A method for calibrating electromagnetic interference data of a hemispherical resonant gyroscope

Abstract

This invention belongs to the field of inertial device testing and calibration technology, specifically involving a calibration method for electromagnetic interference data of a hemispherical resonator gyroscope. The method establishes a unified time reference between the gyroscope controller and the electromagnetic interference device, recording closed-loop state quantities and angular velocity outputs. Before each interference condition, a closed-loop state preset is executed and monitored until the baseline convergence criterion is met. Target interference is applied, and the data is divided into a transition segment, a disturbed steady-state segment, and a disturbance removal recovery segment based on the event and disturbance convergence criterion. In the steady-state segment, a consistency distance is calculated to screen for effective windows, and the mean steady-state angular velocity is calculated. Memory parameters are fitted and calculated for the recovery segment. A memory contamination index is constructed based on the memory parameters, fitting residuals, and consistency distance. When the index exceeds a threshold, the data is discarded, corrected, or retested, and the calibration parameters are output. This reduces the dispersion of calibration results under the same electromagnetic interference condition, significantly improving the reproducibility and reliability of the calibration parameters.

Description

Technical Field

[0001] This invention belongs to the field of inertial device testing and calibration technology, specifically relating to a method for calibrating electromagnetic interference data of a hemispherical resonant gyroscope. Background Technology

[0002] As a high-precision inertial device, the stable operation of a hemispherical resonator gyroscope (HRG) relies on a multi-closed-loop collaborative control architecture: 1. Amplitude Control Loop (AGC): Its core function is to maintain the stability of the resonator's vibration amplitude and avoid amplitude drift that leads to angular velocity measurement errors. Core components include an integrator (used to accumulate amplitude deviation and output control signals), a low-pass filter (to filter out high-frequency noise interference), an amplitude detection module, and a drive amplifier circuit; 2. Phase-Locked Loop / Frequency Tracking Loop (PLL or equivalent loop): Used to track the resonator's vibration phase and natural frequency in real time, ensuring the gyroscope always operates near the resonance point. Core components include a phase detector, a loop filter (containing an integrator), and a frequency tracking loop (PLL or equivalent loop). The system consists of: 1) a voltage-controlled oscillator (or frequency control module) and a status register (which stores the current frequency control word); 2) a quadrature error suppression loop: which suppresses orthogonal coupling interference caused by structural asymmetry and assembly errors during the vibration of the resonator. The core components are a quadrature error detector, an integrator, and a compensation signal generation module. The output of the integrator directly affects the accuracy of the quadrature error suppression; 3) a force feedback loop (FRB): which improves the dynamic measurement range and output linearity of the gyroscope. It detects the vibration imbalance signal of the resonator, accumulates it through an integrator, and outputs a feedback force to offset the disturbance caused by the external angular velocity. The core components include an imbalance detector, a high-gain integrator, and a feedback drive module.

[0003] Each of the aforementioned closed loops includes components with "state retention characteristics," such as integrators, low-pass filters, and state registers. When the HRG faces operating condition changes (such as the on / off state of electromagnetic interference (EMI)) or external disturbances (including EMI, temperature fluctuations, vibration, etc.), the internal states of these components will exhibit "residual effects": for example, the output value of the integrator will not immediately return to zero as the disturbance disappears, the filtering state of the low-pass filter needs to undergo a slow decay process, and the compensation amount, control word, and other parameters stored in the state register will also retain the influence of historical operating conditions. This residual effect manifests as a "slow decay characteristic" on a time scale, namely the "state memory and hysteresis recovery effect" of the closed loop. This effect raises two key issues: first, under the same disturbance conditions, if the controller's initial state is different (such as different residual states from the previous operating condition), its response characteristics will differ; second, after disturbance removal, the closed-loop system needs to undergo a specific time constant to recover to the undisturbed baseline state, during which the output data will be "contaminated" by historical states and cannot truly reflect the inherent performance of the gyroscope.

[0004] The existing HRG EMI calibration process generally adopts the model of "applying interference - waiting for a fixed time - statistical output of results". It relies heavily on statistical quantities such as the mean and variance of angular velocity output as the calibration basis. Its limitations are specifically reflected in the following three aspects: 1. Subjectivity in steady-state determination: The process implicitly assumes that "a fixed waiting time (e.g., 30s, 60s) is sufficient for the system to reach steady state". However, in reality, the intensity and frequency of EMI vary, and the convergence speed of the closed-loop system varies significantly (e.g., convergence time may exceed 100s under strong interference, while it may only require 10s under weak interference). A fixed waiting time cannot adapt to all interference scenarios, resulting in some tests where the system begins statistical analysis before reaching steady state, and the data includes transition noise; 2. "Hidden drift" in steady-state data. 3. "Drift" problem: Even if the system exhibits "apparent steady state" (the range of angular velocity output fluctuation is reduced), the internal state quantities of the closed loop (such as the output of the AGC integrator and the FRB compensation quantity) may still have slow drift (although the drift rate is low, the cumulative effect will affect the measurement accuracy). The existing process does not monitor the closed loop state quantities and only counts the angular velocity output, which causes the drift segment data to be mixed into the steady state statistics, introducing system error; 4. "Cross-condition contamination" of memory effect: After the disturbance is removed, the "memory residue" of the closed loop system is not cleared and directly enters the next condition test, which causes the initial state of the next condition to contain the residual influence of the previous condition. The initial state of repeated measurements under the same disturbance conditions is inconsistent, which ultimately results in large dispersion and non-reproducibility of calibration results. Summary of the Invention

[0005] The technical problem to be solved by this invention is that the existing HRG EMI calibration process has poor consistency and low reproducibility of calibration results.

[0006] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution: A method for calibrating hemispherical resonant gyroscopes using electromagnetic interference data is proposed. This method is used to calibrate the performance of hemispherical resonant gyroscopes under electromagnetic interference environments, and includes the following steps: Step 1: Establish a unified time reference between the hemispherical resonant gyroscope controller and the electromagnetic interference device, and configure a data recording structure for recording the closed-loop state quantities and angular velocity output of the hemispherical resonant gyroscope; Step 2: Before testing under any electromagnetic interference condition, perform closed-loop state standardization processing; closed-loop state standardization processing includes: performing state preset operation on the closed-loop control loop of the hemispherical resonant gyroscope, and monitoring the statistical characteristics of the closed-loop state quantity in real time until the closed-loop state quantity meets the preset baseline convergence criterion. Step 3: Apply target electromagnetic interference; record the operating condition switching events based on a unified time reference, and divide the time response data of the hemispherical resonator gyroscope into a transition segment, a disturbed steady-state segment, and a disturbance removal recovery segment based on the disturbed convergence criterion; Step 4: Perform same-state sampling within the disturbed steady-state segment; same-state sampling includes: calculating the consistency distance between the closed-loop state variables and the reference state for each sampling window, selecting valid windows whose consistency distance meets a preset threshold, and calculating the mean steady-state angular velocity based on the data from the valid windows; Step 5: Identify memory parameters based on the descrambling recovery segment; memory parameter identification includes: fitting the response data of the descrambling recovery segment using a mathematical model to calculate the memory parameters characterizing the closed-loop memory effect; Step 6: Perform memory removal processing; memory removal processing includes: constructing a quantified memory contamination index based on memory parameters, fitting residuals, and consistency distance; when the memory contamination index exceeds the confidence threshold, performing at least one of the following actions on the test data of the current working condition: elimination, numerical correction, or retesting, and outputting the calibration parameters after memory removal processing.

