Silicon gyroscope zero bias drift self-calibration method and system

Abstract

The application discloses a silicon optical gyro zero offset drift self-calibration method and system, collects zero offset, optical power, driving current, time sequence and a silicon optical gyro attitude identifier in a static calibration period, generates a drift candidate record by binding the same sampling trigger, calculates the zero offset jump by attitude grouping difference, generates an adaptive static stability threshold by taking the median jump in the group, divides static stability and disturbance data sets, screens a reference zero offset according to the optical power, driving current and time sequence change degree in the static stability record, finally corrects the zero offset sampling point by point by attitude, and outputs the self-calibration angular velocity. The scheme relies on multi-dimensional data layering linkage screening, can filter out the optical path and driving coupling drift, and greatly improves the long-term output consistency of the gyro.

Description

Technical Field

[0001] This invention relates to the field of industrial data acquisition technology, and in particular to a method and system for zero-bias drift self-calibration of silicon photonic gyroscopes. Background Technology

[0002] Industrial data acquisition technology involves the acquisition, storage, retrieval, and output of industrial equipment operating data, sensor sampling data, calibration parameters, and status records in computer systems. Silicon photonic gyroscopes continuously generate zero-bias drift errors during long-term operation. Conventional self-calibration methods only collect angular velocity and temperature time-series data for uniform compensation, failing to differentiate between stationary and disturbed operating conditions. Errors caused by fluctuations in optical power and changes in drive current are all mixed into the zero-bias sampling, ultimately resulting in continuous fluctuations in angular velocity output and poor reliability of the calibration benchmark. Conventional solutions use fixed judgment thresholds for sampling, with threshold parameters set manually, which cannot adaptively adjust with real-time calibration data. Furthermore, they only perform optical path compensation in a single dimension, without simultaneously combining it with the trend of drive current changes for joint verification. Differences in zero-bias baselines under different stationary postures are not grouped and isolated, making it difficult to effectively filter out multiple coupled drift interferences.

[0003] Existing zero-bias correction methods for silicon photonics and fiber optic gyroscopes are all used independently, either by fixing a threshold to filter abrupt changes or by compensating only for optical power, and simply storing data based on attitude. There is no adaptive dynamic threshold, nor is there a joint verification benchmark that incorporates the trends of optical power and drive current changes. Since the optical path and drive current drift of silicon photonics chips are mutually coupled, traditional single compensation methods cannot simultaneously filter out the two types of superimposed errors, and there is no layered, linked calibration approach in the industry. Summary of the Invention

[0004] This invention differs from traditional fixed threshold and single-dimensional compensation calibration methods. It solves the zero-bias drift problem of optical path and drive coupling in silicon photonics chips by using a four-layer linkage process: simultaneous triggering of multi-source data binding, adaptive static stability screening in attitude grouping, collaborative verification of optical path and drive current changes, and point-by-point correction based on attitude.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a method for zero-bias drift self-calibration of a silicon optical gyroscope, comprising the following steps: Acquire the zero-bias sampling sequence, optical power reading, drive current reading, timing tag, and static attitude identifier within the static calibration period. Based on the timing tag and the silicon photonics gyroscope static attitude identifier, generate drift candidate records under the same sampling trigger. For the drift candidate records under the same static attitude identifier, determine the zero-bias jump variable between two adjacent drift candidate records, take the median jump of this group to generate an adaptive static stability threshold, and select the median zero-bias jump variable according to the numerical order of the zero-bias jump variable. The static stability threshold is determined based on the median zero-bias jump variable. The drift candidate records that do not exceed the static stability threshold are written into the static stability record set, and the drift candidate records that exceed the static stability threshold are written into the disturbance record set. For the set of statically stable records, the reference zero bias is screened by comparing the coordination of adjacent optical power and drive current change directions, the device state matching degree between optical power readings and drive current readings is determined, and the reference zero bias record is selected from the set of statically stable records based on the device state matching degree. A zero-bias drift calibration record is generated based on the reference zero-bias record, and the zero-bias sampling sequence is corrected based on the zero-bias drift calibration record to generate self-calibrated angular velocity output data.

[0006] When acquiring the zero-bias sampling sequence, the optical power reading, the drive current reading, the timing tag, and the static attitude identifier within the static calibration period, the same sampling trigger signal is used as the binding reference. If the zero-bias sampling sequence, the optical power reading, or the drive current reading is missing at any sampling moment, the data group corresponding to the sampling moment shall be discarded. When all data groups exist at the same sampling time, the time sequence label and the stationary attitude identifier corresponding to the sampling time are written into the same drift candidate record.

[0007] When generating the drift candidate record by binding the time series tags in the order of the time series tags, the continuity of the time series tags of the two adjacent data groups to be bound is first checked. When the time tags of two adjacent data groups to be bound meet the preset sampling interval condition, the next data group to be bound is retained and the corresponding drift candidate record is generated. When the timing labels of two adjacent data groups to be bound do not meet the preset sampling interval condition, the next data group to be bound is marked as an interval abnormal data group, and writing the interval abnormal data group into the drift candidate record is prohibited.

[0008] When determining the zero-bias jump variable for the drift candidate records under the same stationary attitude identifier, the drift candidate records under the same stationary attitude identifier are first arranged in order according to the time sequence label; The zero-bias sampling sequences in two adjacent drift candidate records after arrangement are differentially processed to obtain the zero-bias jump variable corresponding to the interval between adjacent records; When the time-series label corresponding to any adjacent record interval is missing and the order is reversed, the generation of the zero-biased jump variable based on that adjacent record interval is prohibited.

[0009] When selecting the median zero-bias jump variable according to the numerical order of the zero-bias jump variable, the multiple zero-bias jump variables generated under the same static attitude identifier are first sorted in ascending order of absolute value. When the number of zero-biased jump variables after sorting is odd, the zero-biased jump variable located in the middle position is selected as the median zero-biased jump variable. When the number of zero-biased jump variables after sorting is even, the one with the larger value among the two adjacent zero-biased jump variables in the middle position is selected as the median zero-biased jump variable.

[0010] When determining the static stability threshold based on the median zero bias jump variable, the median zero bias jump variable is used as the zero bias change benchmark under static conditions. The threshold margin is determined based on the number of zero-bias jump variables under the same static attitude identifier. The zero-bias variation benchmark is superimposed with the threshold margin to obtain the static stability threshold used to screen the drift candidate records.

[0011] When a drift candidate record is written into the static stability record set according to the zero-bias jump variable not exceeding the static stability threshold, the zero-bias jump variable corresponding to each drift candidate record is compared with the static stability threshold; When the zero-bias jump variable is less than or equal to the static stability threshold, the corresponding drift candidate record is written into the static stability record set; When the zero-bias jump variable is greater than the static stability threshold, the corresponding drift candidate record is written into the disturbance record set, and the corresponding time series label is retained as the basis for disturbance backtracking.

