A temperature modeling and compensation method for hemispherical resonant gyroscope based on real-time sliding window
Through the real-time sliding window temperature compensation model, the problems of poor repetition of hemispherical resonant gyro output and the impact of temperature changes are solved, and high-precision prediction and compensation of gyro output are achieved, thereby improving measurement accuracy.
Patent Information
- Application Number
- CN202211204347.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-09-29
- Publication Date
- 2024-11-26
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The output repeatability of the hemispherical resonant gyroscope is poor and is affected by temperature changes, resulting in a decrease in measurement accuracy. It is difficult for the prior art to effectively reduce the impact of temperature changes on the gyroscope.
The real-time sliding window temperature compensation model is used to fit the gyro output data by intercepting the gyro output data, predict the output value at the next moment, and calculate the gyro rotation angle measurement error and zero bias stability based on the prediction error to achieve temperature compensation.
The measurement accuracy of the hemispherical resonant gyroscope is improved, and the output change characteristics are adaptively tracked, which effectively reduces the impact of temperature changes on the gyroscope and improves the regularity and accuracy of the output.
Smart Images

Figure CN115435768B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of inertial instruments, and in particular relates to a temperature modeling and compensation method of a hemispherical resonant gyroscope based on a real-time sliding window. Background Art
[0002] Hemispherical Resonator Gyro (HRG) is a high-precision gyroscope with inertial navigation performance among Columbia vibrating gyroscopes. Its random drift can reach 10° / hr and its life span is up to 15 years. It has many advantages such as high measurement accuracy, high stability, high reliability and long life, and has broad application prospects in aviation, aerospace, missile weapons and other fields.
[0003] However, due to the limitations of the current production process, the output of the hemispherical resonant gyroscope has the following problems: (1) poor repeatability, that is, the gyroscope output is highly correlated with the state when the gyroscope is started, but the output regularity of the gyroscope started multiple times is poor; (2) after the gyroscope is started, the internal temperature of the gyroscope changes to a certain extent due to the operation of the circuit and the vibration of the resonator, which leads to a decrease in the gyroscope measurement accuracy.
[0004] In order to improve the measurement accuracy of the hemispherical resonant gyroscope, how to reduce the influence of temperature change on the hemispherical resonant gyroscope is an urgent problem to be solved. Summary of the invention
[0005] In view of the above-mentioned problems and taking into account the defect of poor repeatability of the hemispherical resonant gyroscope output, the present invention establishes a real-time sliding window temperature compensation model with adaptive output characteristics. The model is essentially a time-varying polynomial compensation model, which can adaptively change the fitting model according to the changes in the gyroscope output, so that the prediction model can track the changing characteristics of the gyroscope output in real time, thereby achieving better temperature compensation effect.
[0006] The core idea of the present invention is: use a time window to intercept gyro output data, fit the input-output relationship in the time window, and predict the gyro output at the next moment based on the fitting model; the time window slides forward one step, the gyro output at the next moment is included in the time window, and it is judged whether the gyro output is a normal value. If it is not a normal value, the predicted value is used to replace the abnormal output. If it is a normal value, the measured value is directly included in the time window; based on the predicted value and the least squares method, a gyro measurement angular velocity is fitted for every 2500 (corresponding to 100s) data; the turntable frame angle true value is subtracted from the predicted value to obtain the turntable frame angle prediction error, and the gyro angle measurement error and gyro zero bias stability can be calculated according to the prediction error.
[0007] The technical solution to achieve the purpose of the present invention is:
[0008] A temperature modeling and compensation method for a hemispherical resonant gyroscope based on a real-time sliding window, characterized by comprising:
[0009] Step 1: Collect multiple groups of test output signal data of the hemispherical resonant gyroscope;
[0010] Step 2: Analyze and select independent variables of the gyro temperature error model based on the output signal of the hemispherical resonant gyro and the shape characteristics of the resonant frequency two-dimensional plane;
[0011] Step 3: According to the selected independent variables, use the real-time sliding window to intercept the data, establish a prediction model based on the data in the time window, and predict the turntable frame angle at the next moment;
[0012] Step 4: Slide the time window forward one step to determine whether the new window data is normal data. If it is not normal data, replace it with the predicted value and then establish a prediction model. If the new window data is normal data, directly establish a prediction model based on the data in the current time window.
[0013] Step 5: Calculate the sensitive angular velocity of the hemispherical resonant gyroscope based on the predicted data;
[0014] Step 6: Verify the compensation effect based on the gyro bias stability.