[0007] Compared with existing technologies, this invention has the following advantages and beneficial effects: It introduces a closed-loop quality control link in the electromagnetic interference calibration process, encompassing "time event alignment, unified closed-loop initial state, segmented disturbed process, steady-state sampling, disturbance removal and recovery modeling, and pollution quantification and handling." By establishing a unified time reference between the controller and the electromagnetic interference equipment and synchronously recording operating condition switching events, the application / removal of interference and the gyroscope output data achieve traceable alignment. Before each operating condition test, the closed-loop control loop is pre-set to a specific state, and convergence is confirmed using baseline convergence criteria, thus reducing the impact of residual integral / filtering states from the previous operating condition on the current measurement. After interference is applied, the response data is objectively divided into transition segments based on the disturbed convergence criteria. The system employs two distinct phases: a disturbed steady-state phase and a de-disturbed recovery phase. This avoids the mixing of transient samples with steady-state statistics. For the disturbed steady-state phase, the consistency distance between the closed-loop state variables and the reference state is calculated based on the sampling window, and valid windows are selected to ensure that the data used to calculate the mean steady-state angular velocity is within a consistent closed-loop state range. Simultaneously, for the de-disturbed recovery phase, a mathematical model is used to extract memory parameters characterizing the closed-loop memory effect, and the recovery quality is characterized by the fitting residuals. Furthermore, a memory contamination index is constructed based on the memory parameters, fitting residuals, and consistency distance, and compared with a reliable threshold. When contamination exceeds the limit, de-memorization measures such as removal, numerical correction, or retesting are implemented. This reduces the dispersion of calibration results under the same electromagnetic interference conditions, significantly improving the reproducibility and reliability of calibration parameters. It also provides a stable and consistent data foundation for subsequent electromagnetic susceptibility assessment and compensation. Attached Figure Description

[0008] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings: Figure 1This is a schematic diagram of a method for calibrating electromagnetic interference data of a hemispherical resonant gyroscope, provided in an embodiment of the present invention. Detailed Implementation

[0009] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. The illustrative embodiments and descriptions of this invention are for illustrative purposes only and are not intended to limit the invention. The embodiments described below are some, but not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0010] In the following description, numerous specific details are set forth to provide a thorough understanding of the invention. However, it will be apparent to those skilled in the art that these specific details are not necessary to practice the invention. In other embodiments, well-known structures, materials, or methods are not specifically described to avoid obscuring the invention. Unless otherwise specified, the materials, instruments, and reagents used in the following embodiments are commercially available. Unless otherwise specified, the techniques used in the embodiments are conventional methods well known to those skilled in the art.

[0011] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0012] Example: To address the issues of poor consistency, low reproducibility, and undiagnostic contamination states in calibration results of hemispherical resonator gyroscopes (HRGs) under electromagnetic interference (EMI) environments due to the "closed-loop state memory and hysteresis recovery effect" (hereinafter referred to as the "memory recovery effect") in the closed-loop control circuit, this paper provides a calibration method for HRG EMI data. Through a full-process calibration process including "closed-loop state standardization, event synchronization segmentation, same-state sampling, memory parameter identification, and contamination quantification," this method reduces the standard deviation of repeated measurements of calibration parameters under the same EMI conditions, quantifies the degree of memory contamination, and ensures the traceability of calibration results (recording contamination index, handling strategies, and other reliable information). This fundamentally suppresses calibration errors caused by the closed-loop state memory and hysteresis recovery effect, solving the technical problems of inconsistent and unreproducible calibration results in existing methods.

[0013] This method is applicable to calibration under single or multiple EMI conditions (point-by-point scanning is performed for multiple conditions). The core process includes: 1. Establish a unified time base and configure a data recording structure to achieve time synchronization between HRG and EMI devices, and configure a full-dimensional data recording structure that includes closed-loop state quantities.

[0014] 2. By using soft reset and convergence monitoring, the convergence criteria between the closed-loop state and the verification baseline are standardized, so that the initial state of each operating condition is unified to the neighborhood of the reference state.

[0015] 3. Based on the operating condition switching event and convergence criteria, accurately divide the transition segment, the disturbed steady-state segment, and the disturbance removal recovery segment.

[0016] 4. By calculating the consistency distance, we can filter out effective windows with consistent closed-loop states to ensure the reliability of statistical data.

[0017] 5. By restoring the model fitting, the memory time constant τ, residual amplitude A and other quantitative parameters are solved.

[0018] 6. Contamination index is constructed based on memory parameters and consistency distance, and contaminated data is processed through elimination, correction, and retesting strategies.

[0019] Based on the above core process, the hemispherical resonant gyroscope electromagnetic interference calibration method includes: Figure 1 The following steps are shown: Step 1: Establish a unified time reference for the HRG controller and the electromagnetic interference device, and configure a data recording structure for recording the closed-loop state quantities and angular velocity output of the hemispherical resonant gyroscope.

[0020] Since HRG sampling data and EMI operating condition switching events often use an "independent timing" mode (HRG samples according to the local clock, and EMI devices record operating condition switching according to their own clock), there is a time axis deviation between the two (which can reach tens of milliseconds or even seconds). This makes it impossible to accurately determine the time correspondence between operating condition switching and HRG response, and the segmentation boundary depends on subjective judgment. Therefore, the purpose of this step is to establish a unified time reference to achieve event-level alignment between EMI operating condition switching events and HRG sampling data, providing an objective time reference for subsequent segmentation, steady-state determination, and memory identification, and avoiding reliance on subjective waiting.

[0021] The specific method is as follows: 1. Establish a local timeline. Integrate a high-precision crystal oscillator (frequency stability ≤ ±0.1ppm, such as a TCXO temperature-compensated crystal oscillator) into the HRG controller to drive a 32-bit monotonically increasing counter (counting frequency matched with the sampling rate, such as a counting step of 1ms when the sampling rate is 1kHz), forming a continuous, non-backtracking local timeline with timestamp format "second.millisecond" (e.g., 1699999999.123).

[0022] 2. Trigger event synchronization. A dual synchronization method is adopted, prioritizing hardware triggers and using software synchronization as a fallback.

[0023] Method 1: Hardware Synchronization Method. The EMI device is configured with a TTL level output interface. When a change in operating condition occurs (e.g., interference on / off, frequency switching from 100MHz to 200MHz, field strength adjustment from 10V / m to 20V / m), a TTL level transition signal (high level 3.3V, low level 0V, rise time ≤10ns) is output and connected to the external interrupt input of the HRG controller. After detecting the TTL edge transition, the HRG controller immediately reads the current counter value as the timestamp ts of the operating condition transition event. event And write it to the event log (log format: ts) event Event type, operating parameters).

[0024] Method 2: Software Synchronization Method. If hardware wiring is not available on-site, time synchronization is achieved using PTP (Precision Time Protocol) or GNSS time synchronization. Both the HRG controller and EMI device are connected to the same PTP server or GNSS module and synchronized to UTC time (time synchronization accuracy ≤10ms). Simultaneously, the EMI device records the UTC timestamp of the operating condition switch and the operating condition parameters, and the HRG controller records the UTC timestamp of the sampled data. The alignment of operating condition events and sampled data is achieved through timestamp matching.

[0025] 3. Define the data record structure. Store the data using binary files or structured database tables (such as SQLite). Each data frame contains the following fields as shown in Table 1.

