A method for modeling temperature errors of fiber optic gyroscopes based on parameter uncertainty

Abstract

The present invention provides a method for modeling the temperature error of a fiber optic gyroscope based on parameter uncertainty. First, the model parameters (or variables) are sorted in order of uncertainty from small to large, and the parameters with small uncertainty are preferentially selected and introduced into the compensation model. At the same time, the contribution of each parameter in the model is tested, and the parameters with large uncertainty are selected first, and the compensation model is eliminated until there are no new introduced items or new eliminated items in the model parameters. For gyros that require an accuracy of ≤0.01° / h after compensation, if the parameter contribution is ≤0.0001° / h, it can be eliminated from the model. This method effectively solves the contradiction between the applicability and accuracy of the model, and can achieve temperature error compensation of high-precision fiber optic gyros.

Description

Technical Field

[0001] The invention belongs to the technical field of instrument testing, and in particular relates to a fiber optic gyroscope temperature error modeling method based on parameter uncertainty. Background Art

[0002] Fiber optic gyros (FOGs), a new generation of inertial devices, offer advantages such as high reliability, long life, short startup time, and a wide dynamic range, and have been widely used in inertial navigation systems. However, the core components of FOGs are sensitive to temperature. Temperature drift occurs when the ambient temperature fluctuates, significantly affecting the gyro's full-temperature accuracy. Due to technical and cost limitations, improvements to the FOG structure and fiber ring winding process can only partially suppress bias drift but cannot completely overcome the effects of temperature. Therefore, temperature modeling is necessary to compensate for difficult-to-control temperature errors.

[0003] Research has shown that gyro temperature error models can be broadly divided into two categories: linear and nonlinear. Considering practical engineering applications, linear models are simpler and more practical. In 2018, Luo Quan et al. segmented data based on temperature variations and employed a linear segmented compensation model to effectively reduce the temperature error of a fiber optic gyro. In 2019, Dai Shaowu et al. established temperature error models for fiber optic gyros based on both the heating and cooling phases. Regarding model parameters, Dai Shaowu et al. proposed introducing a temperature gradient hysteresis term into the temperature drift error model. In 2019, Hou Honglu et al. proposed a polynomial bias drift temperature compensation model based on temperature and temperature change rate to address the issue of real-time temperature compensation in engineering applications. In 2020, Wu Ying et al. used the MAA sliding average algorithm to establish a bias compensation model with temperature hysteresis for rising and falling temperatures, effectively improving the bias stability of the gyro. Analyzing these research results, the challenge for linear models is determining the model parameters. From a mathematical fitting perspective, the more model parameters, the higher the compensation accuracy. However, from the perspective of engineering applications, the more parameters the model has, the worse the applicability of the compensation. Summary of the Invention

[0004] The technical problem solved by the present invention is: to overcome the shortcomings of the existing technology and propose a fiber optic gyroscope temperature error modeling method based on parameter uncertainty, which can accurately and quickly realize high-precision fiber optic gyroscope temperature error modeling to compensate for the fiber optic gyroscope temperature error, and effectively solve the contradiction between the applicability and accuracy of the model.

[0005] The technical solutions of the present invention are as follows:

[0006] A fiber optic gyroscope temperature error modeling method based on parameter uncertainty includes the following steps:

[0007] (1) Obtaining the bias and temperature data of the fiber optic gyroscope;

[0008] (2) Preprocessing the bias and temperature data;

[0009] (3) obtaining a temperature-related item data set, wherein the temperature-related items include a temperature item, a temperature square item, a temperature cube item, a temperature change rate square item, a temperature change rate cube item, a temperature and temperature change rate cross-coupling item, a temperature square and temperature change rate cross-coupling item, and a temperature and temperature change rate square cross-coupling item;

[0010] (4) Select a temperature-related term that meets the requirements based on the uncertainty and contribution, and establish a polynomial constraint relationship between the temperature-related term and the zero bias; remove the selected temperature-related term from the temperature-related term data set;

[0011] (5) Calculate the contribution of all selected temperature-related items and eliminate the temperature-related items whose contribution does not meet the requirements;

[0012] (6) If the temperature-related items are eliminated, proceed to step (7);

[0013] (7) Repeat steps (4) to (6) until there are no temperature-related items that meet the requirements in the temperature-related item data set, and then proceed to step (8);

[0014] (8) Establish a fiber optic gyroscope temperature error model with zero bias as the dependent variable and the selected temperature-related items as independent variables.

