Measurement error correction method based on intelligent sensor and crystal frequency drift compensation

Abstract

The application discloses a measurement error correction method based on intelligent sensors and crystal oscillator frequency drift compensation, which comprises the following steps: step one, obtaining ranging interaction data between a mobile station and multiple base stations, and collecting crystal oscillator running state data of the mobile station and each base station through intelligent sensors; step two, constructing a link crystal oscillator drift feature sequence through residual disturbance gate coding; step three, estimating a crystal oscillator drift trend through an improved grey GM(1, 1) drift prediction model; step four, performing standardized mapping and gate compensation constraint; step five, performing drift integral distribution compensation; step six, calculating a candidate correction distance sequence, limiting and correcting the candidate correction distance sequence through a propagation reachable boundary constraint, and obtaining a corrected measurement distance sequence. Through the improved grey GM(1, 1) drift prediction model and the drift integral distribution compensation, the application improves the accuracy of measurement error correction.

Description

Technical Field

[0001] This invention relates to the field of measurement error correction technology, and in particular to a measurement error correction method based on intelligent sensors and crystal oscillator frequency drift compensation. Background Technology

[0002] With the increasing demand for smart sensors, indoor positioning, industrial inspection, and mobile ranging applications, high-precision error correction technology for wireless ranging data between a mobile station and multiple base stations has received widespread attention. Existing ranging systems typically rely on fixed clock sources, conventional timestamp differential calculations, or simple filtering compensation methods to estimate the distance between the mobile station and base stations. In practical applications, these systems generally suffer from the following problems:

[0003] The crystal oscillators built into mobile stations and base stations are prone to frequency drift under the influence of temperature changes, power supply voltage fluctuations, increased equipment operating time, and vibration interference. This leads to a cumulative time base deviation between the ranging signal transmission time, ranging signal reception time, response signal transmission time, and response signal reception time, making it difficult for the initial time measurement and round-trip timestamp residuals to accurately reflect the true propagation time. The link environment and crystal oscillator disturbance states differ between different base stations and mobile stations. Existing compensation methods often use uniform compensation coefficients or static calibration parameters, making it difficult to dynamically correct for changes in the crystal oscillator's operating state within each ranging interaction cycle, resulting in insufficient stability of the corrected time measurement values. For nonlinear, phased crystal oscillator drift trends, traditional linear prediction or single grey prediction methods struggle to identify drift phase switching points and to reasonably allocate drift contributions in the forward ranging process and return response process. This can cause candidate correction distances to still exceed the actual coverage area and effective ranging boundary of the base station, affecting the reliability of the final corrected measurement distance sequence. Summary of the Invention

[0004] One objective of this invention is to propose a measurement error correction method based on intelligent sensors and crystal oscillator frequency drift compensation. This invention fully utilizes intelligent sensor acquisition, residual disturbance gating coding, improved gray GM(1,1) drift prediction, gating compensation constraints, and propagation reachability boundary constraints to jointly analyze ranging interaction data between the mobile station and multiple base stations and crystal oscillator operating status data. It constructs a link crystal oscillator drift characteristic sequence, predicts the link crystal oscillator drift trend, and generates a link time base drift compensation coefficient sequence. Then, it obtains a correction time measurement value sequence through drift integral allocation compensation, and finally generates a correction measurement distance sequence between the mobile station and each base station. It has the advantages of high ranging error correction accuracy, strong crystal oscillator drift compensation, and good stability of ranging results in complex environments.

[0005] The measurement error correction method based on smart sensors and crystal oscillator frequency drift compensation according to embodiments of the present invention includes the following steps:

[0006] Step 1: Acquire ranging interaction data between the mobile station and multiple base stations, and collect crystal oscillator operation status data of the mobile station and each base station through intelligent sensors;

[0007] Step 2: Based on ranging interaction data and crystal oscillator operating status data, construct the link crystal oscillator drift characteristic sequence between the mobile station and each base station through residual perturbation gating coding;

[0008] Step 3: Input the link crystal oscillator drift feature sequence into the improved gray GM(1,1) drift prediction model to estimate the crystal oscillator drift trend and obtain the link crystal oscillator drift trend parameters between the mobile station and each base station; the improved gray GM(1,1) drift prediction model introduces a residual disturbance gating segmented prediction mechanism;

[0009] Step 4: Based on the link crystal oscillator drift trend parameters, perform standardized mapping and gating compensation constraints to generate a link time base drift compensation coefficient sequence;

[0010] Step 5: Using the link time base drift compensation coefficient sequence, perform drift integral allocation compensation on the ranging interaction cycle between the mobile station and each base station to obtain the corrected time measurement value sequence;

[0011] Step 6: Calculate the candidate calibration distance sequence between the mobile station and each base station based on the calibration time measurement value sequence, and perform amplitude limiting calibration on the candidate calibration distance sequence through propagation reachability boundary constraints to obtain the calibration measurement distance sequence between the mobile station and each base station.

[0012] Optionally, the ranging interaction data includes mobile station identifier, base station identifier, ranging signal transmission time, ranging signal reception time, response signal transmission time, and response signal reception time; the crystal oscillator operating status data includes crystal oscillator temperature data, power supply voltage data, equipment operating time data, and vibration interference data.

[0013] Optionally, step two specifically includes:

[0014] Based on the mobile station identifier and the base station identifier, the ranging signal transmission time, ranging signal reception time, response signal transmission time and response signal reception time formed between the same mobile station and the same base station in the same ranging process are classified into the same ranging interaction cycle.

[0015] In each ranging interaction cycle, for the mobile station and the i-th base station, the differences between the response signal reception time and the ranging signal transmission time, the response signal transmission time and the ranging signal reception time, the ranging signal reception time and the ranging signal transmission time and the response signal reception time and the response signal transmission time are calculated respectively, so as to obtain the total round-trip time difference, the response waiting time difference, the ranging one-way timestamp difference, and the response one-way timestamp difference.

[0016] Half of the difference between the total round-trip time difference and the response waiting time difference is used as the initial time measurement between the mobile station and the i-th base station;

[0017] The difference between the one-way timestamp difference in ranging and the one-way timestamp difference in response is taken as the round-trip timestamp residual between the mobile station and the i-th base station;

[0018] The crystal oscillator operating status data is time-aligned according to the period center time corresponding to each ranging interaction cycle to obtain the crystal oscillator operating status alignment data of the mobile station and the i-th base station in each ranging interaction cycle.

[0019] The perturbation reference is normalized for each dimension of the crystal oscillator operating state alignment data to obtain the crystal oscillator perturbation normalized state quantity.

[0020] Calculate the difference between the normalized state quantity of the crystal oscillator disturbance of the mobile station and the normalized state quantity of the crystal oscillator disturbance of the i-th base station to obtain the crystal oscillator disturbance difference state quantity.

[0021] The perturbation difference state quantity of the crystal oscillator is linearly mapped and Sigmoid activation is applied to obtain the perturbation gating coefficient; the round-trip time stamp residual is gated and modulated based on the perturbation gating coefficient to obtain the perturbation modulation residual;

[0022] The initial time measurement value, round-trip timestamp residual, crystal oscillator disturbance difference state quantity and disturbance modulation residual are spliced ​​together to obtain the crystal oscillator drift state characteristics. The crystal oscillator drift state characteristics are arranged according to the ranging interaction period to obtain the link crystal oscillator drift characteristic sequence between the mobile station and the i-th base station.