[0026] Table 1 shows the data record structure: Field Name Data types unit illustrate Timestamp (ts) Double-precision floating point s Local timeline or UTC time, with an accuracy of ≤1ms Operating Condition Identifier ID unsigned integers - Unique identifier for the current EMI operating condition (e.g., 1 - 100MHz / 10V / m on, 2 - interference off). Frequency point f Double-precision floating point Hz Current EMI interference frequencies (e.g., 1MHz-1GHz) Field strength E Double-precision floating point V / m Current electric field strength of EMI (e.g., 1V / m-100V / m) waveform type unsigned integers - 1-Sine wave, 2-Square wave, 3-Pulse wave (expandable) angular velocity output y Double-precision floating point ° / h HRG angular velocity measurement <![CDATA[AGC integrator output x agc > Double-precision floating point V Output voltage of amplitude control loop integrator <![CDATA[Quadrature suppression loop output x quad > Double-precision floating point V Output voltage of the quadrature error suppression loop integrator <![CDATA[FRB integrator output x frb > Double-precision floating point V The output voltage of the force feedback balance loop integrator <![CDATA[PLL frequency control word x pll > unsigned integers - The frequency control word of the phase-locked loop (corresponding to the resonant frequency). 4. Data storage. Set the sampling rate to 100Hz-5kHz (preferably 1kHz to balance data volume and statistical accuracy). Write data to the storage medium in blocks of 1000 frames each. Each data block contains a frame header (data block number, start timestamp, data length) and a data body for easy subsequent fast reading and window statistics.

[0027] It should be noted that: (1) The preferred sampling rate is 1kHz because the drift time constant of the HRG closed-loop state quantity is usually on the order of seconds. A sampling rate of 1kHz can ensure that each window (such as a 1s window) contains 1000 sampling points, which meets the statistical accuracy requirements. If the sampling rate is lower than 100Hz, the sample size in the window is insufficient and the statistical results fluctuate greatly. If it is higher than 5kHz, the data volume is too large, which increases the storage and computation burden. (2) The crystal oscillator accuracy is selected as a temperature-compensated crystal oscillator with an accuracy of ≤±0.1ppm to ensure that the drift of the local time axis is ≤2.4ms during long-term operation (such as 24-hour test), which meets the event synchronization accuracy requirements. (3) The data storage format adopts binary files. Compared with text files, binary files have higher storage efficiency (saving about 60% of storage space) and faster reading speed, which is suitable for storing large-scale sampled data.

[0028] Step 1 achieves precise time alignment between HRG sampling data and EMI operating events, with a time synchronization accuracy better than 1ms. This provides an objective and unified time benchmark for subsequent steps 3 (response data segmentation), 4 (steady-state window filtering), and 5 (memory parameter fitting). Simultaneously, the complete data recording structure includes closed-loop state variables and output variables, providing data support for subsequent state-based diagnosis and analysis, avoiding the limitations of existing technologies that rely solely on angular velocity output.

[0029] Step 2: Perform closed-loop state normalization before applying any EMI test conditions.

[0030] Closed-loop state standardization refers to: performing state preset operations on the closed-loop control loop of the HRG and monitoring the statistical characteristics of the closed-loop state quantities in real time until the closed-loop state quantities meet the preset baseline convergence criteria. Specifically, before each EMI test, the internal state quantities of each closed loop of the HRG are reset to the preset reference state through active intervention (soft reset), and the process is verified by the convergence criteria to ensure that the initial state of the closed-loop system is consistent.

[0031] in, Soft reset refers to the process of resetting the gyroscope and resonator by writing preset initial values ​​into the closed-loop status register and integrator output through software instructions, which maintains the oscillation state of the resonator and only resets the historical state of the control loop. This is different from hardware power-off reset (which will cause the gyroscope to restart and the resonator to oscillate again, which takes a long time and affects the test efficiency).

[0032] The baseline convergence criterion refers to an objective criterion based on the statistical characteristics of baseline data without EMI, used to determine whether the closed-loop system has recovered to a stable reference state, avoiding errors caused by subjective judgment.

[0033] Since the closed-loop state was not standardized before the operating condition switch, the system only relied on "natural waiting" to recover, which resulted in a long recovery time (up to several minutes) and could not guarantee recovery to the same state. Therefore, the purpose of this step is to quickly pull the closed-loop state back to the reference range through a combination of "soft reset + convergence criteria" and confirm convergence through objective criteria, thereby eliminating the memory residue of historical operating conditions from the source and ensuring that the initial state of different operating conditions and different repeated measurements is consistent.

[0034] The specific method is as follows: 1. Collect and statistically analyze baseline data without EMI. Under an environment without EMI and with HRG stationary (placed on a high-precision vibration isolation table, vibration acceleration ≤0.01g), collect baseline data for 30-120 seconds (preferably 60 seconds to ensure statistical reliability), using the same sampling rate as in step 1 (e.g., 1kHz); analyze the collected closed-loop state variables (x... agc x quad x frb x pll ) Perform statistical analysis and calculate the baseline mean μ for each state variable. i_ref and baseline standard deviation σ i_ref Where i=1 corresponds to x agc i=2 corresponds to x quad i=3 corresponds to x frb i=4 corresponds to x pll The baseline mean and baseline standard deviation of each state variable are used as the core parameters of the reference state.

[0035] 2. Closed-loop soft reset. A soft reset command (command code: 0x0001, transmission protocol: SPI, baud rate: 1Mbps, data bits: 8 bits, parity bit: none) is sent to the HRG controller via a host computer (such as PC control software). After receiving the command, the controller resets each closed-loop state as follows: (1) Write the initial value 0x0000 (corresponding to voltage 0V) to the status register, clear the integrator to zero, and start accumulating amplitude deviation again; (2) Status register writing and μ 2_ref (x) quad The digital code value corresponding to the baseline mean (e.g., μ) 2_ref =0.5V, AD conversion bit depth 16 bits, reference voltage 3.3V, then the digital code value = 0.5 / 3.3×65535≈9929). (3) Status register writing and μ 3_ref (x) frb The numeric code value corresponding to the baseline mean; (4) Maintain the phase-locked state (to avoid re-capturing the resonant frequency and causing an additional transition period, which usually takes 10s-20s), and only reset the status register of its loop filter (write the initial value 0x0000), while retaining the current frequency control word x. pll constant.

[0036] 3. Baseline Convergence Determination. After a soft reset, under EMI-free and HRG-free static conditions, closed-loop state data are continuously acquired. The "sliding window statistical method" is used to determine convergence, specifically including the following steps: (1) Set window parameters. The window length is W=0.5s-2s (preferably W=1s, to match the sampling rate of 1kHz, each window contains 1000 sampling points), and the number of consecutive windows is M=3-10 (preferably M=5, to ensure convergence stability); (2) Calculate window statistics: For each closed-loop state variable i, calculate the window mean μ of the current window by sliding along the window length W. i (k) (k is the window number, starting from 1) and the window standard deviation σ i (k); simultaneously calculate the rate of change Δμ between the current window mean and the previous window mean. i (k)=|μ i (k)-μ i (k-1)| / μ i_ref (Normalized rate of change to avoid the influence of dimensions); Convergence criterion: Baseline convergence is determined when all closed-loop state variables i satisfy the following two conditions simultaneously within M consecutive windows: Condition 1: Rate of change of the mean Δμ i (k)≤Δμ th (Δμ) th Δμ is the threshold determined based on the baseline standard deviation. th =0.05×σ i_ref / μ i_ref That is, the rate of change does not exceed 5% of the baseline relative standard deviation.

[0037] Condition 2: Window standard deviation σ i (k)≤σ th (σ) th =2×σ i_ref That is, the window fluctuation range does not exceed twice the baseline standard deviation.

[0038] Convergence flag output: After the above conditions are met, the HRG controller outputs a "baseline convergence" flag (instruction code: 0x0002) to the host computer, at which point the next working condition test can be entered; if the convergence criterion is not met after 5 minutes of continuous acquisition, an alarm message (instruction code: 0x0003) is output, prompting to check the HRG working status or adjust the soft reset initial value.

