阅读说明：本技术 一种光纤陀螺惯性测量单元信号联合降噪方法 (Signal joint noise reduction method for fiber optic gyroscope inertial measurement unit ) 是由 孙骞 许前昆 王娱萤 叶方 李一兵 田园 于 2021-08-03 设计创作，主要内容包括：本发明提出了一种光纤陀螺惯性测量单元信号联合降噪方法,本方法首先利用快速自适应噪声完整集合经验模态分解方法Fast-CEEMDAN将光纤陀螺IMU输出信号进行本征模态分量IMF分解,不仅大大降低了计算量,同时有效解决了噪声从高频到低频的转移传递问题,提高了信息结构化的精度和效率；随后,计算IMF的排列熵和平均排列熵,实现含噪较多IMF的筛选；最后利用自适应小波包分析,AWPA方法对含噪较多IMF进行去噪,保证信号分解的无冗余性、减少局部信号损失,最大限度提高信号的信噪比,从而得到高精度强鲁棒性测量信息。(The invention provides a method for reducing noise of a fiber-optic gyroscope inertial measurement unit signal combination, which comprises the steps of firstly utilizing a Fast self-adaptive noise complete set empirical mode decomposition method Fast-CEEMDAN to carry out intrinsic mode component IMF decomposition on an IMU output signal of a fiber-optic gyroscope, greatly reducing the calculated amount, effectively solving the problem of transfer of noise from high frequency to low frequency, and improving the precision and efficiency of information structuring; then, calculating the permutation entropy and the average permutation entropy of the IMF to realize the screening of IMF with more noise; and finally, denoising the IMF with more noise by using self-adaptive wavelet packet analysis and an AWPA (active wavelet packet power amplifier) method, ensuring the non-redundancy of signal decomposition, reducing local signal loss, and improving the signal-to-noise ratio of the signal to the maximum extent, thereby obtaining high-precision strong robustness measurement information.)

1. A method for reducing noise of a fiber optic gyroscope inertial measurement unit signal in a combined manner is characterized by comprising the following steps:

the method comprises the following steps:

the method comprises the following steps: the method comprises the following steps of mounting an optical fiber gyroscope IMU on equipment or a carrier, preheating the IMU and collecting output data of an inertial sensitive sensor;

step two: decomposing a sampling signal into a series of intrinsic mode components IMF by using a Fast adaptive noise complete set empirical mode decomposition method Fast-CEEMDAN;

step three: calculating the permutation entropy of each IMF and the average permutation entropy of all IMFs, and screening IMFs with more noise and IMFs with less noise according to the permutation entropy and the average permutation entropy;

step four: threshold denoising is carried out on the screened IMF with more noise by using an Adaptive Wavelet Packet Analysis (AWPA) method, and the IMF with less noise is reserved, so that the integrity of an original signal is ensured;

step five: and performing signal reconstruction on each order of IMF after denoising, and finally obtaining the measurement information after denoising.

2. The method of claim 1, further comprising: in the second step, the first step is carried out,

the Fast-adaptive noise complete set empirical mode decomposition method Fast-CEEMDAN specifically comprises the following steps:

step two, firstly: adding white noise N (t), beta to the IMU's original output signal x (t)₀The signal to be analyzed at the ith time is x_i(t)＝x(t)+β₀N_i(t)；

Step two: defining an operator M (-) to obtain a local mean of a time series, and operating M (x)_i(t)) is found x_i(t) obtaining the upper envelope e described by all the maximum points through cubic spline interpolation for all the maximum points and minimum points₊(t) and a lower envelope e characterized by all minima points_-(t), calculating the upper and lower envelope: m (x)_i(t))＝(e₊(t)+e_-(t))/2；

Step two and step three: calculate the first residual:and further obtaining a first IMF: imf₁＝x(t)-r₁；

Step two, four: adding white noise r to the first residue₁+β₁N_i(t), calculating a second residual:further, a second IMF is obtained: imf₂＝r₁-r₂；

Step two and step five: the K (K — 3, …, K) th residue was calculated:and obtaining the k IMF: imf_k＝r_k-1-r_k；

Step two, step six: repeating the second step and the fifth step until the signals can not be decomposed any more, and obtaining K IMFs.

3. The method of claim 1, further comprising: in the step 3, the process is carried out,

screening out components with more noise and components with less noise according to the permutation entropy and the average permutation entropy, wherein the specific method comprises the following steps:

calculating the permutation entropy of each IMF and the average permutation entropy of all IMFs, comparing the permutation entropies of all IMFs with the average permutation entropy, defining the IMF with the permutation entropy value higher than the average permutation entropy as a component with more noise, and defining the IMF with the permutation entropy value lower than the average permutation entropy as a component with less noise.

