Wheel speed signal self-calibration method based on spoke topology prior

Abstract

This invention discloses a wheel speed signal self-calibration method based on spoke topology prior, belonging to the field of signal correction technology. By introducing spoke topology prior to generate an ideal pulse timing template, the method performs distribution deviation detection on the measured pulse sequence to accurately identify abnormal pulse segments. Based on this, it limits the candidate time interval for missing pulses. Furthermore, by combining the extraction of variation features of the original magnetic response signal and rhythm consistency constraint screening, the position of pulses not identified by the normal decision chain is structurally reconstructed. This enables effective correction of pulse distortion and pulse missingness to address the magnetic field distortion problem caused by the small number of spokes and the small tooth groove depth-to-width ratio. It avoids deviations or jumps in wheel speed calculation results, improves the accuracy of speed statistics, and enhances the stability and reliability of the vehicle control strategy.

Description

Technical Field

[0001] This invention relates to the field of signal correction technology, and more specifically to a self-calibration method for wheel speed signals based on spoke topology priors. Background Technology

[0002] With the continuous improvement of bicycle intelligence and electronic capabilities, wheel speed sensors have been widely used in scenarios such as cycling speed detection, mileage statistics, motion status perception, and overall vehicle control, becoming a crucial basic component for acquiring bicycle operation information. Existing wheel speed detection solutions typically collect magnetic response signals generated during wheel rotation using sensors, then perform pulse judgment, rhythm recognition, and speed conversion on the collected signals to output the corresponding wheel speed result. Furthermore, some solutions incorporate pulse compensation, anomaly detection, or auxiliary correction mechanisms to further improve the stability and accuracy of wheel speed detection. This approach offers advantages such as clear implementation paths, fast detection response, and ease of integration and deployment in small-volume sensors, thus gaining widespread application in current bicycle wheel speed detection.

[0003] However, in the actual application of the above wheel speed detection method, due to the small number of spokes and the small tooth groove depth-to-width ratio of bicycle wheels, magnetic field distortion with scene characteristics is more likely to be formed during wheel rotation, which in turn leads to problems such as pulse distortion and pulse loss in wheel speed signals. This anomaly directly caused by the characteristics of the spokes can cause deviations or even jumps in wheel speed calculation results, thereby affecting the accuracy of speed statistics and the stability of the vehicle control strategy. Summary of the Invention

[0004] The purpose of this invention is to address the problem mentioned in the background art that the rotation of bicycle wheels is more prone to the formation of scene-specific magnetic field distortions, which in turn leads to pulse distortion and pulse loss in the wheel speed signal. Therefore, this invention proposes a wheel speed signal self-calibration method based on spoke topology prior.

[0005] In terms of implementation, this invention provides a self-calibration method for wheel speed signals based on spoke topology priors, the method comprising: Obtain the topology parameters of the current wheel spoke structure and generate the corresponding ideal pulse timing template based on the topology parameters; acquire wheel speed sensing signals and perform pulse decision processing to obtain the measured pulse sequence arranged in time order; The measured pulse sequence is compared with the ideal pulse timing template to calculate the degree of distribution deviation between the two, and the abnormal pulse segment is determined based on the degree of distribution deviation. When identifying abnormal pulse segments, candidate time intervals corresponding to missing pulses are determined based on the differences between the ideal pulse timing template and the measured pulse sequence. The original magnetic response signal is acquired within the candidate time interval. The rate of change of the original magnetic response signal is analyzed to extract multiple change feature points. The preliminary candidate positions are obtained by preliminary screening based on the amplitude of the change feature points. Based on the ideal pulse timing template, the time interval relationship between each preliminary candidate position and the adjacent pulse and the rhythm continuity after insertion are calculated respectively. The preliminary candidate positions are then screened according to the calculation results, and the change feature points that satisfy the rhythm consistency constraint are retained as candidate pulse positions. The candidate pulse position is used as the reconstruction position of the missing pulse, and the reconstruction position is inserted into the measured pulse sequence to obtain the repaired pulse sequence. The wheel speed parameters are calculated based on the repaired pulse sequence, and the calibrated wheel speed signal is output.

[0006] Optionally, the steps for determining the abnormal pulse segment based on the degree of distribution deviation are as follows: The measured pulse sequence and the ideal pulse timing template are aligned and divided according to the same rotation period to obtain multiple corresponding detection periods; Within each detection cycle, the theoretical time interval sequence between adjacent pulses in the ideal pulse timing template and the actual time interval sequence between adjacent pulses in the measured pulse sequence are extracted respectively. The theoretical time interval sequence and the actual time interval sequence are normalized respectively to obtain the ideal interval distribution and the measured interval distribution within the corresponding detection period; Based on the ideal interval distribution and the measured interval distribution, the KL divergence of the measured pulse sequence relative to the ideal pulse time sequence template within the detection period is calculated as the degree of distribution deviation. The distribution deviation degree corresponding to multiple consecutive detection cycles is judged in time sequence. When the distribution deviation degree of three consecutive detection cycles is not less than the preset threshold, it is determined that the current wheel speed signal has a continuous abnormality. The time range corresponding to three consecutive detection cycles is defined as the abnormal pulse segment.

[0007] Optionally, the step of determining the candidate time interval corresponding to the missing pulse based on the difference between the ideal pulse timing template and the measured pulse sequence is as follows: Within the abnormal pulse segment, the theoretical pulse time sequence from the ideal pulse time sequence template and the actual pulse time sequence from the measured pulse sequence are extracted respectively. The actual pulse time sequence is aligned and matched with the theoretical pulse time sequence in chronological order to establish matched pulse pairs and identify the positions of theoretical pulses that have not been matched. Based on the matched pulse pairs, the theoretical time interval between adjacent theoretical pulses and the actual time interval between adjacent actual pulses are calculated respectively, and the correspondence between the theoretical time interval and the actual time interval is analyzed. In the actual time interval, the interval that has abnormally expanded relative to the corresponding theoretical time interval is identified, and the time range between two adjacent actual pulses corresponding to the abnormally expanded interval is determined as the interval where the missing pulse is located. Based on the theoretical pulse time corresponding to the interval where the missing pulse is located, the preset time tolerance range is extended forward and backward with the theoretical pulse time as the center to obtain the candidate time interval corresponding to the missing pulse.

[0008] Optionally, the step of obtaining preliminary candidate positions by preliminary screening based on the amplitude of the changing feature points is as follows: Extract the original magnetic response signal fragments within the candidate time interval; Preprocess the original magnetic response signal segment; The rate of change of the preprocessed magnetic response signal was analyzed to obtain the rate of change sequence. Extract local change peaks from the rate of change sequence as change feature points; Based on the change amplitude of each change feature point, retain the change feature points whose change amplitude is not lower than the preset amplitude threshold; The time locations corresponding to the retained feature points of change are determined as preliminary candidate locations.

[0009] Optionally, the steps of calculating the time interval relationship between each preliminary candidate position and its adjacent pulses, as well as the rhythmic continuity after insertion, and then filtering the preliminary candidate positions based on the calculation results to retain the change feature points that satisfy the rhythmic consistency constraint as candidate pulse positions are as follows: For each preliminary candidate position, the preliminary candidate position is inserted between its adjacent actual pulses to construct the corresponding candidate pulse sequence; Based on the candidate pulse sequence, the distribution of adjacent pulse time intervals before and after the insertion of the initial candidate position is calculated, and the rhythm distribution state before and after the insertion is obtained. Calculate the local rhythm entropy decrease index based on the rhythm distribution state before and after insertion; Based on the candidate pulse sequence, the distribution relationship between adjacent time intervals after the initial candidate position is inserted is calculated, and the time energy distribution index is calculated based on the distribution relationship; Calculate the corresponding rhythm consistency index based on the local rhythm entropy decrease index and the time energy distribution index; The rhythm consistency index is compared with a preset threshold. Initial candidate positions with a rhythm consistency index not lower than the preset threshold are retained as candidate pulse positions, and the remaining initial candidate positions are eliminated.