[0039] It should be noted that: (1) The baseline acquisition time is preferably 60s because the steady-state fluctuations of the HRG closed-loop state variables can be fully reflected within 60s, and the statistically obtained μ i_ref With σ i_ref More reliable; if the duration is less than 30 seconds, the statistical sample size is insufficient and the standard deviation calculation error is large; if it exceeds 120 seconds, the data is redundant and the calculation burden is increased.

[0040] (2) The window length W is set to 1s to match the sampling rate of 1kHz. Each window contains 1000 sampling points, which can effectively smooth high-frequency noise while ensuring time resolution and avoiding the delay in convergence determination caused by excessively long windows.

[0041] (3) Selecting 5 consecutive windows M can avoid misjudgment caused by accidental achievement of the target in a single window and ensure the stability of the convergence state.

[0042] (4) Convergence threshold Δμ th and σ th Based on the dynamic determination of baseline statistical characteristics, it is not a fixed value and can be adapted to individual differences in different HRGs (such as different baseline standard deviations of different gyroscopes), thus improving the universality of the criterion.

[0043] Step 2 uses a soft reset operation to quickly reset the historical state of the closed-loop control loop (reset time ≤100ms, much shorter than the several minutes of a hardware reset), avoiding the long wait for natural recovery; using the baseline convergence criterion, it objectively verifies whether the closed-loop state has recovered to the reference range, ensuring that the initial state is consistent under different operating conditions and different repeated measurements, fundamentally reducing the non-reproducibility of calibration results caused by differences in the initial state; simultaneously, the statistically obtained μ... i_ref With σ i_ref This provides the basic parameters for calculating the consistency distance in step 4 and determining the threshold in step 6.

[0044] Step 3: Apply the target EMI interference and record the operating condition switching events based on a unified time reference. Combine the disturbance convergence criteria to divide the time response data into the transition segment, the disturbance steady-state segment, and the disturbance removal recovery segment.

[0045] The target EMI condition refers to a specific electromagnetic interference condition defined by parameters such as frequency (f), field strength (E), waveform type, and on / off state (e.g., f=100MHz, E=20V / m, waveform=sine wave, on=on).

[0046] Because existing technologies use fixed waiting times to divide the steady-state segment, they cannot adapt to the differences in convergence speed under different interference intensities and frequencies, resulting in the mixing of transition segment data into steady-state statistics. Therefore, the purpose of this step is to use the operating condition switching event as the time node and combine it with dynamically adjusted disturbance convergence criteria to achieve accurate segmentation of response data and ensure clear separation between the transition segment and the steady-state segment.

[0047] The specific method is as follows: 1. Apply target EMI interference. The host computer sends the target EMI parameters (frequency f, field strength E, waveform type, duration t) to the EMI device. emi After receiving the parameters, the EMI device generates an electromagnetic interference signal as required (e.g., total harmonic distortion of a sine wave signal ≤1%) and outputs a TTL level transition signal for the operating condition switching event to the HRG controller (rising edge when interference is enabled, falling edge when interference is disabled); the HRG controller records the timestamp ts of this event. on And began continuously recording response data, ts on To enable interference time.

[0048] 2. Establish a disturbance-induced convergence criterion. Considering that EMI interference can change the noise level of the closed-loop system (e.g., the fluctuation amplitude of closed-loop state variables increases under strong interference), if the baseline convergence criterion without EMI (based on σ) is used... i_ref This can lead to the invalidation of the criterion (such as σ under strong interference). i (k) is always greater than σ th Since convergence cannot be determined, a dynamically adjusted perturbation convergence criterion is constructed, including: (1) Calculate the scale of the disturbed noise: After the operating condition is started, collect the closed-loop state variable data for the first N windows (N=3-5, preferably N=3), and calculate the window standard deviation σ of each state variable i. i_emi _1、σ i_emi _2、σ i_emi _3, take σ i_emi _1、σ i_emi _2、σ i_emi The mean σ of _3 i_emi =(σ i_emi _1+σ i_emi _2+σ i_emi _3) / 3, used as the noise scale after disturbance.

[0049] (2) Setting a dynamic threshold: The threshold of the disturbed convergence criterion is based on σ. i_emi Adjustment, threshold for the rate of change of the mean Δμ emi_th =0.05×σ i_emi / μ i_ref(Consistent with the relative rate of change of the baseline convergence criterion, ensuring consistent criterion logic), window standard deviation threshold σ emi_th =2×σ i_emi (Consistent with the fluctuation factor of the baseline convergence criterion).

[0050] 3. Response data segmentation. Based on TypeScript. on Disturbance convergence criterion and disabling the timestamp ts of the interference event off ,ts off The response data is divided into three stages based on the timestamp corresponding to the TTL transition signal output when the EMI device shuts off interference: (1) Transition segment A: Time interval [ts on ,ts steady ], where ts steady The window start time is the time when the convergence criterion for the first time is met. During this stage, the system starts responding to EMI interference from the initial state. The closed-loop state variables and angular velocity outputs are in a state of rapid change. The data contains a large amount of transition noise and does not participate in steady-state statistics.

[0051] (2) Disturbed steady-state segment B: Time interval [ts steady ,ts off During this stage, the system meets the disturbance convergence criterion, and the closed-loop state variables and angular velocity outputs are in a stable fluctuation state, which is the core interval for steady-state data statistics.

[0052] (3) Disturbance removal and recovery segment C: Time interval [ts off ,ts base ], where ts base This is the window start time for the first time to meet the baseline convergence criterion in step 2 after the interference is removed; during this stage, EMI interference is turned off, and the system gradually recovers from the disturbed state to the undisturbed baseline state, which is used for memory parameter identification in step 5.

[0053] 4. Verify the segmentation results. After segmentation, the time interval, data length, and closed-loop state statistical characteristics (mean, standard deviation) of each segment are automatically output for manual verification of the segmentation's rationality; if the duration of the disturbed steady-state segment B is less than the preset minimum value t... steady_min (Preferably 30s to ensure sufficient steady-state samples), output alarm information, prompting an extension of the EMI operating condition duration t. emi .

[0054] It should be noted that: 1. The preferred number of leading windows N is 3 windows, which can quickly capture the noise level after disturbance, while avoiding the influence of outliers in a single window on σ. i_emi The calculation.

[0055] 2. Minimum duration t of the disturbed steady-state segment steady_minA time of 30 seconds is preferred, which ensures that at least 30 windows (with a window length of 1 second) are included, meeting the sample size requirements for effective window screening and statistics in step 4. If the time is less than 30 seconds, there will be insufficient steady-state samples, and the reliability of the statistical results will be low.

[0056] 3. The dynamic threshold of the disturbed convergence criterion is based on the noise scale σ after disturbance. i_emi Adjustments were made to ensure that the criteria are effective under different interference intensities and frequency points, avoiding the problem that fixed thresholds fail under strong interference or are overly stringent under weak interference.

[0057] Step 3 uses an event-driven approach, with the timestamp of the operating condition switching event as the objective basis, to avoid errors in subjectively dividing the segment boundaries; it adapts to the system noise characteristics under different interference scenarios by dynamically adjusting the disturbance convergence criterion, ensuring the stability and reliability of the segmentation results; it clearly divides the transition segment, the disturbed steady-state segment, and the disturbance removal recovery segment, providing accurate data ranges for the steady-state data statistics in Step 4 and the memory parameter identification in Step 5, avoiding the mixing of transition segment data into steady-state statistics, and improving the accuracy of the calibration results.

[0058] Step 4: Perform same-state sampling within the disturbed steady-state segment.