[0023] Optionally, the residual perturbation gating segmented prediction mechanism specifically includes:

[0024] Based on the link crystal oscillator drift characteristic sequence between the mobile station and the i-th base station, the disturbance modulation residual, crystal oscillator disturbance difference state quantity and disturbance gating coefficient are extracted according to the ranging interaction period;

[0025] The residuals and perturbation gating coefficients are arranged in chronological order to obtain the residual observation sequence and the perturbation gating sequence, respectively.

[0026] The crystal oscillator perturbation difference state quantities are aggregated using the second norm to obtain the crystal oscillator perturbation difference intensity, and the crystal oscillator perturbation difference intensity is arranged in chronological order to generate a perturbation cause sequence;

[0027] Based on the residual observation sequence and the perturbation cause sequence, the drift change intensity between adjacent ranging interaction cycles is calculated by the residual perturbation relative change fusion method to obtain the drift change intensity sequence.

[0028] Construction phase switching threshold;

[0029] The ranging interaction period when the drift change intensity is greater than or equal to the stage switching threshold is determined as the drift stage switching point, and the link crystal oscillator drift feature sequence between the mobile station and the i-th base station is divided into multiple drift trend subsequences based on the drift stage switching point.

[0030] For the s-th drift trend subsequence, extract the sub-segment residual observation sequence, sub-segment disturbance cause sequence, and sub-segment disturbance gating sequence, and perform a cumulative generation on the sub-segment residual observation sequence and the sub-segment disturbance cause sequence to obtain the sub-segment residual cumulative sequence and the sub-segment disturbance cumulative sequence.

[0031] Based on the segment perturbation gated sequence and the segment residual accumulation sequence, construct the s-th drift trend subsequence gated background value sequence;

[0032] Based on the sub-segment residual observation sequence, the gated background value sequence, and the sub-segment disturbance accumulation sequence, a disturbance coupling grey drift prediction equation containing the development coefficient, the basic grey action quantity, and the disturbance coupling coefficient is established through the discretization form of the first-order grey differential equation.

[0033] The perturbation-coupled grey drift prediction equation is converted into matrix form, and the development coefficient, basic grey action and perturbation coupling coefficient are solved by the least squares method.

[0034] Substituting the development coefficient, basic grey action, perturbation coupling coefficient, and sub-segment perturbation accumulation sequence into the perturbation coupling grey drift prediction equation, we obtain the predicted perturbation modulation residual sequence of the s-th drift trend sub-sequence.

[0035] The development coefficient, basic gray action, disturbance coupling coefficient, predicted disturbance modulation residual sequence, and drift stage switching point corresponding to each drift trend subsequence are used as the link crystal oscillator drift trend parameters between the mobile station and the i-th base station.

[0036] Optionally, the process of establishing the perturbation-coupled grey drift prediction equation specifically includes:

[0037] The perturbation modulation residual corresponding to the current ranging interaction period in the sub-segment residual observation sequence is used as the original response term;

[0038] The gated background value corresponding to the current ranging interaction cycle in the gated background value sequence is used as the development background term, the sequence element corresponding to the current ranging interaction cycle in the sub-segment disturbance accumulation sequence is used as the disturbance accumulation value, and the disturbance accumulation value is used as the exogenous disturbance input term.

[0039] The development coefficient, basic grey action quantity, and perturbation coupling coefficient are set as the model parameters to be determined; the original response term is summed with the product of the development coefficient and the development background term, and used as the left-hand side term of the perturbation coupling grey drift prediction equation;

[0040] The product of the basic gray action quantity, the perturbation coupling coefficient, and the exogenous perturbation input term is summed and used as the right-hand side term of the perturbation coupling gray drift prediction equation.

[0041] Set the left-hand side terms to be equal to the right-hand side terms, and establish the perturbation-coupled grey drift prediction equation.

[0042] Optionally, step four specifically includes:

[0043] Read the link crystal drift trend parameters between the mobile station and the i-th base station, and determine the drift trend subsequence to which the current ranging interaction cycle belongs based on the drift phase switching point;

[0044] Extract each predicted perturbation modulation residual from the drift trend subsequence to which the current ranging interaction cycle belongs, and calculate the mean and standard deviation of each predicted perturbation modulation residual to obtain the mean and standard deviation of the sub-segment residuals.

[0045] Based on the mean and standard deviation of the sub-segment residuals, the predicted perturbation modulation residuals corresponding to the current ranging interaction cycle are Z-score standardized to obtain the standard score of the trend residuals.

[0046] The standard score of the trend residual is input into the tanh function for compression mapping, and the compression mapping result is fused with the absolute value of the development coefficient corresponding to the current drift trend subsequence for trend enhancement to obtain the trend residual mapping value.

[0047] Read the disturbance gating coefficient corresponding to the current ranging interaction cycle, and multiply the disturbance gating coefficient with the trend residual mapping value to obtain the gating compensation strength;

[0048] The gating compensation intensity is multiplied by the predicted disturbance modulation residual corresponding to the current ranging interaction cycle to obtain the gating compensation offset.

[0049] Based on the proportional relationship between the gating compensation offset and the initial time measurement value, the link time base drift compensation coefficient corresponding to the current ranging interaction cycle is generated.

[0050] Arrange the link time base drift compensation coefficients according to the ranging interaction cycle to obtain the link time base drift compensation coefficient sequence between the mobile station and the i-th base station.

[0051] Optionally, the drift integral allocation compensation specifically includes:

[0052] Based on the link time-base drift compensation coefficient sequence between the mobile station and the i-th base station, extract the link time-base drift compensation coefficient corresponding to the current ranging interaction period;

[0053] Read the total round-trip time difference, response waiting time difference, one-way distance measurement timestamp difference, and one-way response timestamp difference corresponding to the current distance measurement interaction cycle;

[0054] The response waiting time difference is separated from the total round-trip time difference to obtain the closed-loop propagation difference term corresponding to the current ranging interaction cycle;

[0055] Calculate the total time base compensation amount corresponding to the current ranging interaction cycle based on the closed-loop propagation differential term and the link time base drift compensation coefficient corresponding to the current ranging interaction cycle;

[0056] Within the drift trend subsequence to which the current ranging interaction cycle belongs, the forward drift contribution value is obtained by multiplying the one-way timestamp difference of ranging with the predicted perturbation modulation residual and taking the absolute value; the return drift contribution value is obtained by multiplying the one-way timestamp difference of response with the predicted perturbation modulation residual and taking the absolute value.

[0057] The forward drift contribution value and the return drift contribution value are accumulated and integrated respectively to obtain the accumulated value of the forward drift contribution and the accumulated value of the return drift contribution;

[0058] The forward compensation allocation ratio and the return compensation allocation ratio are obtained by normalizing the cumulative integral values ​​of the forward drift contribution and the cumulative integral values ​​of the return drift contribution.

[0059] The total time base compensation is multiplied by the forward compensation allocation ratio and the return compensation allocation ratio respectively to obtain the forward time base compensation and the return time base compensation.

[0060] The forward time base compensation is used to differentially subtract and compensate the one-way timestamp difference of ranging to obtain the corrected one-way timestamp difference of ranging; the return time base compensation is used to differentially subtract and compensate the one-way timestamp difference of response to obtain the corrected one-way timestamp difference of response.