[0010] Optionally, the steps for calculating the local rhythm entropy decrease exponent based on the rhythm distribution state before and after insertion are as follows: Extract the actual time interval sequence before insertion, the candidate time interval sequence after insertion, and the corresponding theoretical time interval sequence within the local rhythm window where the initial candidate position is located; The two adjacent theoretical time intervals before and after the position corresponding to the missing pulse in the theoretical time interval sequence are merged to obtain the merged theoretical time interval sequence corresponding to the actual time interval sequence before insertion. Normalize the actual time interval sequence before insertion, the merged theoretical time interval sequence, the candidate time interval sequence after insertion, and the theoretical time interval sequence after insertion to obtain the corresponding normalized interval sequences. Calculate the absolute value of the logarithmic deviation between the normalized actual interval before insertion and the corresponding normalized theoretical interval to obtain the local distortion before insertion. Calculate the absolute value of the logarithmic deviation between the normalized candidate interval after insertion and the corresponding normalized theoretical interval to obtain the local distortion after insertion. Calculate the absolute value of the difference between adjacent local distortions before insertion and the absolute value of the difference between adjacent local distortions after insertion to obtain the distortion jump sequence before insertion and the distortion jump sequence after insertion. The distortion jump sequences before and after insertion are normalized to distortion jump distributions, and the corresponding normalized information entropy is calculated to obtain the local distortion propagation entropy before and after insertion. The local distortion before and after insertion are summed to obtain the total distortion before and after insertion, and the effective shrinkage is calculated based on the total distortion before and after insertion. The effective entropy decrease is calculated based on the local distortion propagation entropy before insertion and the local distortion propagation entropy after insertion. The local rhythm entropy decrease index is obtained by multiplying the effective contraction amount by the effective entropy decrease amount and taking the square root.

[0011] Optionally, the calculation steps for the time energy distribution index are as follows: Normalization processing is performed on the actual time interval sequence before insertion, the merged theoretical time interval sequence, the candidate time interval sequence after insertion, and the theoretical time interval sequence after insertion, respectively. The normalization processing is as follows: each time interval is divided by the sum of all time intervals in its sequence to obtain the proportion of each time interval in the corresponding sequence, thereby obtaining the actual time quality distribution before insertion, the theoretical time quality distribution before insertion, the candidate time quality distribution after insertion, and the theoretical time quality distribution after insertion, respectively. Based on the actual time quality distribution and the theoretical time quality distribution before insertion, the cumulative sums are calculated sequentially according to the rhythm boundary positions, and the absolute value of the difference between the two cumulative values ​​is calculated at each rhythm boundary to obtain the cumulative offset corresponding to each rhythm boundary before insertion; at the same time, based on the candidate time quality distribution and the theoretical time quality distribution after insertion, the cumulative offset corresponding to each rhythm boundary after insertion is calculated in the same way. The time energy distribution index is calculated based on the cumulative offset.

[0012] Optionally, the steps for calculating the time energy distribution index based on the cumulative offset are as follows: The cumulative offsets corresponding to each rhythm boundary before insertion are squared, and all squared results are summed sequentially to obtain the total residual time energy before insertion. The cumulative offsets corresponding to each rhythm boundary after insertion are squared, and all squared results are summed sequentially to obtain the total residual time energy after insertion. The total residual time energy before insertion is subtracted from the total residual time energy after insertion, and the difference is divided by the sum of the total residual time energy before and after insertion to obtain the degree of time energy change. When the degree of time energy change is not less than zero, it is used as the time energy release index; when the degree of time energy change is less than zero, the time energy release index is set to zero. Based on the cumulative offset corresponding to each rhythm boundary after insertion, the maximum value is selected, and the maximum value is divided by the sum of all cumulative offsets after insertion to obtain the maximum offset ratio. Then, the result of subtracting the maximum offset ratio is used as the time energy distribution balance index. The time energy release index is obtained by multiplying the time energy distribution equilibrium index by the time energy release index and then taking the square root of the product.

[0013] Optionally, the steps of calculating the corresponding rhythm consistency index based on the local rhythm entropy decrease index and the time energy distribution index, comparing the rhythm consistency index with a preset threshold, retaining preliminary candidate positions with a rhythm consistency index not lower than the preset threshold as candidate pulse positions, and eliminating the remaining preliminary candidate positions are as follows: The corresponding rhythm consistency index is obtained by adding the local rhythm entropy decrease index and the time energy distribution index. Each rhythm consistency index is compared with a preset threshold one by one; When the rhythm consistency index is not lower than the preset threshold, the preliminary candidate position corresponding to the rhythm consistency index is retained; When the rhythm consistency index is lower than a preset threshold, the preliminary candidate position corresponding to the rhythm consistency index is removed. The remaining preliminary candidate positions are determined as the candidate pulse position set.

[0014] Optionally, the steps of using the candidate pulse position as the reconstruction position of the missing pulse, inserting the reconstruction position into the measured pulse sequence to obtain the repaired pulse sequence, and calculating the wheel speed parameters based on the repaired pulse sequence to output the calibrated wheel speed signal are as follows: Extract the corresponding pulse time of each candidate pulse position in the ideal pulse timing template from the candidate pulse position set; The reconstruction order corresponding to each candidate pulse position is determined according to the order of the corresponding pulse times in the ideal pulse timing template; The pulses corresponding to each candidate pulse position are inserted into the measured pulse sequence between their adjacent actual pulses in the reconstruction order. After inserting the pulses corresponding to all candidate pulse positions, the positions of all inserted pulses are reordered according to the chronological order to obtain the repaired pulse sequence. The wheel speed parameters are calculated based on the repaired pulse sequence, and the calibrated wheel speed signal is output.

[0015] The beneficial effects of this invention are: This invention proposes a wheel speed signal self-calibration method based on spoke topology prior. By introducing spoke topology prior to generate an ideal pulse timing template, the method performs distribution deviation detection on the measured pulse sequence to accurately identify abnormal pulse segments. Based on this, it limits the candidate time interval for missing pulses. Furthermore, by combining the extraction of variation features of the original magnetic response signal and the screening of rhythm consistency constraints, the position of pulses not identified by the normal decision chain is structurally reconstructed. This method can effectively correct pulse distortion and pulse missingness, thus addressing the magnetic field distortion problem caused by the small number of spokes and the small tooth groove depth-to-width ratio. It avoids deviations or jumps in wheel speed calculation results, improves the accuracy of speed statistics, and enhances the stability and reliability of the vehicle control strategy. Attached Figure Description

[0016] Figure 1 The flowchart illustrates a wheel speed signal self-calibration method based on spoke topology prior, provided in an embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0018] The present invention provides a method for self-calibrating wheel speed signals based on spoke topology priors. See also Figure 1 , Figure 1 A flowchart illustrating a wheel speed signal self-calibration method based on spoke topology prior, provided in an embodiment of the present invention. The method includes the following steps: S1: Obtain the topology parameters of the current wheel spoke structure and generate the corresponding ideal pulse timing template based on the topology parameters; acquire wheel speed sensing signals and perform pulse decision processing to obtain the measured pulse sequence arranged in time order; S2: Compare the measured pulse sequence with the ideal pulse timing template, calculate the degree of distribution deviation between the two, and determine the abnormal pulse segment based on the degree of distribution deviation; S3: When determining abnormal pulse segments, the candidate time interval corresponding to the missing pulse is determined based on the difference between the ideal pulse timing template and the measured pulse sequence. S4: Obtain the corresponding original magnetic response signal within the candidate time interval, perform rate of change analysis on the original magnetic response signal to extract multiple change feature points, and perform preliminary screening based on the amplitude of the change feature points to obtain preliminary candidate positions; S5: Based on the ideal pulse timing template, calculate the time interval relationship between each preliminary candidate position and the adjacent pulse, as well as the rhythm continuity after insertion. Based on the calculation results, the preliminary candidate positions are screened, and the change feature points that satisfy the rhythm consistency constraint are retained as candidate pulse positions. S6: Use the candidate pulse position as the reconstruction position of the missing pulse, insert the reconstruction position into the measured pulse sequence to obtain the repaired pulse sequence, calculate the wheel speed parameters based on the repaired pulse sequence, and output the calibrated wheel speed signal.

[0019] Based on the wheel speed signal self-calibration method based on spoke topology prior provided in this invention, an ideal pulse timing template is generated by introducing spoke topology prior. The distribution deviation of the measured pulse sequence is detected to accurately identify abnormal pulse segments. On this basis, the candidate time interval of missing pulses is limited. Furthermore, by combining the extraction of the variation features of the original magnetic response signal and the rhythm consistency constraint screening, the pulse positions not identified by the normal decision chain are structurally reconstructed. This can effectively correct the pulse distortion and pulse missingness caused by the small number of spokes and the small tooth groove depth-to-width ratio, avoid deviations or jumps in wheel speed calculation results, improve the accuracy of speed statistics, and enhance the stability and reliability of the vehicle control strategy.