[0059] Specifically, performing same-state sampling within the disturbed steady-state segment involves calculating the consistency distance d(k) between the closed-loop state quantity and the reference state for each sampling window, selecting valid windows whose consistency distance meets a preset threshold, and calculating the mean steady-state angular velocity based on these valid windows. Here, same-state sampling means that within the disturbed steady-state segment, only sampling windows whose closed-loop state quantity meets the consistency requirement with the reference state (without an EMI baseline) are selected for data statistics, ensuring that the windows participating in the statistics are within a consistent closed-loop state interval. The consistency distance d(k) refers to the distance between the closed-loop state quantity of the k-th sampling window and the reference state (μ... i_ref σ i_ref The consistency distance d(k) is a metric that measures the difference between multiple closed-loop state variables, comprehensively reflecting the degree of deviation. The effective window refers to the condition that the consistency distance d(k) satisfies the threshold condition (d(k) ≤ d). max The sampling window has good consistency between its closed-loop state and the reference state, and can be used for steady-state data statistics.

[0060] Since the closed-loop state variables may still drift slowly when the system enters the disturbed steady-state phase (e.g., the output of the AGC integrator may slowly accumulate due to slight fluctuations in EMI), resulting in differences in the closed-loop state of different windows, directly statistically analyzing the data of the entire steady-state phase would introduce system errors caused by drift. Therefore, this step calculates the consistency distance to filter out effective windows with consistent states, achieving "same-state sampling" and ensuring the consistency of statistical samples.

[0061] The specific method is as follows: 1. Divide the disturbed steady-state segment window—divide the disturbed steady-state segment B (time interval [ts]) into windows. steady ,ts off The window is divided into K consecutive windows (window numbers k=1,2,...,K) according to the window length W (consistent with step 2, preferably 1s), and each window contains N. s =W × sampling rate / number of sampling points (e.g., when W = 1s and the sampling rate is 1kHz, N) s =1000).

[0062] 2. Statistical Window State Quantities – For each window k, calculate the n closed-loop state quantities (n≥2, preferably n=2, select x) that participate in the consistency determination. agc With x frb Both reflect the balance between amplitude control and force feedback, and are sensitive to EMI interference but have low correlation. (The window mean μ) i (k) (i=1 corresponds to x) agc i=2 corresponds to x frb ).

[0063] 3. Calculate the consistency distance d(k) – The normalized Euclidean distance is used as the consistency distance, and the calculation formula is: Among them, μ i_ref σ is the baseline mean calculated in step 2. i_ref The baseline standard deviation is calculated in step 2; normalization can eliminate dimensional differences between different state quantities (such as x). agc The units are V and x frb (The unit is V, but the numerical range is different), making the deviation of each state quantity comparable.

[0064] 4. Determine the distance threshold d max — Distance threshold d max Based on statistical analysis of baseline data without EMI, the following steps were performed repeatedly (e.g., 100 times) under EMI-free conditions, calculating K d(k) values ​​each time, for a total of 100×K d(k) samples. The samples were then statistically sorted, and the 95%-99% quantiles were used as the d values. max (Optimal 97.5% quantile, balancing screening rigor with effective sample size); For example, with 100 × K = 10000 samples, the value of the 9750th sample after sorting is d. max .

[0065] 5. Calculate the effective window selection and steady-state parameters.

[0066] (1) Selection criteria: Select those that satisfy d(k)≤d max The window is taken as the valid window, and the set of valid windows is denoted as S={k|d(k)≤dmax ,k=1..K}.

[0067] (2) Calculation of steady-state angular velocity mean: For each window in the effective window set S, calculate the window mean y of the angular velocity output y. k Then for all y k Taking the average, we obtain the mean value of the disturbed steady-state angular velocity y. emi_mean . , where |S| is the number of effective windows.

[0068] (3) Calculate the steady-state noise index: For the angular velocity output data in the effective window set S, calculate the window variance σ. y^2 Alternatively, Allan variance segments (characterizing short-term stability) can be used as a steady-state noise index.

[0069] It should be noted that: 1. The state variable n involved in the decision-making process preferably consists of two state variables (x... agc With x frb The reason is that the two correspond to different closed-loop functions, which can comprehensively reflect the system state, and n=2 can simplify the calculation and avoid the consistency distance being overly sensitive to the deviation of a single state variable due to too much n; 2. Threshold d max The preferred quantile is the 97.5th percentile, because this quantile can eliminate 5% of the abnormal deviation window while retaining 95% of the normal window, thus balancing the effective sample size and state consistency.

[0070] Step 4 quantifies the difference between the window state and the reference state using the consistency distance d(k), achieving "same-state sampling." This filters out window data with excessive state drift or deviation, ensuring that the samples participating in the statistics are in a consistent closed-loop state range and preventing slow-drift windows from entering the statistics and causing bias; the mean steady-state angular velocity y emi_mean The noise index is calculated based on the effective window set, which will be used to calculate the peak consistency distance max(d(k)) in step 6 and provide input for subsequent pollution determination.

[0071] Step 5: Identify memory parameters based on the descrambling recovery segment: Use a mathematical model to fit the response data of the descrambling recovery segment and calculate the memory time constant τ and the memory residual amplitude A.

[0072] Among them, the memory parameters are a set of parameters used to quantify the closed-loop memory effect. The core parameters include the memory time constant τ (reflecting the decay rate of memory residue) and the memory residue amplitude A (reflecting the maximum strength of memory residue). The auxiliary parameters include the fitting residual R (reflecting the model fitting effect). The first-order exponential recovery model is a mathematical model used to describe the exponential decay of memory residue over time and is suitable for the recovery process of a single time constant. The model switching strategy refers to the adaptive strategy of automatically switching to a more complex model (such as a two-segment first-order exponential model or a piecewise linear model) when the fitting effect of the first-order exponential model is poor (the residual is too large), which adapts to the "fast first and slow later" dual-time constant recovery process.

[0073] Since the recovery process after descrambling is a direct manifestation of the memory effect, existing technologies have not quantified this process and cannot objectively assess the intensity and impact of the memory effect. Therefore, this step uses a mathematical model to fit the recovery segment data, transforming the abstract memory effect into quantifiable memory parameters, thus providing a basis for subsequent memory contamination assessment and data correction.

[0074] The specific method is as follows: 1. Selecting the fitting object—Choose a physical quantity that is sensitive to the memory effect and has high data reliability as the fitting object z(t), preferably a closed-loop state quantity (such as x). agc or x frb The reason is that the closed-loop state quantity directly reflects the internal state of the control loop, and the memory effect is more significant; if the closed-loop state quantity data is abnormal (such as excessive noise), the angular velocity output y can be selected as the fitting object.

[0075] 2. Data Preprocessing – For the scrambling recovery segment C (time interval [ts]) off ,ts base The z(t) data of ]) is preprocessed, including: (1) Outlier removal: Outliers are removed using the 3σ criterion (if the mean of a data point deviates from that of the five adjacent data points by more than three times the standard deviation, it is considered an outlier and is replaced by linear interpolation of the adjacent data points). (2) Data downsampling: If the sampling rate is too high (e.g., 5kHz), downsample the data (e.g., downsample to 1kHz) to reduce the amount of computation while preserving the recovery trend.

[0076] 3. Fit the initial model.

[0077] (1) The initial model is a first-order exponential recovery model, and the model expression is: , t≥t1; where t1 is the start time of scrambling (i.e., ts) off ), z inf To restore the steady-state value (approaching the baseline mean μ) i_refA is the residual amplitude of memory (A=z(t1)-z) inf τ is the memory time constant (unit: s; the larger τ is, the slower the memory decays).

[0078] (2) Solve the model parameters z using the least squares method. inf A, τ, the specific method is as follows: First, construct the objective function: Where T is the time set of the scrambling recovery period.

[0079] Then, calculate the initial value: z inf The initial value is taken as the mean of the last 10 data points of the recovery segment, and the initial value of A is taken as z(t1)-z inf_initial The initial value of τ is 10s.

