A method for improving the data accuracy of a gyro array in a stable platform

Abstract

The present invention relates to the field of data processing, and specifically to a method for improving the data accuracy of a gyro array on a stable platform. The method includes the following steps: S1. Primary screening: A low-pass filter is used to smooth and denoise the signal, and the performance of the gyro unit is quantified in the form of variance. The larger the variance, the more severe the gyro jitter; S2. Secondary screening: In the normality test, a frequency distribution histogram is plotted, the fitting Q-Q plot is visualized, and the probability density curve of the normal distribution is fitted; S3. Temperature drift compensation: The gyro drift value is obtained by the encoder differential velocity closed-loop. In the full temperature range, when the velocity obtained by the encoder differential is 0, the gyro data on the stable two-axis platform is sampled, and then the temperature drift curve after acquisition is fitted. The present invention uses the secondary screening algorithm and the temperature drift compensation algorithm to improve the output data accuracy of the gyro array, and greatly improves the control effect of the stable platform.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to a method for improving the data accuracy of a gyro array of a stable platform. Background Art

[0002] In recent years, gyro arrays have been widely used in the control of stable platforms, and can be used as speed feedback devices for speed-loop closed-loop control. Although traditional MEMS array gyros applied to stable platforms can complete relatively accurate speed feedback for stable platforms, there is still a gap between their accuracy and that of higher-precision gyros (such as fiber optic gyros). The main reason is that traditional gyro arrays usually obtain the output angular velocity by simply taking the average of the outputs of each gyro unit to stabilize the platform. This method has the following obvious disadvantages:

[0003] If there are monomers with unstable performance and large jitters in the array, even if they only account for a small part of the array, due to averaging to obtain the final result, they will seriously affect the accuracy of the entire output data, thereby affecting the stability performance of the stable platform. In addition, the data temperature drift of MEMS gyros will ultimately also cause serious deviations in the output data. Summary of the Invention

[0004] The purpose of the present invention is to address the problems in the background art and propose a method for improving the data accuracy of a gyro array of a stable platform, which uses a double screening algorithm and a temperature drift compensation algorithm to improve the accuracy of the output data of the gyro array and greatly enhance the control effect of the stable platform.

[0005] The technical solution of the present invention, a method for improving the data accuracy of a gyro array of a stable platform, includes the following steps:

[0006] S1. Primary screening: A low-pass filter is used to smooth and denoise the signal, and the performance of the gyro unit is quantified in the form of variance. The larger the variance, the more severe the gyro jitter;

[0007] S2. Secondary screening: In the normality test, a frequency distribution histogram is plotted, a visualized fitted Q-Q plot is made, and a normal distribution probability density curve is fitted;

[0008] S3. Temperature drift compensation: The gyro drift value is obtained by encoder differential speed closed-loop. In the full temperature range, when the speed obtained by encoder differential is 0, the gyro data on the stable two-axis platform is sampled, the sampling period is 0.001 s, and then the temperature drift curve after acquisition is fitted. The principle is to use the least squares method to determine its goodness of fit R 2 Threshold is used as the piecewise fitting point and compensation is performed.

[0009] Preferably, in S1, the low-pass filter algorithm formula:

[0010] Y(n) = a * X(n) + (1 - a) * Y(n - 1)

[0011] Where a is the filtering coefficient, X(n) is the current input value of the gyroscope, Y(n - 1) is the output value after the previous gyroscope filtering, and Y(n) is the output value of the current gyroscope filtering.

[0012] Preferably, the variance calculation formula is:

[0013] D(X) = Var(X) = E[X - E(X)] 2

[0014] Adopt the deformation formula of variance: D(X) = E(X 2 ) - [E(X)] 2 , the data set is the gyroscope angular velocity after 10,000 acquisitions. After sending the real-time collected gyroscope data into the low-pass filtering model, the filtered gyroscope value is obtained and the mean differences E(X), E(X 2 ) are calculated. The variances of each gyroscope are obtained through the variance deformation formula, and the gyroscopes with better performance are selected through the quicksort method.

[0015] Preferably, in S2, the process of drawing the Q-Q plot is as follows:

[0016] A1. Sort the data set and calculate the empirical quantiles;

[0017] A2. Compare all the calculated empirical quantiles with the theoretical quantiles of the standard normal distribution;

[0018] A3. Plot the point pairs on the graph, where the x-axis represents the empirical quantiles and the y-axis represents the theoretical quantiles;

[0019] A4. Evaluate the normality of the data by observing the distribution pattern of these points.