[0020] In one embodiment, S1: Obtain the topology parameters of the current wheel spoke structure and generate the corresponding ideal pulse timing template based on the topology parameters; acquire wheel speed sensing signals and perform pulse decision processing to obtain the measured pulse sequence arranged in time order, specifically: Topological parameters are structural parameters used to describe the geometric distribution of spokes in space and their corresponding magnetic response characteristics. They include at least the number of spokes, the angular intervals between adjacent spokes, the location and coverage of magnetic poles, etc. The process of generating an ideal pulse timing template based on topological parameters is essentially mapping the spatial distribution of spokes in one rotation into the rhythm of pulse occurrence in the time domain. That is, given the number of spokes and uniform or non-uniform angular distribution, one rotation is divided into several trigger intervals for corresponding spokes to pass through the sensor. Based on the physical mechanism of the magnetic field change that generates pulses when spokes pass through the sensor, the theoretical time interval between each pulse is determined, thus forming an ideal pulse position sequence arranged in chronological order. For example, when there are 24 spokes and they are evenly distributed, one rotation corresponds to 24 pulse positions with basically equal intervals. When there are local angular deviations or uneven magnetic pole coverage, the pulse intervals are adjusted according to the corresponding angular proportions. Finally, an ideal pulse timing template reflecting the pulse rhythm that the spoke structure should produce under ideal conditions is obtained. In addition, a wheel speed sensor located near the wheel continuously samples the actual magnetic response generated during wheel rotation to obtain a wheel speed sensing signal that changes over time. Since the spokes or magnetic triggering structures that cooperate with the wheel cause periodic changes in magnetic field strength as they pass through the sensor sensing area, the acquired raw signal will form a series of waveform fluctuations corresponding to the wheel rotation rhythm. The raw signal is then processed by pulse decision processing, which may include filtering and denoising the acquired signal, baseline correction, and threshold comparison, thereby converting the continuously changing analog magnetic response into identifiable discrete pulse events. Specifically, when the signal change exceeds a preset judgment condition, the corresponding moment is identified as the trigger moment of a valid pulse, and the time position of each valid pulse is recorded in chronological order, ultimately forming a measured pulse sequence arranged in chronological order.

[0021] In one embodiment, S2: The steps of comparing the measured pulse sequence with the ideal pulse timing template, calculating the degree of distribution deviation between the two, and determining the abnormal pulse segment based on the degree of distribution deviation are as follows: The measured pulse sequence and the ideal pulse timing template are aligned and divided according to the same rotation period to obtain multiple corresponding detection periods; Within each detection cycle, the theoretical time interval sequence between adjacent pulses in the ideal pulse timing template and the actual time interval sequence between adjacent pulses in the measured pulse sequence are extracted respectively. The theoretical time interval sequence and the actual time interval sequence are normalized respectively to obtain the ideal interval distribution and the measured interval distribution within the corresponding detection period; Specifically, the formula for normalization is: , In the formula, This represents the proportion of the ideal interval after normalization (the first in the ideal interval distribution). item), This represents the proportion of the normalized measured intervals (the first interval in the measured interval distribution). item); Indicates the current cycle number (e.g., lap 1, lap 2). This indicates which adjacent pulse interval is in the current period (e.g., the 1st interval, the 2nd interval). This represents the interval number used for summation, and is used to iterate through all intervals within the entire period. Indicates the first The number of intervals within a cycle (usually equal to the number of pulses minus 1). Indicates the first In the ideal pulse timing template within the nth cycle, the first... The theoretical time interval between adjacent pulses Indicates the first The measured pulse timing template within the period of the first period The actual time interval between adjacent pulses; Based on the ideal interval distribution and the measured interval distribution, the KL divergence of the measured pulse sequence relative to the ideal pulse time sequence template within this detection period is calculated as the degree of distribution deviation; the calculation formula is: In the formula, Indicates the first The degree of deviation of the measured pulse sequence from the ideal pulse timing template within each cycle; The distribution deviation degree corresponding to multiple consecutive detection cycles is judged in time sequence. When the distribution deviation degree of three consecutive detection cycles is not less than the preset threshold, it is determined that the current wheel speed signal has a continuous abnormality. The time range corresponding to three consecutive detection cycles is defined as the abnormal pulse segment.

[0022] It should be noted that the reason for using KL divergence as the degree of distribution deviation to determine whether there is an anomaly in the current wheel speed signal is that KL divergence can directly characterize the overall structural shift of the measured pulse interval distribution relative to the ideal pulse interval distribution. It focuses not on the local differences of a single interval, but on the degree of change of the entire rhythm distribution in a probabilistic sense. Therefore, when pulses are missing or distorted, the phenomenon of a certain interval being stretched or the proportion of adjacent intervals being unbalanced will be amplified at the distribution level, making the anomaly easier to identify stably. Compared with simple threshold comparison or single-point detection, KL divergence can maintain sensitivity and robustness to rhythm structure anomalies even in the presence of speed fluctuations and small jitters. On the other hand, using three consecutive detection cycles with values ​​not less than a preset threshold to determine anomalies is to avoid misjudging instantaneous noise, sporadic interference, or sampling jitter as genuine anomalies. For example, a pulse interval may be briefly abnormal within a single cycle due to road vibration or signal interference, but such anomalies usually do not persist across multiple consecutive cycles. However, when magnetic field distortion caused by wheel spoke structure characteristics leads to pulse loss, this anomaly is persistent and will be repeatedly manifested across multiple consecutive cycles. Therefore, determining anomalies across three consecutive cycles can effectively distinguish between sporadic fluctuations and structural anomalies, significantly reducing the false judgment rate while ensuring detection sensitivity. This ensures that subsequent pulse reconstruction is triggered only when a genuine anomaly exists, improving the accuracy of wheel speed calculation and avoiding unnecessary compensation interventions that could affect normal signals.

[0023] In one embodiment, S3: When determining abnormal pulse segments, the step of determining the candidate time interval corresponding to the missing pulse based on the difference between the ideal pulse timing template and the measured pulse sequence is as follows: Within the abnormal pulse segment, the theoretical pulse time sequence from the ideal pulse time sequence template and the actual pulse time sequence from the measured pulse sequence are extracted respectively. The actual pulse time sequence is aligned and matched with the theoretical pulse time sequence in chronological order to establish matched pulse pairs and identify the positions of theoretical pulses that have not been matched. Based on the matched pulse pairs, the theoretical time interval between adjacent theoretical pulses and the actual time interval between adjacent actual pulses are calculated respectively, and the correspondence between the theoretical time interval and the actual time interval is analyzed. In the actual time interval, the interval that has abnormally expanded relative to the corresponding theoretical time interval is identified, and the time range between two adjacent actual pulses corresponding to the abnormally expanded interval is determined as the interval where the missing pulse is located. Based on the theoretical pulse time corresponding to the interval where the missing pulse is located, the preset time tolerance range is extended forward and backward with the theoretical pulse time as the center to obtain the candidate time interval corresponding to the missing pulse.

[0024] It should be noted that the corresponding ideal pulse timing template segment and the measured pulse sequence segment are extracted from the abnormal pulse segment. The ideal pulse timing template segment reflects the theoretical time positions where each pulse in this abnormal segment should appear under the current spoke topology prior. The measured pulse sequence segment reflects the actual detected pulse time positions. Since it has been determined in step S2 that there is continuous abnormality in this segment, it is then defaulted that a certain theoretical pulse in the ideal template is not correctly detected in the measured sequence, or although there is a response, no effective pulse is formed, resulting in a discrepancy between the number of ideal pulses and the number of measured pulses.