[0061] For each ranging interaction cycle, the average of the correction ranging one-way timestamp difference and the correction response one-way timestamp difference is taken as the correction time measurement value for the corresponding ranging interaction cycle, and arranged to obtain the sequence of correction time measurement values ​​between the mobile station and the i-th base station.

[0062] Optionally, step six specifically includes:

[0063] For each ranging interaction cycle between the mobile station and the i-th base station, the correction time measurement value of the corresponding ranging interaction cycle is multiplied by the signal propagation speed to obtain the candidate correction distance;

[0064] Read the base station ranging configuration parameter table corresponding to the i-th base station. The base station ranging configuration parameter table includes the base station coverage radius, minimum effective ranging distance and maximum effective ranging distance.

[0065] The minimum effective ranging distance is taken as the lower boundary of the propagation reachable distance, and the smaller value between the maximum effective ranging distance and the base station coverage radius is taken as the upper boundary of the propagation reachable distance.

[0066] Based on the lower and upper boundaries of the propagation reachability distance, the candidate correction distance is constrained by the propagation reachability boundary, specifically as follows:

[0067] If the candidate correction distance is less than the lower boundary of the reachable propagation distance, then the candidate correction distance is limited to the lower boundary of the reachable propagation distance.

[0068] If the candidate correction distance is greater than the upper boundary of the propagation reachable distance, then the candidate correction distance is limited to the upper boundary of the propagation reachable distance;

[0069] If the candidate correction distance is located between the lower boundary of the reachable distance and the upper boundary of the reachable distance, then the candidate correction distance is used as the correction measurement distance for the corresponding ranging interaction cycle.

[0070] The corrected measurement distances are arranged according to the ranging interaction cycle to obtain the corrected measurement distance sequence between the mobile station and the i-th base station.

[0071] The beneficial effects of this invention are:

[0072] This invention synchronously collects crystal oscillator operating status data from mobile stations and base stations using intelligent sensors. It then combines this data with ranging interaction data between the mobile station and multiple base stations for joint modeling. This allows for the establishment of a correspondence between ranging signal transmission time, ranging signal reception time, response signal transmission time, and response signal reception time, and crystal oscillator temperature data, power supply voltage data, equipment operating time data, and vibration interference data. This avoids relying solely on ranging timestamps for static error correction, improving the completeness of crystal oscillator drift cause identification and the specificity of ranging error analysis. Furthermore, by fusing timestamp residuals and crystal oscillator disturbance difference state quantities through residual disturbance gating coding, a disturbance modulation residual is generated, enhancing the correlation between crystal oscillator disturbances and ranging errors. In the crystal oscillator drift trend estimation stage, an improved grey GM(1,1) drift prediction model is introduced, employing a residual disturbance gating segmented prediction mechanism. The drift stage switching point is determined based on the drift change intensity sequence, and the link crystal oscillator drift characteristic sequence is divided into multiple drift trend sub-sequences. This adapts to the nonlinear and segmented variation characteristics of crystal oscillator frequency drift in different ranging stages. By using a gated background value sequence and a disturbance-coupled grey drift prediction equation, the disturbance modulation residual, gated background value, and disturbance accumulation value are jointly used for drift trend prediction, improving the prediction accuracy and stage adaptability of the link crystal oscillator drift trend parameters. In the compensation processing stage, a link time-base drift compensation coefficient sequence is generated through standardized mapping and gated compensation constraints, and drift integral allocation compensation is performed to improve the time-base drift compensation accuracy of the bidirectional ranging link. Finally, propagation reachability boundary constraints are used to perform amplitude limiting correction on the candidate correction distance sequence to prevent the correction results from exceeding the effective ranging range of the base station, improving the rationality and stability of the corrected measurement distance. Therefore, this invention can improve the accuracy, stability, and reliability of ranging error correction between the mobile station and each base station in scenarios where crystal oscillator frequency drift, environmental disturbances, and differences in multi-base station ranging links coexist. Attached Figure Description

[0073] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0074] Figure 1 This is a schematic diagram of the measurement error correction method based on intelligent sensors and crystal oscillator frequency drift compensation proposed in this invention;

[0075] Figure 2 This is a flowchart of the improved gray GM(1,1) drift prediction model in the measurement error correction method based on intelligent sensors and crystal oscillator frequency drift compensation proposed in this invention.

[0076] Figure 3 This is a flowchart of the drift integral allocation compensation process in the measurement error correction method based on intelligent sensors and crystal oscillator frequency drift compensation proposed in this invention. Detailed Implementation

[0077] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0078] refer to Figures 1-3 The measurement error correction method based on smart sensors and crystal oscillator frequency drift compensation includes the following steps:

[0079] Step 1: Acquire ranging interaction data between the mobile station and multiple base stations, and collect crystal oscillator operation status data of the mobile station and each base station through intelligent sensors;

[0080] Step 2: Based on ranging interaction data and crystal oscillator operating status data, construct the link crystal oscillator drift characteristic sequence between the mobile station and each base station through residual perturbation gating coding;

[0081] Step 3: Input the link crystal oscillator drift feature sequence into the improved gray GM(1,1) drift prediction model to estimate the crystal oscillator drift trend and obtain the link crystal oscillator drift trend parameters between the mobile station and each base station; among them, the improved gray GM(1,1) drift prediction model introduces a residual disturbance gating segmented prediction mechanism.

[0082] Step 4: Based on the link crystal oscillator drift trend parameters, perform standardized mapping and gating compensation constraints to generate a link time base drift compensation coefficient sequence;

[0083] Step 5: Using the link time base drift compensation coefficient sequence, perform drift integral allocation compensation on the ranging interaction cycle between the mobile station and each base station to obtain the corrected time measurement value sequence;

[0084] Step 6: Calculate the candidate calibration distance sequence between the mobile station and each base station based on the calibration time measurement value sequence, and perform amplitude limiting calibration on the candidate calibration distance sequence through propagation reachability boundary constraints to obtain the calibration measurement distance sequence between the mobile station and each base station.

[0085] In this embodiment, the ranging interaction data includes mobile station identifier, base station identifier, ranging signal transmission time, ranging signal reception time, response signal transmission time, and response signal reception time; the crystal oscillator operating status data includes crystal oscillator temperature data, power supply voltage data, equipment operating time data, and vibration interference data.

[0086] In this embodiment, step two specifically includes:

[0087] Based on the mobile station identifier and the base station identifier, the ranging signal transmission time, ranging signal reception time, response signal transmission time and response signal reception time formed between the same mobile station and the same base station in the same ranging process are classified into the same ranging interaction cycle.

[0088] In each ranging interaction cycle, for the mobile station and the i-th base station, the differences between the response signal reception time and the ranging signal transmission time, the response signal transmission time and the ranging signal reception time, the ranging signal reception time and the ranging signal transmission time and the response signal reception time and the response signal transmission time are calculated respectively, so as to obtain the total round-trip time difference, the response waiting time difference, the ranging one-way timestamp difference, and the response one-way timestamp difference.

[0089] Where i represents the index of the base station;

[0090] Half of the difference between the total round-trip time difference and the response waiting time difference is used as the initial time measurement between the mobile station and the i-th base station;

[0091] The difference between the one-way timestamp difference in ranging and the one-way timestamp difference in response is taken as the round-trip timestamp residual between the mobile station and the i-th base station;

[0092] The crystal oscillator operating status data is time-aligned according to the period center time corresponding to each ranging interaction cycle to obtain the crystal oscillator operating status alignment data of the mobile station and the i-th base station in each ranging interaction cycle.