[0012] When determining the device state matching degree between the optical power reading and the drive current reading for the static stable record set, the optical power reading and the drive current reading in the same drift candidate record are extracted according to the time series label; A reading coordination marker is generated based on the coordinated direction of changes in optical power and drive current readings corresponding to adjacent time tags; When the direction of change of the optical power reading is consistent with the direction of change of the drive current reading, the device state matching degree is increased; When the direction of change of the optical power reading is opposite to the direction of change of the drive current reading, or when either reading is missing, the device state matching degree is reduced and the corresponding drift candidate record is prohibited from being used as the reference zero bias record.

[0013] When generating the zero-bias drift calibration record based on the reference zero-bias record, the reference zero-bias value, the reference timing label, and the corresponding static attitude identifier are extracted from the reference zero-bias record. Calculate the difference between the baseline zero bias value and the sampling points of the zero bias sampling sequence under the same static attitude identifier to generate the calibration offset; Write the calibration offset, the reference timing label, and the stationary attitude identifier into the zero-bias drift calibration record; The zero-bias sampling sequence under the same stationary attitude identifier is corrected point by point according to the zero-bias drift calibration record, and the corrected zero-bias sampling sequence is converted into the self-calibrated angular velocity output data.

[0014] A silicon photonic gyroscope zero-bias drift self-calibration system, relying on a four-layer linked hardware unit architecture, sequentially completes same-source data binding, same-attitude median adaptive threshold division, collaborative verification of optical power and drive current changes, and attitude-specific zero-bias point-by-point correction. This achieves layered filtering of coupled drift at the hardware level, unlike traditional single-dimensional compensation hardware structures. It includes: Drift candidate record generation unit: acquires the zero-bias sampling sequence, optical power reading, drive current reading, timing tag and static attitude identifier within the static calibration period, and generates drift candidate records by binding them in the order of the timing tags; under the same sampling trigger, all readings are bound to generate drift candidate records; Median zero-bias jump variable determination unit: For the drift candidate records under the same static attitude identifier, determine the zero-bias jump variable between two adjacent drift candidate records, select the median zero-bias jump variable according to the numerical order of the zero-bias jump variable, and calculate the adaptive static stability threshold using the median jump variable of this group; Static and disturbance record division unit: Determine the static stability threshold based on the median zero-bias jump variable; write the drift candidate records whose zero-bias jump variable does not exceed the static stability threshold into the static stability record set; write the drift candidate records that exceed the static stability threshold into the disturbance record set. Reference zero-bias record selection unit: For the statically stable record set, the device matching degree is determined by the coordination degree of adjacent optical power and drive current change direction; the device state matching degree between optical power reading and drive current reading is determined, and a reference zero-bias record is selected from the statically stable record set according to the device state matching degree; Zero-bias drift self-calibration output unit: grouped according to different static postures, the zero-bias sampling sequence is corrected point by point; a zero-bias drift calibration record is generated according to the reference zero-bias record, the zero-bias sampling sequence is corrected according to the zero-bias drift calibration record, and self-calibrated angular velocity output data is generated.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, zero bias, optical path power, drive current, timing and attitude information are uniformly bound and stored within a single calibration cycle. The stable and disturbed datasets are adaptively divided based on the sampling difference results. The reliable zero bias reference is then selected by combining the synchronous change trends of optical power and drive current. Invalid samples that are mismatched with instantaneous disturbances and device states are filtered out, reducing the interference of multiple coupling drift on angular velocity output and improving the stability of long-term calibration output.

[0016] This invention is not a simple combination of three conventional methods. The three layers of logic—adaptive thresholding, optical flow collaborative filtering, and attitude-based correction—are mutually constrained and indispensable: dynamic thresholding adapts to slow zero drift, power and current synchronization eliminates coupling errors, and attitude-based calibration eliminates baseline differences. For fiber optics and discrete gyroscopes with silicon-free optical chips, there is no incentive in the art to combine these three methods to address the optical path-drive coupling drift problem. Using thresholding or optical compensation alone can only solve a single error and cannot simultaneously suppress multiple drifts; the overall stabilization effect cannot be achieved by superimposing individual methods. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is the main flowchart of the zero-bias drift self-calibration of the silicon photonic gyroscope of the present invention, which reflects the innovative logic of hierarchical screening, including data binding from the same source, adaptive threshold, and device collaborative matching. Figure 2 This is a schematic diagram illustrating the homogeneous binding effect of drift candidate records in this invention, demonstrating the innovative preprocessing logic of synchronous binding of multiple types of data and time-series verification under the same sampling trigger. Figure 3 This is a schematic diagram illustrating the effect of dividing the statically stable record set and the perturbation record set in this invention, demonstrating the core screening innovation of generating an adaptive statically stable threshold from the same attitude median jump. Figure 4 This is a schematic diagram of the device state matching degree screening effect of the present invention, demonstrating the unique verification mechanism based on the screening criteria of optical power and drive current change coordination. Figure 5 This is a schematic diagram of the correction effect of the zero-bias drift calibration record of the present invention, which illustrates the hierarchical calibration logic of correcting the zero bias point by point according to different static postures. Detailed Implementation

[0019] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0020] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments. Example 1

[0021] Please see Figures 1 to 5 This embodiment provides a method and system for self-calibrating zero-bias drift of a silicon photonic gyroscope. During the actual operation of the silicon photonic gyroscope in static calibration and continuously generating zero-bias, optical power, drive current and attitude-related data, the calibration data enters the zero-bias drift processing flow according to the same timing link, and completes static stability screening, device state matching and zero-bias drift correction within the calibration interval where the attitude remains unchanged. It includes a method and system for self-calibrating zero-bias drift of a silicon photonic gyroscope.

[0022] During the zero-bias drift calibration of silicon photonic gyroscopes, zero-bias sampling is easily affected by the combined effects of static attitude switching, transient disturbances, optical power drift, and drive current fluctuations. If all zero-bias samples within the calibration period are directly used for calibration, zero-bias jumps under disturbance conditions will enter the reference selection process, making the calibration record unable to stably reflect the device drift under static conditions. This embodiment first binds the zero-bias sampling sequence, optical power reading, drive current reading, timing label, and static attitude identifier at the same sampling time as drift candidate records. Then, it filters statically stable records based on the zero-bias jump variables of adjacent records and selects the reference zero-bias record by combining the device state matching degree between the optical power reading and the drive current reading. This ensures that subsequent zero-bias drift calibration records can be derived from data that meets the calibration requirements under both static and device states.