[0015] Furthermore, the specific operation steps of step 1 include:
[0016] Step 1.1: Install the hemispherical resonant gyro on the turntable loop of the inertial platform, where the gyro input axis is installed in the vertical direction, and the gyro output axis and the other axis are installed in the horizontal direction, in the same horizontal plane, and the stabilization loop of the inertial platform is not powered on at this time;
[0017] Step 1.2: Set the sampling frequency to 0.04 Hz, and collect the turntable frame angle, gyro standing wave angle, and gyro resonant frequency at the moment the gyro is powered on;
[0018] Step 1.3: Repeat steps 1.1-1.2 6 times to conduct 6 groups of experiments. The interval between two experiments is at least 24 hours. The data collection time for a single experiment is at least 6 hours. Finally, 6 groups of experimental data are obtained, and each group of experimental data includes the turntable frame angle, gyro standing wave angle, and gyro resonant frequency.
[0019] Furthermore, the specific operation steps of step 2 include:
[0020] Step 2.1: Taking the resonant frequency f as the independent variable and the turntable frame angle y as the dependent variable, obtain the turntable frame angle-resonant frequency diagram of the six groups of experiments;
[0021] Step 2.2: According to the shape of the turntable frame angle-resonance frequency graph, the independent variable of the gyro temperature error model is determined as Where Δf is the resonant frequency gradient.
[0022] Furthermore, the specific operation steps of step 3 include:
[0023] Step 3.1: Use a time window of length 10 to intercept the turntable frame angle and gyro resonant frequency, and calculate the independent variable based on the resonant frequency
[0024] Step 3.2: Based on the data in the time window, use the second-order model to calculate the turntable frame angle y and the independent variable The relationship between is regressed and the regression equation is obtained as follows:
[0025]
[0026] Where: a1, a2, a3 are the coefficients of the first-order terms, a 11 is the coefficient of the quadratic term, a 12 、a 13 is the coupling term coefficient;
[0027] Step 3.3: Based on formula (1) and the resonant frequency f(t+1) collected at the next moment, the turntable frame angle y(t+1) at the next moment is predicted, and the turntable frame angle at the next moment is obtained as:
[0028]
[0029] Where: is the turntable frame angle at the next moment.
[0030] Furthermore, the specific operation steps of step 4 include:
[0031] Step 4.1: Slide the time window forward by one step, so that the turntable frame angle y(t+1) at the next moment is included in the current time window;
[0032] Step 4.2: Set the gyro output change threshold Δy. When y(t+1)-y(t)≥Δy, y(t+1) is output as an abnormality. At this time, the predicted value is used. Replace y(t+1), return to step 3 for fitting and prediction, and output the predicted value; when y(t+1)-y(t)<Δy, take y(t+1) as the normal output, directly return to step 3 for fitting and prediction, and output the predicted value.
[0033] Furthermore, the specific operation steps of step 5 include:
[0034] Based on the predicted value obtained in step 4.2, a first-order linear fit is performed on every 2500 data points using the least squares method. The fitted slope is the angular velocity to which the gyroscope is sensitive:
[0035] Ω=(X T X) -1 XY (3)
[0036] Where: Ω is the gyro output angular velocity obtained by fitting, is the matrix constructed from the sampling time, is the matrix constructed from the predicted values of the turntable frame angle.
[0037] Furthermore, the specific operation steps of step 6 include:
[0038] Step 6.1: subtract the predicted value from the actually collected turntable frame angle output value to obtain the prediction error of the turntable frame angle;
[0039] Step 6.2: Obtain the gyro measurement angular velocity error corresponding to the prediction error according to formula (3);
[0040] Step 6.3: Use a time window with a length of 36 and a step size of 1, calculate a standard deviation for every 36 angular velocity errors, and obtain the gyro bias stability at that moment;
[0041] Step 6.4: Slide the time window forward by one step, and repeat steps 6.3 and 6.4 until all data are calculated, and the variation law of the zero bias stability of the hemispherical resonant gyroscope is obtained;
[0042] Step 6.5: Compare the obtained zero bias stability variation law of the hemispherical resonant gyroscope with the zero bias stability variation law obtained by the polynomial-based compensation method to verify the compensation effect.