[0080] Finally, iterative optimization is performed: the Levenberg-Marquardt algorithm (which combines the advantages of the Gauss-Newton method and the steepest descent method, with fast convergence and good stability) is used to minimize the objective function J, iterating until the residual change is less than 1×10⁻⁶. -6 Alternatively, the iteration can be stopped after reaching 1000 iterations to obtain parameter estimates. , , .

[0081] 4. Calculate the fitting residual: Calculate the normalized mean square error (NMSE) as the fitting residual R, using the following formula: ,in, These are the fitted values ​​for the model. To recover the mean of segment z(t), R is a dimensionless quantity with a value range of [0,1]. The smaller R is, the better the fitting effect.

[0082] 5. Model switching. The specific method is as follows: (1) Determine the residual threshold R th —Under EMI-free conditions, repeat steps 1 to 5 (e.g., repeat 20 times). After each fitting, calculate the fitting residual R, and take the 95th percentile of the 20 fitting residual R values ​​as the residual threshold R. th (i.e., the maximum acceptable residual under EMI-free conditions).

[0083] (2) Model switching condition - if the current fitting residual R ≤ R th This demonstrates that the first-order exponential model can effectively describe the recovery process while preserving parameters. , , If R > R thThis indicates that the recovery process exhibits a "fast at first, slow later" dual-time-constant characteristic, automatically switching to a two-segment first-order exponential model, with the model expression as follows: , t≥t1; where τ1 is the fast decay time constant (τ1<τ2), A1 is the amplitude of the fast decay component; τ2 is the slow decay time constant, A2 is the amplitude of the slow decay component; t2 is the boundary time between the two models (usually taken as t1+3τ1).

[0084] (3) Solve for the fitting parameters after switching—resolve the parameters of the two first-order exponential models using the least squares method, and calculate the fitting residual R'; if R' is still greater than R th Then, switch to a piecewise linear model (divided into time intervals, with each interval fitted using a linear model) to ensure that the fitting effect meets the requirements.

[0085] 6. Output memory parameters: Output the memory parameters of the final model, including core parameters (τ, A) and auxiliary parameters (R): If a first-order exponential model is used, output... , If a two-segment first-order exponential model is used, output τ1, τ2, A1, A2, R'; if a piecewise linear model is used, output the time constant, slope, intercept, and residual of each segment.

[0086] It should be noted that: 1. The fitting object z(t) is preferably x. agc or x frb The reason is that these two state variables are directly affected by the integrator, resulting in a more significant memory effect, smaller data fluctuations, and higher fitting accuracy; while the angular velocity output y is affected by various factors (such as mechanical noise and electromagnetic coupling noise), and the memory effect signal is easily submerged. 2. Residual threshold R th Based on statistical analysis of baseline data without EMI, the model switching is determined to ensure objectivity and avoid inappropriate model selection due to subjective judgment.

[0087] Step 5 uses a mathematical model to fit the descrambling recovery segment data, quantifying the abstract memory effect into specific memory parameters (τ, A, R), so that the intensity (A) and decay rate (τ) of the memory effect can be objectively evaluated; the model switching strategy adapts to different types of recovery processes, ensuring the accuracy and reliability of memory parameter identification; the memory parameters provide the core basis for constructing the memory contamination index and data correction in step 6, so that the impact of the memory effect can be quantified and controlled.

[0088] Step 6: Construct the memory contamination index MI and compare it with the reliable threshold MIok; when MI exceeds the threshold, perform removal, numerical correction or retesting, and output the calibration parameters after memory removal.

[0089] Among them, the memory contamination index (MI) is a quantitative index constructed based on memory parameters (τ, R) and the peak consistency distance (max(d(k))), used to assess the degree of contamination of calibration results by memory effects; the confidence threshold (MI) ok It is a threshold determined based on statistical analysis of EMI-free baseline data, used to determine whether memory contamination of the current calibration data is within acceptable limits; memory removal refers to: when MI exceeds MI ok The measures taken to eliminate or reduce the impact of memory contamination include three types of strategies: data removal, numerical correction, and retesting.

[0090] Because existing technologies do not assess the degree of memory contamination in calibration data and directly output statistical results, data with excessive contamination is misused as a calibration basis, affecting calibration accuracy. Therefore, this step constructs a memory contamination index to objectively determine the reliability of the data and develops handling strategies for data with excessive contamination to ensure the consistency and reproducibility of the output calibration parameters.

[0091] The specific method is as follows: 1. Determine the confidence threshold MI ok This includes the following steps: (1) Test baseline – Under EMI-free conditions, repeat steps 1 to 5 (e.g., repeat 20 times), and record the memory parameters (τ) after each test. j R j The peak value of the consistency distance between the j=1..20) and the disturbed steady-state segment is max(d j (k))(j=1..20).

[0092] (2) Statistical Analysis – Statistical analysis was performed on 20 sets of data, and the following were calculated: mean baseline memory time constant, mean baseline fit residual, and mean peak baseline consistency distance. Among these, the mean baseline memory time constant τ base The calculation formula is: , τ base Represents the memory time constant of j baseline tests; Rij is the mean of the baseline fit residuals. base The calculation formula is: R base Represents the fitting residuals of j baseline tests; the mean peak value of the baseline consistency distance is max. d_base The calculation formula is: ,max(d j (k) represents the peak consistency distance of the j-th baseline test.

[0093] (3) Calculate the confidence threshold MI ok Calculation—Calculate 20 memory contamination indices (MI) using the memory contamination index model for 20 sets of data. j Value, for the memory contamination index MIj Sort the data and use the 95th percentile as the confidence threshold (MI). ok (i.e., the maximum acceptable memory contamination index under EMI-free conditions).

[0094] The memory pollution index model is constructed using a weighted summation method to ensure a balanced contribution from each indicator. The model expression is as follows: Among them, w1, w2, and w3 are weighting coefficients that satisfy w1 + w2 + w3 = 1. Preferred values ​​are w1 = 0.4 (the memory time constant has the greatest influence because it directly reflects the duration of memory retention), w2 = 0.3 (the fitting residual reflects the reliability of memory parameters), and w3 = 0.3 (the peak consistency distance reflects the degree of deviation from the steady-state state). ε is a small quantity (ε = 1 × 10⁻⁶). -6 ), used to avoid the denominator being zero (such as τ). base =0 in the extreme case); max(d(k)) is the peak value of the consistency distance (i.e., the maximum deviation) of all windows in the disturbed steady-state segment; d max This is the consistency distance threshold determined in step 4.

[0095] 2. De-memory processing for contamination determination (1) Judgment criteria: The memory contamination index MI of the current working condition is compared with the confidence threshold MI. ok Comparisons include the following two cases: Case 1: MI≤MI ok This indicates that memory contamination is within an acceptable range, and the calibration parameters, including the mean steady-state angular velocity y, are directly output. emi_mean Steady-state noise parameters (such as variance σ) y^2 ), memory parameters (τ, A, R); Scenario 2: MI > MI ok This indicates that memory contamination exceeds the standard, and memory removal treatment should be performed, selecting at least one of the following three strategies: Strategy 1: Data Removal Strategy – Sort the windows of the disturbed steady-state segments from largest to smallest according to d(k), and remove the first M... del windows (M) del =5%×K, where K is the total number of windows, or the top 3 largest d(k) windows are removed. The effective window set S' and the mean steady-state angular velocity y are then recalculated. emi_mean If after removal, MI' = MI(y) emi_mean ')≤MI ok Output y emi_mean ';If MI'>MI ok Increase the rejection ratio (e.g., M) del =10%×K), until MI'≤MI okOr the elimination ratio reaches 30% (if the elimination ratio is still not met after 30% elimination, the elimination strategy is abandoned).