[0093] The perturbation reference is normalized for each dimension of the crystal oscillator operating status alignment data to obtain the crystal oscillator perturbation normalized state quantity. Specifically, the perturbation reference normalization involves: reading the crystal oscillator calibration parameter tables corresponding to the mobile station and the i-th base station, which include the nominal crystal oscillator temperature, crystal oscillator temperature normalization scale, nominal supply voltage, supply voltage normalization scale, operating time normalization scale, and vibration interference normalization scale; and dividing the difference between the crystal oscillator temperature data in the crystal oscillator operating status alignment data and the nominal crystal oscillator temperature by the crystal oscillator temperature normalization scale. The following steps are performed: First, obtain the normalized component of the crystal oscillator temperature. Second, divide the difference between the supply voltage data and the nominal supply voltage by the normalized supply voltage scale to obtain the normalized supply voltage component. Third, divide the equipment running time data by the normalized running time scale to obtain the normalized running time component. Fourth, divide the vibration interference data by the normalized vibration interference scale to obtain the normalized vibration interference component. Finally, combine the normalized components of the crystal oscillator temperature, supply voltage, running time, and vibration interference to obtain the normalized state quantity of the crystal oscillator disturbance.

[0094] Calculate the difference between the normalized state quantity of the crystal oscillator disturbance of the mobile station and the normalized state quantity of the crystal oscillator disturbance of the i-th base station to obtain the crystal oscillator disturbance difference state quantity.

[0095] The perturbation difference state quantity of the crystal oscillator is linearly mapped and Sigmoid activation is applied to obtain the perturbation gating coefficient; the round-trip time stamp residual is gated and modulated based on the perturbation gating coefficient to obtain the perturbation modulation residual;

[0096] The initial time measurement value, round-trip timestamp residual, crystal oscillator disturbance difference state quantity and disturbance modulation residual are spliced ​​together to obtain the crystal oscillator drift state characteristics. The crystal oscillator drift state characteristics are arranged according to the ranging interaction period to obtain the link crystal oscillator drift characteristic sequence between the mobile station and the i-th base station.

[0097] In this embodiment, the residual perturbation gating segmented prediction mechanism specifically includes:

[0098] Based on the link crystal oscillator drift characteristic sequence between the mobile station and the i-th base station, the disturbance modulation residual, crystal oscillator disturbance difference state quantity and disturbance gating coefficient are extracted according to the ranging interaction period;

[0099] The residuals and perturbation gating coefficients are arranged in chronological order to obtain the residual observation sequence and the perturbation gating sequence, respectively.

[0100] The crystal oscillator perturbation difference state quantities are aggregated using the second norm to obtain the crystal oscillator perturbation difference intensity, and the crystal oscillator perturbation difference intensity is arranged in chronological order to generate a perturbation cause sequence;

[0101] Based on the residual observation sequence and the perturbation cause sequence, the drift change intensity between adjacent ranging interaction cycles is calculated by the residual perturbation relative change fusion method to obtain the drift change intensity sequence.

[0102] The residual disturbance relative change fusion method is as follows: calculate the relative change value of disturbance modulation residual between adjacent ranging interaction cycles, and calculate the relative change value of crystal oscillator disturbance difference intensity between adjacent ranging interaction cycles. Then, sum the relative change value of disturbance modulation residual and the relative change value of crystal oscillator disturbance difference intensity to obtain the drift change intensity.

[0103] The stage switching threshold is constructed as follows: the median of the drift change intensity sequence is calculated to obtain the drift change median; the median of the absolute deviation between each drift change intensity and the drift change median is calculated to obtain the drift change intensity dispersion; the drift change intensity dispersion is converted into dispersion compensation amount according to the adjustment coefficient, and the drift change intensity median is added to the dispersion compensation amount to obtain the stage switching threshold.

[0104] The ranging interaction period when the drift change intensity is greater than or equal to the stage switching threshold is determined as the drift stage switching point, and the link crystal oscillator drift feature sequence between the mobile station and the i-th base station is divided into multiple drift trend subsequences based on the drift stage switching point.

[0105] For the s-th drift trend subsequence, the sub-segment residual observation sequence, sub-segment disturbance cause sequence, and sub-segment disturbance gating sequence are extracted. The sub-segment residual observation sequence and sub-segment disturbance cause sequence are then accumulated once to generate the sub-segment residual accumulated sequence and sub-segment disturbance accumulated sequence. Here, s represents the index of the drift trend subsequence; the sub-segment residual observation sequence is the subsequence of the residual observation sequence within the s-th drift trend subsequence; the sub-segment disturbance cause sequence is the subsequence of the disturbance cause sequence within the s-th drift trend subsequence; and the sub-segment disturbance gating sequence is the subsequence of the disturbance gating sequence within the s-th drift trend subsequence.

[0106] Based on the segment perturbation-gated sequence and the segment residual accumulation sequence, the gated background value sequence of the s-th drift trend subsequence is constructed. Specifically, the perturbation gating coefficient and residual accumulation value corresponding to each ranging interaction cycle in the s-th drift trend subsequence are read. The residual accumulation value corresponding to the current ranging interaction cycle is taken as the current residual accumulation term, and the residual accumulation value corresponding to the previous ranging interaction cycle is taken as the historical residual accumulation term. The proportion of the current residual accumulation term in the background value construction is determined based on the perturbation gating coefficient, and the proportion of the historical residual accumulation term in the background value construction is determined based on the difference between the constant 1 and the perturbation gating coefficient. The current residual accumulation term and the historical residual accumulation term are gated and weighted to obtain the gated background value corresponding to the current ranging interaction cycle. The gated background values ​​are arranged according to the time order of the ranging interaction cycles to obtain the gated background value sequence of the s-th drift trend subsequence.

[0107] Based on the sub-segment residual observation sequence, the gated background value sequence, and the sub-segment disturbance accumulation sequence, a disturbance coupling grey drift prediction equation containing the development coefficient, the basic grey action quantity, and the disturbance coupling coefficient is established through the discretization form of the first-order grey differential equation.

[0108] The perturbation-coupled grey drift prediction equation is converted into matrix form, and the development coefficient, basic grey action and perturbation coupling coefficient are solved by the least squares method.

[0109] Substituting the development coefficient, basic grey action, perturbation coupling coefficient, and sub-segment perturbation accumulation sequence into the perturbation coupling grey drift prediction equation, we obtain the predicted perturbation modulation residual sequence of the s-th drift trend sub-sequence.

[0110] The development coefficient, basic gray action, disturbance coupling coefficient, predicted disturbance modulation residual sequence, and drift stage switching point corresponding to each drift trend subsequence are used as the link crystal oscillator drift trend parameters between the mobile station and the i-th base station.

[0111] In this embodiment, the process of establishing the perturbation-coupled grey drift prediction equation specifically includes:

[0112] The perturbation modulation residual corresponding to the current ranging interaction period in the sub-segment residual observation sequence is used as the original response term;

[0113] The gated background value corresponding to the current ranging interaction cycle in the gated background value sequence is used as the development background term, the sequence element corresponding to the current ranging interaction cycle in the sub-segment disturbance accumulation sequence is used as the disturbance accumulation value, and the disturbance accumulation value is used as the exogenous disturbance input term.