[0015] Preferably, in step (1), the zero bias and temperature data of the fiber optic gyroscope are obtained by:

[0016] Connect the power box to the fiber optic gyroscope through a test cable to supply power to the fiber optic gyroscope; electrically connect the computer running the acquisition software to the fiber optic gyroscope to collect data from the fiber optic gyroscope;

[0017] The fiber optic gyroscope is placed in a temperature chamber with a vibration isolation foundation, which is used to simulate the temperature environment; the power box and computer are located outside the temperature chamber.

[0018] Preferably, in step (2), performing data preprocessing on the zero bias and temperature data refers to smoothing the zero bias data and the temperature data, and at the same time, differentiating the temperature data to obtain temperature change rate data.

[0019] Preferably, in step (4), a temperature-related item that meets the requirements is selected based on the uncertainty and contribution, and the implementation method is as follows:

[0020] Calculate the uncertainty of each temperature-related data;

[0021] Sort the temperature-related items in order of increasing uncertainty;

[0022] Select the temperature-related item with the smallest uncertainty and add it to the temperature error compensation model;

[0023] Calculate the contribution of the temperature-related term. If the difference between the temperature compensation accuracy of the fiber optic gyroscope after adding the temperature-related term and the temperature compensation accuracy without adding the temperature-related term is less than 0.0001° / h, the temperature-related term does not meet the requirements. If the difference between the temperature compensation accuracy of the fiber optic gyroscope after adding the temperature-related term and the temperature compensation accuracy without adding the temperature-related term is greater than 0.0001° / h, the temperature-related term meets the requirements.

[0024] Preferably, the contribution of the temperature-related term is calculated by adding the temperature-related term to the temperature error compensation model to form a new model. Compared with the original compensation model without the temperature-related term, the accuracy of the fiber optic gyroscope improved after compensation with the new model is the contribution of the current temperature-related term, that is, the contribution of the temperature-related term = the accuracy of the fiber optic gyroscope after compensation with the new model minus the accuracy of the fiber optic gyroscope after compensation with the original compensation model.

[0025] Preferably, the implementation of step (5) is as follows: calculate the contribution of all selected temperature-related items, and remove the temperature-related items whose contribution does not meet the requirements.

[0026] Sort all selected temperature-related items in descending order of uncertainty;

[0027] The contribution of each temperature-related item is calculated in sequence according to the above order, and the temperature-related items whose contribution does not meet the requirements are eliminated.

[0028] Preferably, the method for calculating the contribution of the i-th temperature-related term is as follows:

[0029] After the i-th temperature-related term is removed from the temperature error compensation model, a new model is formed, and the new model is used for temperature compensation. The difference in accuracy between the fiber optic gyroscope after compensation with the new model and the original model without the temperature-related term is the contribution of the i-th temperature-related term.

[0030] Preferably, the method of eliminating temperature-related items whose contribution does not meet the requirements is as follows:

[0031] Select the temperature-related term with the smallest contribution;

[0032] If the temperature compensation accuracy of the FOG after removing the temperature-related term with the smallest contribution differs from the temperature compensation accuracy of the FOG after adding the temperature-related term by more than 0.0001° / h, the temperature-related term is not removed; otherwise, the temperature-related term is removed.

[0033] Preferably, the temperature compensation accuracy of the fiber optic gyroscope is expressed by zero bias stability (100s, 1σ).

[0034] Preferably, the uncertainty ΔZ of a temperature-related term satisfies:

[0035]

[0036] T is the temperature data, ΔT is the measurement error of the temperature data, and f is the functional relationship between the temperature-related term and the temperature data. Find the partial derivative of the function f with respect to T.