[0096] Strategy 2: Numerical Correction Strategy – Correcting the mean steady-state angular velocity based on the residual amplitude A, the correction formula is: y emi_corr =y emi_mean -α·A; where α is a correction factor, determined by repeated measurements under the same EMI condition. Specifically: the number of repetitions N is N for the same EMI condition. rep =10 times, each measurement yields y emi_mean_m With A m (m=1..10), traverse the candidate values ​​within the preset interval [0,2] of α (step size 0.01), and calculate the standard deviation σ(α)=std(y) of the corrected data corresponding to each α. emi_corr_m ), select the α that minimizes σ(α) as the optimal correction coefficient (α). opt ); MI calculated after correction corr =MI(y emi_corr If MI corr ≤MI ok Output y emi_corr If the levels still exceed the limit, further processing will be carried out using a removal strategy.

[0097] Strategy 3: Retest Strategy – Return to step 2, re-execute closed-loop state standardization processing (soft reset + baseline convergence determination), and then re-execute steps 3 to 5 for the current EMI operating condition; if MI is still greater than MI after retesting. ok Adjust the calibration parameters (e.g., increase the window length W to 2s, increase the number of consecutive windows M to 8), and repeat the measurement until MI ≤ MI. ok Or the number of retests reaches 3 (if the 3 retests do not meet the requirement, an alarm message will be output, prompting you to check the status of the HRG or EMI device).

[0098] 3. Reproducibility self-check Reproducibility self-test includes the following steps: First, the same EMI condition was measured three times to obtain the number of reproducibility tests, N. test =3 times, obtaining 3 calibration results y1, y2, y3. Calculate the standard deviation σ based on the calibration results y1, y2, y3. test =std([y1,y2,y3]) and the range R test =max([y1,y2,y3])-min([y1,y2,y3]).

[0099] Then, under EMI-free conditions, the measurement was repeated 20 times, and the corresponding 20 baseline calibration results were obtained. base1 y base2 ... ybase20 ; Calculate the baseline standard deviation σ based on the above 20 baseline calibration results. base =std([y base1 y base2 ... y base20 ]) and baseline range R base =max([y base1 y base2 ... y base20 ])-min([y base1 y base2 ... y base20 ]), take the acceptance threshold σ accept =2×σ base R accept =2×R base .

[0100] Finally, a self-check is performed—if σ test ≤σ accept And R test ≤R accept If the reproducibility meets the requirements, output the final calibration parameters; if not, increase the window length W (e.g., from 1s to 1.5s) or increase the number of consecutive windows M (e.g., from 5 to 7), and repeat steps 2 to 6 until the reproducibility meets the requirements.

[0101] 4. Output the calibrated fitting parameters. These include: EMI operating parameters (f, E, waveform), mean steady-state angular velocity (original / corrected value), steady-state noise index (variance / Allan variance), memory parameters (τ, A, R), memory contamination index (MI), and reproducibility index (standard deviation σ). test Range R test ), handling strategy description (none / removal / correction / retesting).

[0102] It should be noted that: 1. The preferred weighting coefficients w1, w2, and w3 are w1=0.4, w2=0.3, and w3=0.3, respectively. This is because the memory time constant τ directly affects the duration of memory residue and has the greatest impact on subsequent operating conditions, thus giving it the highest weight. The fitting residual R reflects the reliability of the memory parameters, and the peak consistency distance max(d(k)) reflects the deviation from the steady-state state; both are weighted in a balanced manner. 2. The preset interval for the correction coefficient α is [0,2], which can cover the correction requirements in most scenarios. A step size of 0.01 ensures that α... opt To achieve the desired accuracy while avoiding excessive iterations. 3. Regarding the selection of the number of repeated measurements: N rep =10 times can guarantee α opt Reliability, N test =3 times can quickly assess reproducibility, N base=20 times can accurately calculate the baseline reproducibility index. 4. Acceptance threshold σ accept Take 2×σ base R accept Take 2×R base This allows for a certain degree of measurement fluctuation while effectively filtering out results with poor reproducibility.

[0103] Step 6 quantifies the degree of memory contamination into an objective indicator using the Memory Contamination Index (MI), avoiding errors in subjective judgment of data reliability. The memory removal and treatment strategies (removal, correction, and retesting) provide effective solutions for different contamination scenarios, significantly reducing the impact of memory effects. The reproducibility self-test further verifies the consistency of calibration results, forming a closed-loop quality control process of "judgment-treatment-self-testing". The final output calibration parameters not only include measurement results but also information such as contamination assessment and reproducibility indicators, thereby reducing the dispersion of results from multiple measurements under the same EMI conditions and improving reproducibility.

[0104] Example 2: Based on the method provided in Example 1 above, this example provides a computer device that executes the method described in Example 1 or any other method that may involve the method described in Example 1. The device includes a memory, a processor, and a transceiver connected in sequence. The memory stores a computer program, the transceiver sends and receives messages, and the processor reads the computer program and executes the method described in Example 1 or any other method that may involve the method described in Example 1. Specifically, the memory may include, but is not limited to, random-access memory (RAM), read-only memory (ROM), flash memory, first-in-first-out (FIFO) memory, and / or last-in-first-out (FILO) memory, etc.; the processor may include, but is not limited to, a microprocessor of the STM32F105 series. Furthermore, the computer device may also include, but is not limited to, a power module, a display screen, and other necessary components.

[0105] The working process, working details and technical effects of the aforementioned computer device provided in this embodiment can be found in the method described in Embodiment 1 or any method that may involve the method described in Embodiment 1, and will not be repeated here.

[0106] Example 3: This example provides a computer-readable storage medium that stores instructions that include the method described in Example 1 or any other method that may involve the method described in Example 1. Specifically, the computer-readable storage medium stores instructions that, when executed on a computer, perform the method described in Example 1 or any other method that may involve the method described in Example 1. The computer-readable storage medium refers to a data storage medium, which may include, but is not limited to, floppy disks, optical disks, hard disks, flash memory, USB flash drives, and / or Memory Sticks. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0107] The working process, working details and technical effects of the aforementioned computer-readable storage medium provided in this embodiment can be found in the method described in Embodiment 1 or any method that may be related to Embodiment 1, and will not be repeated here.

[0108] Example 4: This example provides a computer program product containing instructions that, when executed on a computer, cause the computer to perform the method described in Example 1 or any method that may involve the method described in Example 1. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device.

[0109] It should be understood that the terms "system," "device," "unit," and / or "module" as used in this specification are a method of distinguishing different components, elements, parts, sections, or assemblies at different levels. However, if other terms can achieve the same purpose, they may be replaced by other expressions.

[0110] As indicated in this specification and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" do not specifically refer to the singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of expressly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0111] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0112] It should be noted that the structures, proportions, sizes, etc., illustrated in the accompanying drawings are merely for illustrative purposes to aid those skilled in the art and are not intended to limit the scope of the invention. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in proportions, or adjustments to size, without affecting the effectiveness and purpose of the invention, should still fall within the scope of the disclosed technical content. Furthermore, terms such as "upper," "lower," "left," "right," and "middle" used in this specification are merely for clarity and not intended to limit the scope of the invention. Changes or adjustments to their relative relationships, without substantially altering the technical content, should also be considered within the scope of the invention.