[0114] The development coefficient, basic grey action quantity, and perturbation coupling coefficient are set as the model parameters to be determined; the original response term is summed with the product of the development coefficient and the development background term, and used as the left-hand side term of the perturbation coupling grey drift prediction equation;

[0115] The product of the basic gray action quantity, the perturbation coupling coefficient, and the exogenous perturbation input term is summed and used as the right-hand side term of the perturbation coupling gray drift prediction equation.

[0116] Set the left-hand side terms to be equal to the right-hand side terms, and establish the perturbation-coupled grey drift prediction equation.

[0117] In this invention, the improved grey GM(1,1) drift prediction model inherits the basic prediction structure of the grey GM(1,1) prediction model. Both are based on the original observation sequence, weakening the random fluctuations of the original sequence through a single accumulation generation method, and establishing prediction relationships based on the first-order grey differential equation. During the model solution process, both require constructing a background value sequence, converting the grey differential equation into matrix form, solving for the model parameters using the least squares method, and then performing subsequent trend prediction based on the solved model parameters.

[0118] The improved grey GM(1,1) drift prediction model of this invention does not directly perform overall grey prediction on a single residual sequence, but introduces a residual perturbation-gated segmented prediction mechanism based on the grey GM(1,1) prediction model. Specifically, the improved grey GM(1,1) drift prediction model first extracts the perturbation modulation residual, the crystal oscillator perturbation difference state quantity, and the perturbation gating coefficient from the link crystal oscillator drift feature sequence, and constructs the residual observation sequence, the perturbation cause sequence, and the perturbation gating sequence, respectively; then, it calculates the drift change intensity through the residual perturbation relative change fusion method, and uses the stage switching threshold to determine the drift stage switching point, dividing the link crystal oscillator drift feature sequence into multiple drift trend subsequences. Meanwhile, the improved grey GM(1,1) drift prediction model introduces a disturbance gating coefficient in the background value construction stage, and obtains the gating background value by gating weighted fusion of the current residual accumulation term and the historical residual accumulation term; further, a disturbance coupling coefficient and an exogenous disturbance input term are introduced into the prediction equation to establish a disturbance coupled grey drift prediction equation, so that the crystal oscillator disturbance difference state quantity participates in the drift trend prediction process.

[0119] Through the above improvements, the improved grey GM(1,1) drift prediction model avoids the problem of insensitivity to drift stage changes when the grey GM(1,1) prediction model models the entire sequence at once. The model can automatically identify drift stage switching points based on the intensity of drift changes and make predictions for different drift trend subsequences separately, improving its adaptability to nonlinear, staged crystal oscillator drift trends. Furthermore, by introducing the disturbance gating coefficient into the background value construction process through a gating background value sequence, the dynamic trade-off between the current drift state and historical drift states can be enhanced. Introducing an exogenous disturbance input term into the disturbance-coupled grey drift prediction equation allows the crystal oscillator temperature, voltage, operating time, and vibration interference-induced crystal oscillator disturbance difference state quantities to participate in prediction correction. Therefore, this invention can improve the prediction accuracy and stability of link crystal oscillator drift trend parameters, providing a more reliable trend basis for the subsequent generation of link time-base drift compensation coefficient sequences.

[0120] In this embodiment, step four specifically includes:

[0121] Read the link crystal drift trend parameters between the mobile station and the i-th base station, and determine the drift trend subsequence to which the current ranging interaction cycle belongs based on the drift phase switching point;

[0122] Extract each predicted perturbation modulation residual from the drift trend subsequence to which the current ranging interaction cycle belongs, and calculate the mean and standard deviation of each predicted perturbation modulation residual to obtain the mean and standard deviation of the sub-segment residuals.

[0123] Based on the mean and standard deviation of the sub-segment residuals, the predicted perturbation modulation residuals corresponding to the current ranging interaction cycle are Z-score standardized to obtain the standard score of the trend residuals.

[0124] The standard score of the trend residual is input into the tanh function for compression mapping, and the compression mapping result is fused with the absolute value of the development coefficient corresponding to the current drift trend subsequence for trend enhancement to obtain the trend residual mapping value.

[0125] Read the disturbance gating coefficient corresponding to the current ranging interaction cycle, and multiply the disturbance gating coefficient with the trend residual mapping value to obtain the gating compensation strength;

[0126] The gating compensation intensity is multiplied by the predicted disturbance modulation residual corresponding to the current ranging interaction cycle to obtain the gating compensation offset.

[0127] Based on the proportional relationship between the gating compensation offset and the initial time measurement value, the link time base drift compensation coefficient corresponding to the current ranging interaction cycle is generated.

[0128] Arrange the link time base drift compensation coefficients according to the ranging interaction cycle to obtain the link time base drift compensation coefficient sequence between the mobile station and the i-th base station.

[0129] In this embodiment, drift integral allocation compensation specifically includes:

[0130] Based on the link time-base drift compensation coefficient sequence between the mobile station and the i-th base station, extract the link time-base drift compensation coefficient corresponding to the current ranging interaction period;

[0131] Read the total round-trip time difference, response waiting time difference, one-way distance measurement timestamp difference, and one-way response timestamp difference corresponding to the current distance measurement interaction cycle;

[0132] The response waiting time difference is separated from the total round-trip time difference to obtain the closed-loop propagation difference term corresponding to the current ranging interaction cycle;

[0133] Based on the closed-loop propagation differential term and the link time base drift compensation coefficient corresponding to the current ranging interaction period, the total time base compensation amount corresponding to the current ranging interaction period is calculated. Specifically, the difference between the link time base drift compensation coefficient and constant 1 is calculated to obtain the time base drift compensation ratio corresponding to the current ranging interaction period; the closed-loop propagation differential term is multiplied by the time base drift compensation ratio to obtain the total time base compensation amount corresponding to the current ranging interaction period.

[0134] Within the drift trend subsequence to which the current ranging interaction cycle belongs, the forward drift contribution value is obtained by multiplying the one-way timestamp difference of ranging with the predicted perturbation modulation residual and taking the absolute value; the return drift contribution value is obtained by multiplying the one-way timestamp difference of response with the predicted perturbation modulation residual and taking the absolute value.

[0135] The forward drift contribution value and the return drift contribution value are accumulated and integrated respectively to obtain the accumulated value of the forward drift contribution and the accumulated value of the return drift contribution;

[0136] The forward compensation allocation ratio and the return compensation allocation ratio are obtained by normalizing the cumulative integral values ​​of the forward drift contribution and the cumulative integral values ​​of the return drift contribution.

[0137] The total time base compensation is multiplied by the forward compensation allocation ratio and the return compensation allocation ratio respectively to obtain the forward time base compensation and the return time base compensation.

[0138] The forward time base compensation is used to differentially subtract and compensate the one-way timestamp difference of ranging to obtain the corrected one-way timestamp difference of ranging; the return time base compensation is used to differentially subtract and compensate the one-way timestamp difference of response to obtain the corrected one-way timestamp difference of response.

[0139] For each ranging interaction cycle, the average of the correction ranging one-way timestamp difference and the correction response one-way timestamp difference is taken as the correction time measurement value for the corresponding ranging interaction cycle, and arranged to obtain the sequence of correction time measurement values ​​between the mobile station and the i-th base station.