[0037] The present invention has the following beneficial effects:

[0038] The present invention provides a method for modeling the temperature error of an optical fiber gyroscope based on parameter uncertainty. First, according to the error propagation law, the uncertainty of temperature, temperature change rate, cross-coupling terms between temperature and temperature change rate, and their higher-order terms are sorted. According to the size of the parameter uncertainty, parameters with small uncertainty are preferentially selected and added to the compensation model. According to the size of the parameter uncertainty, parameters with large uncertainty are first selected and the compensation model is eliminated. This method avoids the problem of poor subsequent adaptability of the model due to too many parameters. Secondly, it avoids the problem of poor subsequent adaptability of the model due to too few parameters. Thirdly, the modeling method is simple and easy to implement and can be extended to different parameter sets. It can accurately and quickly realize the temperature error modeling and compensation of the optical fiber gyroscope with high precision, and effectively solve the contradiction between the applicability and accuracy of the model. Zero bias modeling and compensation are carried out using the temperature data of the optical fiber gyroscope, and the effectiveness of the model is verified. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 This is a flow chart of the fiber optic gyroscope temperature error modeling based on parameter uncertainty of the present invention;

[0040] Figure 2 This is the fiber optic gyroscope test curve at room temperature of the present invention;

[0041] Figure 3 This is the fiber optic gyroscope test curve at a constant temperature point (-40°C) of the present invention;

[0042] Figure 4 This is the fiber optic gyroscope test curve at a constant temperature point (+60°C) of the present invention;

[0043] Figure 5 This is the test curve of the fiber optic gyroscope under variable temperature of the present invention (first test);

[0044] Figure 6 This is the test curve of the fiber optic gyroscope under variable temperature of the present invention (second test);

[0045] Figure 7 This is the test curve of the fiber optic gyroscope under variable temperature of the present invention (the third test);

[0046] Figure 8 This is the test curve of the fiber optic gyroscope under variable temperature of the present invention (the fourth test);

[0047] Figure 9 This is the verification curve of the fiber optic gyroscope after compensation under variable temperature of the present invention. DETAILED DESCRIPTION

[0048] The present invention is described in detail below, and the features and advantages of the present invention will become clearer and more distinct with the following description.

[0049] The word "exemplary" is used exclusively herein to mean "serving as an example, example, or illustration." Any embodiment described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments. Although various aspects of the embodiments are shown in the drawings, the drawings are not necessarily drawn to scale unless otherwise indicated.

[0050] In order to overcome the deficiencies in the prior art, the present invention provides a fiber optic gyroscope temperature error modeling method based on parameter uncertainty. In view of the difficulty in selecting parameters in the multivariate linear regression model, based on the size of the parameter uncertainty, parameters with small uncertainty are preferentially selected and added to the compensation model. For a gyroscope with a speed of 0.01° / h, if the parameter contribution is less than 0.0001° / h, then the parameter does not need to be added. Based on the size of the parameter uncertainty, parameters with large uncertainty are selected first and the compensation model is eliminated. For a gyroscope with a speed of 0.01° / h, if the parameter contribution is less than 0.0001° / h, then the parameter can be eliminated. The compensation model obtained by adopting this modeling method can effectively avoid the problem of poor subsequent adaptability of the model or poor accuracy after compensation due to too many or too few parameters.

[0051] The parameters in this invention refer to the temperature-related terms involved in the compensation model, including temperature, temperature change rate, and cross-coupling terms between temperature and temperature change rate. The parameter uncertainty refers to the uncertainty inherent in the function formed by the observed quantity, which, according to the error propagation law, implies that if the observed quantity has an error, then the observed quantity also has an error. Typically, only temperature data T is directly measured in an experiment, while other temperature-related data Z is obtained through calculations and can be expressed as a function of Z = f(T, t), where t represents time.

[0052] Since the variable T has a measurement error ΔT, its function also has an error ΔZ. According to the error propagation law, the error of the variable and the error of the function can be approximately expressed by the total differential of the function: is the partial derivative of the function with respect to the variable. Parameter uncertainties are ranked according to the magnitude of the ΔZ uncertainty. When introducing model parameters, prioritize those with small uncertainties. When eliminating model parameters, prioritize those with large uncertainties. Finally, a model with high compensation accuracy and strong environmental adaptability is developed. Changes in the fiber optic gyroscope test environment can cause changes in gyro data and temperature data. Poor model adaptability can lead to poor subsequent compensation accuracy.