[0020] Preferably, when fitting the normal distribution probability density curve, first preprocess the gyroscope data set to remove outliers;

[0021] Use the maximum likelihood estimation to estimate the mean μ and standard deviation σ of the normal distribution. By constructing the likelihood function of the data set and using the gradient descent method to maximize the likelihood function, the optimal mean and standard deviation are solved;

[0022] The probability density function of the normal distribution is:

[0023]

[0024] The data set X = {x1, x2,..., x n} and the likelihood function is the joint probability density function, and the joint probability density function is the product of the probability density functions of each data point:

[0025]

[0026] Log-likelihood function:

[0027]

[0028] Set the objective function:

[0029] Maximize the log-likelihood function -l(μ,σ 2 ) In the optimization problem, it is usually transformed into minimizing its negative value, that is, maximizing l is equivalent to minimizing -l:

[0030]

[0031] Use the gradient descent method to calculate the partial derivatives of J with respect to μ and σ 2 :

[0032]

[0033] Update μ and σ 2 :

[0034]

[0035]

[0036] where α is the learning rate, used to control the step size of the update;

[0037] The fitting process is an iterative optimization process. Starting from the initial parameter values, the parameter values are continuously optimized through multiple iterations until the preset convergence conditions are met.

[0038] Preferably, in S3, for the encoder differential speed closed-loop to obtain the gyro drift value, when the speed is 0, collect the gyro data, which is the current gyro drift angular velocity.

[0039] Preferably, the optimal solutions of k and b are determined by calculating the optimal extreme points for compensation, and the algorithm formula is:

[0040] The linear function is:

[0041] where x i is the original data collected by the gyro, k is the slope of the fitting linear function, and b is the intercept of the function;

[0042] The mean squared error of the linear loss function:

[0043] where, represents finding the independent variable value that makes the function value reach the minimum;

[0044] According to the principle of the mean square error of the above linear loss function, it is transformed into finding the minimum value. Let Find the extreme points of the derivative of L and calculate the optimal solutions of k and b. The calculation formulas for their respective partial derivatives are as follows:

[0045]

[0046] Compared with the prior art, the present invention has the following beneficial technical effects:

[0047] The present invention uses a double screening algorithm and a temperature drift compensation algorithm to improve the accuracy of the output data of the gyroscope array. Through experimental verification, the gyro angular velocity jitter amplitude of the final output of this screening algorithm is reduced by about 10 times. In the static state, the true angular velocity after temperature drift compensation is also stable near zero. The results show that after processing the gyro data through the screening algorithm and temperature drift compensation, the control effect of the stable platform can be greatly improved. Description of the Drawings

[0048] Figure 1 is the structural flowchart of the embodiment of the present invention;

[0049] Figure 2 is the original output data diagram after the fusion of 8 gyro data;

[0050] Figure 3 is the original data diagram and the data diagram after filtering and denoising of gyro 1;

[0051] Figure 4 is the original data diagram and the data diagram after filtering and denoising of gyro 2;

[0052] Figure 5 is the original data diagram and the data diagram after filtering and denoising of gyro 3;

[0053] Figure 6 is the original data diagram and the data diagram after filtering and denoising of gyro 4;

[0054] Figure 7 is the original data diagram and the data diagram after filtering and denoising of gyro 5;

[0055] Figure 8 is the original data diagram and the data diagram after filtering and denoising of gyro 6;

[0056] Figure 9 is the original data diagram and the data diagram after filtering and denoising of gyro 7;

[0057] Figure 10 is the original data diagram and the data diagram after filtering and denoising of gyro 8;

[0058] Figure 11 For the frequency distribution histogram, Q-Q plot, and normal distribution plot of gyro 1;

[0059] Figure 12 For the frequency distribution histogram, Q-Q plot, and normal distribution plot of gyro 2;

[0060] Figure 13 For the frequency distribution histogram, Q-Q plot, and normal distribution plot of gyro 3;

[0061] Figure 14 For the frequency distribution histogram, Q-Q plot, and normal distribution plot of gyro 4;

[0062] Figure 15 For the frequency distribution histogram, Q-Q plot, and normal distribution plot of gyro 5;

[0063] Figure 16 For the frequency distribution histogram, Q-Q plot, and normal distribution plot of gyro 6;