[0140] In this invention, drift integral allocation compensation calculates the forward drift contribution and return drift contribution based on the one-way timestamp difference in ranging, the one-way timestamp difference in response, and the predicted disturbance modulation residual, and determines the forward compensation allocation ratio and the return compensation allocation ratio by accumulating the integrals. Conventional error correction methods typically employ fixed deviation deduction, uniform proportional compensation, averaging filtering, Kalman filtering, or static crystal oscillator calibration compensation to uniformly correct the overall ranging time or overall ranging distance, making it difficult to distinguish the drift contribution differences between the forward ranging process and the return response process. This invention allocates the total time base compensation according to the actual drift contribution in the two-way ranging link, avoiding the two-way link compensation imbalance caused by uniform deduction, and improving the accuracy and stability of correcting the one-way timestamp difference in ranging, correcting the one-way timestamp difference in response, and correcting the time measurement value.

[0141] In this embodiment, step six specifically includes:

[0142] For each ranging interaction cycle between the mobile station and the i-th base station, the measured value of the correction time for the corresponding ranging interaction cycle is multiplied by the signal propagation speed to obtain the candidate correction distance; where the signal propagation speed is the propagation speed of the wireless ranging signal in the air medium.

[0143] Read the base station ranging configuration parameter table corresponding to the i-th base station. The base station ranging configuration parameter table includes the base station coverage radius, minimum effective ranging distance, and maximum effective ranging distance. Among them, the base station coverage radius is determined by the deployment configuration data of the i-th base station, and the minimum effective ranging distance and maximum effective ranging distance are determined by the ranging hardware configuration parameters and ranging protocol configuration parameters of the i-th base station.

[0144] The minimum effective ranging distance is taken as the lower boundary of the propagation reachable distance, and the smaller value between the maximum effective ranging distance and the base station coverage radius is taken as the upper boundary of the propagation reachable distance.

[0145] Based on the lower and upper boundaries of the propagation reachability distance, the candidate correction distance is constrained by the propagation reachability boundary, specifically as follows:

[0146] If the candidate correction distance is less than the lower boundary of the reachable propagation distance, then the candidate correction distance is limited to the lower boundary of the reachable propagation distance.

[0147] If the candidate correction distance is greater than the upper boundary of the propagation reachable distance, then the candidate correction distance is limited to the upper boundary of the propagation reachable distance;

[0148] If the candidate correction distance is located between the lower boundary of the reachable distance and the upper boundary of the reachable distance, then the candidate correction distance is used as the correction measurement distance for the corresponding ranging interaction cycle.

[0149] The corrected measurement distances are arranged according to the ranging interaction cycle to obtain the corrected measurement distance sequence between the mobile station and the i-th base station.

[0150] Example 1: To verify the feasibility of this invention in practice, the method was applied to a mobile station positioning and ranging scenario in a smart warehousing park. Eight UWB base stations were deployed in the park, with the mobile station mounted on top of an automated guided vehicle (AGV). The AGV moved continuously between rack aisles, turning areas, and loading / unloading areas. Due to the dense metal racks in the warehousing environment, the AGV's equipment temperature increased after prolonged operation. Additionally, localized areas experienced motor vibration, power supply voltage fluctuations, and differences in base station installation height, making the ranging results between the mobile station and each base station prone to time-base drift errors. The existing ranging system primarily used fixed deviation deduction and moving average filtering for error correction. However, when the vehicle approached the edge of the rack or made a rapid turn, the ranging results still showed jumps, causing the mobile station's positioning trajectory to deviate from the actual driving trajectory, affecting the AGV's path control and obstacle avoidance.

[0151] In the implementation scenario of this invention, a mobile station continuously engages in bidirectional ranging interaction with multiple base stations. Each ranging interaction cycle records the ranging signal transmission time, ranging signal reception time, response signal transmission time, and response signal reception time. Simultaneously, the intelligent sensors built into the mobile station and each base station collect crystal oscillator temperature data, power supply voltage data, equipment runtime data, and vibration interference data. First, the timestamp data between the same mobile station and the same base station are grouped into the same ranging interaction cycle, and the round-trip time difference, response waiting time difference, one-way ranging timestamp difference, and one-way response timestamp difference are calculated to obtain the initial time measurement value and the round-trip timestamp residual. Then, the crystal oscillator operating status data is time-aligned according to the cycle center time, and the crystal oscillator disturbance normalized state quantity is obtained through disturbance reference normalization. Further, the crystal oscillator disturbance difference state quantity between the mobile station and the corresponding base station is calculated, and the disturbance gating coefficient is obtained through linear mapping and Sigmoid activation processing. The round-trip timestamp residual is then gated and modulated using the disturbance gating coefficient to obtain the disturbance modulation residual, thereby constructing a link crystal oscillator drift characteristic sequence.

[0152] In actual operation, the link crystal oscillator drift characteristic sequence is input into the improved grey GM(1,1) drift prediction model. The drift change intensity is calculated based on the residual observation sequence and the disturbance cause sequence to determine the drift stage switching point and divide the model into multiple drift trend sub-sequences. For each drift trend sub-sequence, a gated background value sequence is constructed, and a disturbance-coupled grey drift prediction equation is established. The development coefficient, basic grey action quantity, and disturbance coupling coefficient are solved to obtain the predicted disturbance modulation residual sequence. Subsequently, based on the link crystal oscillator drift trend parameters, standardized mapping and gated compensation constraints are performed to generate the link time base drift compensation coefficient sequence, and the total time base compensation is calculated in each ranging interaction cycle. Unlike the ordinary unified deduction method, this invention determines the forward compensation allocation ratio and the return compensation allocation ratio based on the forward drift contribution value and the return drift contribution value, so that the total time base compensation is applied to the forward ranging process and the return response process respectively, obtaining a corrected time measurement value sequence. Finally, the corrected time measurement values ​​are converted into candidate corrected distances, and amplitude limiting correction is performed through propagation reachability boundary constraints to generate a corrected measurement distance sequence.

[0153] To further verify the practical effect of the present invention, a comparative experiment was conducted with the fixed deviation subtraction method, the moving average filtering method, and the static crystal oscillator calibration compensation method. The fixed deviation subtraction method is an error correction method that uniformly subtracts the original ranging result corresponding to each ranging interaction cycle based on a pre-calibrated fixed ranging deviation value. The moving average filtering method is an error suppression method that constructs a sliding window according to multiple consecutive ranging interaction cycles and uses the average value of the ranging results within the window as the correction result for the current ranging interaction cycle. The static crystal oscillator calibration compensation method is an error correction method that compensates the ranging time value by a fixed proportion based on the crystal oscillator frequency deviation parameters obtained by the mobile station and base station during the calibration phase. The experimental results are shown in Table 1.

[0154] Table 1. Comparison of the effects of different error correction methods in the mobile station ranging scenario in a warehouse park.