[0053] The parameters mentioned above refer to the temperature-related terms involved in the compensation model, including temperature, temperature change rate, cross-coupling terms between temperature and temperature change rate, and their higher-order terms. Temperature data is a directly observable quantity. Other related terms are obtained through calculations on temperature data. Generally, the highest order of temperature-related quantities is no greater than 3, and the order of parameter uncertainty from largest to smallest is:

[0054]

[0055] Therefore, prioritizing parameters with small uncertainties and entering them into the compensation model can effectively improve the subsequent adaptability of the model.

[0056] The present invention comprises the steps of:

[0057] The modeling method based on parameter uncertainty includes obtaining zero bias, temperature data, data preprocessing, reading zero bias and temperature-related data, introducing variables, eliminating variables, checking whether variables need to be eliminated, checking whether variables need to be added, and establishing a polynomial model.

[0058] To obtain bias and temperature data, Fiber Optic Gyro 1_1 is electrically connected to a power supply box 1_4 and a computer 1_3 running acquisition software 1_2, powering Fiber Optic Gyro 1_1 and acquiring data. Fiber Optic Gyro 1_1 is placed in a temperature chamber 1_6 with a vibration-isolating foundation to simulate a temperature environment.

[0059] Smoothing of gyro bias data and temperature-related data is performed, with selectable smoothing times of 10s or 100s, to reduce the impact of transient noise in the raw data on model parameters. Simultaneously, temperature data is differentiated to obtain temperature change rate data.

[0060] The way to introduce variables is as follows:

[0061] Parameters with the smallest uncertainty are prioritized to form the compensation model, starting with the largest uncertainty. The contribution of each parameter in the new model is examined, and the parameter with the largest contribution is selected for inclusion in the compensation model. For gyros with accuracy better than 0.01° / h, if the parameter contribution is within 0.0001° / h, the contribution is too small and can be omitted. The contribution of a variable is examined by comparing the accuracy difference between the model with and without the variable for temperature compensation. This difference is expressed as bias stability (100s, 1σ). The parameter with the largest contribution is prioritized. If the difference in accuracy between the compensated model and the model without the variable is less than 0.0001° / h, the contribution is small and can be omitted. If the difference in accuracy between the compensated model and the model without the variable is greater than 0.0001° / h, the variable is added.

[0062] Variables were eliminated as follows: Parameters with the largest uncertainties were prioritized, and the contribution of each variable was examined, with the smallest contribution selected. For gyros with a parameter contribution better than 0.01° / h, if it was within 0.0001° / h, the contribution was too small and could be eliminated.

[0063] To examine the contribution of each variable, we use a model with and without that variable for temperature compensation. The difference in accuracy after compensation is expressed as bias stability (100s, 1σ). The parameter with the smallest contribution is selected and removed. If the difference in accuracy between the compensation with and without that variable is greater than 0.0001° / h, the variable's contribution is significant and is not removed.

[0064] Each variable in the compensation model needs to be tested for its contribution, as the contribution of a variable will change with the addition of new variables or the removal of old variables. Avoid parameters with low contribution, which may lead to poor model adaptability.

[0065] The contribution of each variable not included in the model needs to be tested, because the contribution of variables will change with the addition of new variables or the removal of old variables. Avoid not including parameters with high contribution, which will lead to poor model accuracy.

[0066] Let T represent the temperature observation value, and Z = f(T, t) be a function of the observation value. Then the temperature, the temperature change rate, the cross-coupling term between the temperature change rate and the temperature change rate, and their higher-order terms are defined as:

[0067]

[0068] If there is a measurement error ΔT in the variable T, its function also has an error ΔZ, which can be approximately expressed by the total differential of the function in is the partial derivative of the function with respect to the variable. Therefore, the temperature error, the temperature change rate error, the cross-coupling term error between temperature and temperature change rate, and its higher-order term error are defined as:

[0069]

[0070] The order of uncertainty from greatest to least is as follows:

[0071] ΔZ6>ΔZ9>ΔZ8>ΔZ3>ΔZ5>ΔZ7>ΔZ2>ΔZ4>ΔZ1

[0072] Right now

[0073]

[0074] T——temperature data; ΔT——the difference between the true value and the measured value of temperature data;

[0075] T 2 ——Square of temperature data; T 3 ——the cube of temperature data;

[0076] ——Temperature change rate data, equivalent to

[0077] ——Square of temperature change rate data; ——the cube of the temperature change rate data;

[0078] ——The product of temperature and temperature change rate data;

[0079] ——The product of the square of the temperature and the temperature change rate data;

[0080] ——The product of temperature and the square of the temperature change rate data;

[0081] ——Find the partial derivative of the rate of temperature change with respect to temperature.