[0064] Figure 17 For the frequency distribution histogram, Q-Q plot, and normal distribution plot of gyro 7;

[0065] Figure 18 For the frequency distribution histogram, Q-Q plot, and normal distribution plot of gyro 8;

[0066] Figure 19 For the gyro fusion output data graph after screening;

[0067] Figure 20 For the comparison graph of the original output data and the output data after screening;

[0068] Figure 21 For the graphs before and after temperature compensation, their comparison graph, and the data graph of temperature varying with time. Specific implementation manner

[0069] As Figure 1 shown, a method for improving the data accuracy of the gyro array of a stable platform proposed in this embodiment includes the following steps:

[0070] Step 1, primary screening: Use a low-pass filter to perform smoothing and denoising processing on the signal, and quantify the performance of the gyro unit in the form of variance. This can not only roughly screen out gyro units with poor noise resistance and self-performance, but also greatly reduce the computational amount of subsequent normal distribution curve fitting. The low-pass filter algorithm formula:

[0071] Y(n) = a * X(n) + (1 - a) * Y(n - 1)

[0072] Among them, a is the filtering coefficient, X(n) is the current input value of the gyroscope, Y(n - 1) is the output value after the previous gyroscope filtering, and Y(n) is the output value of the current gyroscope filtering.

[0073] Variance calculation formula:

[0074] D(X) = Var(X) = E[X - E(X)] 2

[0075] Adopt the deformation formula of variance: D(X) = E(X 2 ) - [E(X)] 2 , the data set is the gyroscope angular velocity after 10,000 acquisitions. After sending the real-time acquired gyroscope data into the low-pass filtering model, the filtered gyroscope value is obtained and the mean differences E(X) and E(X 2 ) are calculated. Through the variance deformation formula, the variances of each gyroscope are obtained, and the gyroscopes with better performance are selected by the quicksort method. The larger the variance, the more severe the jitter of the gyroscope.

[0076] Step 2: Double screening: In statistics, normality testing is crucial for the analysis of data sets because it can ensure the effectiveness of many statistical methods. In this paper, two methods are used for the data visualization fitting of normality testing; in normality testing, a frequency distribution histogram and a visualization fitting Q-Q plot are drawn. The Q-Q plot (Q represents quantile) is a probability plot;

[0077] The drawing process of the Q-Q plot is as follows:

[0078] A1. Sort the data set and calculate the empirical quantiles;

[0079] A2. Compare all the calculated empirical quantiles with the theoretical quantiles of the standard normal distribution;

[0080] A3. Plot the point pairs on the graph, where the x-axis represents the empirical quantiles and the y-axis represents the theoretical quantiles;

[0081] A4. Evaluate the normality of the data by observing the distribution pattern of these points.

[0082] Fitting the probability density curve of the normal distribution:

[0083] Data preprocessing: First, preprocess the gyroscope data set to remove outliers. The outliers are removed by deleting the measured values whose absolute value of the difference from their mean exceeds 3 times the standard deviation (|x - μ| > 3σ);

[0084] Optimization algorithm selection: Maximum Likelihood Estimation (MLE) is used to estimate the mean μ and standard deviation σ of the normal distribution. By constructing the likelihood function of the dataset and using the gradient descent method to maximize the likelihood function, the optimal mean and standard deviation are solved;

[0085] The probability density function of the normal distribution is:

[0086]

[0087] The dataset X = {x1, x2, …, x n}, and the likelihood function is the joint probability density function, which is the product of the probability density functions of each data point:

[0088]

[0089] Log-likelihood function:

[0090]

[0091] Set the objective function:

[0092] Maximize the log-likelihood function -l(μ, σ 2 ). In the optimization problem, it is usually transformed into minimizing its negative value, that is, maximizing l is equivalent to minimizing -l:

[0093]

[0094] Use the gradient descent method to calculate the partial derivatives of J with respect to μ and σ 2 :

[0095]

[0096] Update μ and σ 2 :

[0097]

[0098] where α is the learning rate, taking a small positive number (such as 0.01), which is used to control the step size of the update;

[0099] The fitting process is an iterative optimization process. Starting from the initial parameter values, the parameter values are continuously optimized through multiple iterations until the preset convergence conditions are met (such as the change in parameter values is less than a certain threshold or the increase in the likelihood function value is less than a certain threshold). After each iteration, check the change in the parameter values. If the convergence conditions are met, stop the iteration and output the final mean and standard deviation as the fitting result.