[0025] Specifically, the theoretical pulse time sequence in the ideal pulse timing template within the abnormal segment and the actual pulse time sequence in the measured pulse sequence can be extracted separately first. Suppose there are M theoretical pulses in the abnormal segment of the ideal pulse timing template, and their time positions are ; there are L actual pulses in the abnormal segment of the measured pulse sequence, and their time positions are . Among them, represents the occurrence time of the m-th theoretical pulse in the ideal pulse timing template within the abnormal segment, and m represents the serial number of the theoretical pulse; represents the occurrence time of the n-th actual pulse in the measured pulse sequence within the abnormal segment, and n represents the serial number of the actual pulse; M represents the total number of theoretical pulses in the abnormal segment, and L represents the total number of actual pulses in the abnormal segment; When there is a pulse missing, usually L < M. The measured pulse sequence and the ideal pulse timing template are aligned in sequence to determine which theoretical pulse has no corresponding item in the measured sequence, and the time deviation between the ideal pulse and the measured pulse is calculated. Suppose the deviation between the ideal pulse time and its matched measured pulse time is , then there is: , where represents the time deviation of the r-th matched pulse, represents the r-th actual pulse time, represents the corresponding r-th theoretical pulse time, and r represents the serial number of the matched pulse pair. By examining the deviation changes of consecutive matched pulses, it can be judged whether the current abnormality is closer to "slight overall timing drift" or "local pulse missing". If the matching deviations before and after a certain position are continuously overall, but there is no one-to-one correspondence near a certain theoretical pulse, it indicates that this theoretical pulse is probably the missing pulse; To more directly judge the missing position, the adjacent pulse intervals are further compared. Suppose the theoretical time interval between the m-th and the (m + 1)-th theoretical pulses in the ideal pulse timing template is: , and suppose the actual time interval between the n-th and the (n + 1)-th actual pulses in the measured pulse sequence is: ,in, This represents the theoretical time interval between the m-th adjacent theoretical pulses in the ideal pulse timing template. This represents the actual time interval between the nth adjacent actual pulses in the measured pulse sequence. When a certain actual time interval is significantly larger than its corresponding theoretical time interval and close to the sum of two adjacent theoretical time intervals, it can be considered that the actual time interval likely contains an undetected pulse. To quantify the degree of aberration in the interval, the interval expansion ratio can be defined: ,in, This indicates the extent of expansion of the nth actual interval relative to the corresponding theoretical interval. represents the nth actual pulse interval, and represents the corresponding theoretical pulse interval. When A value significantly greater than 1 indicates that the actual interval has deviated from the normal rhythm; if A value close to 2 indicates that a pulse is likely missing within that interval, because the two theoretical intervals were combined into a larger actual interval in the actual measurement. For example, if the interval between two adjacent theoretical pulses in the ideal template is approximately 5 milliseconds, but a certain actual interval in the actual measurement becomes close to 10 milliseconds, it can be preliminarily determined that a pulse that should be located near the middle is missing in this 10-millisecond actual interval. Once the actual interval of this abnormal expansion is determined, the candidate time interval for the missing pulse can be identified. Specifically, the steps are: first, find the two actual pulse times preceding and following the actual interval of the abnormal expansion, and let them be... and If the missing pulse must lie between these two actual pulses, then, combining this with the corresponding theoretical pulse position in the ideal pulse timing template, a time range near the time of the missing theoretical pulse is taken as a candidate time interval. Let the starting time of this candidate time interval be denoted as... The termination time is recorded as Then it can be written as: , In the formula, Indicates the start time of the candidate time interval. Indicates the end time of the candidate time interval. This indicates the theoretical pulse time corresponding to the pulse that was determined to be missing. This represents the time tolerance range extending forward and backward around the theoretical pulse moment. This time tolerance range is used to take into account the effects of wheel speed fluctuations, sampling errors, and slight local rhythm drifts, ensuring that the candidate time interval is neither too wide, which would increase the burden of subsequent searches, nor too narrow, which would miss the physical traces of real missing pulses.

[0026] Therefore, by first extracting the corresponding segments of the ideal pulse timing template and the measured pulse sequence within the abnormal pulse region, and then identifying the actual interval where the abnormal expansion occurred by aligning the theoretical pulse time with the actual pulse time and comparing the intervals of adjacent pulses, the missing pulse can be located between two actual pulses. Finally, a candidate time interval is constructed using the corresponding theoretical pulse time as the center and a preset time tolerance range. The advantage of this approach is that it avoids blindly searching for missing pulses within the entire abnormal pulse region. Instead, it narrows the search range to a more precise and reasonable local time window based on the structural differences between the ideal pulse timing template and the measured pulse sequence. This provides a clear basis for subsequently extracting the original magnetic response signal, performing rate of change analysis, and screening candidate pulse locations within this candidate time interval.

[0027] In one embodiment, S4: The steps of acquiring the corresponding original magnetic response signal within the candidate time interval, performing rate-of-change analysis on the original magnetic response signal to extract multiple change feature points, and performing preliminary screening based on the amplitude of the change feature points to obtain preliminary candidate positions are as follows: Extract the original magnetic response signal fragments within the candidate time interval; Preprocess the original magnetic response signal segment; The rate of change of the preprocessed magnetic response signal was analyzed to obtain the rate of change sequence. Extract local change peaks from the rate of change sequence as change feature points; Based on the change amplitude of each change feature point, retain the change feature points whose change amplitude is not lower than the preset amplitude threshold; The time locations corresponding to the retained feature points of change are determined as preliminary candidate locations.

[0028] It should be noted that within the candidate time interval obtained in the previous step, the corresponding original magnetic response signal segment is extracted. Let the starting time of the candidate time interval be... The termination time is The original magnetic response signal acquired within this time interval is denoted as . ,in, This represents the amplitude of the original magnetic response as a function of time within the candidate time interval. Represents a time variable, and Since the raw magnetic response signal typically contains noise, minor jitter, or baseline drift, it can be preprocessed before rate-of-change analysis. This preprocessing, such as smoothing or denoising, yields a more stable signal curve. The preprocessed signal is denoted as... ,in, This represents the original magnetic response signal after denoising or smoothing. After obtaining the preprocessed original magnetic response signal, a rate of change analysis is performed on the signal to reflect the degree of local variation in the signal along the time axis. If a continuous form is used, the rate of change can be expressed as: ,in, This indicates the original magnetic response signal after preprocessing at time [time value missing]. rate of change at that point Indicates the signal relative to time The first derivative. Since practical systems generally use discrete sampling, under discrete sampling conditions, if N sampling points are collected within the candidate time interval, the first derivative is... The sampling time of each sampling point is denoted as The corresponding preprocessed signal amplitude is denoted as The rate of change can then be calculated using the difference method between adjacent sampling points: ,in, Indicates the first The discrete rate of change within each sampling interval can be calculated using the above method to obtain the rate of change sequence corresponding to each sampling position within the candidate time interval. Although the missing pulse is not recognized as a valid pulse by the normal decision chain, the corresponding real magnetic field disturbance usually still leaves a relatively obvious edge or transition process in the original magnetic response signal. Therefore, it is necessary to extract multiple change feature points from the rate of change sequence. To take into account both rising and falling changes, the absolute value of the rate of change can be taken first to obtain the change intensity sequence. Then, local peak points are identified in the change intensity sequence. If a certain position satisfies the condition that its change intensity is greater than the change intensity of the previous position and greater than the change intensity of the next position, then this position can be regarded as a local change peak point, that is, satisfying: and ,in, Indicates the first The intensity of change within each sampling interval Indicates the first The intensity of change within each sampling interval Indicates the first The intensity of change within a sampling interval. A location satisfying the above conditions indicates a local bulge in the rate of change before and after that point, meaning the signal exhibits a relatively prominent change process near that location, and therefore can be considered a change feature point. After extracting multiple variation feature points, they cannot all be directly used as the basis for subsequent pulse reconstruction, because these variation feature points may simultaneously contain magnetic response traces corresponding to the real missing pulse, spikes caused by noise, and pseudo-edges caused by local jitter. Therefore, preliminary screening based on the amplitude of the variation feature points is still required. Let the extracted feature points be... The change intensity corresponding to each change feature point is: Then we have: in, Indicates the first The amplitude of each characteristic point of change Indicates the first The position index of each change feature point in the change intensity sequence This indicates the intensity of change at that location. Set a threshold for filtering change amplitude. If a certain characteristic point of change satisfies: If the condition is not met, the change feature point is retained; otherwise, it is discarded. This represents the preset amplitude threshold used to distinguish effective change features from weak noise fluctuations. After this screening, the time positions corresponding to the retained change feature points can be used as preliminary candidate positions. Let the number of retained points after screening be... The initial candidate positions are denoted as These positions together constitute the initial candidate position set, where, Indicates the first The time coordinates of the initial candidate locations Indicates the sequence number of the preliminary candidate positions; For example, assuming the candidate time interval obtained in the previous step is 8 to 12 milliseconds, after extracting the original magnetic response signal within this interval, rate of change analysis reveals local change peaks around 9.1 milliseconds, 10.0 milliseconds, and 11.3 milliseconds. However, the change amplitude at 9.1 milliseconds is smaller, more closely resembling background noise fluctuations, while the change amplitudes at 10.0 milliseconds and 11.3 milliseconds are larger, indicating that these two positions are more likely to correspond to the actual magnetic response transition traces. Therefore, 10.0 milliseconds and 11.3 milliseconds can be retained as preliminary candidate positions for further rhythm consistency screening using an ideal pulse timing template. In other words, this step does not directly determine that "the missing pulse must be at 10.0 milliseconds," but rather further narrows the candidate range from the entire candidate time interval to a few positions with obvious physical change characteristics.

