Multi-channel sampling clutter processing method, system and device for an optical fiber F-P sensor

Abstract

The present invention relates to a multi-channel sampling clutter processing method, system and device for an optical fiber F-P sensor. The method includes: parallelly collecting the reflected spectral signals of the optical fiber F-P sensor to obtain multiplexed reflected spectral data; taking the data points corresponding to each reflected spectral data at the same moment as a set of first data sets to obtain multiple sets of first data sets at consecutive moments; sequentially obtaining the center points of each set of first data sets and using the center points as a second data set; performing Kalman filtering on the center points of the second data set to obtain wavelength signal data, and plotting the wavelength signal data into a waveform curve for output. The present invention effectively eliminates the waveform burrs of the output signal of the multi-channel sampling clutter, making the demodulated output curve smoother, thereby improving the demodulation accuracy.

Description

Technical Field

[0001] The present invention belongs to the technical field of fiber optic sensor demodulation, and relates to a multi-channel sampling clutter processing method, system and device for fiber optic F-P sensors. Background Technique

[0002] Fiber optic F-P sensors are sensors based on the Fabry-Perot interference principle, used to measure and monitor changes in physical quantities. These sensors utilize the Fabry-Perot interference phenomenon outside the optical fiber to convert optical signals into corresponding physical quantity signals for measurement. Fiber optic F-P sensors consist of a pair of mirrors at both ends of the optical fiber. The mirrors can be two reflecting surfaces at the ends of the optical fiber, or metal or dielectric reflective layers evaporated or welded onto the optical fiber. When light enters from one end of the optical fiber, a part of the light is reflected back by the first reflecting surface and then reflected back by the second reflecting surface to form interference. When an external physical quantity (such as temperature, pressure, or deformation, etc.) changes, it will cause changes in the length or refractive index of the optical fiber, and the position or intensity of its interference peak will also change accordingly. By measuring the movement or intensity change of the interference peak, the change in the physical quantity can be inferred.

[0003] When demodulating fiber optic F-P sensors, since the frame rate of the spectrometer can only reach about 25 kHz, if a demodulation speed of 100 kHz is to be achieved, 4 spectrometers need to be used. Then, by performing time-sharing exposure and parallel sampling on the 4 spectrometers, the spectral sampling frame rate can be quadrupled. Therefore, a high-speed demodulator needs to drive 4 spectrometers simultaneously. Since the output signal of the spectrometer is an analog voltage signal, after the drive circuits of the 4 spectrometers, 4 high-speed AD sampling circuits are connected in parallel. Due to the time shift and offset between the 4 analog voltage signals, the cavity length output curve obtained after splicing the 4 analog voltage signals is locally jagged with burrs and is not a smooth curve. Therefore, there will be a certain error in the information obtained after demodulation.

[0004] A method for filtering pulse clutter in the demodulation of fiber Bragg grating sensors disclosed in Publication No. CN 106706011 A. It builds an optical path and a circuit hardware platform for a fiber grating demodulator, performs fiber grating demodulation, uses a field-programmable gate array for high-speed AD acquisition to read AD sampling values, and sets a "peak judgment threshold". When the current sampling value starts to be less than the threshold, calculate the distance from the beginning of being greater than the threshold to the beginning of being less than the threshold. Perform filtering on pulse interference signals and set a "width expansion threshold". Through the above steps, filtering of pulse interference existing in the demodulation process is achieved, and the effect of improving the demodulation accuracy and stability of fiber Bragg gratings is achieved. This scheme uses threshold judgment to determine whether target information needs to be filtered to achieve filtering. However, for the problem of multi-channel sampling clutter, due to the randomness of spectrometer sampling, it is not easy to set the threshold for filtering clutter. Therefore, this method cannot solve the multi-channel sampling clutter problem well.

[0005] A fiber grating signal demodulation method disclosed in Publication No. CN114353844 A. By setting parameters of a laser generating device to generate light waves of different wavelengths, an original signal containing noise is obtained. For the original signal containing noise, a wavelet denoising method is used to obtain a signal with ordinary noise removed. Perform optimal predictor variable threshold processing on the signal with ordinary noise removed to obtain a signal with special clutter removed. Perform 5-point smoothing processing on the signal with special clutter removed to obtain a smoothed signal. Take the derivative of the smoothed signal to obtain a peak differential signal. When the peak differential signal is greater than the threshold, retain the signal. When it is less than the threshold, it indicates that the signal drifts and early warning processing is performed. This scheme uses a wavelet denoising method for noise reduction. However, due to the need for signal processing at an extremely high speed of 100 kHz in high-speed demodulation, and the wavelet functions for extracting signals in each time period will occupy more resources of the demodulation chip, it also cannot handle the multi-channel sampling clutter problem well. Summary of the Invention

[0006] In order to solve the above technical problems, the purpose of the present invention is to provide a multi-channel sampling clutter processing method, system and device for fiber F-P sensors. The present invention calculates the center points by grouping the sampling data points, and then performs Kalman filtering processing on the new data set composed of the center points of all groups, which can solve the multi-channel sampling clutter problem during 100 kHz high-speed demodulation, and further eliminate the waveform burr problem caused by time shift and offset between sampling data.

