a frequency offset correction method in HPLC+HRF dual-mode communication to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, a frequency offset correction method for HPLC+HRF dual-mode communication is provided, comprising the following steps:

[0007] S1. Acquire phase sequence data from the HPLC communication link, acquire frequency offset estimation data and corresponding confidence data from the HRF communication link, and adaptively adjust the preset detection threshold of the phase sequence data based on the confidence data.

[0008] S2. Based on the adjusted preset detection threshold, pulse abnormal sample points are detected and removed from the phase sequence data, and the phase sequence data after removing abnormal sample points and the gradient constraint data of the neighborhood of the abnormal sample points are output.

[0009] S3. Based on the frequency offset estimation data as a guide, and combined with the gradient constraint data, perform weighted interpolation compensation on the phase sequence data after removing abnormal samples to generate an interpolation compensation sequence.

[0010] S4. Using the interpolation compensation sequence and the frequency offset estimation data as input, calculate the frequency offset estimate and the corresponding frequency offset estimation error of HPLC through the time-varying phase difference filtering method.

[0011] S5. Using the frequency offset estimation data as the prior distribution, the frequency offset estimation value and the corresponding frequency offset estimation error as the observation likelihood, combined with physical constraints, Bayesian fusion calculation is performed to output the fused frequency offset estimation value and fusion confidence.

[0012] S6. Perform feedback control filtering on the fused frequency offset estimate based on the fused confidence level, and output the frequency offset correction value.

[0013] As a further improvement to this technical solution, the acquisition of phase sequence data from the HPLC communication link and the acquisition of frequency offset estimation data and corresponding confidence level data from the HRF communication link specifically include:

[0014] Phase sequence data were acquired from the HPLC communication link using a high-speed phase modem;

[0015] Frequency offset estimation data and corresponding confidence data are collected from the HRF communication link using a built-in frequency tracker.

[0016] The phase sequence data, frequency offset estimation data, and corresponding confidence data are time-aligned according to the acquisition timestamp.

[0017] As a further improvement to this technical solution, the step of adaptively adjusting the preset detection threshold of the phase sequence data based on the confidence level data specifically includes:

[0018] A threshold adjustment function is predefined, and the confidence data is mapped to a threshold adjustment coefficient through the threshold adjustment function;

[0019] The adjusted preset detection threshold is obtained by multiplying the threshold adjustment coefficient by the preset detection threshold.

[0020] As a further improvement to this technical solution, the step of detecting and removing pulse anomaly samples from the phase sequence data according to the adjusted preset detection threshold, and outputting the phase sequence data after removing anomaly samples and the gradient constraint data of the neighborhood of the anomaly samples, specifically includes:

[0021] Calculate the absolute value of the phase difference between adjacent sampling points for the phase sequence data, and determine whether the absolute value of the phase difference is greater than the adjusted preset detection threshold. If it is, determine that the sampling point corresponding to the absolute value of the phase difference is an abnormal sampling point.

[0022] Remove outlier samples to obtain the phase sequence data after removing outlier samples;

[0023] The gradient difference gradient between the left and right adjacent sampling points of the abnormal sampling point is calculated to obtain the gradient constraint data of the neighborhood of the abnormal sampling point.

[0024] As a further improvement to this technical solution, the step of using frequency offset estimation data as a guiding basis and combining the gradient constraint data to perform weighted interpolation compensation on the phase sequence data after removing outlier samples, and generating an interpolation compensation sequence specifically includes:

[0025] For missing points in the phase sequence data after removing outlier samples, the interpolation weights of each missing point are obtained through nonlinear calculation based on the frequency offset estimation data.

[0026] Based on the interpolation weight of the missing point and the phase values ​​of the sampling points to the left and right of the missing point in the phase sequence data after removing outlier points, the phase value of the missing point is obtained by linear calculation.

[0027] Calculate the gradient before and after the phase value of the missing point;

[0028] If the gradient calculation results before and after are greater than the gradient constraint data of the neighborhood of the missing point, then perform gradient constraint correction processing on the phase value of the missing point, and generate an interpolation compensation sequence based on the phase value of the missing point after correction.

[0029] As a further improvement to this technical solution, the step of calculating the frequency offset estimate and corresponding frequency offset estimation error of HPLC using the interpolation compensation sequence and the frequency offset estimation data as input through a time-varying phase differential filtering method specifically includes:

[0030] For the interpolation compensation sequence, the time-varying phase difference value of its adjacent missing points is calculated; the time-varying phase difference value is obtained by calculating the complex difference value of the phase difference between adjacent missing points;

[0031] The filter coefficients of the time-varying phase differential filter are dynamically adjusted based on the confidence level data.

[0032] The time-varying phase difference value is subjected to a weighted moving average filter, and the filter weight is determined by the dynamically adjusted filter coefficients mentioned above, and the frequency deviation estimate of HPLC is output.