[0023] Traditional calibration methods use manually fixed screening thresholds, which cannot adapt to slow, static drift. They only compensate for optical power without verifying the benchmark based on the trend of drive current changes, and the zero baseline differences between different attitudes are not isolated, allowing disturbance sampling to be directly mixed into the benchmark data. This solution generates dynamic static stability thresholds independently based on attitude, relies on synchronous changes in the optical path and drive to screen reliable benchmarks, and removes disturbance data in layers, significantly reducing long-term zero bias fluctuations. Example 2

[0024] 1. Acquire the zero-bias sampling sequence, optical power reading, drive current reading, timing tag, and stationary attitude identifier within the static calibration period, and generate drift candidate records according to the order of the timing tags. The static calibration period refers to the continuous calibration phase in which the silicon photonics gyroscope is without active angular velocity input and used to acquire zero-bias drift data. The zero-bias sampling sequence refers to the ordered data formed by the zero-bias values ​​obtained within the static calibration period according to the sampling order. The optical power reading refers to the power reading corresponding to the optical path state of the silicon photonics gyroscope at the sampling time. The drive current reading refers to the current reading corresponding to the sampling time when maintaining the drive state of the silicon photonics gyroscope. The timing tag is a time sequence marker that identifies the sampling order. The stationary attitude identifier is a status marker used to distinguish different placement attitudes of the silicon photonics gyroscope during the static calibration period. The drift candidate record refers to a record composed of zero-bias sampling data, optical power reading, drive current reading, timing tag, and stationary attitude identifier under the same sampling trigger relationship, awaiting subsequent screening. When generating drift candidate records, the time sequence labels are first used as the record sorting basis. Data fields corresponding to the same sampling time are written into the same record, and then arranged according to the order of the time sequence labels to form a candidate record chain that can be directly called by subsequent zero-bias jump variable processing. Through this process, zero bias, optical power, drive current, and attitude identifiers are already bound to the same source before entering drift analysis, thereby avoiding incorrect matching based on data from different sampling times in subsequent steps. In this embodiment, the preset sampling interval is set to 10ms for example.

[0025] When acquiring the zero-bias sampling sequence, optical power reading, drive current reading, timing tag, and resting attitude identifier within the static calibration period, the same sampling trigger signal is used as the binding reference. The same sampling trigger signal refers to a trigger reference formed when the same sampling action occurs, indicating that the zero-bias sampling, optical power reading, and drive current reading belong to the same sampling moment. Based on this sampling trigger signal, the fields of each reading under the same trigger relationship are aligned to confirm that the current sampling point, optical power reading, drive current reading, timing tag, and resting attitude identifier in the zero-bias sampling sequence all point to the same calibration state. These are then treated as the same data group to be bound. The data group to be bound refers to a set of data that has not yet passed integrity and continuity checks but already has the same sampling trigger source.

[0026] When the zero-bias sampling sequence, optical power reading, and drive current reading are missing at any given sampling time, the data set corresponding to that sampling time is discarded. "Missing" means that the sampling trigger signal has been generated, but the corresponding field has not formed valid readable content, or the content of that field cannot establish a correspondence with the current timing label for the same sampling time. After discarding, this data set will not enter the drift candidate record, nor will it participate in the subsequent generation of zero-bias jump variables, static stability thresholds, and device state matching degrees. This process ensures that all data entering the candidate stage possesses the three basic readings of zero bias, optical power, and drive current, avoiding untraceable screening results in subsequent static stability assessments due to incomplete fields.

[0027] When multiple data sets exist at the same sampling time, the timing tag and stationary attitude identifier corresponding to each sampling time are written into the same drift candidate record. During writing, the drift candidate record includes at least the zero-bias sampling field, optical power field, drive current field, timing field, and attitude field. The zero-bias sampling field carries the zero-bias value corresponding to that sampling time, the optical power field carries the optical path reading, the drive current field carries the drive state reading, the timing field carries the record's position within the calibration period, and the attitude field carries the corresponding stationary attitude. After the drift candidate record is generated, it serves as input for subsequent sorting, differential processing, and state matching processing within the same attitude.

[0028] When generating drift candidate records by binding them according to the order of time sequence tags, the continuity of time sequence tags for two adjacent data groups to be bound is first checked. This continuity check determines whether two adjacent data groups to be bound maintain a normal sampling order based on a preset sampling interval condition. The preset sampling interval condition is a time sequence constraint given in advance by the sampling triggering rules of the static calibration period, used to identify sampling interruptions, repeated triggers, and sequence anomalies. This condition is entered into the candidate record generation process as a configurable decision field. Its source is the sampling rules of the static calibration period, and it is adjusted synchronously with the calibration rules during updates. It is subsequently used to determine whether the data groups to be bound are allowed to be written into the drift candidate record. For example, the sampling interval is set to a fixed 10ms, and this value is used as the preset sampling interval condition to determine whether the time sequence is continuous during verification.

[0029] When the time stamps of two adjacent data groups to be bound meet the preset sampling interval condition, the latter data group to be bound is retained and a corresponding drift candidate record is generated. Meeting the preset sampling interval condition means that the latter data group to be bound is sequentially connected to the former data group to be bound, and there are no interruptions, order reversals, or duplicate entries identified by continuity checks between them. After retaining the latter data group to be bound, it is converted into a drift candidate record, which, together with the previously retained record, forms the basis for processing subsequent adjacent record intervals.

[0030] When the time series labels of two adjacent data groups to be bound do not meet the preset sampling interval condition, the latter data group to be bound is marked as an interval aberration data group, and writing the interval aberration data group into the drift candidate record is prohibited. An interval aberration data group refers to a data group whose time series label cannot form a normal adjacent relationship with the preceding data group to be bound. After being marked as an interval aberration data group, this data group does not participate in the candidate record sorting, does not generate zero-bias jump variables, and does not enter the statically stable record set or the perturbed record set. This processing ensures that the candidate record chain maintains temporal continuity before entering adjacent difference, reducing the interference caused by temporal breaks in the zero-bias jump judgment.

[0031] In this embodiment, by using the same sampling trigger signal and timing continuity verification to process the multi-source readings within the static calibration period, the zero-bias sampling field, optical power field, drive current field, timing field, and attitude field can form a closed loop of recording at the same sampling time, thereby providing candidate data with consistent sources and reliable order for subsequent static stability screening.

[0032] 2. For drift candidate records under the same stationary attitude identifier, determine the zero-bias jump variable between two adjacent drift candidate records, and select the median zero-bias jump variable according to the numerical order of the zero-bias jump variables. The zero-bias jump variable refers to the change in the corresponding zero-bias value between two adjacent drift candidate records under the same stationary attitude identifier, used to describe whether the zero-bias change between adjacent sampling states is stable. The median zero-bias jump variable refers to the centrally representative change selected according to the change amplitude of the zero-bias jump variable under the same stationary attitude identifier, used as the benchmark for zero-bias change in the stationary state. During processing, drift candidate records are first grouped according to the stationary attitude identifier to avoid mixing zero-bias states under different attitudes in the same comparison process. Then, within each attitude group, they are arranged according to the time sequence label, and the zero-bias jump variables corresponding to the interval between adjacent records are generated. Finally, the median zero-bias jump variable is selected according to the sorting rules and output to the static stability threshold determination step.