[0082] Example:

[0083] like Figure 1 As shown, this embodiment provides a method for modeling temperature errors of a fiber optic gyroscope based on parameter uncertainty, comprising the following steps:

[0084] S1. Rank the uncertainty of all model parameters using the method described above. In this example, the model parameters used to model gyro temperature data include the temperature term, the temperature squared term, the temperature cubed term, the temperature change rate squared term, the temperature change rate squared term, the temperature change rate cubed term, the cross-coupling term between temperature and temperature change rate, the cross-coupling term between temperature squared and temperature change rate, and the cross-coupling term between temperature and temperature change rate squared, for a total of nine terms. The temperature sensor is a DS18B20 located near the center of the fiber optic loop. The output frequency of the gyro and temperature sensor is 1 Hz.

[0085] S2: Build a test system and test the accuracy of the system at room temperature and constant temperature points to ensure that the accuracy index is met. Only after the index is met can temperature modeling be carried out.

[0086] Test at room temperature, power on for 6 hours, test the fiber optic gyroscope curve as follows Figure 2 ,Using the data of the last two hours, the bias stability is calculated to be 0.0033° / h, which reflects the gyro's noise level at room temperature.

[0087] In the constant temperature point test, the temperature chamber 14 is set to constant temperature of -40℃ and +60℃ respectively, and the power is turned on for 6 hours. The curve of the fiber optic gyroscope is as follows Figure 3 、 Figure 4 Using the data of the last two hours, the zero bias stability is calculated to be 0.0065° / h and 0.0041° / h, which reflects the low high temperature noise level of the fiber optic gyroscope and the vibration resistance of the temperature box 14.

[0088] The full-temperature test was conducted with a set temperature range of -40°C to +60°C, a temperature ramp rate of 1°C / min, and a three-hour hold time. The calculated full-temperature bias stability was 0.1523° / h, reflecting the drift of the fiber optic gyroscope with temperature changes.

[0089] S3 uses acquisition software to collect and save the FOG output and temperature signals. These signals undergo preprocessing, including data smoothing and temperature-related calculations. There's no direct way to measure the rate of temperature change. Instead, the ratio of the temperature difference to the time taken is used as the temperature rate of change.

[0090]

[0091] Where T is the temperature, t is the sampling time, and Δt is the time interval. The unit of temperature change rate is generally °C / min.

[0092] If the fluctuation of the temperature signal at a certain moment is ΔT, the fluctuation of the temperature change rate may be 2ΔT, which is greater than the fluctuation of the temperature signal. Therefore, it is necessary to smooth the temperature data and then calculate the temperature change rate. As

[0093]

[0094] Where N is the number of data smoothing, which corresponds to the data smoothing time. T(k) is the temperature value at time k.

[0095] S4. Sort the temperature-related data in order of parameter uncertainty from small to large, give priority to parameters with small uncertainty to add to the compensation model, and test the degree of parameter contribution. If the parameter contribution is ≤0.0001° / h, it means the contribution is too small and it is chosen not to be added.

[0096] The contribution degree of the parameters in the model is tested in descending order of parameter uncertainty. If the contribution degree of the parameter is ≤0.0001° / h, it means that the contribution degree is too small and it is selected to be eliminated.

[0097] In practical problems, there is often more than one variable that affects the result y. Generally, let x1, x2, ..., x m variables, the multiple linear regression model is expressed as

[0098] y=β0+β1x1+β2x2+…+β m x m +ε (5)

[0099] where β0,β1,…β m There are m+1 unknowns. If we have n sets of observations, the linear regression model is

[0100]

[0101] Written in matrix form as

[0102]

[0103] in

[0104]

[0105] The unknowns of the multivariate linear regression equation are usually estimated by least squares.

[0106]

[0107] The sample regression model under ordinary least squares method is

[0108]

[0109] The compensated data is Seek Standard deviation To compensate for the gyro bias stability.