[0033] The frequency deviation estimate is calculated by comparing the frequency deviation estimate with the frequency deviation estimate data from the HPLC to obtain the filtering residual. The frequency deviation estimate error is then calculated based on the sum of squares of the filtering residuals. The frequency deviation estimate error is obtained by statistically analyzing the variance of the squared deviations between the frequency deviation estimate data before and after filtering and the frequency deviation estimate data.

[0034] As a further improvement to this technical solution, the step of using the frequency offset estimation data as the prior distribution, using the frequency offset estimation value and the corresponding frequency offset estimation error as the observation likelihood, and combining physical constraints to perform Bayesian fusion calculation, and outputting the fused frequency offset estimate and fusion confidence specifically includes:

[0035] The frequency offset estimation data is used as the prior distribution for Bayesian fusion; the prior distribution is formed by modeling the probability density function of the frequency offset estimation data.

[0036] The frequency offset estimate and the corresponding frequency offset estimate error are used as the observation likelihood; the observation likelihood adopts a Gaussian distribution model.

[0037] Based on the prior distribution and observed likelihood, and combined with the physical constraints of the communication system, including the frequency offset rate of change limit and the frequency offset range boundary constraint, the fused frequency offset posterior probability distribution is calculated using the Bayesian recursive formula.

[0038] Calculate the mean of the fused frequency offset posterior probability distribution and use it as the fused frequency offset estimate. Calculate the variance of the fused frequency offset posterior probability distribution and use its reciprocal as the fused confidence level.

[0039] As a further improvement to this technical solution, the step of performing feedback control filtering on the fused frequency offset estimate based on the fused confidence level, and outputting a frequency offset correction value specifically includes:

[0040] Based on the fusion confidence level, the adaptive gain coefficient of the feedback control filter is calculated using a nonlinear mapping function;

[0041] Based on the adaptive gain coefficient, a weighted moving average filtering operation is performed on the fused frequency offset estimate. Specifically, the current fused frequency offset estimate and the filtered output value at the previous moment are weighted and summed according to the gain coefficient and its complement to generate the filtered frequency offset correction value.

[0042] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0043] 1. In this frequency offset correction method in HPLC+HRF dual-mode communication, the detection threshold is dynamically adjusted by confidence data, so that the abnormal detection process can match the current interference environment in real time, thereby maintaining the integrity of the phase sequence data. This solves the problem of insufficient adaptability of fixed detection threshold in the scenario of dynamic change of electromagnetic interference, and provides a more reliable phase data basis for subsequent interpolation compensation and frequency offset estimation, thereby improving the frequency offset correction accuracy of the dual-mode communication system.

[0044] 2. In this frequency offset correction method for HPLC+HRF dual-mode communication, the phase distortion in the interpolation compensation process is effectively suppressed through the dual mechanism of dynamic weight adjustment and gradient constraint correction. This solves the problem of frequency offset estimation distortion caused by ignoring phase gradient constraints in the existing technology. It can generate interpolation sequences that are more in line with the real signal characteristics in the strong interference environment of underground substations, thereby improving the accuracy and reliability of frequency offset correction in dual-mode communication systems. Attached Figure Description

[0045] Figure 1 This is a flowchart illustrating the overall process of the present invention. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] Please see Figure 1 As shown, the purpose of this embodiment is to achieve highly reliable data transmission in strong electromagnetic interference environments such as underground substations by combining the advantages of HPLC and HRF links in the existing technology. However, the fixed detection threshold cannot adapt to the instantaneous changes in electromagnetic pulse intensity, resulting in inaccurate rejection of abnormal samples. The interpolation compensation process ignores the neighborhood phase gradient constraint, and the frequency offset estimation accuracy is limited. The advantages of dual-mode data collaboration are not fully utilized.

[0048] To address the aforementioned issues, this paper proposes a method for dynamically adjusting the detection threshold based on confidence level to address the insufficient adaptability of fixed thresholds. To address the problem of insufficient interpolation compensation accuracy, a weighted interpolation method under gradient constraints is proposed. Furthermore, to address the low efficiency of dual-mode data fusion, a fusion mechanism based on a Bayesian framework is designed. Through the synergistic effect of confidence feedback, gradient constraint iterative compensation, time-varying filtering, and probabilistic fusion, a closed-loop correction system is formed. Therefore, a frequency offset correction method for HPLC+HRF dual-mode communication is provided, comprising the following steps:

[0049] S1. Acquire phase sequence data from the HPLC communication link, acquire frequency offset estimation data and corresponding confidence level data from the HRF communication link, and adaptively adjust the preset detection threshold of the phase sequence data based on the confidence level data.

[0050] S2. Based on the adjusted preset detection threshold, pulse abnormal sample points are detected and removed from the phase sequence data, and the phase sequence data after removing abnormal sample points and the gradient constraint data of the neighborhood of the abnormal sample points are output.

[0051] S3. Based on the frequency offset estimation data as a guide, and combined with the gradient constraint data, perform weighted interpolation compensation on the phase sequence data after removing abnormal samples to generate an interpolation compensation sequence.