[0029] The advantage of doing this is that it allows for the direct extraction of potential missing pulse traces from the underlying raw magnetic response without relying on the final digital pulse decision result. Furthermore, it enables the early removal of obviously unreliable pseudo-feature points through amplitude screening, thereby providing a more accurate and reliable input basis for subsequent rhythm consistency verification based on the ideal pulse timing template.

[0030] In one embodiment, S5: Based on the ideal pulse timing template, the time interval relationship between each preliminary candidate position and adjacent pulses and the rhythm continuity after insertion are calculated respectively. The preliminary candidate positions are then screened according to the calculation results, and the change feature points that satisfy the rhythm consistency constraint are retained as candidate pulse positions. For each preliminary candidate position, the preliminary candidate position is inserted between its adjacent actual pulses to construct the corresponding candidate pulse sequence; Based on the candidate pulse sequence, the distribution of adjacent pulse time intervals before and after the insertion of the initial candidate position is calculated, and the rhythm distribution state before and after the insertion is obtained. Based on the rhythm distribution state before and after insertion, the local rhythm entropy decrease index is calculated to characterize the degree of influence of the insertion of the initial candidate position on the local rhythm order. Based on the candidate pulse sequence, the distribution relationship between adjacent time intervals after the initial candidate position is inserted is calculated, and the time energy distribution index is calculated according to the distribution relationship to characterize the rationality of the rhythm energy distribution after the initial candidate position is inserted. Calculate the corresponding rhythm consistency index based on the local rhythm entropy decrease index and the time energy distribution index; The rhythm consistency index is compared with a preset threshold. Initial candidate positions with a rhythm consistency index not lower than the preset threshold are retained as candidate pulse positions, and the remaining initial candidate positions are eliminated.

[0031] It should be noted that after obtaining multiple preliminary candidate positions in the previous step, the first step is to determine which pair of adjacent actual pulses each preliminary candidate position lies between. Then, this preliminary candidate position is temporarily inserted as a "pulse moment to be verified" between these adjacent actual pulses, thus forming a new candidate pulse sequence. Specifically, if the time positions of two adjacent actual pulses in the measured pulse sequence are the previous and next actual pulse moments, and the time coordinate of a certain preliminary candidate position falls between these two actual pulses, then the order of other actual pulses in the original measured pulse sequence remains unchanged. Only the preliminary candidate position is inserted between the previous and next actual pulse moments to obtain a sequence containing the preliminary candidate pulse moment. A new sequence is generated for each candidate position, which is the candidate pulse sequence corresponding to the initial candidate position. For example, if a local segment in the original measured pulse sequence consists of the first actual pulse, the second actual pulse, and the third actual pulse, and if an initial candidate position is located between the second and third actual pulses, then the initial candidate position is inserted between the second and third actual pulses to form a candidate pulse sequence of "first actual pulse - second actual pulse - initial candidate position - third actual pulse". Subsequently, based on the candidate pulse sequence, the local rhythm change, rhythm entropy change, and temporal energy distribution matching degree after insertion are calculated to determine whether the initial candidate position is truly suitable as a missing pulse position.

[0032] In one implementation, the steps for calculating the local rhythm entropy decrease exponent are as follows: Extract the actual time interval sequence before insertion within the local rhythm window where the preliminary candidate positions are located. Candidate time interval sequence after insertion and the corresponding theoretical time interval sequence ;in, This represents the actual time interval sequence obtained within the local rhythm window where the initial candidate position is located, before the initial candidate position is inserted; This represents each actual time interval in the actual time interval sequence before insertion; Indicates the number of actual time intervals before insertion; This represents the sequence of candidate time intervals formed after inserting preliminary candidate positions between adjacent actual pulses; This represents each candidate time interval in the candidate time interval sequence after insertion; Indicates the number of candidate time intervals after insertion; This represents the theoretical time interval sequence that corresponds one-to-one with the candidate time interval sequence after insertion. This represents each theoretical time interval in the theoretical time interval sequence; The two adjacent theoretical time intervals before and after the position corresponding to the missing pulse in the theoretical time interval sequence are merged, and the corresponding merged theoretical time interval is: This yields the merged theoretical time interval sequence corresponding to the actual time interval sequence before insertion. In the formula, This represents the combined theoretical time interval obtained by merging the two adjacent theoretical time intervals before and after the position corresponding to the missing pulse; This indicates the position number of the previous theoretical time interval corresponding to the missing pulse in the theoretical time interval sequence; This represents the merged theoretical time interval sequence corresponding to the actual time interval sequence before insertion. Normalize the actual time interval sequence before insertion, the merged theoretical time interval sequence, the candidate time interval sequence after insertion, and the theoretical time interval sequence after insertion to obtain the corresponding normalized interval sequences. Specifically, the normalized result of the actual time interval sequence before insertion is: ; The normalized result of merging the theoretical time interval sequences is: ; The normalized result of the candidate time interval sequence after insertion is: ; The normalized result of the theoretical time interval sequence after insertion is: ; in, Indicates the first insertion. The percentage of each actual time interval in the total number of actual time intervals before all insertions; Indicates the position number in the actual time interval sequence before insertion; Indicates the first The proportion of each theoretical merger time interval in the total sum of all theoretical merger time intervals; Indicates the position number in the theoretical time interval sequence; Indicates the number after insertion. The proportion of each candidate time interval in the total number of candidate time intervals after all insertions; Indicates the position number in the candidate time interval sequence after insertion; Indicates the number after insertion. The proportion of each theoretical time interval in the total theoretical time interval after all insertions; Indicates the position number in the theoretical time interval sequence after insertion; Calculate the absolute value of the logarithmic deviation between the normalized actual interval and the corresponding normalized theoretical interval before insertion to obtain the local distortion before insertion. The formula for calculation is: ,in, Indicates the first insertion. Local distortion at each location; This represents the corresponding position index of the normalized actual interval before insertion and the normalized merged theoretical interval; and the absolute value of the logarithmic deviation between the normalized candidate interval after insertion and the corresponding normalized theoretical interval is calculated to obtain the local distortion after insertion. The calculation formula is as follows: ;in, Indicates the number after insertion. Local distortion at each location; This indicates the corresponding position number between the normalized candidate margin and the normalized theoretical margin after insertion; Calculate the absolute value of the difference between adjacent local distortion values ​​before insertion and the absolute value of the difference between adjacent local distortion values ​​after insertion to obtain the distortion jump sequence before insertion and the distortion jump sequence after insertion; the terms in the distortion jump sequence before insertion are: The terms in the distorted jump sequence after insertion are: ;in, Indicates the first insertion. Distortion jump variables between adjacent positions; Indicates the adjacent position number in the local distortion sequence before insertion; Indicates the number after insertion. Distortion jump variables between adjacent positions; Indicates the adjacent position number in the local distortion sequence after insertion; The distortion jump sequences before and after insertion are normalized to distortion jump distributions, and the corresponding normalized information entropy is calculated to obtain the local distortion propagation entropy before and after insertion; specifically: Pre-insertion distortion jump distribution for: ; Post-insertion distortion jump distribution for: Pre-insertion local distortion propagation entropy for: ; Post-insertion local distortion propagation entropy for: ; In the formula, Indicates the first insertion. The proportion of each distorted jump variable in the total number of distorted jump variables before all insertions; Indicates the position number in the distorted jump sequence before insertion; Indicates the number after insertion. The percentage of each distorted jump variable in the total number of distorted jump variables after all insertions; Indicates the position number in the distorted jump sequence after insertion; This represents the local distortion propagation entropy before insertion; This represents the local distortion propagation entropy after insertion; the natural logarithm term in the denominator is used to normalize the information entropy result so that the result falls between zero and one. The total distortion before insertion is obtained by summing the local distortion before and after insertion. and total distortion after insertion The effective shrinkage is calculated based on the total distortion before and after insertion. The calculation formula is: In the formula, The degree of reduction in total distortion is used to ensure that the output of this step is not less than zero; the calculation formula is: ; The decrease in effective entropy is calculated based on the local distortion propagation entropy before and after insertion. The calculation formula is: In the formula, Let be the entropy decrease, used to ensure that the entropy decrease is not less than zero. The formula for calculation is: ; The square root of the product of the effective contraction and the effective entropy decrease is used to obtain the local rhythm entropy decrease index. The calculation formula is: Since both the effective contraction and the effective entropy decrease are not less than zero and are both limited to the range of zero to one, the local rhythm entropy decrease index is also a dimensionless quantity between zero and one.