[0110] x1,x2,…,x m ——Dependent variable in the model, in this invention, refers to the temperature data related item;

[0111] y——independent variable in the model, which refers to the gyro bias value in the present invention;

[0112] ——model estimates; ——the difference between the measured and estimated values ​​of the model;

[0113] —— The standard deviation of , which refers to the gyro bias stability in this invention, is expressed as (100s, 1σ);

[0114] β0,β1,…β m ——coefficients in the model;

[0115] In this example, 1) calculate T, T 2 、 T 3 、 The contribution of each parameter, that is, the gyro accuracy after compensation is 0.01503° / h, 0.01105° / h, 0.1519° / h, 0.1336° / h, 0.1486° / h, 0.1522° / h, 0.1174° / h, 0.1472° / h, 0.05489° / h, and the zero bias stability before compensation is 0.1523° / h. In comparison, the one with the greatest contribution is Add compensation model:

[0116]

[0117] 2) Calculate T and T in the order of uncertainty from small to large. 2 、 T 3 、 The contribution of each parameter, that is, After the new model is compensated, the gyro accuracy is 0.01008° / h, 0.01012° / h, 0.01018° / h, 0.01064° / h, 0.01066° / h, 0.01101° / h, 0.01105° / h, and 0.01088° / h. The zero bias stability after the last compensation is 0.01105° / h. In comparison, T has the greatest contribution. Add the compensation model:

[0118]

[0119] 3) Calculate T in order of uncertainty from small to large 2 、 T 3 、 The contribution of each parameter, that is, the contribution to T, After the new model is compensated, the gyro accuracy is 0.00989° / h, 0.01008° / h, 0.01047° / h, 0.01043° / h, 0.01083° / h, 0.01080° / h, and 0.00999° / h. The zero bias stability after the last compensation is 0.01008° / h. In comparison, the one with the greatest contribution is T 2 , add compensation model:

[0120]

[0121] 4) Test the contribution of existing parameters and choose to exclude temperature change rate data After compensation, the gyro accuracy is 0.1515° / h, so the model cannot be eliminated. If the temperature data T is eliminated, the gyro accuracy after compensation is 0.01012° / h, so the model cannot be eliminated.

[0122] 5) Calculate the uncertainty in ascending order. T 3 、 The contribution of each parameter, that is, the contribution to T, T 2 After the new model is compensated, the gyro accuracy is 0.00931° / h, 0.00989° / h, 0.00983° / h, 0.00979° / h, 0.00986° / h, and 0.00947° / h. The zero bias stability after the last compensation is 0.00989° / h. In comparison, the one with the highest contribution is Add compensation model:

[0123]

[0124] 6) Test the contribution of existing parameters and select the temperature square term data T to be eliminated in descending order of uncertainty. 2 After compensation, the gyro accuracy is 0.01008° / h, and the temperature change rate data is selected to be eliminated. The gyro accuracy after compensation is 0.1330° / h. If the temperature data T is removed, the gyro accuracy after compensation is 0.00933° / h, and then all parameters cannot be removed from the model.

[0125] 7) Calculate the uncertainty in ascending order. T 3 、 The contribution of each parameter, that is, the contribution to T, T 2 、 After the new model is compensated, the gyro accuracy is 0.00901° / h, 0.00879° / h, 0.009716° / h, 0.009002° / h, and 0.00988° / h. The zero bias stability after the last compensation is 0.00931° / h. In comparison, the one with the greatest contribution is T 3 , add compensation model:

[0126]

[0127] 8) Test the contribution of existing parameters and select and eliminate the temperature square term data in descending order of uncertainty. After compensation, the gyro accuracy is 0.00984° / h, and the temperature square term data T is selected to be eliminated. 2 After compensation, the gyro accuracy is 0.009468° / h, and the temperature change rate data is selected to be eliminated. The gyro accuracy after compensation is 0.1327° / h. If the temperature data T is removed, the gyro accuracy after compensation is 0.00902° / h, and the parameters cannot be removed from the model.