[0100] Step 3. Temperature drift compensation: On the two-axis platform, the temperature drift compensation calculates the angular velocity through the angular differential of the angle encoder to replace the MEMS gyro for the speed loop closed-loop control. The reason is that the angular velocity calculated by the encoder is the speed value in the base coordinate system. When the angular velocity is zero, it can be considered that the theoretical output of the gyro should be 0. However, due to temperature changes, the speed in the base coordinate system is not zero at this time. Therefore, the encoder is used to temporarily replace the gyro for the speed loop closed-loop control. When the speed is 0, the gyro data is collected, which is the current drift angular velocity of the gyro.

[0101] Drift algorithm: In the full temperature range, when the speed obtained by the encoder differential is 0, the gyro data on the stable two-axis platform is sampled with a sampling period of 0.001 s. Then, the temperature drift curve after acquisition is fitted. The principle is to use the least squares method to determine its goodness of fit R 2 The threshold is used as the piecewise fitting point, and the optimal solutions of k and b are determined by calculating the optimal extreme points for compensation. The algorithm formula is:

[0102] The linear function is:

[0103] where x i is the original data collected by the gyro, k is the slope of the fitted linear function, and b is the intercept of the function;

[0104] Mean square error of the linear loss function:

[0105] where represents finding the value of the independent variable that makes the function value reach the minimum. Since the sum of the squares of the deviations is the smallest, it can ensure that each deviation is not very large. Thus, the problem is reduced to determining the constants k and b in it to make it the smallest. The method of determining the coefficients k and b in this way is called the least squares method;

[0106] According to the principle of the mean square error of the above linear loss function, transform it into finding the minimum value of . Let Find the extreme points of the derivative of L and calculate the optimal solutions of k and b. The calculation formulas for their respective partial derivatives are as follows:

[0107]

[0108] Similarly, we get:

[0109]

[0110] Experimental verification:

[0111] To verify the proposed method for improving the data accuracy of the gyroscope array, an array consisting of 8 MEMS gyroscopes of the same model was adopted. They have the same manufacturing process and materials, as well as the same acquisition environment. Since the gyroscopes used in this experiment are two-axis gyroscopes and their two axes are orthogonal, only the data of one axis was collected to verify the conclusion. After directly taking the mean of the original data of the 8 gyroscopes measured in the experiment, as Figure 2 shown, its variance is 1.043492. (All the charts in this experiment are data of the gyroscope sampled 10,000 times, the sampling period is 0.001 s, the unit of the gyroscope angular velocity is rad / s, and the unit of the abscissa of the chart is 0.001 s).

[0112] Table 1. Variance of the original data of each gyroscope and the variance after filtering and denoising

[0113] Gyroscope label Variance Variance after low-pass filtering Gyroscope 1 0.219589 0.080640 Gyroscope 2 0.0756 0.1101 Gyroscope 3 0.595654 0.175323 Gyroscope 4 0.206116 0.003668 Gyroscope 5 0.642437 0.094620 Gyroscope 6 0.054598 0.008973 Gyroscope 7 0.060056 0.010126 Gyroscope 8 0.880444 0.012839

[0114] The above data shows that for the variance of the unfiltered gyroscopes, gyroscope 8 has the worst performance and gyroscope 6 has the best performance. However, it is far from enough to sort and screen the gyroscopes only by this method. For the low-frequency jitter of gyroscope 2, sorting and screening only by variance will result in the performance of gyroscope 2 being second only to gyroscopes 6 and 7. From the data fluctuation diagram, it can be seen that the performance of gyroscope 4 is much better than that of gyroscope 2. Therefore, only relying on variance can initially and roughly screen out gyroscopes with poor anti-noise ability and severe high-frequency jitter, but it may not be able to reflect the performance of the gyroscope under low-frequency jitter. In the variance data after low-pass filtering, low-pass filtering does play a very good role, and gyroscope 2 is ranked behind. However, the data shows that the performance of gyroscopes 7 and 8 is about the same. From the data fluctuation diagram, it can be seen that the jitter degree of gyroscope 7 is much smaller than that of gyroscope 8. And because of the low-frequency jitter of gyroscope 8, the calculated variance is very small. To sum up, it is inaccurate to judge and screen the gyroscopes only relying on variance. Therefore, the second screening algorithm is necessary. By setting the gyroscope elimination rate under its variance to 12.5%, the second gyroscope is eliminated first.