[0033] When determining zero-bias jump variables for drift candidate records under the same stationary attitude identifier, the drift candidate records under the same stationary attitude identifier are first arranged sequentially according to their time series labels. Before arrangement, it is confirmed that each drift candidate record contains both an attitude field and a time series field. After arrangement, each record only establishes an adjacent record interval with time-adjacent records within the same attitude group, and does not establish zero-bias jump variable relationships with records in other attitude groups. Through this processing, the zero-bias jump variables only reflect the zero-bias changes under the same stationary attitude, and do not include state differences caused by attitude changes.

[0034] The zero-bias sampling sequences of two adjacent drift candidate records after arrangement are differentially processed to obtain the zero-bias jump variable corresponding to the interval between adjacent records. In this embodiment, the differential processing refers to using the zero-bias sampling fields of the preceding and following records as the processing object, identifying the direction and magnitude of the zero-bias change of the subsequent record relative to the preceding record, and writing the change result into the zero-bias jump variable field corresponding to the interval between adjacent records. The zero-bias jump variable field establishes a correspondence with the subsequent drift candidate record, which is used to determine whether the subsequent drift candidate record remains statically stable. If the first record in the same attitude group has no corresponding preceding record, the first record does not generate a zero-bias jump variable for the adjacent interval, and can be used to determine whether to participate in the benchmark candidate based on the generated adjacent record relationship during subsequent screening.

[0035] When any adjacent record interval has a missing time sequence label or is inverted, generating zero-biased jump variables based on that interval is prohibited. A missing time sequence label means that the time sequence field of at least one of the adjacent records cannot be read or its order cannot be confirmed; an inverted order means that the time sequence state of the later record cannot follow that of the previous record. In such cases, the adjacent record interval is excluded from the zero-biased jump variable generation process, and records related to that interval are not included in the static stability assessment, thus preventing abnormal time sequence relationships from being misidentified as zero-biased abrupt changes.

[0036] When selecting the median zero-bias jump variable according to its numerical order, multiple zero-bias jump variables generated under the same static attitude identifier are first sorted by their absolute values ​​from smallest to largest. In this embodiment, the absolute value is used to characterize the magnitude of the change in the zero-bias jump variable, without distinguishing the direction of change; it is only used to determine the relative order of the magnitude of the change. During sorting, the correlation between each zero-bias jump variable and its corresponding adjacent record interval and drift candidate record is preserved, so that the sorting result can not only determine the median zero-bias jump variable but also be used to backtrack to the corresponding time series label during subsequent filtering.

[0037] When the number of zero-bias jump variables after sorting is odd, the zero-bias jump variable located in the middle position is selected as the median zero-bias jump variable. The middle position refers to the position where, after arranging the variables in ascending order of change magnitude, there are the same number of candidates on both sides. The zero-bias jump variable corresponding to this position serves as the baseline output of the zero-bias change under the same static attitude, and is used in the static stability threshold determination step.

[0038] When the number of zero-bias jump variables after sorting is even, the variable with the larger value among the two adjacent zero-bias jump variables in the middle position is selected as the median zero-bias jump variable. This rule ensures that the median selection bias under an even number of variables can cover zero-bias jump variables with larger changes in the middle adjacent positions, avoiding the normal stationary fluctuations from being mistakenly classified as disturbance states due to an overly narrow selection in the center of the static stability threshold. After selection, the median zero-bias jump variable and its corresponding stationary attitude indicator are entered into the subsequent static stability threshold determination step.

[0039] In this embodiment, by sorting the temporal sequence within the same static posture and extracting adjacent zero-bias changes, the drift candidate records are processed so that the median zero-bias jump variable can be derived from the interval between adjacent records with consistent posture and effective temporal sequence, thereby providing a stable zero-bias change benchmark for the subsequent division of statically stable records and disturbed records.

[0040] 3. Determine the static stability threshold based on the median zero-bias jump variable. Drift candidate records whose zero-bias jump variables do not exceed the static stability threshold are written into the static stability record set, while drift candidate records exceeding the static stability threshold are written into the disturbance record set. The static stability threshold is the decision boundary used to determine whether a drift candidate record is in a static stable state. The static stability record set is the set of records that have passed the static stability threshold screening and can participate in the selection of the baseline zero-bias record. The disturbance record set is the set of records that are excluded from the baseline selection process because their zero-bias jump variables exceed the static stability threshold. The static stability threshold is formed by the median zero-bias jump variable and the threshold margin. The threshold margin is a screening tolerance field determined based on the number of zero-bias jump variables under the same static attitude identifier. Its source is the threshold margin rule preset in the static calibration process, which is used to maintain the stability of the screening boundary when the number of effective zero-bias jump variables is different.

[0041] When determining the static stability threshold based on the median zero-bias jump variable, the median zero-bias jump variable is used as the zero-bias change benchmark under the static state. The zero-bias change benchmark carries the representative amplitude of adjacent zero-bias changes under the same static attitude and serves as the basic field of the static stability threshold. This basic field is valid within the same static attitude identifier and is not called across attitudes; when different static attitudes generate median zero-bias jump variables, each attitude generates a corresponding zero-bias change benchmark.

[0042] The threshold margin is determined based on the number of zero-bias jump variables under the same static attitude identifier. The number of zero-bias jump variables refers to the number of zero-bias jump variables generated through effective adjacent record intervals and allowed to enter the sorting. The threshold margin is determined using a preset margin lookup table rule. The preset margin lookup table rule uses the textual state to which the number of zero-bias jump variables belongs as the lookup condition. The textual state includes states such as insufficient effective items, effective items meeting the requirements for static stability judgment, and effective items having a stable statistical basis. When there are insufficient effective items, the threshold margin adopts a conservative screening state and restricts the corresponding record from entering the selection of the baseline zero-bias record. When the effective items meet the requirements for static stability judgment, the threshold margin adopts a normal screening state. When the effective items have a stable statistical basis, the threshold margin adopts a convergent screening state. The above states are not disclosed with specific values, but are determined by the sampling integrity requirements and attitude grouping validity requirements in the static calibration rules. The threshold margin is divided into three levels: conservative, normal, and convergent, which automatically switches according to the number of effective jumps. The optical path intensity fluctuation judgment threshold is set to 3%; exceeding this value indicates that the optical power reading is abnormal.

[0043] The zero-bias variation baseline and the threshold margin are superimposed to obtain the static stability threshold used for screening drift candidate records. In this embodiment, superposition means combining the zero-bias variation baseline with the screening tolerance state corresponding to the threshold margin to form the final static stability judgment boundary. The static stability threshold is written into the threshold field corresponding to the same static attitude identifier and is only used for drift candidate records within that attitude group in subsequent comparisons. By forming static stability thresholds separately for each attitude, the impact of zero-bias baseline differences under different static attitudes on static stability screening can be avoided.