[0128] 9) Calculate the uncertainty in ascending order. The contribution of each parameter, that is, the contribution to T, T 2 、 T 3 After the new model is compensated, the gyro accuracy is 0.00879° / h, 0.00878° / h, 0.00872° / h, and 0.00875° / h. Compared with the zero bias stability of 0.00879° / h after the previous compensation, the contribution is less than 0.0001° / h, so it is not included in the compensation model.

[0129] After checking, there are no parameters to be introduced in the parameter set and no parameters to be eliminated in the model, so the final compensation model is formula (14). Considering the applicability of the model, the gyro and temperature data are collected at four different times, as shown in the following example: Figure 5 、 Figure 6 、 Figure 7 and Figure 8 , respectively, establish the temperature error compensation model, the model coefficients and compensation accuracy are shown in Table 1.

[0130] Table 1

[0131]

[0132]

[0133] The final polynomial compensation model can be obtained by averaging the four sets of model coefficients (see Table 2).

[0134] Table 2

[0135] Model coefficients Model coefficient average C -9.3257 T -2.3903e-4 <![CDATA[T 2 ]]> -1.1977e-5 <![CDATA[T 3 ]]> 1.7017e-7 dT 0.3179 T·dT 4.1142e-4

[0136] In order to verify the compensation effect of the model, a random temperature change experiment was set up. The original data curve of the gyro and the data curve after compensation are shown in Figure 9 In the verification experiment, the gyro accuracy before compensation was 0.0886° / h, and after compensation using this model, the gyro accuracy was 0.0087° / h, which is better than 0.01° / h.

[0137] The present invention addresses the problem of being unable to accurately select model parameters when using a multivariate linear regression model to model and compensate for fiber optic gyroscope temperature errors. A method for selecting model parameters based on the magnitude of parameter uncertainty is proposed, effectively avoiding the problem of poor adaptability or accuracy caused by too many or too few model parameters. First, the model parameters are sorted in ascending order of uncertainty, with parameters with low uncertainty being preferentially selected and introduced into the compensation model. Simultaneously, each parameter in the model is subjected to an importance check, with parameters with high uncertainty being selected first and removed from the compensation model until no new terms are introduced or removed from the model. For gyros requiring an accuracy of ≤0.01° / h after compensation, if the parameter contribution is ≤0.0001° / h, the gyroscope can be removed from the model. This method effectively resolves the conflict between model applicability and accuracy and provides guidance for temperature error compensation in high-precision fiber optic gyros.

[0138] The present invention has been described in detail above with reference to specific embodiments and exemplary examples. However, these descriptions should not be construed as limiting the present invention. Those skilled in the art will appreciate that various equivalent substitutions, modifications, or improvements may be made to the technical solutions and implementations of the present invention without departing from the spirit and scope of the present invention, all of which fall within the scope of the present invention. The scope of protection of the present invention shall be determined by the appended claims.

[0139] The contents not described in detail in the specification of the present invention belong to the common knowledge of those skilled in the art.

Claims

1. A fiber optic gyroscope temperature error modeling method based on parameter uncertainty, characterized in that: The steps include: (1) Obtaining the bias and temperature data of the fiber optic gyroscope; (2) Preprocessing the bias and temperature data; (3) obtaining a temperature-related item data set, wherein the temperature-related items include a temperature item, a temperature square item, a temperature cube item, a temperature change rate square item, a temperature change rate cube item, a temperature and temperature change rate cross-coupling item, a temperature square and temperature change rate cross-coupling item, and a temperature and temperature change rate square cross-coupling item; (4) Select a temperature-related term that meets the requirements based on the uncertainty and contribution, and establish a polynomial constraint relationship between the temperature-related term and the zero bias; remove the selected temperature-related term from the temperature-related term data set; (5) Calculate the contribution of all selected temperature-related items and eliminate the temperature-related items whose contribution does not meet the requirements; (6) If the temperature-related items are eliminated, proceed to step (7); (7) Repeat steps (4) to (6) until there are no temperature-related items that meet the requirements in the temperature-related item data set, and then proceed to step (8); (8) Establishing a fiber optic gyro temperature error model with zero bias as the dependent variable and the selected temperature-related items as the independent variables; In step (4), a temperature-related term that meets the requirements is selected based on the uncertainty and contribution, and the implementation method is as follows: Calculate the uncertainty of each temperature-related data; Sort the temperature-related items in order of increasing uncertainty; Select the temperature-related item with the smallest uncertainty and add it to the temperature error compensation model; Calculate the contribution of the temperature-related term. If the difference between the temperature compensation accuracy of the fiber optic gyroscope after adding the temperature-related term and the temperature compensation accuracy without adding the temperature-related term is less than 0.0001° / h, the temperature-related term does not meet the requirements. If the difference between the temperature compensation accuracy of the fiber optic gyroscope after adding the temperature-related term and the temperature compensation accuracy without adding the temperature-related term is greater than 0.0001° / h, the temperature-related term meets the requirements.