[0115] As Figure 11 - Figure 18 shown, visual data fitting is performed on the gyroscope data after low-pass filtering. It can be seen from both the frequency distribution histogram and the Q-Q plot that the normality test of the gyroscope data after noise reduction is reasonable. Except for gyroscope 2, the other gyroscopes are very close to the normal distribution. Therefore, gyroscopes with better performance can be screened out by setting the threshold of the error interval of its normal distribution. Let the standard deviation interval be ±0.1, then the probability of each gyroscope data in this error interval can be obtained, as shown in Table 2 below:

[0116] Table 2. Probability of each gyroscope data in the ±0.1 error interval

[0117] Gyroscope label Variance after 10hz filtering Probability (±0.1 error) Gyroscope 1 0.080640 27.27％ Gyroscope 2 0.1101 23.69％ Gyroscope 3 0.175323 16.64％ Gyroscope 4 0.003668 86.02％ Gyroscope 5 0.094620 20.44％ Gyroscope 6 0.008973 71.98％ Gyroscope 7 0.010126 59.98％ Gyroscope 8 0.012839 9.74％

[0118] As can be seen from Table 2, the performance of gyro 4 is the best, and the worst is gyro 8. From the variance after low-pass filtering, it can be seen that the variances of gyro 7 and gyro 8 are approximately equal, while the probability magnitudes of the error intervals of the two gyros are quite different, and the performance is vastly different. The probability magnitude of the normal distribution curve interval can precisely reflect the fluctuation degree of the data near the zero position. The results show that by fitting the normal distribution curve, gyros with good performance can be well selected. For the gyros after screening and data fusion, the comparison between the original output data and the screened and fused data is as Figure 20 shown, and its variance is 0.000961.

[0119] By setting the gyro rejection rate under the normal distribution curve to 57%, it can be obtained that except for the second gyro, gyros 1, 3, 5, and 8 are first rejected, leaving three gyros with better performance, namely gyro 4, 6, and 7.

[0120] Before temperature drift compensation, after compensation, the comparison diagram of the two, and the temperature change over time are as Figure 21 shown. It can be seen from the figure that for the gyro data after compensation in the stationary state compared with the uncompensated data, the data value is stable near the zero position line.

[0121] The following conclusions can be drawn from the above experimental verification process:

[0122] In order to improve the accuracy of the gyro array data of the stable platform, reduce the gyro jitter and drift under temperature changes, for the array composed of 8 gyros, a double screening algorithm of low-pass filtering variance and normal curve fitting is used to screen the gyros and fit the temperature drift curve, effectively reducing the gyro jitter and drift problems. It can be seen from the experimental data that the standard deviation is reduced by 32.97 times in the static case, and the data after temperature drift compensation is also stable at the zero position. The experiment shows that after screening and compensation, the output data of the gyro array is more accurate. The experimental results show that the present screening method and temperature drift compensation method are effective.

[0123] In this embodiment, the double screening algorithm and the temperature drift compensation algorithm are used to improve the accuracy of the output data of the gyroscope array, and the control effect of the stable platform is greatly improved. First, through the double screening algorithm, the performance indicators of the gyroscope unit are quantitatively evaluated, and the gyroscope units with low static jitter are screened out. Secondly, combined with the law of the change of MEMS gyroscope data with temperature, a temperature drift compensation algorithm based on least squares curve fitting is proposed. By modeling and predicting the trend of temperature change, the effective suppression of gyroscope drift is achieved. Finally, through experimental verification, the screening algorithm proposed in this embodiment reduces the amplitude of the gyroscope angular velocity jitter in the final output by about 10 times. In the static state, the true angular velocity after temperature drift compensation is also stable near zero. The results show that the control effect of the stable platform can be greatly improved by processing the gyroscope data through the screening algorithm and temperature drift compensation.

[0124] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to this. Various changes can be made without departing from the spirit of the present invention within the scope of knowledge possessed by those skilled in the art to which the present invention pertains.