[0052] S4. Using the interpolation compensation sequence and frequency offset estimation data as input, calculate the frequency offset estimate and corresponding frequency offset estimation error of HPLC through the time-varying phase difference filtering method.

[0053] S5. Using the frequency offset estimation data as the prior distribution, the frequency offset estimation value and the corresponding frequency offset estimation error as the observation likelihood, combined with physical constraints, Bayesian fusion calculation is performed to output the fused frequency offset estimation value and fusion confidence.

[0054] S6. Perform feedback control filtering on the fused frequency offset estimate based on the fused confidence level, and output the frequency offset correction value.

[0055] Among them, phase sequence data refers to the phase change sequence obtained by demodulation through the HPLC link, which can be acquired using a high-speed phase modem;

[0056] Confidence data refers to the reliability assessment index of HRF link frequency offset estimation, which can be generated by the error statistics module of the frequency tracker;

[0057] Gradient constraint data refers to the rate constraint of phase change within the neighborhood of an anomaly point, which is obtained by calculating the phase difference between the two sides of the anomaly point;

[0058] Weighted interpolation compensation refers to assigning interpolation weights based on frequency offset estimation data, combining linear interpolation with gradient constraint iterative correction to generate a compensation sequence, specifically achieved through nonlinear weight calculation and gradient difference feedback adjustment.

[0059] Bayesian fusion computation refers to using HRF frequency offset data as the prior distribution and HPLC frequency offset estimation as the observation likelihood, combining frequency variation rate constraints to perform probability fusion, and calculating the optimal estimate through the posterior probability distribution.

[0060] Feedback control filtering refers to dynamically adjusting the filter gain coefficient based on the fusion confidence level and using a weighted moving average method to suppress estimation fluctuations.

[0061] Specifically, a dual-mode collaborative foundation is established by time-aligned HPLC phase sequence data and HRF frequency offset estimation data. Confidence data is converted into threshold coefficients through a threshold adjustment function to dynamically optimize anomaly detection sensitivity. Anomaly detection employs a real-time comparison mechanism between the absolute value of adjacent phase differences and dynamic thresholds to ensure accurate removal of interfering pulses. Gradient constraint data is generated by calculating the phase difference on both sides of the anomaly point, providing a boundary for the phase change rate of interpolation compensation. Interpolation weights are nonlinearly allocated based on the time correlation of HRF frequency offset estimation data and iteratively corrected in conjunction with gradient constraints to ensure that the compensated phase conforms to the actual signal characteristics. Time-varying phase differential filtering uses complex difference calculation and dynamic gain adjustment to achieve rapid tracking and noise suppression of frequency offset estimation. The Bayesian fusion process effectively balances the confidence differences between the two-mode data through Gaussian distribution modeling and physical constraint integration. Feedback filtering adaptively adjusts the smoothing intensity according to the fusion confidence, ultimately outputting a stable and reliable frequency offset correction value.

[0062] Through the above technical solutions, this application effectively solves the problem of inaccurate anomaly rejection caused by fixed detection thresholds. It achieves dynamic optimization of detection sensitivity through a confidence feedback mechanism, and the gradient constraint interpolation method ensures the smoothness and rationality of the phase compensation curve, avoiding the secondary error introduced by traditional interpolation. The dual-mode data fusion mechanism fully leverages the complementary advantages of HRF high-frequency accuracy and HPLC phase continuity, achieving high-precision frequency offset estimation under strong electromagnetic interference. The closed-loop correction system, through the combination of confidence feedback and physical constraints, significantly improves communication reliability in harsh environments such as underground substations.

[0063] The acquisition of phase sequence data from the HPLC communication link and frequency offset estimation data and corresponding confidence level data from the HRF communication link specifically includes:

[0064] Phase sequence data were acquired from the HPLC communication link using a high-speed phase modem;

[0065] Frequency offset estimation data and corresponding confidence data are collected from the HRF communication link using a built-in frequency tracker.

[0066] The phase sequence data, frequency offset estimation data, and corresponding confidence data are time-series aligned based on the acquisition timestamp.

[0067] Among them, the high-speed phase modem refers to a demodulation device that can capture high-frequency phase changes in real time, which can be achieved by using multi-channel parallel sampling technology;

[0068] The built-in frequency tracker refers to the dynamic frequency monitoring module integrated into the HRF receiver, which can be implemented using a phase-locked loop combined with an adaptive filtering algorithm.

[0069] Timing alignment refers to synchronizing data across different communication links based on a unified time reference. This can be achieved using interpolation matching or sliding window calibration methods.

[0070] Specifically, in the HPLC communication link, the high-speed phase modem continuously captures the instantaneous phase value of the carrier signal at a fixed sampling interval to form phase sequence data. In the HRF communication link, the built-in frequency tracker generates frequency offset estimation data by analyzing the frequency offset of the radio frequency signal, and outputs confidence data based on the signal quality assessment results. Subsequently, the two types of data are aligned according to the acquisition timestamp to ensure the time synchronization of the dual-mode data in subsequent processing steps.