[0033] It should be noted that the local rhythm entropy reduction index is essentially an evaluation metric used to measure whether the local pulse rhythm shifts from an abnormally disordered diffusion state to an ordered convergence state closer to the ideal pulse timing template after the insertion of a certain preliminary candidate position. It reflects both whether the total local rhythm distortion decreases after the insertion of the preliminary candidate position and whether the distortion propagation originally dispersed across multiple adjacent intervals becomes more concentrated and regular. Therefore, this index does not simply measure the magnitude of the time interval error, but rather the strength of the candidate position's ability to restore the local rhythm structure. A larger local rhythm entropy reduction index indicates a greater degree of contraction in the total distortion before and after the insertion of the preliminary candidate position. The higher the degree, the closer the local candidate time interval sequence after insertion is to the theoretical time interval sequence than before insertion. On the other hand, the greater the decrease in local distortion propagation entropy before and after insertion, the more effectively the local rhythm disorder and distortion spread caused by the missing pulse are compressed after the insertion of this preliminary candidate position. The local rhythm is restored from a more dispersed and chaotic state to a more stable and orderly state. Therefore, the greater the decrease in local rhythm entropy index, the more likely the preliminary candidate position is to correspond to the actual missing pulse, and the stronger the rhythm consistency between this position and the ideal pulse timing template. Consequently, the corresponding rhythm consistency index should also be larger. Conversely, if a certain preliminary... While inserting a candidate position formally adds a pulse, it doesn't necessarily improve the local interval relationships. Instead, it may exacerbate imbalances in certain local intervals. Alternatively, although the total local distortion may slightly change, the distortion propagation remains scattered across multiple adjacent positions, failing to achieve a clear orderly recovery. In such cases, the local rhythm entropy decrease index at that position will be small, indicating a weak reconstruction effect on the missing pulse and making it unsuitable as a final candidate pulse position. For example, within an abnormally elongated actual time interval, there might be two initial candidate positions. Inserting one of these positions might split the abnormal interval into two sub-intervals closer to the theoretical template, and also improve the surrounding intervals. As the distortion of the interval decreases simultaneously, the total distortion decreases and the distortion propagation entropy decreases. The corresponding local rhythm entropy decrease index is larger, indicating that this position is more likely to be the location of the true missing pulse. If the other position is inserted, it only splits the original abnormally elongated interval into an excessively short interval and an excessively long interval. In this case, the total local distortion is difficult to decrease effectively, and the distortion propagation will not converge significantly. Its local rhythm entropy decrease index is smaller. Therefore, from both the physical meaning and the rhythm structure restoration effect, the larger the local rhythm entropy decrease index, the stronger the restoration effect of the initial candidate position on the local rhythm order, the more consistent it is with the ideal rhythm template, and the more suitable it is as a candidate position for the true missing pulse.

[0034] In one embodiment, the steps for calculating the time energy distribution index are as follows: For each preliminary candidate position, extract the actual time interval sequence before insertion, the candidate time interval sequence after insertion, and the theoretical time interval sequence that corresponds one-to-one with the candidate time interval sequence after insertion within the local rhythm window where the preliminary candidate position is located. At the same time, merge the two adjacent theoretical time intervals before and after the position corresponding to the missing pulse in the theoretical time interval sequence to obtain the merged theoretical time interval sequence that corresponds one-to-one with the actual time interval sequence before insertion. Normalization processing is performed on the actual time interval sequence before insertion, the merged theoretical time interval sequence, the candidate time interval sequence after insertion, and the theoretical time interval sequence after insertion, respectively. The normalization processing is as follows: each time interval is divided by the sum of all time intervals in its sequence to obtain the proportion of each time interval in the corresponding sequence, thereby obtaining the actual time quality distribution before insertion, the theoretical time quality distribution before insertion, the candidate time quality distribution after insertion, and the theoretical time quality distribution after insertion, respectively. Based on the actual time quality distribution and the theoretical time quality distribution before insertion, the cumulative sums are calculated sequentially according to the rhythm boundary positions, and the absolute value of the difference between the two cumulative values ​​is calculated at each rhythm boundary to obtain the cumulative offset corresponding to each rhythm boundary before insertion; at the same time, based on the candidate time quality distribution and the theoretical time quality distribution after insertion, the cumulative offset corresponding to each rhythm boundary after insertion is calculated in the same way. The cumulative offsets corresponding to each rhythm boundary before insertion are squared, and all squared results are summed sequentially to obtain the total residual time energy before insertion. The cumulative offsets corresponding to each rhythm boundary after insertion are squared, and all squared results are summed sequentially to obtain the total residual time energy after insertion. Based on this, the total residual time energy before insertion is subtracted from the total residual time energy after insertion, and the difference is divided by the sum of the total residual time energy before and after insertion to obtain the degree of time energy change. When the degree of time energy change is not less than zero, it is used as the time energy release index; when the degree of time energy change is less than zero, the time energy release index is set to zero. Based on the cumulative offset corresponding to each rhythm boundary after insertion, the maximum value is selected, and the maximum value is divided by the sum of all cumulative offsets after insertion to obtain the maximum offset ratio. Then, the result of subtracting the maximum offset ratio is used as the time energy distribution balance index. Multiply the time energy release index by the time energy distribution equilibrium index and then take the square root of the product to obtain the time energy distribution index. Since both the time energy release index and the time energy distribution equilibrium index are values ​​between zero and one, the time energy distribution index is also a dimensionless quantity between zero and one.

[0035] It's important to note that the temporal energy distribution index is essentially an evaluation metric used to measure whether the time resources abnormally accumulated in local time intervals due to missing pulses are reasonably released and redistributed after the insertion of a preliminary candidate position. It doesn't focus on whether a single time interval simply increases or decreases, but rather on whether the proportion of time within each time interval in the local rhythm window after the insertion of the preliminary candidate position is closer to the time allocation structure corresponding to the ideal pulse timing template. Specifically, this index reflects whether the abnormally accumulated time resources have been effectively separated and released through the change in total residual energy before and after insertion. It also reflects whether the released time resources are still excessively concentrated near a certain rhythm boundary through the maximum offset percentage of the cumulative offset after insertion. Therefore, the larger the temporal energy distribution index, the more sufficient, balanced, and closer to the distribution state corresponding to the ideal rhythm template the local time resources are redistributed after the insertion of the preliminary candidate position. This means that the preliminary candidate position is more likely to be the position corresponding to the actual missing pulse. In this case, the rhythm consistency corresponding to this position is obviously stronger, so the rhythm consistency index should also be larger, not smaller. In other words, a larger temporal energy distribution index indicates that after insertion at that position, not only are the locally abnormally long intervals more reasonably divided, but the sub-intervals after the division do not form new obvious time accumulation points. The overall rhythm distribution is smoother, more natural, and more in line with the pulse rhythm that should be under the spoke topology prior. For example, if a certain actual time interval was obviously too long, it means that the time that should have been shared by two pulses was abnormally compressed into one interval. If a certain preliminary candidate position happens to be the actual missing pulse position, then after inserting at that position, this excessively long interval will be divided into two sub-intervals that are closer to the theoretical template. At this time, the total residual time energy before and after the insertion will decrease significantly. At the same time, the cumulative offset of each rhythm boundary after the insertion is no longer concentrated at a certain position, and the temporal energy distribution index will be larger. This indicates that at this position... The insertion of a pseudo-candidate position has a strong effect on local rhythm recovery. Conversely, if a pseudo-candidate position is inserted, although the original abnormally long interval is formally split into two sub-intervals, the time allocation after the split is not reasonable. One sub-interval may be too short and the other may still be too long, resulting in local time resources being transferred from one accumulation point to another. The total residual time energy decreases not significantly after insertion, and the offset after insertion may still be highly concentrated near a certain rhythm boundary. In this case, the corresponding time energy distribution index will be small. Therefore, from both the physical meaning and the rhythm structure recovery effect, the larger the time energy distribution index, the more reasonable the local time allocation formed after the insertion of the initial candidate position is, the more consistent it is with the ideal rhythm template, and the higher the credibility of the reconstruction of missing pulses. The final rhythm consistency index should also be increased.