4. The method according to claim 1, wherein in step four,

the specific method for carrying out threshold denoising on the screened IMF with more noise comprises the following steps:

step four, firstly: selecting an optimal wavelet basis function of a wavelet packet according to the logarithmic energy entropy of the IMF signal;

step four and step two: selecting the optimal number of wavelet packet decomposition layers according to the composite evaluation indexes of the root mean square error, the signal-to-noise ratio and the smoothness of the IMF signal;

step four and step three: realizing the self-adaptive adjustment of the wavelet threshold according to the power spectral density of the signal;

step four: and (3) utilizing a wavelet packet analysis method to reduce noise of the IMF, thereby improving the signal-to-noise ratio of the IMF.

Technical Field

The invention belongs to the field of noise reduction, and particularly relates to a signal joint noise reduction method for an inertial measurement unit of a fiber-optic gyroscope.

Background

Under the limitation of technologies such as a fiber ring winding process and integrated packaging, the conventional miniaturized fiber optic gyroscope IMU is sensitive to electronic and mechanical noises, and the output signal contains large high-frequency random noise which seriously affects the measurement precision of the fiber optic gyroscope IMU, so that the output signal needs to be subjected to noise reduction treatment.

The current fiber optic gyroscope IMU signal denoising processing Mode mainly comprises a Kalman filtering method, a wavelet analysis method, an Empirical Mode Decomposition (EMD) method and the like

The Kalman filtering denoising method is a recursive filtering method, and can obtain the optimal estimation value of a dynamic system from a measurement signal containing noise through self-correction and pre-estimation processing of a Kalman filter. However, the premise of Kalman filtering optimal estimation is that a system model is accurately known, and the noise of the fiber-optic gyroscope IMU influenced by the use environment has strong uncertainty, so that the system cannot be accurately modeled, and the noise reduction performance is influenced.

In the denoising method based on wavelet analysis, firstly, signals are decomposed through wavelet transformation, then coefficients after wavelet decomposition are judged by using a threshold value, the coefficients containing noise are removed, the coefficients containing signal components are reserved, and finally, information is reconstructed by using inverse wavelet transformation. Although the method obtains a certain effect in the signal denoising field by virtue of the characteristic of multi-resolution, the quality of the denoising effect completely depends on the selection of the wavelet basis function, so that the method is limited in practical application.

The EMD method performs multiple times of self-adaptive decomposition on the signal depending on the time scale of the data, and does not need to set a basis function in the decomposition process in advance, so the method has great advantages in processing nonlinear non-stationary random signals. The generation of the envelope is the key of the EMD algorithm, and the quality of the envelope fitting directly influences the decomposition result; and the method has the defect of easy generation of modal aliasing phenomenon, and influences the noise reduction performance of signals.

In recent years, in order to further improve the denoising performance of the EMD method, many scholars have improved it. Among them, the researchers of bars, Torres, etc. have proposed the CEEMDAN method, which can not only reduce the modal aliasing phenomenon but also accurately reconstruct the original signal by adding the adaptive white noise. However, in the method, adaptive white noise needs to be added for many times in the process of solving each level of Intrinsic Mode component (IMF), the calculation is complex, and the efficiency of the method is reduced.

Therefore, in a practical system, the random drift of the fiber optic gyroscope IMU has the characteristics of non-stationarity and nonlinearity, and in order to improve the precision and robustness of the output signal of the fiber optic gyroscope IMU, the invention provides a fiber optic gyroscope IMU signal combined noise reduction method based on Fast-CEEMDAN and AWPA. According to the method, firstly, a Fast-CEEMDAN method is utilized to carry out structuralization processing on nonlinear and non-stable IMU output signals of the fiber-optic gyroscope, so that the calculated amount is greatly reduced, the problem of transfer transmission of noise from high frequency to low frequency is effectively solved, and the precision and the efficiency of information structuralization are improved; then, in order to ensure the non-redundancy of signal decomposition, reduce the local signal loss and maximize the data characteristics, the adaptive setting is carried out on the wavelet packet parameters by utilizing the IMU signal characteristics, and the signal-to-noise ratio of the signal is improved to the maximum extent; and finally, denoising the IMU output signal by a wavelet packet analysis method, thereby obtaining high-precision strong-robustness measurement information.

Disclosure of Invention

The invention provides a signal joint Noise reduction method for an inertial measurement unit of a fiber optic gyroscope, which utilizes a Fast Adaptive Noise Complete set Empirical Mode Decomposition (Fast-CEEMDAN) method and an Adaptive Wavelet Packet Analysis (AWPA) method to realize Noise reduction of IMU inertial sensing information, thereby improving the measurement accuracy and robustness of the IMU of the fiber optic gyroscope on the premise of not changing the accuracy of an inertial device.