[0071] Through the above technical solution, this application can effectively solve the problem of frequency offset estimation error accumulation caused by asynchronous data acquisition, improve the accuracy of abnormal sample point detection, and ensure the collaborative processing effect of dual-mode data through time alignment, providing reliable data input for frequency offset correction in strong interference environment of underground substation.

[0072] Based on the confidence level data, the preset detection threshold for adaptively adjusting the phase sequence data specifically includes:

[0073] A predefined threshold adjustment function is used to map confidence data to threshold adjustment coefficients. Its specific expression is as follows:

[0074] ;

[0075] Where K(C) represents the threshold adjustment coefficient, α represents the adjustment factor of the threshold adjustment coefficient, C represents the confidence level data, and β represents the confidence level adjustment factor.

[0076] The adjusted preset detection threshold is obtained by multiplying the threshold adjustment coefficient by the preset detection threshold.

[0077] The threshold adjustment function refers to the mathematical relationship that converts confidence data into threshold adjustment coefficients, which can be implemented using exponential or linear functions.

[0078] The threshold adjustment coefficient refers to the scaling factor generated based on the confidence data, which can be calculated by multiplying the adjustment factor with the confidence data.

[0079] Confidence data refers to a quantitative indicator that reflects the reliability of HRF frequency offset estimation, which can be obtained by statistically analyzing the variance or stability of the frequency offset estimation.

[0080] Specifically, by using a predefined threshold adjustment function, the confidence data collected by the HRF link is mapped to a threshold adjustment coefficient. The adjusted detection threshold is generated by multiplying the original detection threshold by the threshold adjustment coefficient. This process enables the detection threshold to be adaptively adjusted according to the dynamic changes in the intensity of electromagnetic pulse interference, thereby improving the accuracy of abnormal sample detection.

[0081] Based on the adjusted preset detection threshold, pulse anomaly samples are detected and removed from the phase sequence data. The output includes the phase sequence data after removing anomaly samples and the gradient constraint data of the neighborhood of the anomaly samples.

[0082] Calculate the absolute value of the phase difference between adjacent sampling points for the phase sequence data, and determine whether the absolute value of the phase difference is greater than the adjusted preset detection threshold. If it is, the sampling point corresponding to the absolute value of the phase difference is determined to be an abnormal sampling point.

[0083] Remove outlier samples to obtain the phase sequence data after removing outlier samples;

[0084] The gradient difference gradient between the left and right adjacent sampling points of the abnormal sampling point is calculated to obtain the gradient constraint data of the neighborhood of the abnormal sampling point.

[0085] The absolute value of the phase difference refers to the absolute value of the phase change between adjacent sampling points, which can be achieved by differential calculation using the phase sequence data output by the phase demodulator.

[0086] Abnormal sample point detection refers to determining whether a phase change is caused by electromagnetic pulse interference by setting a preset threshold. Specifically, this can be achieved by comparing a dynamically adjusted detection threshold with the absolute value of the phase difference.

[0087] Specifically, in the phase sequence data processing, the absolute value of the phase difference between adjacent sampling points is first calculated and compared with the dynamically adjusted detection threshold. If the value exceeds the threshold, it is determined to be an abnormal sample and removed. After the abnormal sample is removed, the phase gradient is calculated in the neighborhood of the missing point. For example, the phase difference between the previous sampling point and the next sampling point is divided by the time interval to generate gradient constraint data. This gradient constraint data reflects the phase change law of the normal signal in the neighborhood of the abnormal sample and can be used as the boundary condition for subsequent interpolation compensation to ensure that the interpolated phase sequence still conforms to the smooth characteristics of the actual signal in the abrupt change region.

[0088] Through the above technical solution, this application can accurately identify and eliminate abnormal samples in the phase sequence under the electromagnetic pulse interference environment of underground substations. At the same time, by using gradient constraint data, the true trend of signal change is preserved, providing a highly reliable data foundation for subsequent interpolation compensation and frequency offset estimation, thereby improving the anti-interference capability and frequency offset correction accuracy of the dual-mode communication system.

[0089] Based on frequency offset estimation data as a guide, and combined with gradient constraint data, weighted interpolation compensation is performed on the phase sequence data after removing outlier samples to generate the interpolation compensation sequence, specifically including:

[0090] For missing points in the phase sequence data after removing outlier samples, the interpolation weights of each missing point are obtained through nonlinear calculation based on the frequency offset estimation data. The specific calculation formula is as follows:

[0091] ;

[0092] in, Indicates missing point m k The interpolation weights, where γ represents the weight decay coefficient. Indicates missing point m k The timestamp of the previous sampling point, t represents the timestamp of the missing point, f HRF (τ) represents the frequency offset estimation data, τ is the integral variable, and dτ is the derivative of the integral variable;