[0043] Compared with the prior art, this method has the following beneficial effects:
[0044] The present invention adopts a temperature modeling compensation method of a real-time sliding window, which can adaptively track the changing output law of a gyroscope, thereby realizing high-precision prediction and compensation of the gyroscope output. First, a time window with a length of 10 is used to intercept data, and an input-output model regression is performed based on the data in the window, and the gyroscope output at the next moment is predicted based on the regression model; then, the time window slides forward one step, and if the output value at the next moment is a normal value, it is directly included in the time window, and if the output is an abnormal value, it is replaced with a predicted value, and the replaced predicted value is included in the time window; model regression and prediction are performed based on the data in the time window, and the gyroscope is rolled forward in this way until the gyroscope is powered off. The temperature compensation model based on the real-time sliding window is essentially a time-varying polynomial model, which can adaptively track the real-time changing system output law, thereby effectively improving the measurement accuracy of the hemispherical resonant gyroscope. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] Figure 1A flow chart of the temperature modeling compensation method based on real-time sliding window;
[0046] Figure 2 This is a schematic diagram of the installation of a hemispherical resonant gyroscope;
[0047] Figure 3 (a)-(f) are the curves of the change of the frame angle-resonance frequency of six groups of turntables;
[0048] Figure 4 (a)-(f) are the angular velocities measured by the gyro after compensation;
[0049] Figure 5 (a)-(f) are the gyro bias stability after compensation. DETAILED DESCRIPTION
[0050] In order to enable those skilled in the art to better understand the technical solution of the present invention, the technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.
[0051] The present invention proposes a temperature modeling compensation method for a hemispherical resonant gyroscope based on a real-time sliding window. Figure 1 As shown, the specific steps include:
[0052] Step 1: Collect gyro output signal data
[0053] Step 1.1: Install the 431# gyro on the turntable circuit of the inertial platform. The stabilization circuit of the inertial platform is not powered on. The gyro input shaft is installed in the vertical direction to sense the vertical component of the local ground speed. The gyro output shaft and another shaft are installed in the horizontal plane, as shown in the attached figure. Figure 2 As shown;
[0054] Step 1.2: Set the sampling frequency to 0.04 Hz, and start collecting data when the gyro is powered on. The collected parameters include the turntable frame angle, gyro standing wave angle, and gyro resonant frequency.
[0055] Step 1.3: Carry out 6 tests according to the above gyro installation and data collection method, with at least 24 hours between each test and at least 6 hours of data collection time for a single test. Analysis of the raw data shows that: (1) the gyro has an output drift process during the startup phase, which is caused by the temperature change caused by the gyro startup; (2) the gyro output has a "burr" phenomenon.
[0056] Step 2: Analysis and selection of independent variables based on the multiple output characteristics of 431# gyro
[0057] Step 2.1: According to the disclosed prior art, there is a strong linear relationship between the internal temperature T and the resonant frequency f of the hemispherical resonant gyroscope. Therefore, the resonant frequency f is used as the basic independent variable to replace the temperature T to establish the gyroscope temperature error model.
[0058] Step 2.2: Take the resonant frequency as the independent variable and the turntable frame angle as the dependent variable, and draw the turntable frame angle y-resonant frequency f graph for the six groups of experiments, as shown in the attached figure. Figure 3 As shown;
[0059] Step 2.3: Analysis of the six groups of turntable frame angle-resonance frequency graphs shows that: (1) The repeatability of the six outputs is poor, which means that it is difficult to establish a unified turntable angle-resonance frequency model; (2) The input-output relationship of the gyroscope should be time-varying, and it is difficult to use a time-invariant model to obtain a better compensation effect; (3) There is a one-to-many mapping relationship between the turntable angle y and the resonant frequency f, so it is impossible to establish a mapping relationship between the two by simply relying on the resonant frequency as a variable; (4) Except for Figure (b), there are "check mark" shaped areas in the other figures, so consider changing As an independent variable; (5) Based on the resonant frequency f, the differential component Δf can also be considered as an independent variable. Based on the above analysis, the temperature model independent variable is finally selected as Used to build a new gyro temperature error model.
[0060] Step 3: Use the real-time sliding window to intercept data, establish a prediction model based on the data in the window, and predict the gyro output value at the next moment, which includes the following specific steps:
[0061] Step 3.1: Use a time window of length 10 to intercept the turntable frame angle and gyro resonant frequency, and calculate the independent variable based on the resonant frequency
[0062] Step 3.2: Based on the data in the time window, use the second-order model to calculate the turntable frame angle y and the independent variable The relationship between is regressed to obtain a new gyro temperature error model:
[0063]
[0064] Where: a1, a2, a3 are the coefficients of the first-order terms, a 11 is the coefficient of the quadratic term, a 12 、a 13 is the coupling coefficient.