Claims

1. A method for improving the data accuracy of a gyro array in a stable platform, characterized in that It includes the following steps: S1. Primary screening: A low-pass filter is used to smooth and denoise the signal, and the performance of the gyro unit is quantified in the form of variance. The larger the variance, the more severe the gyro jitter. S2. Secondary screening: In the normality test, a frequency distribution histogram is plotted, the Q-Q plot is visualized, and the normal distribution probability density curve is fitted. S3. Temperature drift compensation: The gyro drift value is obtained from the encoder differential speed closed-loop. In the full temperature range, when the speed obtained by the encoder differential is 0, the gyro data on the stable two-axis platform is sampled with a sampling period of 0.001 s. Then, the temperature drift curve after acquisition is fitted, and the principle is to use the least squares method to determine its goodness of fit R 2 The threshold is used as the piecewise fitting point and compensated; First, through the secondary screening algorithm, the performance indicators of the gyro unit are quantitatively evaluated, and the gyro units with low static jitter are screened out. Secondly, combined with the law of the change of MEMS gyro data with temperature, a temperature drift compensation algorithm based on least squares curve fitting is proposed. By modeling and predicting the trend of temperature change, the effective suppression of gyro drift is achieved.

2. The method for improving the data accuracy of the gyro array of the stable platform according to claim 1, characterized in that In S1, the low-pass filter algorithm formula: Y(n) = a * X(n) + (1 - a) * Y(n - 1) where a is the filter coefficient, X(n) is the current input value of the gyro, Y(n - 1) is the output value of the gyro after the previous filtering, and Y(n) is the output value of the gyro after the current filtering.

3. A method for improving the data accuracy of a gyro array in a stable platform according to claim 2, characterized in that, Variance calculation formula: D(X) = Var(X) = E[X - E(X)] 2 Adopt the deformation formula of variance: D(X) = E(X 2 ) - [E(X)] 2 , the data set is the gyro angular velocity after 10,000 acquisitions. After sending the real-time acquired gyro data into the low-pass filter model, the filtered gyro values are obtained and the mean differences E(X) and E(X 2 ) are calculated. The variances of each gyro are obtained through the variance deformation formula, and the gyros with better performance are selected by the quicksort method.

4. A method for improving the data accuracy of a gyro array in a stable platform according to claim 3, characterized in that, In S2, the process of drawing the Q-Q plot is as follows: A1. Sort the data set and calculate the empirical quantiles. A2. Compare all the calculated empirical quantiles with the theoretical quantiles of the standard normal distribution. A3. Plot the point pairs on the graph, where the x-axis represents the empirical quantiles and the y-axis represents the theoretical quantiles. A4. Evaluate the normality of the data by observing the distribution pattern of these points.

5. A method for improving the data accuracy of a gyro array of a stable platform according to claim 4, characterized in that, When fitting the normal distribution probability density curve, first preprocess the gyro data set to remove outliers. The maximum likelihood estimation is used to estimate the mean μ and standard deviation σ of the normal distribution. By constructing the likelihood function of the data set and using the gradient descent method to maximize the likelihood function, the optimal mean and standard deviation are solved. The probability density function of the normal distribution is: Data set X = {x1, x2, …, x n}, and the likelihood function is the joint probability density function, which is the product of the probability density functions of individual data points: Log-likelihood function: Set the objective function: Maximize the log-likelihood function -l(μ, σ 2 ) In an optimization problem, it is usually transformed into minimizing its negative value, that is, maximizing l is equivalent to minimizing -l: Calculate the partial derivatives of J with respect to μ and σ using the gradient descent method 2 : Update μ and σ 2 : where α is the learning rate, which is used to control the update step size. The fitting process is an iterative optimization process. Starting from the initial parameter values, the parameter values are continuously optimized through multiple iterations until the preset convergence conditions are met.

6. A method for improving the data accuracy of a gyro array in a stable platform, according to claim 5, characterized in that In S3, for the encoder differential velocity closed-loop, the gyro drift value is obtained. When the velocity is 0, the gyro data is collected, which is the current gyro drift angular velocity.

7. A method for improving the data accuracy of a gyro array in a stable platform according to claim 6, characterized in that, The optimal solutions of k and b are determined by calculating the optimal extreme points for compensation. The algorithm formula is: The linear function is as follows: where x i is the raw data collected by the gyroscope, k is the slope of the fitted linear function, and b is the intercept of the function; Mean squared error of linear loss function: Among them, represents finding the value of the independent variable that makes the function value reach the minimum; According to the principle of the mean squared error of the above linear loss function, transform it into finding the minimum value. Let Find the extreme points of the derivative of L and calculate the optimal solutions of k and b. The calculation formulas for their respective partial derivative values are as follows:

Images