[0093] Based on the interpolation weights of the missing points and the phase values ​​of the adjacent sampling points to the left and right of the missing points in the phase sequence data after removing outliers, the phase value of the missing point is obtained through linear calculation. The specific calculation formula is as follows:

[0094] ;

[0095] In the formula, Indicates missing point t d The phase value, φ(t) d-1 ) represents the phase value of the sampling point preceding the missing point, φ(t) d+1 w(t) represents the phase value of the sampling point following the missing point. d ) represents the missing point t d Interpolation weights;

[0096] The gradient before and after the phase value of the missing point is calculated using the following formula:

[0097] ;

[0098] ;

[0099] Among them, g left g represents the phase gradient between the missing point and the previous sampling point. right t represents the phase gradient between the missing point and the next sampling point. d t represents the timestamp corresponding to the missing point. d-1 t represents the timestamp corresponding to the sampling point preceding the missing point. d+1 This represents the timestamp corresponding to the next sampling point after the missing point;

[0100] If the gradient calculation results before and after the missing point are greater than the gradient constraint data of the neighborhood of the outlier sample corresponding to the missing point, then gradient constraint correction processing is performed on the phase value of the missing point. An interpolation compensation sequence is generated based on the phase value of the missing point after correction processing, as follows:

[0101] The difference between the maximum value of the gradient before and after and the gradient constraint data of the neighborhood of the outlier sample corresponding to the missing point is calculated.

[0102] Based on the difference calculation results, the interpolation weights of the missing points are dynamically adjusted. The specific calculation formula is as follows:

[0103] ;

[0104] In the formula, w(t) d ) new Indicates the missing point t after adjustment d The interpolation weights are ΔG, which represents the phase gradient difference calculation result, and δ, which represents the weighting factor of the phase gradient difference calculation result.

[0105] Based on the adjusted interpolation weights of the missing points, the phase values ​​of the missing points are recalculated until the gradients before and after the phase values ​​of the missing points are no greater than the gradient constraint data of the neighborhood of the corresponding outlier sample point.

[0106] Among them, the interpolation weight of missing points refers to the weight coefficient that is dynamically adjusted based on the frequency offset estimation data. Specifically, it can be calculated by combining the exponential decay function with the timestamp difference.

[0107] The missing point phase value refers to the compensation value generated by linear interpolation combined with dynamic weights. Specifically, it can be calculated by using a weighted average method combined with the phase values ​​of the previous and subsequent sampling points.

[0108] Gradient calculation refers to quantifying the phase change rate between the interpolation point and the adjacent sampling point, which can be achieved by calculating the phase difference normalized by timestamp difference.

[0109] Gradient constraint correction refers to adjusting the interpolation weights to make the phase gradient of the interpolation point conform to the neighborhood constraint conditions. Specifically, it can be achieved by iterative weight updates combined with gradient threshold comparison.

[0110] Specifically, when generating the interpolation compensation sequence, the interpolation weights of missing points are first calculated based on the temporal correlation of the frequency offset estimation data using a nonlinear function. For example, an exponential decay function is used to map timestamp differences to weight coefficients, giving higher weights to sampling points closer to the missing points. Then, the phase values ​​of adjacent sampling points are linearly interpolated using these weights to generate preliminary compensation values. Further, the phase gradient between the interpolation point and the preceding and following sampling points is calculated and compared with the gradient constraint data pre-calculated in the neighborhood of the outlier sampling points. If the gradient exceeds the constraint range, the interpolation weights are dynamically adjusted and the phase values ​​are recalculated, for example, through difference calculation and iterative updates of the weight factors, until the phase gradient of the interpolation point satisfies the neighborhood constraint conditions. The resulting interpolation compensation sequence retains the trend characteristics of the frequency offset estimation data and meets the local smoothness requirements of the phase sequence.

[0111] Through the above technical solution, this application solves the problem of frequency offset estimation distortion caused by neglecting phase gradient constraints in interpolation compensation in the prior art. It can generate interpolation sequences that are more in line with the real signal characteristics in the strong interference environment of underground substations, thereby improving the accuracy and reliability of frequency offset correction of dual-mode communication systems.

[0112] Using interpolated compensation sequences and frequency offset estimation data as input, the frequency offset estimate and corresponding frequency offset estimation error of HPLC are calculated through a time-varying phase differential filtering method. Specifically, this includes:

[0113] For the interpolated compensation sequence, calculate the time-varying phase difference value of its adjacent missing points; the time-varying phase difference value is obtained by calculating the complex difference value of the phase difference between adjacent missing points;

[0114] The filter coefficients of the time-varying phase differential filter are dynamically adjusted based on the confidence level data.

[0115] Specifically, by mapping confidence data to filter gain parameters, the filter becomes smoother when the confidence is increased and more sensitive when the confidence is decreased, in order to adapt to rapid changes in frequency offset;

[0116] The time-varying phase difference value is subjected to a weighted moving average filter, and the filter weight is determined by the dynamically adjusted filter coefficients mentioned above, and the frequency deviation estimate of the HPLC is output.