[0155] Average distance measurement error / m 0.46 0.39 0.31 0.14 Maximum ranging error / m 1.28 1.05 0.82 0.36 Standard deviation of ranging error / m 0.27 0.22 0.18 0.07 Number of trajectory jumps in the turning area / 1000 cycles 37 29 21 6 Crystal oscillator error increase after temperature rise / % 42.6 35.8 24.3 8.7 Effective ranging rate (%) under voltage fluctuation scenarios 91.4 93.2 95.1 98.6 Number of times the effective ranging boundary was exceeded / 1000 cycles 18 14 9 1 Corrected positioning trajectory continuity score / point 82.3 85.7 89.4 96.8 Average processing time per cycle / ms 7.2 8.5 9.1 13.6

[0156] As shown in Table 1, under the same warehousing park ranging data, the average ranging error of the method of this invention is 0.14m, significantly lower than 0.46m of the fixed deviation subtraction method, 0.39m of the moving average filtering method, and 0.31m of the static crystal oscillator calibration compensation method. This indicates that the present invention can more accurately compensate for the time base error caused by crystal oscillator frequency drift. Regarding the maximum ranging error, the present invention controls the error to within 0.36m, a reduction of approximately 71.9% compared to the fixed deviation subtraction method. This demonstrates that even under conditions of rapid turning of the mobile station, partial obstruction, and vibration interference, the present invention can still maintain good ranging stability.

[0157] Furthermore, the error increase after crystal oscillator temperature rise was only 8.7%, indicating that by using crystal oscillator temperature data, power supply voltage data, equipment runtime data, and vibration interference data in residual disturbance gating coding, the impact of crystal oscillator state changes on ranging results can be effectively reduced. Under voltage fluctuation scenarios, the effective ranging rate increased to 98.6%, and the number of times exceeding the effective ranging boundary decreased to 1 per 1000 cycles, indicating that the propagation reachability boundary constraint can prevent the corrected measurement distance from exceeding the actual coverage range of the base station. Although the average processing time per cycle of this invention is 13.6ms, slightly higher than ordinary error correction methods, it still meets the real-time ranging requirements of automated guided vehicles (AGVs) in the park. Therefore, this invention can improve the accuracy, stability, and physical rationality of the corrected measurement distance sequence in actual ranging environments where crystal oscillator frequency drift, equipment temperature rise, voltage fluctuation, and vibration interference coexist.

[0158] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A measurement error correction method based on intelligent sensors and crystal oscillator frequency drift compensation, characterized in that, Includes the following steps: Step 1: Acquire ranging interaction data between the mobile station and multiple base stations, and collect crystal oscillator operation status data of the mobile station and each base station through intelligent sensors; Step 2: Based on ranging interaction data and crystal oscillator operating status data, construct the link crystal oscillator drift characteristic sequence between the mobile station and each base station through residual perturbation gating coding; Step 3: Input the link crystal oscillator drift feature sequence into the improved gray GM(1,1) drift prediction model to estimate the crystal oscillator drift trend and obtain the link crystal oscillator drift trend parameters between the mobile station and each base station; The improved gray GM(1,1) drift prediction model introduces a residual perturbation-gated segmented prediction mechanism; Step 4: Based on the link crystal oscillator drift trend parameters, perform standardized mapping and gating compensation constraints to generate a link time base drift compensation coefficient sequence; Step 5: Using the link time base drift compensation coefficient sequence, perform drift integral allocation compensation on the ranging interaction cycle between the mobile station and each base station to obtain the correction time measurement value sequence; Step 6: Calculate the candidate calibration distance sequence between the mobile station and each base station based on the calibration time measurement value sequence, and perform amplitude limiting calibration on the candidate calibration distance sequence through propagation reachability boundary constraints to obtain the calibration measurement distance sequence between the mobile station and each base station.

2. The measurement error correction method based on intelligent sensors and crystal oscillator frequency drift compensation according to claim 1, characterized in that, The ranging interaction data includes mobile station identifier, base station identifier, ranging signal transmission time, ranging signal reception time, response signal transmission time, and response signal reception time; the crystal oscillator operating status data includes crystal oscillator temperature data, power supply voltage data, equipment operating time data, and vibration interference data.

3. The measurement error correction method based on intelligent sensors and crystal oscillator frequency drift compensation according to claim 1, characterized in that, Step two specifically includes: Based on the mobile station identifier and the base station identifier, the ranging signal transmission time, ranging signal reception time, response signal transmission time and response signal reception time formed between the same mobile station and the same base station in the same ranging process are classified into the same ranging interaction cycle. In each ranging interaction cycle, for the mobile station and the i-th base station, the differences between the response signal reception time and the ranging signal transmission time, the response signal transmission time and the ranging signal reception time, the ranging signal reception time and the ranging signal transmission time and the response signal reception time and the response signal transmission time are calculated respectively, so as to obtain the total round-trip time difference, the response waiting time difference, the ranging one-way timestamp difference, and the response one-way timestamp difference. Half of the difference between the total round-trip time difference and the response waiting time difference is used as the initial time measurement between the mobile station and the i-th base station; The difference between the one-way timestamp difference in ranging and the one-way timestamp difference in response is taken as the round-trip timestamp residual between the mobile station and the i-th base station; The crystal oscillator operating status data is time-aligned according to the period center time corresponding to each ranging interaction cycle to obtain the crystal oscillator operating status alignment data of the mobile station and the i-th base station in each ranging interaction cycle. The perturbation reference is normalized for each dimension of the crystal oscillator operating state alignment data to obtain the crystal oscillator perturbation normalized state quantity. Calculate the difference between the normalized state quantity of the crystal oscillator disturbance of the mobile station and the normalized state quantity of the crystal oscillator disturbance of the i-th base station to obtain the crystal oscillator disturbance difference state quantity. The perturbation difference state quantity of the crystal oscillator is linearly mapped and Sigmoid activation is applied to obtain the perturbation gating coefficient; the round-trip time stamp residual is gated and modulated based on the perturbation gating coefficient to obtain the perturbation modulation residual; The initial time measurement value, round-trip timestamp residual, crystal oscillator disturbance difference state quantity and disturbance modulation residual are spliced ​​together to obtain the crystal oscillator drift state characteristics. The crystal oscillator drift state characteristics are arranged according to the ranging interaction period to obtain the link crystal oscillator drift characteristic sequence between the mobile station and the i-th base station.

4. The measurement error correction method based on intelligent sensors and crystal oscillator frequency drift compensation according to claim 1, characterized in that, The residual perturbation gated piecewise prediction mechanism specifically includes: Based on the link crystal oscillator drift characteristic sequence between the mobile station and the i-th base station, the disturbance modulation residual, crystal oscillator disturbance difference state quantity and disturbance gating coefficient are extracted according to the ranging interaction period; The residuals and perturbation gating coefficients are arranged in chronological order to obtain the residual observation sequence and the perturbation gating sequence, respectively. The crystal oscillator perturbation difference state quantities are aggregated using the second norm to obtain the crystal oscillator perturbation difference intensity, and the crystal oscillator perturbation difference intensity is arranged in chronological order to generate a perturbation cause sequence; Based on the residual observation sequence and the perturbation cause sequence, the drift change intensity between adjacent ranging interaction cycles is calculated by the residual perturbation relative change fusion method to obtain the drift change intensity sequence. Construction phase switching threshold; The ranging interaction period when the drift change intensity is greater than or equal to the stage switching threshold is determined as the drift stage switching point, and the link crystal oscillator drift feature sequence between the mobile station and the i-th base station is divided into multiple drift trend subsequences based on the drift stage switching point. For the s-th drift trend subsequence, extract the sub-segment residual observation sequence, sub-segment disturbance cause sequence, and sub-segment disturbance gating sequence, and perform a cumulative generation on the sub-segment residual observation sequence and the sub-segment disturbance cause sequence to obtain the sub-segment residual cumulative sequence and the sub-segment disturbance cumulative sequence. Based on the segment perturbation gated sequence and the segment residual accumulation sequence, construct the s-th drift trend subsequence gated background value sequence; Based on the sub-segment residual observation sequence, the gated background value sequence, and the sub-segment disturbance accumulation sequence, a disturbance coupling grey drift prediction equation containing the development coefficient, the basic grey action quantity, and the disturbance coupling coefficient is established through the discretization form of the first-order grey differential equation. The perturbation-coupled grey drift prediction equation is converted into matrix form, and the development coefficient, basic grey action and perturbation coupling coefficient are solved by the least squares method. Substituting the development coefficient, basic grey action, perturbation coupling coefficient, and sub-segment perturbation accumulation sequence into the perturbation coupling grey drift prediction equation, we obtain the predicted perturbation modulation residual sequence of the s-th drift trend sub-sequence. The development coefficient, basic gray action, disturbance coupling coefficient, predicted disturbance modulation residual sequence, and drift stage switching point corresponding to each drift trend subsequence are used as the link crystal oscillator drift trend parameters between the mobile station and the i-th base station.