[0044] When writing drift candidate records whose zero-bias jump variables do not exceed the static stability threshold into the static stability record set, the zero-bias jump variable corresponding to each drift candidate record is compared with the static stability threshold. During the comparison, the zero-bias jump variable field corresponding to the drift candidate record is used as input, and the static stability threshold field corresponding to the same static attitude identifier is used as the decision boundary. If a drift candidate record has no available zero-bias jump variables, it is first checked whether it has not generated an adjacent interval because it is the first record in the attitude group; if so, it is only retained as a timing connection record and not directly used as a baseline zero-bias record; if there are no available zero-bias jump variables due to abnormal timing or missing fields, it is excluded from the static stability record set.

[0045] When the zero-bias jump variable is less than or equal to the static stability threshold, the corresponding drift candidate record is written to the static stability record set. During writing, the zero-bias sampling field, optical power field, drive current field, timing field, attitude field, and corresponding zero-bias jump variable field of the drift candidate record are retained for subsequent device state matching determination steps. This record maintains its original timing order in the static stability record set, allowing the direction of change of adjacent optical power readings and drive current readings to be continuously identified.

[0046] When the zero-bias jump variable exceeds the static stability threshold, the corresponding drift candidate record is written into the disturbance record set, and the corresponding time series label is retained as the basis for disturbance backtracking. The disturbance backtracking basis refers to the time series field used to locate the location of the disturbance, which can point to the excluded record during subsequent calibration record generation and calibration process verification. The records written into the disturbance record set do not participate in the selection of the reference zero-bias record, nor are they used as the reference source for generating calibration offsets, but their time series labels and attitude identifiers are retained to confirm the boundary relationship between the disturbance record and the static stability record.

[0047] In this embodiment, a static stability threshold is formed by combining the median zero-bias jump variable and the threshold margin. The zero-bias jump variables corresponding to drift candidate records are compared one by one according to the same attitude, ensuring that the statically stable record set can exclude zero-bias jumps caused by instantaneous disturbances. This allows subsequent selection of benchmark zero-bias records to be performed only within the stable data range. In this embodiment, the duration of a single calibration sampling window is 5 seconds, and the minimum number of effective sampling groups under a single attitude is no less than 20 groups.

[0048] 4. For the statically stable record set, determine the device state matching degree between the optical power reading and the drive current reading, and select a reference zero-bias record from the statically stable record set based on the device state matching degree. The device state matching degree refers to the state evaluation result generated within the statically stable record set based on the coordinated state between the direction of change of the optical power reading and the direction of change of the drive current reading, used to determine whether the record simultaneously meets the reference selection requirements for both the optical path state and the drive state. The reference zero-bias record refers to the reference record selected from the statically stable record set for generating the zero-bias drift calibration record. When determining the device state matching degree, different physical quantities are not directly subtracted. Instead, the direction of change of the optical power reading and the drive current reading relative to adjacent records is first identified within the same timing link, and then a reading coordination marker is formed based on whether the direction of change is consistent. In this embodiment, if the number of consecutive determinations of device state mismatch exceeds 5, the current attitude reference selection is abandoned.

[0049] When determining the device state matching degree between optical power readings and drive current readings for a set of statically stable records, optical power readings and drive current readings are extracted from the same drift candidate record according to the time series label. The extraction range is limited to the set of statically stable records, and adjacent statically stable records under the same static orientation identifier are used as the basis for comparison. Optical power readings and drive current readings retain their own physical meaning and are only used to generate their respective directions of change; no direct comparison of cross-dimensional values ​​is performed. If there are time series discontinuities in the set of statically stable records, no relationship between the reading change directions is established between the records on both sides of the discontinuity.

[0050] A reading coordination marker is generated based on the direction of change in optical power readings and drive current readings corresponding to adjacent timing tags. The direction of change in optical power readings refers to the directional state of the optical power reading in the later stable record relative to the previous stable record; the direction of change in drive current readings refers to the directional state of the drive current reading in the later stable record relative to the previous stable record. The reading coordination marker is a state field used to record whether the two types of change directions are consistent. The reading coordination marker includes consistent direction state, opposite direction state, and undeterminable reading state. A consistent direction state enters the device state matching degree increase process; an opposite direction state enters the device state matching degree decrease process; and an undeterminable reading state disallows the corresponding drift candidate record from becoming a reference zero-bias record.

[0051] When the direction of change in optical power reading is consistent with the direction of change in drive current reading, the device state matching degree is increased. In this embodiment, increasing the device state matching degree means adjusting the state evaluation result corresponding to the current record to a state level that is closer to the benchmark selection requirements, and associating this level with the current statically stable record. The state level can be determined by a preset level mapping rule, which originates from the judgment requirements for the coordinated changes in optical path state and drive state during the static calibration process of the silicon photonic gyroscope. The level boundaries are distinguished by textual states, including states such as suitable as a benchmark candidate, requiring further observation, and unsuitable as a benchmark candidate. This state level serves as the direct basis for subsequent benchmark zero-bias record selection.

[0052] When the direction of change in optical power reading is opposite to the direction of change in drive current reading, or when either reading is missing, the device state matching degree is reduced and the corresponding drift candidate record is prohibited from being used as the reference zero-bias record. Reducing the device state matching degree means adjusting the state evaluation result of the current record to a state level that does not meet the reference selection requirements. Any missing reading includes situations such as missing optical power field, missing drive current field, adjacent records failing to form a change direction, and unconfirmed recording timing. Drift candidate records that are prohibited from being used as reference zero-bias records can continue to be retained in the statically stable record set for timing tracing, but will not enter the reference zero-bias record output range.

[0053] When selecting a reference zero-bias record from the statically stable record set based on device state matching degree, records identified as having opposite directions or undeterminable readings by the reading coordination marker are first excluded. Then, records with suitable device state matching degrees are selected from the remaining records. If multiple records satisfying the reference candidate state exist under the same static attitude identifier, the temporal correlation of the records is maintained, and records connecting to continuous statically stable intervals in the statically stable record set are preferentially selected as the reference zero-bias record. If no record satisfying the reference candidate state exists under the same static attitude identifier, no reference zero-bias record is generated for that attitude group, and the corresponding attitude identifier is output as a missing reference state. Subsequent zero-bias drift calibration records do not generate calibration offsets based on this attitude group.

[0054] In this embodiment, by converting the optical power reading and the drive current reading into the direction of change and then forming the reading coordination mark, the device state matching degree can avoid direct comparison across dimensions, and can exclude statically stable records where the optical path state and the drive state are not coordinated from the reference selection, so that the reference zero-bias record can simultaneously satisfy the zero-bias static stability condition and the device state coordination condition.