[0117] The frequency deviation estimate is calculated by comparing the frequency deviation estimate with the frequency deviation estimate data to obtain the filtering residual. The frequency deviation estimate error is then calculated based on the sum of squares of the filtering residuals. The frequency deviation estimate error is obtained by statistically analyzing the variance of the squared deviations between the frequency deviation estimate data before and after filtering and the frequency deviation estimate data.

[0118] Among them, the time-varying phase difference value refers to the amount of phase change between adjacent missing points, which can be specifically realized by calculating the phase difference in the complex domain.

[0119] Dynamically adjusting the filter coefficients refers to changing the filter response characteristics in real time based on the confidence data. Specifically, a nonlinear mapping function can be used to convert the confidence level into a gain parameter, thereby enhancing the filter's ability to capture sudden signals at low confidence levels and improving filter stability at high confidence levels.

[0120] Weighted moving average filtering refers to smoothing the phase difference sequence according to time-varying weights, which can be achieved by multiplying the weights of each point in the sliding window with the filtering coefficients.

[0121] Frequency offset estimation error refers to the statistical deviation between the filtered output and the reference data. Specifically, it can be calculated using the variance of the residual sum of squares. The filtering accuracy is evaluated by quantifying the degree of deviation between the estimated value and the actual value.

[0122] Specifically, based on the interpolation compensation sequence, the phase difference between adjacent missing points is calculated using complex interpolation, effectively preserving phase abrupt change information. Based on the confidence data provided by the HRF link, a mapping relationship between confidence and filter gain is established. When the confidence is high, the filter smoothing coefficient is increased to suppress noise; when the confidence is low, the smoothing coefficient is decreased to maintain the ability to track rapid frequency offset changes. A variable-weight moving average algorithm is used during filtering, with the weight of each sampling point determined by the dynamically adjusted filter coefficient, achieving adaptive noise suppression and signal tracking. The filter residual is obtained by comparing the difference between the HPLC estimate and the HRF reference value. The variance calculation of the residual sum of squares objectively reflects the fluctuation of the frequency offset estimate, providing a quantitative basis for error quantification in subsequent fusion calculations.

[0123] Through the above technical solutions, this application effectively solves the problem that fixed filter parameters cannot adapt to dynamic changes in electromagnetic interference intensity. By using a confidence-driven filter dynamic adjustment mechanism, the tracking accuracy of sudden signals is improved while ensuring the stability of frequency offset estimation. The collaborative processing of complex phase difference calculation and weighted moving average reduces the frequency offset estimation error and can still maintain reliable frequency offset correction capability in strong pulse interference scenarios, providing an accurate frequency offset estimation basis for the collaborative optimization of dual-mode communication systems.

[0124] Using the frequency offset estimation data as the prior distribution, and the frequency offset estimate and its corresponding frequency offset estimation error as the observation likelihood, combined with physical constraints, Bayesian fusion calculation is performed to output the fused frequency offset estimate and fusion confidence, specifically including:

[0125] The frequency offset estimation data is used as the prior distribution for Bayesian fusion; the prior distribution is formed by modeling the probability density function of the frequency offset estimation data.

[0126] The frequency offset estimate and the corresponding frequency offset estimate error are used as the observation likelihood; the observation likelihood adopts a Gaussian distribution model.

[0127] Based on the prior distribution and observed likelihood, and combined with the physical constraints of the communication system, including the frequency offset rate of change limit and the frequency offset range boundary constraint, the fused frequency offset posterior probability distribution is calculated using the Bayesian recursive formula.

[0128] Calculate the mean of the fused frequency offset posterior probability distribution and use it as the fused frequency offset estimate. Calculate the variance of the fused frequency offset posterior probability distribution and use its reciprocal as the fused confidence level.

[0129] Among them, the prior distribution refers to the statistical model established based on the frequency offset estimation data, which can be implemented by the probability density function modeling method.

[0130] Observational likelihood refers to the probabilistic model of the HPLC link frequency offset estimation results, which can be implemented using a Gaussian distribution model.

[0131] Physical constraints refer to the objective limitations on frequency offset variation in a communication system, which can be achieved by limiting the rate of frequency offset variation and the boundary constraints of the frequency offset range.

[0132] Bayesian recursion is a mathematical method for updating probabilities based on prior distribution and observation likelihood, which can be implemented using a posterior probability distribution calculation model.

[0133] Specifically, the Bayesian fusion calculation process consists of four stages: First, the frequency offset estimation data collected by the HRF link is modeled as a probability density function to form a prior distribution. Second, the frequency offset estimation value and its error obtained by the HPLC link through the time-varying phase difference filtering method are constructed into a Gaussian observation likelihood. Then, the physical constraints of the communication system are transformed into the limiting boundary of the probability distribution, for example, by truncating the probability distribution or introducing a penalty function, to ensure that the fused frequency offset estimation value meets the frequency offset change rate limit and boundary range requirements. Finally, the mean and variance of the posterior probability distribution are calculated using the Bayesian recursive formula, where the mean is used as the fused frequency offset estimation value and the reciprocal of the variance is used as the fusion confidence level. In this process, the physical constraints are modified by correcting the shape of the posterior probability distribution to force the fusion result to conform to the physical realizability of the actual system.