[0065] Step 3.3: Based on the model fitted in step 3.2 and the resonant frequency f(t+1) collected at the next moment, the turntable frame angle y(t+1) at the next moment is predicted as follows:
[0066]
[0067] Where: is the turntable frame angle at the next moment.
[0068] Step 4: The time window slides forward one step to determine whether the new window data is normal data. If it is not normal data, it is replaced with the predicted value, and then a prediction model is established; if the new window data is normal data, a prediction model is directly established based on the data in the window, which includes the following specific steps:
[0069] Step 4.1: The time window slides forward by one step, and the turntable frame angle y(t+1) at the next moment is included in the time window;
[0070] Step 4.2: Considering that the 431# hemispherical resonant gyro output has a burr problem, it is necessary to determine whether the gyro output is an abnormal output. Set a gyro output change threshold Δy=0.01. When y(t+1)-y(t)≥Δy, y(t+1) is considered as an abnormal output. At this time, use Replace y(t+1), and then return to step 3 to perform fitting and prediction according to equations (1)-(2); when y(t+1)-y(t)<Δy, y(t+1) is regarded as a normal output, and directly return to step 3 for fitting and prediction.
[0071] Step 5: Based on the predicted data, the sensitive angular velocity of the 431# hemispherical resonant gyroscope is calculated, which includes the following specific steps:
[0072] Based on the predicted data, a first-order linear fit is performed every 2500 data (i.e. 100s). The fitting slope is the sensitive angular velocity of the gyroscope. The fitting slope is obtained by the least squares method as follows:
[0073] Ω=(X T X) -1 XY (3)
[0074] Where: Ω is the gyro output angular velocity obtained by fitting, is the matrix constructed from the sampling time, is the matrix constructed from the predicted values of the turntable frame angle.
[0075] Step 6: Using gyro bias stability as a standard, verify the temperature modeling compensation method for the real-time sliding window, which includes the following specific steps:
[0076] Step 6.1: Subtract the predicted value from the turntable frame angle output value to obtain the prediction error of the turntable frame angle;
[0077] Step 6.2: According to the least square method in step 5, the gyro measurement angular velocity error corresponding to the predicted error is obtained, such as Figure 4 As shown;
[0078] Step 6.3: Use a time window with a length of 36 and a step size of 1, and calculate a standard deviation of every 36 (corresponding to 1 hour of data) angular velocity errors in a windowed manner according to the root mean square error formula in the prior art to obtain the gyro bias stability at that moment;
[0079] Step 6.4: Slide the time window forward by one data point, repeat steps 6.3 and 6.4 until all data points are calculated, and obtain the gyro bias stability at each moment, and plot it accordingly. Figure 5 ,according to Figure 5 The variation law of the bias stability of the 431# hemispherical resonant gyro can be obtained: the smaller the bias stability, the higher the gyro measurement accuracy. In the field of gyroscopes, bias stability is an extremely important indicator for evaluating gyroscope measurement accuracy. The bias stability is used for subsequent gyroscope measurement accuracy evaluation.
[0080] Example
[0081] In order to further verify the effect of the system, the method proposed in the present invention is compared with the compensation method based on polynomials. The comparison results are as follows: Figure 5 As shown. Figure 5 The compensation effect given shows that after temperature modeling compensation based on real-time sliding window, the gyro zero bias stability can quickly converge to the range of 10-3, and its compensation accuracy is much higher than the compensation method based on polynomial. This is because the real-time sliding window method can track the changes in gyro output characteristics in real time, while the polynomial is a fixed, time-invariant model that is difficult to adapt to the real-time changing gyro output characteristics. Therefore, the temperature modeling compensation method based on real-time sliding window has higher accuracy.