[0055] 5. Generate a zero-bias drift calibration record based on the baseline zero-bias record, correct the zero-bias sampling sequence based on the zero-bias drift calibration record, and generate self-calibrated angular velocity output data. The zero-bias drift calibration record refers to the calibration basis consisting of the baseline zero-bias value, baseline timing label, static attitude identifier, and calibration offset extracted from the baseline zero-bias record. The self-calibrated angular velocity output data refers to the angular velocity output result calculated after the zero-bias sampling sequence has undergone drift correction. When generating the zero-bias drift calibration record, the baseline zero-bias record that has passed static stability screening and device state matching is used as the source to avoid perturbation records and records with incompatible states from entering the calibration offset generation process.

[0056] When generating a zero-bias drift calibration record based on a baseline zero-bias record, the baseline zero-bias value, baseline timing tag, and corresponding stationary attitude identifier are extracted from the baseline zero-bias record. The baseline zero-bias value refers to the zero-bias sampling field content in the baseline zero-bias record that can serve as a calibration reference under the same stationary attitude. The baseline timing tag refers to the timing field corresponding to the baseline zero-bias record, and the stationary attitude identifier is used to define the attitude range in which the baseline zero-bias value can be applied. After extraction, these three elements are used as the basic fields of the calibration record, enabling the calibration record to be traced back to the original baseline source.

[0057] The calibration offset is generated based on the difference between the reference zero-bias value and the zero-bias values ​​of the sampling points in the zero-bias sampling sequence. The calibration offset is an offset field used to correct the zero-bias state of a sampling point under the same stationary attitude identifier towards the reference zero-bias state. During generation, the same stationary attitude identifier is used as a constraint; only zero-bias sampling points within that attitude group are compared with their corresponding reference zero-bias values, and offset fields are not generated across attitudes. If a sampling point belongs to a disturbance record set, and its corresponding device state matching degree does not meet the reference candidate requirements, then that sampling point is not used as a reference source, but can be corrected based on the zero-bias drift calibration records already generated under the same attitude.

[0058] The calibration offset, reference timing label, and stationary attitude identifier are written to the zero-bias drift calibration record. The zero-bias drift calibration record also retains the calibration source status, which identifies that the calibration record originates from the reference zero-bias record in the statically stable record set. If a certain attitude group is in a reference-missing state, that attitude group is not written to a new zero-bias drift calibration record, and the corresponding stationary attitude identifier is retained as an uncalibrated state to prevent calibration offsets without a reference source from entering the output process.

[0059] The zero-bias sampling sequence under the same stationary attitude identifier is corrected point-by-point based on the zero-bias drift calibration record, and the corrected zero-bias sampling sequence is converted into self-calibrated angular velocity output data. During point-by-point correction, the zero-bias drift calibration record is retrieved using the stationary attitude identifier corresponding to the sampling point. The calibration offset consistent with that attitude identifier is used to process the zero-bias value of that sampling point, resulting in the corrected zero-bias sampling point. When converting to self-calibrated angular velocity output data, the corrected zero-bias sampling sequence is used as the source of zero-bias correction for the angular velocity output, ensuring that the angular velocity output is no longer directly affected by uncalibrated zero-bias drift. If a sampling point cannot be matched with a zero-bias drift calibration record under the same stationary attitude identifier, that sampling point is marked as an output state to be calibrated, and is not mixed with already calibrated output data.

[0060] In this embodiment, by extracting the reference zero-bias value, reference timing label, and stationary attitude identifier from the reference zero-bias record, and writing the calibration offset into the zero-bias drift calibration record, the zero-bias sampling sequence can be corrected point by point within a consistent attitude range, thereby forming self-calibrated angular velocity output data with drift compensation basis. Most existing gyroscope zero-bias correction schemes employ three independent methods: fixed threshold elimination of abrupt sampling, single optical path intensity compensation, and simple stationary acquisition. These three techniques are only used to address a single source of error and lack a processing approach that integrates median jump adaptive threshold, coordinated matching of optical power and drive current, and attitude-based grouping and screening.

[0061] Traditional fixed thresholds do not automatically adjust the judgment boundary according to the current calibration sampling data, and can only cope with instantaneous and drastic changes, and cannot adapt to the slow drifting conditions during the static process; simple optical power compensation only corrects the deviation of a single dimension of the optical path, and does not combine it with the synchronous fluctuation state of the driving circuit current for joint screening; static attitude acquisition only performs data classification and storage, and is not used to distinguish the zero bias reference difference under different attitudes.

[0062] Existing technologies do not offer any technical insights into combining three methods to simultaneously suppress the combined error of attitude switching drift and synchronous fluctuations in optical power / drive current. Relying on any single processing method alone cannot simultaneously eliminate the interference caused by these two types of coupled drift on the selection of the zero-bias reference. The layered linkage screening logic of this solution is not a simple superposition of conventional technologies in this field. Example 3

[0063] This embodiment also provides a silicon photonic gyroscope zero-bias drift self-calibration system. The system implements the aforementioned silicon photonic gyroscope zero-bias drift self-calibration method and includes a drift candidate record generation unit, a median zero-bias jump variable determination unit, a static stability and disturbance record division unit, a reference zero-bias record selection unit, and a zero-bias drift self-calibration output unit. Each module processes data generated within the same static calibration period sequentially. The records, variables, sets, and calibration records output by the previous module serve as inputs to the next module, forming a processing chain from candidate record generation to self-calibration angular velocity output data generation. The system does not change the technical meaning of each field in the above method; the modules only transmit the zero-bias sampling sequence, optical power reading, drive current reading, timing tag, static attitude identifier, drift candidate record, zero-bias jump variable, median zero-bias jump variable, static stability threshold, static stability record set, disturbance record set, device state matching degree, reference zero-bias record, and zero-bias drift calibration record as defined in the claims.

[0064] The drift candidate record generation unit acquires the zero-bias sampling sequence, optical power reading, drive current reading, timing tag, and resting attitude identifier within the static calibration period, and generates drift candidate records by binding them according to the order of the timing tags. This module receives the zero-bias sampling field, optical power field, drive current field, timing field, and attitude field formed under the same sampling trigger signal, and performs integrity and timing continuity checks on data groups at the same sampling time. If the zero-bias sampling sequence, optical power reading, or drive current reading is missing at any sampling time, the module discards the corresponding data group; if all data groups at the same sampling time exist, the module writes the timing tag and resting attitude identifier along with the corresponding reading into the same drift candidate record. If adjacent data groups to be bound meet the preset sampling interval condition, the module retains the next data group to be bound and generates a drift candidate record; if not, the module marks the next data group to be bound as an interval abnormal data group and prohibits it from being written into the drift candidate record. The drift candidate records output by this module are arranged according to the timing tags and then transmitted to the median zero-bias jump variable determination unit.