[0134] Through the above technical solution, this application effectively solves the problem of fusion deviation caused by ignoring the statistical characteristics of data and physical constraints in the prior art. It can generate frequency offset estimates that conform to the actual system operating conditions under strong electromagnetic interference environment. At the same time, by quantifying the reliability of the estimation results through fusion confidence, it provides accurate input parameters for subsequent feedback control filtering, thereby improving the anti-interference capability and frequency offset correction accuracy of the dual-mode communication system.

[0135] The fused frequency offset estimate is subjected to feedback control filtering based on the fused confidence level, and the output frequency offset correction value specifically includes:

[0136] Based on the fusion confidence, the adaptive gain coefficient of the feedback control filter is calculated using a nonlinear mapping function;

[0137] Based on the adaptive gain coefficient, a weighted moving average filtering operation is performed on the fused frequency offset estimate. Specifically, the current fused frequency offset estimate and the previous time-lapse filter output value are weighted and summed according to the gain coefficient and its complement to generate the filtered frequency offset correction value.

[0138] Among them, the fusion confidence is a quantitative index that characterizes the reliability of the frequency offset estimation result, which can be achieved by using the inverse of the variance of the Bayesian fusion posterior probability distribution.

[0139] A nonlinear mapping function refers to the transformation relationship that maps confidence level to gain coefficient. Specifically, it can be implemented using an exponential function or a piecewise linear function. For example, different gain coefficient slopes can be set by setting confidence level threshold intervals.

[0140] The adaptive gain coefficient refers to the dynamically adjusted filter weight parameter, which can be calculated by the ratio of the confidence level to the preset benchmark value.

[0141] The weighted moving average filtering operation is a time series data smoothing method based on variable weights, which can be implemented using a first-order recursive filtering structure.

[0142] Specifically, when electromagnetic pulse interference causes fluctuations in the fusion confidence level, the nonlinear mapping function maps low confidence levels to smaller gain coefficients, making the filtering process more reliant on historical output values ​​to suppress anomalous jumps. When the confidence level is high, the gain coefficient is increased, allowing the filtering result to quickly track the current fusion estimate and making the current estimate dominant. Through this dynamic weight allocation mechanism, both the stability of frequency offset correction and tracking capability are ensured.

[0143] Through the above technical solution, this application can automatically adjust the response characteristics of the filter when electromagnetic pulse interference causes drastic fluctuations in the fusion confidence level, suppress the influence of abnormal frequency offset jumps on the correction results, and at the same time maintain the ability to quickly track normal frequency offset changes, significantly improving the stability and dynamic response accuracy of frequency offset correction in the complex electromagnetic environment of underground substations.

[0144] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A frequency offset correction method in HPLC+HRF dual-mode communication, characterized in that, The method comprises the following steps: S1, collecting phase sequence data from an HPLC communication link, collecting frequency offset estimation data and corresponding confidence data from an HRF communication link, and adaptively adjusting a preset detection threshold of the phase sequence data according to the confidence data; S2, performing pulse abnormal sample detection and elimination on the phase sequence data according to the adjusted preset detection threshold, and outputting the phase sequence data after the abnormal samples are eliminated and gradient constraint data of the abnormal sample neighborhood; S3, performing weighted interpolation compensation on the phase sequence data after the abnormal samples are eliminated according to the frequency offset estimation data as a guide basis and in combination with the gradient constraint data, and generating an interpolation compensation sequence; S4, taking the interpolation compensation sequence and the frequency offset estimation data as inputs, and calculating a frequency offset estimation value of the HPLC and corresponding frequency offset estimation error through a time-varying phase difference filtering method; S5, taking the frequency offset estimation data as a prior distribution, taking the frequency offset estimation value and the corresponding frequency offset estimation error as an observation likelihood, and performing Bayesian fusion calculation in combination with physical constraints to output a fusion frequency offset estimation value and a fusion confidence; S6, performing feedback control filtering on the fusion frequency offset estimation value based on the fusion confidence, and outputting a frequency offset correction value.

2. The frequency offset correction method in HPLC+HRF dual-mode communication according to claim 1, characterized in that: The collecting of the phase sequence data from the HPLC communication link and the collecting of the frequency offset estimation data and the corresponding confidence data from the HRF communication link specifically comprise: collecting the phase sequence data from the HPLC communication link through a high-speed phase modulation demodulator; collecting the frequency offset estimation data and the corresponding confidence data from the HRF communication link through a built-in frequency tracker; performing time sequence alignment on the phase sequence data, the frequency offset estimation data and the corresponding confidence data according to collection time stamps.