[0082] The contents not described in detail in this specification belong to the prior art known to the professional and technical personnel in this field. Although the present invention patent is described in detail with reference to the above embodiments, for the technical personnel in this field, it is still possible to modify the technical solutions recorded in the above embodiments, or to replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A temperature modeling and compensation method for a hemispherical resonant gyroscope based on a real-time sliding window, characterized in that: include: Step 1: Collect multiple groups of test output signal data of the hemispherical resonant gyroscope; Step 2: Analyze and select independent variables of the gyro temperature error model based on the output signal of the hemispherical resonant gyro and the shape characteristics of the resonant frequency two-dimensional plane; Step 3: According to the selected independent variables, use the real-time sliding window to intercept the data, establish a prediction model based on the data in the time window, and predict the turntable frame angle at the next moment; Step 4: Slide the time window forward one step to determine whether the new window data is normal data. If it is not normal data, replace it with the predicted value and then establish a prediction model. If the new window data is normal data, directly establish a prediction model based on the data in the current time window. Step 5: Calculate the sensitive angular velocity of the hemispherical resonant gyroscope based on the predicted data; Step 6: Verify the compensation effect based on the gyro bias stability. The specific operation steps of step 2 include: Step 2.1: Taking the resonant frequency f as the independent variable and the turntable frame angle y as the dependent variable, obtain the turntable frame angle-resonant frequency diagram of the six groups of experiments; Step 2.2: According to the shape of the turntable frame angle-resonance frequency graph, the independent variable of the gyro temperature error model is determined as Where △f is the resonant frequency gradient; The specific steps of step 3 include: Step 3.1: Use a time window of length 10 to intercept the turntable frame angle and gyro resonant frequency, and calculate the independent variable based on the resonant frequency Step 3.2: Based on the data in the time window, use the second-order model to calculate the turntable frame angle y and the independent variable The relationship between is regressed and the regression equation is obtained as follows: Where: a1, a2, a3 are the coefficients of the first-order terms, a 11 is the coefficient of the quadratic term, a 12 、a 13 is the coupling term coefficient; Step 3.3: Based on formula (1) and the resonant frequency f(t+1) collected at the next moment, the turntable frame angle y(t+1) at the next moment is predicted, and the turntable frame angle at the next moment is obtained as: Where: is the turntable frame angle at the next moment.
2. The temperature modeling and compensation method of a hemispherical resonant gyro based on a real-time sliding window according to claim 1, characterized in that: The specific steps of step 1 include: Step 1.1: Install the hemispherical resonant gyro on the turntable loop of the inertial platform, where the gyro input axis is installed in the vertical direction, and the gyro output axis and the other axis are installed in the horizontal direction, in the same horizontal plane, and the stabilization loop of the inertial platform is not powered on at this time; Step 1.2: Set the sampling frequency to 0.04 Hz, and collect the turntable frame angle, gyro standing wave angle, and gyro resonant frequency at the moment the gyro is powered on; Step 1.3: Repeat steps 1.1-1.2 6 times to conduct 6 groups of experiments. The interval between two experiments is at least 24 hours. The data collection time for a single experiment is at least 6 hours. Finally, 6 groups of experimental data are obtained, and each group of experimental data includes the turntable frame angle, gyro standing wave angle, and gyro resonant frequency.
3. The temperature modeling and compensation method of a hemispherical resonant gyro based on a real-time sliding window according to claim 2, characterized in that: The specific steps of step 4 include: Step 4.1: Slide the time window forward by one step, so that the turntable frame angle y(t+1) at the next moment is included in the current time window; Step 4.2: Set the gyro output change threshold △y. When y(t+1)-y(t)≥△y, y(t+1) is output as an abnormality. At this time, the predicted value is used. Replace y(t+1), return to step 3 for fitting and prediction, and output the predicted value; when y(t+1)-y(t)<△y, take y(t+1) as the normal output, directly return to step 3 for fitting and prediction, and output the predicted value.
4. The temperature modeling and compensation method of a hemispherical resonant gyro based on a real-time sliding window according to claim 3, characterized in that: The specific steps of step 5 include: Based on the predicted value obtained in step 4.2, a first-order linear fit is performed on every 2500 data points using the least squares method. The fitted slope is the angular velocity to which the gyroscope is sensitive: Ω=(X T X) -1 XY (3) Where: Ω is the gyro output angular velocity obtained by fitting, is the matrix constructed from the sampling time, is the matrix constructed from the predicted values of the turntable frame angle.
5. The method for temperature modeling and compensation of a hemispherical resonant gyro based on a real-time sliding window according to claim 4, characterized in that: The specific steps of step 6 include: Step 6.1: subtract the predicted value from the actually collected turntable frame angle output value to obtain the prediction error of the turntable frame angle; Step 6.2: Obtain the gyro measurement angular velocity error corresponding to the prediction error according to formula (3); Step 6.3: Use a time window with a length of 36 and a step size of 1, calculate a standard deviation for every 36 angular velocity errors, and obtain the gyro bias stability at that moment; Step 6.4: Slide the time window forward by one step, and repeat steps 6.3 and 6.4 until all data are calculated, and the variation law of the zero bias stability of the hemispherical resonant gyroscope is obtained; Step 6.5: Compare the obtained zero bias stability variation law of the hemispherical resonant gyroscope with the zero bias stability variation law obtained by the polynomial-based compensation method to verify the compensation effect.