The invention is realized by the following scheme:

a method for reducing noise of a fiber optic gyroscope inertial measurement unit signal in a combined manner comprises the following steps:

the method comprises the following steps:

the method comprises the following steps: the method comprises the following steps of mounting an optical fiber gyroscope IMU on equipment or a carrier, preheating the IMU and collecting output data of an inertial sensitive sensor;

step two: decomposing a sampling signal into a series of intrinsic mode components IMF by using a Fast adaptive noise complete set empirical mode decomposition method Fast-CEEMDAN;

step three: calculating the permutation entropy of each IMF and the average permutation entropy of all IMFs, and screening IMFs with more noise and IMFs with less noise according to the permutation entropy and the average permutation entropy;

step four: threshold denoising is carried out on the screened IMF with more noise by using an Adaptive Wavelet Packet Analysis (AWPA) method, and the IMF with less noise is reserved, so that the integrity of an original signal is ensured;

step five: and performing signal reconstruction on each order of IMF after denoising, and finally obtaining the measurement information after denoising.

Further, in the second step, the first step,

the Fast-adaptive noise complete set empirical mode decomposition method Fast-CEEMDAN specifically comprises the following steps:

step two, firstly: adding white noise N (t), beta to the IMU's original output signal x (t)₀The signal to be analyzed at the ith time is x_i(t)＝x(t)+β₀N_i(t)；

Step two: defining an operator M (-) to obtain a local mean of a time series, and operating M (x)_i(t)) is found x_i(t) obtaining the upper envelope e described by all the maximum points through cubic spline interpolation for all the maximum points and minimum points₊(t) and a lower envelope e characterized by all minima points_-(t), calculating the upper and lower envelope: m (x)_i(t))＝(e₊(t)+e_-(t))/2；

Step two and step three: calculate the first residual:and further obtaining a first IMF: imf₁＝x(t)-r₁；

Step two, four: adding white noise r to the first residue₁+β₁N_i(t), calculating a second residual:further, a second IMF is obtained: imf₂＝r₁-r₂；

Step two and step five: the K (K — 3, …, K) th residue was calculated:and obtaining the k IMF: imf_k＝r_k-1-r_k；

Step two, step six: repeating the second step and the fifth step until the signals can not be decomposed any more, and obtaining K IMFs;

further, in step 3,

screening out components with more noise and components with less noise according to the permutation entropy and the average permutation entropy, wherein the specific method comprises the following steps:

calculating the permutation entropy of each IMF and the average permutation entropy of all IMFs, comparing the permutation entropies of all IMFs with the average permutation entropy, defining the IMF with the permutation entropy value higher than the average permutation entropy as a component with more noise, and defining the IMF with the permutation entropy value lower than the average permutation entropy as a component with less noise.

Further, in the fourth step,

the specific method for carrying out threshold denoising on the screened IMF with more noise comprises the following steps:

step four, firstly: selecting an optimal wavelet basis function of a wavelet packet according to the logarithmic energy entropy of the IMF signal;

step four and step two: selecting the optimal number of wavelet packet decomposition layers according to the composite evaluation indexes of the root mean square error, the signal-to-noise ratio and the smoothness of the IMF signal;

step four and step three: realizing the self-adaptive adjustment of the wavelet threshold according to the power spectral density of the signal;

step four: and (3) utilizing a wavelet packet analysis method to reduce noise of the IMF, thereby improving the signal-to-noise ratio of the IMF.

The invention has the beneficial effects

The invention utilizes Fast-CEEMDAN method to carry out self-adaptive decomposition on IMU output signals, which not only can effectively reduce modal aliasing phenomenon, but also can improve the self-adaptive decomposition efficiency of nonlinear and non-stable IMU output signals; meanwhile, the signal is denoised by using an AWPA method, so that not only can full-band denoising be realized, but also all parameters can be set in a self-adaptive manner, the non-redundancy of signal decomposition is ensured to the maximum extent, the local signal loss is reduced, the data characteristics are maximized, the external environment interference is finally inhibited, and the measurement precision and the robustness of the IMU signal are improved.