2. The fiber optic gyroscope temperature error modeling method based on parameter uncertainty according to claim 1, characterized in that: In the step (1), the zero bias and temperature data of the fiber optic gyroscope are obtained by: A power supply box (1_4) is connected to a fiber optic gyroscope (1_1) via a test cable to supply power to the fiber optic gyroscope (1_1); a computer (1_3) running acquisition software (1_2) is electrically connected to the fiber optic gyroscope (1_1) to collect data from the fiber optic gyroscope (1_1); The fiber optic gyroscope (1_1) is placed in a temperature box (1_6) with a vibration isolation foundation, and the temperature box is used to simulate a temperature environment; the power box (1_4) and the computer (1_3) are located outside the temperature box (1_6).

3. The fiber optic gyroscope temperature error modeling method based on parameter uncertainty according to claim 1, characterized in that: In the step (2), performing data preprocessing on the zero bias and temperature data refers to smoothing the zero bias data and the temperature data, and at the same time, differentiating the temperature data to obtain temperature change rate data.

4. The method for modeling temperature error of a fiber optic gyroscope based on parameter uncertainty according to claim 1, wherein: Calculating the contribution of the temperature-related term refers to the process of adding the temperature-related term to the temperature error compensation model to form a new model. Compared with the original compensation model without the temperature-related term, the improved accuracy of the fiber optic gyroscope after compensation with the new model is equal to the contribution of the current temperature-related term. That is, the contribution of the temperature-related term = the accuracy of the fiber optic gyroscope after compensation with the new model minus the accuracy of the fiber optic gyroscope after compensation with the original compensation model.

5. The fiber optic gyroscope temperature error modeling method based on parameter uncertainty according to claim 1, characterized in that: The implementation method of step (5) is as follows: calculate the contribution of all selected temperature-related items, and eliminate the temperature-related items whose contribution does not meet the requirements Sort all selected temperature-related items in descending order of uncertainty; The contribution of each temperature-related item is calculated in sequence according to the above order, and the temperature-related items whose contribution does not meet the requirements are eliminated.

6. The method for modeling temperature error of a fiber optic gyroscope based on parameter uncertainty according to claim 5, characterized in that: The method for calculating the contribution of the i-th temperature-related term is as follows: After the i-th temperature-related term is removed from the temperature error compensation model, a new model is formed, and the new model is used for temperature compensation. The difference in accuracy between the fiber optic gyroscope after compensation with the new model and the original model without the temperature-related term is the contribution of the i-th temperature-related term.

7. The method for modeling temperature error of a fiber optic gyroscope based on parameter uncertainty according to claim 6, characterized in that: The method for eliminating temperature-related items whose contribution does not meet the requirements is as follows: Select the temperature-related term with the smallest contribution; If the temperature compensation accuracy of the FOG after removing the temperature-related term with the smallest contribution differs from the temperature compensation accuracy of the FOG after adding the temperature-related term by more than 0.0001° / h, the temperature-related term is not removed; otherwise, the temperature-related term is removed.

8. A fiber optic gyroscope temperature error modeling method based on parameter uncertainty according to claim 4 or 6, characterized in that: The temperature compensation accuracy of the fiber optic gyroscope is expressed by zero bias stability (100s, 1σ).

9. A fiber optic gyroscope temperature error modeling method based on parameter uncertainty according to claim 1 or 5, characterized in that: The uncertainty ΔZ of a temperature-related term satisfies: T is the temperature data, ΔT is the measurement error of the temperature data, and f is the functional relationship between the temperature-related term and the temperature data. Find the partial derivative of the function f with respect to T.

Images