[0036] In one implementation, the steps for calculating the corresponding rhythm consistency index based on the local rhythm entropy decrease index and the time energy distribution index are as follows: The corresponding rhythm consistency index is obtained by adding the local rhythm entropy decrease index and the time energy distribution index.

[0037] In one embodiment, the step of comparing the rhythm consistency index with a preset threshold, retaining preliminary candidate positions with a rhythm consistency index not lower than the preset threshold as candidate pulse positions, and eliminating the remaining preliminary candidate positions is as follows: Each rhythm consistency index is compared with a preset threshold one by one; When the rhythm consistency index is not lower than the preset threshold, the preliminary candidate position corresponding to the rhythm consistency index is retained; When the rhythm consistency index is lower than a preset threshold, the preliminary candidate position corresponding to the rhythm consistency index is removed. The remaining preliminary candidate positions are determined as the candidate pulse position set.

[0038] It should be noted that after calculating the corresponding rhythm consistency index for each preliminary candidate position, each rhythm consistency index is compared with a pre-set rhythm consistency judgment threshold according to the one-to-one correspondence with its corresponding preliminary candidate position. If the rhythm consistency index corresponding to a preliminary candidate position is not less than the preset threshold, it means that the insertion of this preliminary candidate position can restore the local rhythm order and make the local time resource allocation closer to the ideal pulse timing template, so the preliminary candidate position is retained. If the rhythm consistency index corresponding to a preliminary candidate position is less than the preset threshold, it means that the insertion of this preliminary candidate position can restore the local rhythm order and make the local time resource allocation closer to the ideal pulse timing template, so the preliminary candidate position is retained. If the repair effect on the local rhythm structure is insufficient, or the time allocation formed after insertion is still unreasonable, the initial candidate position is eliminated. After comparing all the initial candidate positions one by one, all the retained initial candidate positions are summarized to form a candidate pulse position set. For example, if there are three initial candidate positions in a certain candidate time interval, and the corresponding rhythm consistency indices are 0.82, 0.57 and 0.76, and the preset threshold is 0.7, then the positions corresponding to 0.82 and 0.76 that meet the condition of not less than 0.7 are retained, and the position corresponding to 0.57 is eliminated, and finally the candidate pulse position set in the candidate time interval is obtained.

[0039] In one embodiment, S6: The steps of using the candidate pulse position as the reconstruction position of the missing pulse, inserting the reconstruction position into the measured pulse sequence to obtain the repaired pulse sequence, calculating the wheel speed parameters based on the repaired pulse sequence, and outputting the calibrated wheel speed signal are as follows: Extract the corresponding pulse time of each candidate pulse position in the ideal pulse timing template from the candidate pulse position set; The reconstruction order corresponding to each candidate pulse position is determined according to the order of the corresponding pulse times in the ideal pulse timing template; The pulses corresponding to each candidate pulse position are inserted into the measured pulse sequence between their adjacent actual pulses in the reconstruction order. After inserting the pulses corresponding to all candidate pulse positions, the positions of all inserted pulses are reordered according to the chronological order to obtain the repaired pulse sequence. The wheel speed parameters are calculated based on the repaired pulse sequence, and the calibrated wheel speed signal is output.

[0040] It should be noted that after obtaining the candidate pulse position set in the previous step, each candidate pulse position is first matched with the corresponding theoretical pulse time in the ideal pulse timing template to clarify the sequential position of each candidate pulse position in the entire ideal rhythm. Then, according to the arrangement order of these corresponding theoretical pulse times in the ideal pulse timing template, the reconstruction order of each candidate pulse position is determined. On this basis, the pulses corresponding to each candidate pulse position are sequentially inserted into the measured pulse sequence between their corresponding adjacent actual pulses. After all insertions are completed, all the inserted pulse positions are rearranged according to the time sequence of each pulse to obtain a complete repaired pulse sequence. Subsequently, the time interval between adjacent pulses is recalculated based on the repaired pulse sequence, and the wheel speed parameters are calculated accordingly. Finally, the calibrated wheel speed signal is output. For example, if the original measured pulse sequence in an abnormal segment only detects the previous and next actual pulses, and a candidate pulse position is selected in the middle through the aforementioned steps, the pulse corresponding to the candidate pulse position is inserted between the previous and next actual pulses to fill in the missing pulses back into the measured pulse sequence, making the wheel speed result calculated later closer to the true wheel speed state.

[0041] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention should still fall within the scope of the claims of the present invention.

Claims

1. A wheel speed signal self-calibration method based on spoke topology prior, characterized in that, Includes the following steps: Obtain the topology parameters of the current wheel spoke structure and generate the corresponding ideal pulse timing template based on the topology parameters; acquire wheel speed sensing signals and perform pulse decision processing to obtain the measured pulse sequence arranged in time order; The measured pulse sequence is compared with the ideal pulse timing template to calculate the degree of distribution deviation between the two, and the abnormal pulse segment is determined based on the degree of distribution deviation. When identifying abnormal pulse segments, candidate time intervals corresponding to missing pulses are determined based on the differences between the ideal pulse timing template and the measured pulse sequence. The original magnetic response signal is acquired within the candidate time interval. The rate of change of the original magnetic response signal is analyzed to extract multiple change feature points. The preliminary candidate positions are obtained by preliminary screening based on the amplitude of the change feature points. Based on the ideal pulse timing template, the time interval relationship between each preliminary candidate position and the adjacent pulse and the rhythm continuity after insertion are calculated respectively. The preliminary candidate positions are then screened according to the calculation results, and the change feature points that satisfy the rhythm consistency constraint are retained as candidate pulse positions. The candidate pulse position is used as the reconstruction position of the missing pulse, and the reconstruction position is inserted into the measured pulse sequence to obtain the repaired pulse sequence. The wheel speed parameters are calculated based on the repaired pulse sequence, and the calibrated wheel speed signal is output.

2. The wheel speed signal self-calibration method based on spoke topology prior according to claim 1, characterized in that, The steps for determining abnormal pulse segments based on the degree of distribution deviation are as follows: The measured pulse sequence and the ideal pulse timing template are aligned and divided according to the same rotation period to obtain multiple corresponding detection periods; Within each detection cycle, the theoretical time interval sequence between adjacent pulses in the ideal pulse timing template and the actual time interval sequence between adjacent pulses in the measured pulse sequence are extracted respectively. The theoretical time interval sequence and the actual time interval sequence are normalized respectively to obtain the ideal interval distribution and the measured interval distribution within the corresponding detection period; Based on the ideal interval distribution and the measured interval distribution, the KL divergence of the measured pulse sequence relative to the ideal pulse time sequence template within the detection period is calculated as the degree of distribution deviation. The distribution deviation degree corresponding to multiple consecutive detection cycles is judged in time sequence. When the distribution deviation degree of three consecutive detection cycles is not less than the preset threshold, it is determined that the current wheel speed signal has a continuous abnormality. The time range corresponding to three consecutive detection cycles is defined as the abnormal pulse segment.

3. The wheel speed signal self-calibration method based on spoke topology prior according to claim 1, characterized in that, The steps to determine the candidate time interval corresponding to the missing pulse based on the difference between the ideal pulse timing template and the measured pulse sequence are as follows: Within the abnormal pulse segment, the theoretical pulse time sequence from the ideal pulse time sequence template and the actual pulse time sequence from the measured pulse sequence are extracted respectively. The actual pulse time sequence is aligned and matched with the theoretical pulse time sequence in chronological order to establish matched pulse pairs and identify the positions of theoretical pulses that have not been matched. Based on the matched pulse pairs, the theoretical time interval between adjacent theoretical pulses and the actual time interval between adjacent actual pulses are calculated respectively, and the correspondence between the theoretical time interval and the actual time interval is analyzed. In the actual time interval, the interval that has abnormally expanded relative to the corresponding theoretical time interval is identified, and the time range between two adjacent actual pulses corresponding to the abnormally expanded interval is determined as the interval where the missing pulse is located. Based on the theoretical pulse time corresponding to the interval where the missing pulse is located, the preset time tolerance range is extended forward and backward with the theoretical pulse time as the center to obtain the candidate time interval corresponding to the missing pulse.