[0065] In this embodiment, the drift candidate record generation unit binds, excludes missing data, and verifies temporal continuity for data under the same sampling trigger relationship, so that the output drift candidate records can maintain field integrity and temporal validity, thereby providing a record basis that can be directly sorted and grouped for subsequent determination of zero-biased jump variables.

[0066] The median zero-bias jump variable determination unit is used to determine the zero-bias jump variable between two adjacent drift candidate records for the same stationary attitude identifier, and selects the median zero-bias jump variable according to the numerical order of the zero-bias jump variables. This module receives the drift candidate records output by the drift candidate record generation unit, first groups them according to the stationary attitude identifier, and then arranges the records within the same attitude group in order according to the time sequence label. For two records that are temporally adjacent and have valid fields, this module performs differential processing on the zero-bias sampling field to form the zero-bias jump variable for the corresponding interval between adjacent records; for adjacent record intervals with missing time sequence labels or reversed order, this module prohibits the generation of zero-bias jump variables. After generating multiple zero-bias jump variables, this module sorts them in ascending order of change magnitude, and selects the median zero-bias jump variable according to the odd or even state of the number of sorted items, where the middle position is selected in odd states, and the zero-bias jump variable with the larger change magnitude among the middle adjacent positions is selected in even states. This module transmits the median zero-bias jump variable and its corresponding stationary attitude identifier to the static stability and disturbance record division unit.

[0067] In this embodiment, the median zero-bias jump variable determination unit defines the difference and sorting rules of adjacent records within the same attitude, so that the output median zero-bias jump variable can reflect the zero-bias change benchmark under the condition of consistent static attitude, thereby providing attitude-corresponding input for the formation of static stability threshold.

[0068] The static stability and disturbance record segmentation unit determines the static stability threshold based on the median zero-bias jump variable. Drift candidate records whose zero-bias jump variables do not exceed the static stability threshold are written into the static stability record set, while those exceeding the threshold are written into the disturbance record set. This module receives the median zero-bias jump variable and its corresponding static attitude identifier from the median zero-bias jump variable determination unit. It uses the median zero-bias jump variable as the zero-bias change benchmark for that attitude and determines the threshold margin based on the number of zero-bias jump variables under the same static attitude identifier. The threshold margin is given by a preset margin lookup table rule. This rule determines the screening tolerance state based on the textual state corresponding to the number of valid zero-bias jump variables, and together with the zero-bias change benchmark, forms the static stability threshold. This module compares the zero-bias jump variable corresponding to each drift candidate record with the static stability threshold under the same attitude. When the zero-bias jump variable is less than or equal to the static stability threshold, the corresponding drift candidate record is written into the static stability record set; when the zero-bias jump variable is greater than the static stability threshold, the corresponding drift candidate record is written into the disturbance record set, and the corresponding time tag is retained as the basis for disturbance backtracking. This module outputs the static stability record set to the baseline zero-bias record selection unit, and retains the disturbance record set for exclusion and backtracking.

[0069] In this embodiment, the zero-bias jump variable, threshold margin and attitude identifier are bound together by the static and disturbance record division unit, so that the drift candidate record can be divided into static and disturbance records under the same attitude conditions, thereby preventing the disturbance record from entering the reference zero-bias record selection process.

[0070] The reference zero-bias record selection unit is used to determine the device state matching degree between optical power readings and drive current readings for a set of statically stable records, and selects reference zero-bias records from the set of statically stable records based on the device state matching degree. This module receives the statically stable record set output by the statically stable and disturbed record division unit, extracts the optical power readings and drive current readings from the same drift candidate record according to the time sequence label, and generates the direction of change of optical power readings and drive current readings based on adjacent time sequence labels under the same static attitude identifier. This module generates reading coordination markers based on the two types of change directions. When the change directions are consistent, the device state matching degree is increased and the corresponding record is marked as a suitable state for reference candidate status; when the change directions are opposite or either reading is missing, the device state matching degree is decreased, and the corresponding record is prohibited from being used as a reference zero-bias record. This module excludes records that do not meet the reading coordination requirements from the statically stable record set, selects records that meet the device state matching requirements as reference zero-bias records, and transmits the reference zero-bias records to the zero-bias drift self-calibration output unit.

[0071] In this embodiment, the reference zero-bias record selection unit first generates reading coordination markers and then determines the device state matching degree, so that the optical power reading and drive current reading participate in the screening in a coordinated relationship of changing direction, so that the reference zero-bias record not only comes from the static steady record set, but also meets the device state coordination requirements.

[0072] The zero-bias drift self-calibration output unit is used to generate a zero-bias drift calibration record based on the reference zero-bias record, correct the zero-bias sampling sequence based on the zero-bias drift calibration record, and generate self-calibrated angular velocity output data. This module receives the reference zero-bias record output by the reference zero-bias record selection unit, extracts the reference zero-bias value, reference timing tag, and stationary attitude identifier from it, calculates the difference between the reference zero-bias value and the sampling points of the zero-bias sampling sequence under the same stationary attitude identifier, generates a calibration offset, and writes the calibration offset, reference timing tag, and stationary attitude identifier into the zero-bias drift calibration record. When correcting the zero-bias sampling sequence, this module retrieves the zero-bias drift calibration record based on the stationary attitude identifier corresponding to the sampling point, corrects the zero-bias sampling points under the same attitude point by point, and converts the corrected zero-bias sampling sequence into self-calibrated angular velocity output data. If no reference zero-bias record is formed under a certain stationary attitude identifier, this module marks the corresponding sampling point as an output state to be calibrated, and does not mix it with the already calibrated angular velocity output data.

[0073] In this embodiment, the reference zero bias value, reference timing label, stationary attitude identifier and calibration offset are written into the same zero bias drift calibration record by the zero bias drift self-calibration output unit, so that the zero bias sampling sequence can be corrected point by point according to the calibration basis with consistent attitude, thereby forming self-calibrated angular velocity output data that is connected with stationary calibration data, static stability screening results and device state matching results.

[0074] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of protection of the technical solution.

Claims

1. A method for self-calibrating zero-bias drift of a silicon optical gyroscope, characterized in that, Includes the following steps: Acquire the zero-bias sampling sequence, optical power reading, drive current reading, timing tag, and static attitude identifier within the static calibration period. Based on the timing tag and the silicon photonics gyroscope static attitude identifier, generate drift candidate records under the same sampling trigger. For the drift candidate records under the same static attitude identifier, determine the zero-bias jump variable between two adjacent drift candidate records, take the median jump of this group to generate an adaptive static stability threshold, and select the median zero-bias jump variable according to the numerical order of the zero-bias jump variable. The static stability threshold is determined based on the median zero-bias jump variable. The drift candidate records that do not exceed the static stability threshold are written into the static stability record set, and the drift candidate records that exceed the static stability threshold are written into the disturbance record set. For the set of statically stable records, the reference zero bias is screened by comparing the coordination of adjacent optical power and drive current change directions, the device state matching degree between optical power readings and drive current readings is determined, and the reference zero bias record is selected from the set of statically stable records based on the device state matching degree. A zero-bias drift calibration record is generated based on the reference zero-bias record, and the zero-bias sampling sequence is corrected based on the zero-bias drift calibration record to generate self-calibrated angular velocity output data.