Drawings

FIG. 1 is a schematic flow diagram of the process of the present invention;

FIG. 2 is an output signal of the X-axis fiber optic gyroscope before denoising;

FIG. 3 is a diagram of an output signal of an X-axis fiber optic gyroscope after being processed using the combined noise reduction method of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments; all other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

In conjunction with the figures 1 to 3,

a method for reducing noise of a fiber optic gyroscope inertial measurement unit signal in a combined manner comprises the following steps:

the method comprises the following steps:

the method comprises the following steps: an optical fiber gyroscope (Inertial measurement unit) IMU is arranged on equipment or a carrier, and is preheated and output data of an Inertial sensitive sensor is acquired;

step two: decomposing a sampling signal into a series of Intrinsic Mode components (IMF) by using a Fast Adaptive Noise Complete set Empirical Mode Decomposition (Fast-CEEMDAN) method; the method not only greatly reduces the calculated amount, but also effectively solves the problem of transfer of noise from high frequency to low frequency, and improves the precision and efficiency of information structuring;

step three: calculating the permutation entropy of each IMF and the average permutation entropy of all IMFs, and screening IMFs with more noise and IMFs with less noise according to the permutation entropy and the average permutation entropy;

step four: threshold denoising is carried out on the screened IMF with more noise by using an Adaptive Wavelet Packet Analysis (AWPA) method (adaptive Wavelet Packet analysis), and the IMF with less noise is reserved, so that the integrity of an original signal is ensured;

step five: and performing signal reconstruction on each order of IMF after denoising, and finally obtaining the measurement information after denoising.

In the second step: the Fast-adaptive noise complete set empirical mode decomposition method Fast-CEEMDAN specifically comprises the following steps:

step two, firstly: adding white noise N (t), beta to the IMU's original output signal x (t)₀The signal to be analyzed at the ith time is x_i(t)＝x(t)+β₀N_i(t)；

Step two: defining an operator M (-) to obtain a local mean of a time series, and operating M (x)_i(t)) is found x_i(t) obtaining the upper envelope e described by all the maximum points through cubic spline interpolation for all the maximum points and minimum points₊(t) and a lower envelope e characterized by all minima points_-(t), calculating the upper and lower envelope: m (x)_i(t))＝(e₊(t)+e_-(t))/2；

Step two and step three: calculate the first residual:and further obtaining a first IMF: imf₁＝x(t)-r₁；

Step two, four: adding white noise r to the first residue₁+β₁N_i(t) calculating a second residueQuantity:further, a second IMF is obtained: imf₂＝r₁-r₂；

Step two and step five: the K (K — 3, …, K) th residue was calculated:and obtaining the k IMF: imf_k＝r_k-1-r_k；

Step two, step six: repeating the second step and the fifth step until the signals can not be decomposed any more, and obtaining K IMFs;

in the third step, the first step is carried out,

screening out components with more noise and components with less noise according to the permutation entropy and the average permutation entropy, wherein the specific method comprises the following steps:

calculating the permutation entropy of each IMF and the average permutation entropy of all IMFs, comparing the permutation entropies of all IMFs with the average permutation entropy, defining the IMF with the permutation entropy value higher than the average permutation entropy as a component with more noise, and defining the IMF with the permutation entropy value lower than the average permutation entropy as a component with less noise.

In the fourth step of the method, the first step of the method,

the specific method for carrying out threshold denoising on the screened IMF with more noise comprises the following steps:

step four, firstly: selecting an optimal wavelet basis function of a wavelet packet according to the logarithmic energy entropy of the IMF signal;

step four and step two: selecting the optimal number of wavelet packet decomposition layers according to the composite evaluation indexes of the root mean square error, the signal-to-noise ratio and the smoothness of the IMF signal;

step four and step three: realizing the self-adaptive adjustment of the wavelet threshold according to the power spectral density of the signal;

step four: and (3) utilizing a wavelet packet analysis method to reduce noise of the IMF, thereby improving the signal-to-noise ratio of the IMF.

The effect of the invention can be verified by the following tests:

firstly, a static test environment is set up, a middle-precision fiber optic gyroscope IMU developed in a laboratory is utilized and fixedly installed on a vibration isolation metal installation plate of the laboratory, a reinforced notebook is utilized to collect output information of the IMU, the sampling frequency is 100Hz, the total data length is 1800s, and an output signal of an X-axis fiber optic gyroscope is taken as an example to verify the effect of the invention.

First, the acquired original signal of the X-axis fiber optic gyroscope is analyzed, as shown in fig. 2. The method for reducing noise of original data by using the joint noise reduction method provided by the invention obtains the signal after noise reduction as shown in fig. 3. In order to further analyze the denoising effect of the invention, the RMSE is utilized to evaluate the denoising performance. Respectively calculating the RMSE values of the original signal and the denoised signal, wherein the RMSE values are respectively as follows: 1.0640deg/h and 0.1881deg/h, it can be seen that the RMSE value of the signal after denoising is significantly less than that of the signal before denoising. Therefore, the denoising method provided by the invention can effectively reduce the measurement noise of the fiber optic gyroscope IMU and effectively improve the measurement precision and robustness of the IMU inertial device.

The method for reducing noise of the inertial measurement unit signal combination of the fiber-optic gyroscope provided by the invention is introduced in detail, the principle and the implementation mode of the invention are explained, and the description of the embodiment is only used for helping to understand the method and the core idea of the invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.