4. The wheel speed signal self-calibration method based on spoke topology prior according to claim 1, characterized in that, The steps for obtaining preliminary candidate locations by initially screening based on the amplitude of the changing feature points are as follows: Extract the original magnetic response signal fragments within the candidate time interval; Preprocess the original magnetic response signal segment; The rate of change of the preprocessed magnetic response signal was analyzed to obtain the rate of change sequence. Extract local change peaks from the rate of change sequence as change feature points; Based on the change amplitude of each change feature point, retain the change feature points whose change amplitude is not lower than the preset amplitude threshold; The time locations corresponding to the retained feature points of change are determined as preliminary candidate locations.

5. The wheel speed signal self-calibration method based on spoke topology prior according to claim 1, characterized in that, The steps for calculating the time interval relationship between each preliminary candidate position and its adjacent pulses, as well as the rhythmic continuity after insertion, and then filtering the preliminary candidate positions based on the calculation results, retaining the change feature points that satisfy the rhythmic consistency constraint as candidate pulse positions, are as follows: For each preliminary candidate position, the preliminary candidate position is inserted between its adjacent actual pulses to construct the corresponding candidate pulse sequence; Based on the candidate pulse sequence, the distribution of adjacent pulse time intervals before and after the insertion of the initial candidate position is calculated, and the rhythm distribution state before and after the insertion is obtained. Calculate the local rhythm entropy decrease index based on the rhythm distribution state before and after insertion; Based on the candidate pulse sequence, the distribution relationship between adjacent time intervals after the initial candidate position is inserted is calculated, and the time energy distribution index is calculated based on the distribution relationship; Calculate the corresponding rhythm consistency index based on the local rhythm entropy decrease index and the time energy distribution index; The rhythm consistency index is compared with a preset threshold. Initial candidate positions with a rhythm consistency index not lower than the preset threshold are retained as candidate pulse positions, and the remaining initial candidate positions are eliminated.

6. The wheel speed signal self-calibration method based on spoke topology prior according to claim 5, characterized in that, The steps for calculating the local rhythm entropy decrease exponent based on the rhythm distribution before and after insertion are as follows: Extract the actual time interval sequence before insertion, the candidate time interval sequence after insertion, and the corresponding theoretical time interval sequence within the local rhythm window where the initial candidate position is located; The two adjacent theoretical time intervals before and after the position corresponding to the missing pulse in the theoretical time interval sequence are merged to obtain the merged theoretical time interval sequence corresponding to the actual time interval sequence before insertion. Normalize the actual time interval sequence before insertion, the merged theoretical time interval sequence, the candidate time interval sequence after insertion, and the theoretical time interval sequence after insertion to obtain the corresponding normalized interval sequences. Calculate the absolute value of the logarithmic deviation between the normalized actual interval before insertion and the corresponding normalized theoretical interval to obtain the local distortion before insertion. Calculate the absolute value of the logarithmic deviation between the normalized candidate interval after insertion and the corresponding normalized theoretical interval to obtain the local distortion after insertion. Calculate the absolute value of the difference between adjacent local distortions before insertion and the absolute value of the difference between adjacent local distortions after insertion to obtain the distortion jump sequence before insertion and the distortion jump sequence after insertion. The distortion jump sequences before and after insertion are normalized to distortion jump distributions, and the corresponding normalized information entropy is calculated to obtain the local distortion propagation entropy before and after insertion. The local distortion before and after insertion are summed to obtain the total distortion before and after insertion, and the effective shrinkage is calculated based on the total distortion before and after insertion. The effective entropy decrease is calculated based on the local distortion propagation entropy before insertion and the local distortion propagation entropy after insertion. The local rhythm entropy decrease index is obtained by multiplying the effective contraction amount by the effective entropy decrease amount and taking the square root.

7. The wheel speed signal self-calibration method based on spoke topology prior according to claim 6, characterized in that, The calculation steps for the time energy distribution index are as follows: Normalization processing is performed on the actual time interval sequence before insertion, the merged theoretical time interval sequence, the candidate time interval sequence after insertion, and the theoretical time interval sequence after insertion, respectively. The normalization processing is as follows: each time interval is divided by the sum of all time intervals in its sequence to obtain the proportion of each time interval in the corresponding sequence, thereby obtaining the actual time quality distribution before insertion, the theoretical time quality distribution before insertion, the candidate time quality distribution after insertion, and the theoretical time quality distribution after insertion, respectively. Based on the actual time quality distribution and the theoretical time quality distribution before insertion, the cumulative sums are calculated sequentially according to the rhythm boundary positions, and the absolute value of the difference between the two cumulative values ​​is calculated at each rhythm boundary to obtain the cumulative offset corresponding to each rhythm boundary before insertion; at the same time, based on the candidate time quality distribution and the theoretical time quality distribution after insertion, the cumulative offset corresponding to each rhythm boundary after insertion is calculated in the same way. The time energy distribution index is calculated based on the cumulative offset.

8. The wheel speed signal self-calibration method based on spoke topology prior according to claim 7, characterized in that, The steps for calculating the time energy distribution index based on the cumulative offset are as follows: The cumulative offsets corresponding to each rhythm boundary before insertion are squared, and all squared results are summed sequentially to obtain the total residual time energy before insertion. The cumulative offsets corresponding to each rhythm boundary after insertion are squared, and all squared results are summed sequentially to obtain the total residual time energy after insertion. The total residual time energy before insertion is subtracted from the total residual time energy after insertion, and the difference is divided by the sum of the total residual time energy before and after insertion to obtain the degree of time energy change. When the degree of time energy change is not less than zero, it is used as the time energy release index; when the degree of time energy change is less than zero, the time energy release index is set to zero. Based on the cumulative offset corresponding to each rhythm boundary after insertion, the maximum value is selected, and the maximum value is divided by the sum of all cumulative offsets after insertion to obtain the maximum offset ratio. Then, the result of subtracting the maximum offset ratio is used as the time energy distribution balance index. The time energy release index is obtained by multiplying the time energy distribution equilibrium index by the time energy release index and then taking the square root of the product.

9. The wheel speed signal self-calibration method based on spoke topology prior according to claim 5, characterized in that, Based on the local rhythm entropy decrease index and the time energy distribution index, the corresponding rhythm consistency index is calculated. The rhythm consistency index is compared with a preset threshold, and preliminary candidate positions with a rhythm consistency index not lower than the preset threshold are retained as candidate pulse positions. The remaining preliminary candidate positions are then eliminated as follows: The corresponding rhythm consistency index is obtained by adding the local rhythm entropy decrease index and the time energy distribution index. Each rhythm consistency index is compared with a preset threshold one by one; When the rhythm consistency index is not lower than the preset threshold, the preliminary candidate position corresponding to the rhythm consistency index is retained; When the rhythm consistency index is lower than a preset threshold, the preliminary candidate position corresponding to the rhythm consistency index is removed. The remaining preliminary candidate positions are determined as the candidate pulse position set.

10. The wheel speed signal self-calibration method based on spoke topology prior according to claim 1, characterized in that, The steps for using candidate pulse positions as reconstruction positions for missing pulses, inserting these reconstruction positions into the measured pulse sequence to obtain the repaired pulse sequence, and calculating wheel speed parameters based on the repaired pulse sequence to output the calibrated wheel speed signal are as follows: Extract the corresponding pulse time of each candidate pulse position in the ideal pulse timing template from the candidate pulse position set; The reconstruction order corresponding to each candidate pulse position is determined according to the order of the corresponding pulse times in the ideal pulse timing template. The pulses corresponding to each candidate pulse position are inserted into the measured pulse sequence between their adjacent actual pulses according to the reconstruction order. After inserting the pulses corresponding to all candidate pulse positions, the positions of all inserted pulses are reordered according to the chronological order to obtain the repaired pulse sequence. The wheel speed parameters are calculated based on the repaired pulse sequence, and the calibrated wheel speed signal is output.

Images