2. The silicon optical gyroscope zero-bias drift self-calibration method according to claim 1, characterized in that, When acquiring the zero-bias sampling sequence, the optical power reading, the drive current reading, the timing tag, and the static attitude identifier within the static calibration period, the same sampling trigger signal is used as the binding reference. If the zero-bias sampling sequence, the optical power reading, or the drive current reading are missing at any sampling moment, the data group corresponding to the sampling moment shall be discarded. When all data groups exist at the same sampling time, the time sequence label and the stationary attitude identifier corresponding to the sampling time are written into the same drift candidate record.

3. The silicon optical gyroscope zero-bias drift self-calibration method according to claim 2, characterized in that, When generating the drift candidate record by binding the time series tags in the order of the time series tags, the continuity of the time series tags of the two adjacent data groups to be bound is first checked. When the time-series labels of two adjacent data groups to be bound meet the preset sampling interval condition, the next data group to be bound is retained and the corresponding drift candidate record is generated. When the timing labels of two adjacent data groups to be bound do not meet the preset sampling interval condition, the next data group to be bound is marked as an interval abnormal data group, and writing the interval abnormal data group into the drift candidate record is prohibited.

4. The silicon optical gyroscope zero-bias drift self-calibration method according to claim 1, characterized in that, When determining the zero-bias jump variable for the drift candidate records under the same stationary attitude identifier, the drift candidate records under the same stationary attitude identifier are first arranged in order according to the time sequence label; The zero-bias sampling sequences in two adjacent drift candidate records after arrangement are differentially processed to obtain the zero-bias jump variable corresponding to the interval between adjacent records; When the time-series label corresponding to any adjacent record interval is missing and the order is reversed, the generation of the zero-biased jump variable based on that adjacent record interval is prohibited.

5. The silicon optical gyroscope zero-bias drift self-calibration method according to claim 4, characterized in that, When selecting the median zero-bias jump variable according to the numerical order of the zero-bias jump variable, the multiple zero-bias jump variables generated under the same static attitude identifier are first sorted in ascending order of absolute value. When the number of zero-biased jump variables after sorting is odd, the zero-biased jump variable located in the middle position is selected as the median zero-biased jump variable. When the number of zero-biased jump variables after sorting is even, the one with the larger value among the two adjacent zero-biased jump variables in the middle position is selected as the median zero-biased jump variable.

6. The silicon optical gyroscope zero-bias drift self-calibration method according to claim 1, characterized in that, When determining the static stability threshold based on the median zero bias jump variable, the median zero bias jump variable is used as the zero bias change benchmark under static conditions. The threshold margin is determined based on the number of zero-bias jump variables under the same static attitude identifier; the zero-bias change benchmark is superimposed with the threshold margin to obtain the static stability threshold used to screen the drift candidate records.

7. The silicon optical gyroscope zero-bias drift self-calibration method according to claim 6, characterized in that, When a drift candidate record is written into the static stability record set according to the zero-bias jump variable not exceeding the static stability threshold, the zero-bias jump variable corresponding to each drift candidate record is compared with the static stability threshold; When the zero-bias jump variable is less than or equal to the static stability threshold, the corresponding drift candidate record is written into the static stability record set; When the zero-bias jump variable is greater than the static stability threshold, the corresponding drift candidate record is written into the disturbance record set, and the corresponding time series label is retained as the basis for disturbance backtracking.

8. The silicon optical gyroscope zero-bias drift self-calibration method according to claim 1, characterized in that, When determining the device state matching degree between the optical power reading and the drive current reading for the static stable record set, the optical power reading and the drive current reading in the same drift candidate record are extracted according to the time series label; A reading coordination marker is generated based on the coordinated direction of changes in optical power and drive current readings corresponding to adjacent time tags; When the direction of change of the optical power reading is consistent with the direction of change of the drive current reading, the device state matching degree is increased; When the direction of change of the optical power reading is opposite to the direction of change of the drive current reading, or when either reading is missing, the device state matching degree is reduced and the corresponding drift candidate record is prohibited from being used as the reference zero bias record.

9. The silicon optical gyroscope zero-bias drift self-calibration method according to claim 1, characterized in that, When generating the zero-bias drift calibration record based on the reference zero-bias record, the reference zero-bias value, the reference timing label, and the corresponding static attitude identifier are extracted from the reference zero-bias record. Calculate the difference between the baseline zero bias value and the sampling points of the zero bias sampling sequence under the same static attitude identifier to generate the calibration offset; Write the calibration offset, the reference timing label, and the stationary attitude identifier into the zero-bias drift calibration record; The zero-bias sampling sequence under the same stationary attitude identifier is corrected point by point according to the zero-bias drift calibration record, and the corrected zero-bias sampling sequence is converted into the self-calibrated angular velocity output data.

10. A silicon optical gyroscope zero-bias drift self-calibration system, characterized in that, The system is used to implement the silicon optical gyroscope zero-bias drift self-calibration method according to any one of claims 1-9, comprising: Drift candidate record generation unit: acquires the zero-bias sampling sequence, optical power reading, drive current reading, timing tag and static attitude identifier within the static calibration period, and generates drift candidate records by binding them in the order of the timing tags; under the same sampling trigger, all readings are bound to generate drift candidate records; Median zero-bias jump variable determination unit: For the drift candidate records under the same static attitude identifier, determine the zero-bias jump variable between two adjacent drift candidate records, select the median zero-bias jump variable according to the numerical order of the zero-bias jump variable, and calculate the adaptive static stability threshold using the median jump variable of this group; Static and disturbance record division unit: Determine the static stability threshold based on the median zero-bias jump variable; write the drift candidate records whose zero-bias jump variable does not exceed the static stability threshold into the static stability record set; write the drift candidate records that exceed the static stability threshold into the disturbance record set. Reference zero-bias record selection unit: For the statically stable record set, the device matching degree is determined by the coordination degree of adjacent optical power and drive current change direction; the device state matching degree between optical power reading and drive current reading is determined, and a reference zero-bias record is selected from the statically stable record set according to the device state matching degree; Zero-bias drift self-calibration output unit: grouped according to different static postures, the zero-bias sampling sequence is corrected point by point; a zero-bias drift calibration record is generated according to the reference zero-bias record, the zero-bias sampling sequence is corrected according to the zero-bias drift calibration record, and self-calibrated angular velocity output data is generated.

Images