5. The measurement error correction method based on intelligent sensors and crystal oscillator frequency drift compensation according to claim 4, characterized in that, The process of establishing the perturbation-coupled grey drift prediction equation specifically includes: The perturbation modulation residual corresponding to the current ranging interaction cycle in the sub-segment residual observation sequence is used as the original response term; The gated background value corresponding to the current ranging interaction cycle in the gated background value sequence is used as the development background term, the sequence element corresponding to the current ranging interaction cycle in the sub-segment disturbance accumulation sequence is used as the disturbance accumulation value, and the disturbance accumulation value is used as the exogenous disturbance input term. The development coefficient, basic grey action quantity, and perturbation coupling coefficient are set as the model parameters to be determined; the original response term is summed with the product of the development coefficient and the development background term, and used as the left-hand side term of the perturbation coupling grey drift prediction equation; The product of the basic gray action quantity, the perturbation coupling coefficient, and the exogenous perturbation input term is summed and used as the right-hand side term of the perturbation coupling gray drift prediction equation. Set the left-hand side terms to be equal to the right-hand side terms, and establish the perturbation-coupled grey drift prediction equation.

6. The measurement error correction method based on intelligent sensors and crystal oscillator frequency drift compensation according to claim 1, characterized in that, Step four specifically includes: Read the link crystal drift trend parameters between the mobile station and the i-th base station, and determine the drift trend subsequence to which the current ranging interaction cycle belongs based on the drift phase switching point; Extract each predicted perturbation modulation residual from the drift trend subsequence to which the current ranging interaction cycle belongs, and calculate the mean and standard deviation of each predicted perturbation modulation residual to obtain the mean and standard deviation of the sub-segment residuals. Based on the mean and standard deviation of the sub-segment residuals, the predicted perturbation modulation residuals corresponding to the current ranging interaction cycle are Z-score standardized to obtain the standard score of the trend residuals. The standard score of the trend residual is input into the tanh function for compression mapping, and the compression mapping result is fused with the absolute value of the development coefficient corresponding to the current drift trend subsequence for trend enhancement to obtain the trend residual mapping value. Read the disturbance gating coefficient corresponding to the current ranging interaction cycle, and multiply the disturbance gating coefficient with the trend residual mapping value to obtain the gating compensation strength; The gating compensation intensity is multiplied by the predicted disturbance modulation residual corresponding to the current ranging interaction cycle to obtain the gating compensation offset. Based on the proportional relationship between the gating compensation offset and the initial time measurement value, the link time base drift compensation coefficient corresponding to the current ranging interaction cycle is generated. Arrange the link time base drift compensation coefficients according to the ranging interaction cycle to obtain the link time base drift compensation coefficient sequence between the mobile station and the i-th base station.

7. The measurement error correction method based on intelligent sensors and crystal oscillator frequency drift compensation according to claim 1, characterized in that, The drift integral allocation compensation specifically includes: Based on the link time-base drift compensation coefficient sequence between the mobile station and the i-th base station, extract the link time-base drift compensation coefficient corresponding to the current ranging interaction period; Read the total round-trip time difference, response waiting time difference, one-way distance measurement timestamp difference, and one-way response timestamp difference corresponding to the current distance measurement interaction cycle; The response waiting time difference is separated from the total round-trip time difference to obtain the closed-loop propagation difference term corresponding to the current ranging interaction cycle; Calculate the total time base compensation amount corresponding to the current ranging interaction cycle based on the closed-loop propagation differential term and the link time base drift compensation coefficient corresponding to the current ranging interaction cycle; Within the drift trend subsequence to which the current ranging interaction cycle belongs, the forward drift contribution value is obtained by multiplying the one-way timestamp difference of ranging with the predicted perturbation modulation residual and taking the absolute value; the return drift contribution value is obtained by multiplying the one-way timestamp difference of response with the predicted perturbation modulation residual and taking the absolute value. The forward drift contribution value and the return drift contribution value are accumulated and integrated respectively to obtain the accumulated value of the forward drift contribution and the accumulated value of the return drift contribution; The forward compensation allocation ratio and the return compensation allocation ratio are obtained by normalizing the cumulative integral values ​​of the forward drift contribution and the cumulative integral values ​​of the return drift contribution. The total time base compensation is multiplied by the forward compensation allocation ratio and the return compensation allocation ratio respectively to obtain the forward time base compensation and the return time base compensation. The forward time base compensation is used to differentially subtract and compensate the one-way timestamp difference of ranging to obtain the corrected one-way timestamp difference of ranging; the return time base compensation is used to differentially subtract and compensate the one-way timestamp difference of response to obtain the corrected one-way timestamp difference of response. For each ranging interaction cycle, the average of the correction ranging one-way timestamp difference and the correction response one-way timestamp difference is taken as the correction time measurement value for the corresponding ranging interaction cycle, and arranged to obtain the sequence of correction time measurement values ​​between the mobile station and the i-th base station.

8. The measurement error correction method based on intelligent sensors and crystal oscillator frequency drift compensation according to claim 1, characterized in that, Step six specifically includes: For each ranging interaction cycle between the mobile station and the i-th base station, the correction time measurement value of the corresponding ranging interaction cycle is multiplied by the signal propagation speed to obtain the candidate correction distance; Read the base station ranging configuration parameter table corresponding to the i-th base station. The base station ranging configuration parameter table includes the base station coverage radius, minimum effective ranging distance and maximum effective ranging distance. The minimum effective ranging distance is taken as the lower boundary of the propagation reachable distance, and the smaller value between the maximum effective ranging distance and the base station coverage radius is taken as the upper boundary of the propagation reachable distance. Based on the lower and upper boundaries of the propagation reachability distance, the candidate correction distance is constrained by the propagation reachability boundary, specifically as follows: If the candidate correction distance is less than the lower boundary of the reachable propagation distance, then the candidate correction distance is limited to the lower boundary of the reachable propagation distance. If the candidate correction distance is greater than the upper boundary of the propagation reachable distance, then the candidate correction distance is limited to the upper boundary of the propagation reachable distance; If the candidate correction distance is located between the lower boundary of the reachable distance and the upper boundary of the reachable distance, then the candidate correction distance is used as the correction measurement distance for the corresponding ranging interaction cycle. The corrected measurement distances are arranged according to the ranging interaction cycle to obtain the corrected measurement distance sequence between the mobile station and the i-th base station.

Images