Claims

1. A method for calibrating electromagnetic interference data of a hemispherical resonant gyroscope, characterized in that, The following steps are included for calibrating the performance of a hemispherical resonant gyroscope in an environment with electromagnetic interference: Step 1: Establish a unified time reference between the hemispherical resonant gyroscope controller and the electromagnetic interference device, and configure a data recording structure for recording the closed-loop state quantities and angular velocity output of the hemispherical resonant gyroscope; Step 2: Before testing under any electromagnetic interference condition, perform closed-loop state standardization processing; closed-loop state standardization processing includes: performing state preset operation on the closed-loop control loop of the hemispherical resonant gyroscope, and monitoring the statistical characteristics of the closed-loop state quantity in real time until the closed-loop state quantity meets the preset baseline convergence criterion. Step 3: Apply target electromagnetic interference; record the operating condition switching events based on a unified time reference, and divide the time response data of the hemispherical resonator gyroscope into a transition segment, a disturbed steady-state segment, and a disturbance removal recovery segment based on the disturbed convergence criterion; Step 4: Perform same-state sampling within the disturbed steady-state segment; same-state sampling includes: calculating the consistency distance between the closed-loop state variables and the reference state for each sampling window, selecting valid windows whose consistency distance meets a preset threshold, and calculating the mean steady-state angular velocity based on the data from the valid windows; Step 5: Identify memory parameters based on the descrambling recovery segment; memory parameter identification includes: fitting the response data of the descrambling recovery segment using a mathematical model to calculate the memory parameters characterizing the closed-loop memory effect; Step 6: Perform memory removal processing; memory removal processing includes: constructing a quantified memory contamination index based on memory parameters, fitting residuals, and consistency distance; when the memory contamination index exceeds the confidence threshold, performing at least one of the following actions on the test data of the current working condition: elimination, numerical correction, or retesting, and outputting the calibration parameters after memory removal processing.

2. The method for calibrating electromagnetic interference data of a hemispherical resonant gyroscope according to claim 1, characterized in that, Step 1 establishes a unified time base, including the following steps: In the hemispherical resonant gyroscope controller, a monotonically increasing counter based on a high-precision crystal oscillator is configured as the local time axis; When the electromagnetic interference device performs operating condition switching, frequency point change, or intensity change, it sends a level jump signal to the hemispherical resonant gyroscope controller through a physical trigger line. After detecting the level transition signal, the hemispherical resonant gyroscope controller reads the current counter value as the timestamp of the operating condition switching event.

3. The method for calibrating electromagnetic interference data of a hemispherical resonant gyroscope according to claim 1, characterized in that, Step 2, the state preset operation, includes the following steps: By writing preset initial values ​​into the state register of the hemispherical resonant gyroscope controller, software zeroing or initialization operations are performed on the amplitude control loop integrator, the quadrature error suppression loop integrator, and the force feedback back-balance loop integrator; at the same time, the phase-locked state of the phase-locked loop (PLL) is maintained, and only the filter state register of the PLL is reset.

4. The method for calibrating electromagnetic interference data of a hemispherical resonant gyroscope according to claim 2, characterized in that, In step 2, the baseline convergence criterion adopts the sliding window statistical method, which includes the following steps: Set the sliding window length and the number of consecutive windows M; Real-time calculation of the mean rate of change and window standard deviation of the closed-loop state variables within the current window; When the mean change rate and window standard deviation of M consecutive windows are both less than the first threshold Δμth and the window standard deviation is less than the second threshold σth, the state is considered to have converged and the standardization process ends.

5. The method for calibrating electromagnetic interference data of a hemispherical resonant gyroscope according to claim 3, characterized in that, Step 3 also includes the following steps: When establishing the disturbance convergence criterion, the judgment threshold of the disturbance convergence criterion is dynamically adjusted in real time based on the state variable noise scale in the early stage of the disturbance steady-state segment; specifically, it includes the following steps: Calculate the mean standard deviation of the state variables within a preset number of windows in the initial stage of the disturbed steady-state phase, and use it as a measure of the disturbed noise. A disturbance convergence criterion adapted to the current disturbance intensity is constructed by replacing the standard deviation of the non-electromagnetic interference baseline with the disturbed noise scale.

6. The method for calibrating electromagnetic interference data of a hemispherical resonant gyroscope according to claim 4, characterized in that, In step 4, the formula for calculating the consistency distance is: Where n is the number of closed-loop state variables involved in the decision, n≥2, μ i (k) represents the mean value of the i-th closed-loop state variable within the k-th sampling window, μ i_ref Let σ be the mean of the i-th closed-loop state variable under baseline conditions without electromagnetic interference. i_ref Let be the standard deviation of the i-th closed-loop state variable under baseline conditions without electromagnetic interference.

7. The method for calibrating electromagnetic interference data of a hemispherical resonant gyroscope according to claim 5, characterized in that, In step 5, the memory parameters include: memory time constant τ and memory residual amplitude A; The mathematical model adopts a first-order exponential model, a two-segment first-order exponential model, or a piecewise linear model. The identification process includes a model switching strategy. The model switching strategy is as follows: first-order exponential model is used for fitting first, and the fitting residual is calculated. If the fitting residual is greater than a preset threshold, the model is automatically switched to two segments of first-order exponential model or piecewise linear model to re-identify the parameters.

8. A method for calibrating electromagnetic interference data of a hemispherical resonant gyroscope according to claim 5 or 7, characterized in that, In step 6, the formula for calculating the Memory Contamination Index (MI) is: ;in, τ The memory time constant for the current operating condition. R To fit the residuals, max( d ( k )) represents the consistency distance of each sampling window within the disturbed steady-state segment. d ( k The maximum value of ). τ base The average of the multiple memory time constants obtained by repeatedly executing steps 1 to 5 under conditions of no electromagnetic interference; R base The mean of multiple fitting residuals obtained by repeatedly executing steps 1 to 5 under conditions of no electromagnetic interference; d max The consistency distance threshold determined in step 4 based on baseline data without electromagnetic interference; ε To avoid small quantities where the denominator is zero; w 1. w 2. w 3 is the weighting coefficient, and satisfies w 1+ w 2+ w 3 = 1.

9. A method for calibrating electromagnetic interference data of a hemispherical resonant gyroscope according to claim 8, characterized in that, In step 6, the method for numerical correction is as follows: based on formula y emi_corr =y emi_mean -α·A calculates the corrected calibration value; where y emi_corr The corrected mean steady-state angular velocity, y emi_mean The mean steady-state angular velocity is given by α, which is a correction factor. The method for determining the correction factor α is as follows: collect multiple sets of repeated measurement data under the same electromagnetic interference conditions, traverse candidate α values ​​within a preset value range, and select the α value that minimizes the repeated measurement standard deviation of the corrected data as the final correction factor.

10. A method for calibrating electromagnetic interference data of a hemispherical resonant gyroscope according to claim 4, characterized in that, Before outputting calibration parameters, a reproducibility self-test step is included; the reproducibility self-test step includes: Perform repeated measurements under the same electromagnetic interference condition, and calculate the standard deviation and range of the repeated measurement results. If the standard deviation or range exceeds the acceptance threshold determined based on the statistical characteristics of the baseline without electromagnetic interference, the convergence criterion parameters in step 2 are adjusted and a retest is triggered; the adjusted parameters include increasing the sliding window length or increasing the number of consecutive windows.

11. A computer device, characterized in that, The system includes a memory, a processor, and a transceiver; the memory is used to store a computer program; the transceiver is used to transmit and receive data or control commands generated by a hemispherical resonator gyroscope controller, an electromagnetic interference device, or a host computer; the processor is used to read and execute the computer program to implement a hemispherical resonator gyroscope electromagnetic interference data calibration method as described in any one of claims 1 to 10.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program or instructions; when the computer program or instructions are executed by a processor, a method for calibrating electromagnetic interference data of a hemispherical resonant gyroscope as described in any one of claims 1 to 10 is implemented.

13. A computer program product, characterized in that, The computer program product includes a computer program or instructions; when the computer program or instructions are executed by a computer device, the computer device implements a hemispherical resonant gyroscope electromagnetic interference data calibration method as described in any one of claims 1 to 10.

Images