3. The frequency offset correction method in HPLC+HRF dual-mode communication according to claim 1, characterized in that: The adaptive adjustment of the preset detection threshold of the phase sequence data according to the confidence data specifically comprises: predefining a threshold adjustment function, mapping the confidence data into a threshold adjustment coefficient through the threshold adjustment function; multiplying the threshold adjustment coefficient by the preset detection threshold to obtain the adjusted preset detection threshold.

4. The frequency offset correction method in HPLC+HRF dual-mode communication according to claim 1, characterized in that: The pulse abnormal sample detection and elimination on the phase sequence data according to the adjusted preset detection threshold, and the output of the phase sequence data after the abnormal samples are eliminated and the gradient constraint data of the abnormal sample neighborhood specifically comprise: calculating the phase difference absolute value of adjacent sampling points of the phase sequence data, judging whether the phase difference absolute value is greater than the adjusted preset detection threshold, and determining that the sampling point corresponding to the phase difference absolute value is an abnormal sample if yes; eliminating the abnormal sample to obtain the phase sequence data after the abnormal sample is eliminated; calculating the phase difference gradient of the left and right adjacent sampling points of the abnormal sample to obtain the gradient constraint data of the abnormal sample neighborhood.

5. The frequency offset correction method in HPLC+HRF dual-mode communication according to claim 1, characterized in that: The weighted interpolation compensation on the phase sequence data after the abnormal samples are eliminated according to the frequency offset estimation data as a guide basis and in combination with the gradient constraint data, and the generation of the interpolation compensation sequence specifically comprise: calculating the interpolation weight of each missing point in the phase sequence data after the abnormal samples are eliminated through a nonlinear calculation according to the frequency offset estimation data. According to the interpolation weight of the missing point and the phase value of the left and right adjacent sampling points of the missing point in the phase sequence data after the abnormal sample points are removed, the phase value of the missing point is obtained through linear calculation; The front and back gradients of the missing point phase value are calculated; It is judged whether the front and back gradient calculation results are greater than the gradient constraint data of the neighborhood of the missing point corresponding to the abnormal sample points. If yes, gradient constraint correction processing is performed on the missing point phase value, and an interpolation compensation sequence is generated according to the missing point phase value after the correction processing.

6. The frequency offset correction method in HPLC+HRF dual-mode communication according to claim 1, characterized in that: The interpolation compensation sequence and the frequency offset estimation data are input, and the frequency offset estimation value and the corresponding frequency offset estimation error of the HPLC are calculated through the time-varying phase difference filtering method, which specifically includes: The time-varying phase difference value of the adjacent missing point of the interpolation compensation sequence is calculated. The time-varying phase difference value is obtained by calculating the complex difference value of the adjacent missing point phase difference; According to the confidence data, the filter coefficient of the time-varying phase difference filter is dynamically adjusted; The time-varying phase difference value is weighted and slidingly averaged filtered, and the filter weight is determined by the above dynamically adjusted filter coefficient. The frequency offset estimation value of the HPLC is output. The frequency offset estimation value of the HPLC is subtracted from the frequency offset estimation data to obtain the filtering residual error. The frequency offset estimation error is calculated according to the square sum of the filtering residual error. The frequency offset estimation error is obtained by calculating the variance of the square of the deviation between the frequency offset estimation data before and after filtering and the frequency offset estimation data.

7. The frequency offset correction method in HPLC+HRF dual-mode communication according to claim 1, characterized in that: The frequency offset estimation data is used as the prior distribution, the frequency offset estimation value and the corresponding frequency offset estimation error are used as the observation likelihood, and the Bayesian fusion calculation is performed combined with the physical constraint to output the fusion frequency offset estimation and the fusion confidence, which specifically includes: The frequency offset estimation data is used as the prior distribution of Bayesian fusion. The prior distribution is formed by modeling the probability density function of the frequency offset estimation data; The frequency offset estimation value and the corresponding frequency offset estimation error are used as the observation likelihood. The observation likelihood adopts a Gaussian distribution model; According to the prior distribution and the observation likelihood, combined with the physical constraint of the communication system, including the frequency offset change rate limit and the frequency offset range boundary constraint, the Bayesian recursive formula is used to calculate the fused frequency offset posterior probability distribution; The mean of the fused frequency offset posterior probability distribution is calculated and used as the fusion frequency offset estimation value. The variance of the fused frequency offset posterior probability distribution is calculated and its reciprocal is used as the fusion confidence.

8. The frequency offset correction method in HPLC+HRF dual-mode communication according to claim 1, characterized in that: The fusion frequency offset estimation value is subjected to feedback control filtering based on the fusion confidence, and the frequency offset correction value is output, which specifically includes: According to the fusion confidence, the adaptive gain coefficient of the feedback control filter is calculated by using a nonlinear mapping function; According to the adaptive gain coefficient, the weighted sliding average filtering operation is performed on the fusion frequency offset estimation value. Specifically, the current fusion frequency offset estimation value and the filtering output value at the previous moment are weighted and summed according to the gain coefficient and its complement to generate the filtered frequency offset correction value.

Images