[0007] In order to achieve the above purpose, the technical scheme adopted by the present invention is as follows:

[0008] The present invention provides a multi-channel sampling clutter processing method for fiber F-P sensors, and the method includes the following steps:

[0009] Step S1, collect the reflected spectral signals of the fiber optic F-P sensor in parallel to obtain multiplexed reflected spectral data;

[0010] Step S2, take the data points corresponding to each moment of each set of the reflected spectral data as a set of first data sets, and obtain multiple sets of the first data sets at consecutive moments;

[0011] Step S3, sequentially obtain the center points of each set of the first data sets, and use the center points as the second data set;

[0012] Step S4, perform Kalman filtering on the center points of the second data set to obtain wavelength signal data, and plot the wavelength signal data as a waveform curve for output.

[0013] Further, in step S1, collecting the reflected spectral signals of the fiber optic F-P sensor in parallel to obtain multiplexed reflected spectral data specifically includes:

[0014] Input broadband light from the first port of the fiber optic circulator into the fiber optic F-P sensor. The reflected light of the fiber optic F-P sensor is output from the third port of the fiber optic circulator. The output reflected light is split and enters multiple spectrometers respectively. Each spectrometer simultaneously obtains the electrical signals of the reflected spectral data of the fiber optic F-P sensor. After converting the electrical signals of each path into digital signals, they are sent to the FPGA controller for signal processing to obtain multiplexed reflected spectral data.

[0015] Further, in step S3, obtaining the center point of each set of the first data sets specifically includes:

[0016] Define the center point to the square of the distance of each data point in the first data set as:

[0017] ;

[0018] Wherein, represents the distance from the th point to the center point , , are the coordinates of the center point , , are the coordinates of the th data point;

[0019] Set the function of the distance between the center point and each data point as:

[0020] ;

[0021] Solve the minimum value of the function using the iterative formula , and the iterative formula is as follows:

[0022] ;

[0023] where is the learning rate, and are the partial derivatives of the function with respect to the coordinate and the coordinate respectively. Then the partial derivative of is:

[0024] ;

[0025] Substitute the partial derivatives into the iterative formula to obtain the coordinates of the center point , .

[0026] Furthermore, in step S4, the Kalman filter specifically includes:

[0027] Process the center point sequentially through the Kalman state prediction equation, and the Kalman state prediction equation is:

[0028] ;

[0029] where is the state prediction value at time, is the state transition matrix, is the optimal state estimate value at time, is the control input matrix, is the control input at time;

[0030] The prediction error covariance is:

[0031] ;

[0032] where is the prediction error covariance at time, is the optimal estimation error covariance at time, is the transpose matrix of, is the process noise covariance.

[0033] Further, the Kalman filtering further includes: after establishing and processing the Kalman state prediction equation, the center point is then updated by establishing a state update equation, and the state update equation is:

[0034] ;

[0035] wherein, is the Kalman gain at time is the observation matrix, is the observation noise covariance, is the optimal state estimate value at time is the observation value at time is the optimal estimation error covariance at time is the identity matrix.

[0036] Further, the Kalman state prediction equation is established by an FPGA controller.

[0037] Further, the state update equation is established by an FPGA controller.

[0038] Further, in step S4, the wavelength signal data is plotted as a waveform curve and output, specifically: the FPGA controller sends the wavelength signal data to the host computer software, and the host computer software plots the wavelength signal data as a waveform curve and displays it.

[0039] The present invention also provides a multi-channel sampling clutter processing system for an optical fiber F-P sensor, including:

[0040] a reflected spectrum data acquisition module, configured to collect the reflected spectrum signals of the optical fiber F-P sensor in parallel and obtain multiple channels of reflected spectrum data;

[0041] a first data set module, configured to use the data points corresponding to the same time of each path of the reflected spectrum data as a group of first data sets, and obtain multiple groups of the first data sets at consecutive times;

[0042] a second data set module, configured to sequentially obtain the center points of each group of the first data sets, and use the center points as the second data set;

[0043] a waveform curve output module, configured to perform Kalman filtering on the center points of the second data set to obtain wavelength signal data, and plot the wavelength signal data as a waveform curve and output it.

[0044] The present invention also provides an electronic device, including at least one processor; and a memory communicatively connected to the processor; wherein, the memory stores instructions executable by the processor, and when the instructions are executed by the processor, the processor is enabled to execute a multi-channel sampling clutter processing method for an optical fiber F-P sensor as described above.

[0045] Due to the above technical solutions adopted by the present invention, it has the following advantages and effects:

[0046] (1) For the multi-channel sampling clutter processing method, system and device of an optical fiber F-P sensor of the present invention, mainly when performing high-speed demodulation at 100 kHz, due to the need for spectrometers to perform parallel sampling, there are time shifts and offsets between them. The present invention effectively eliminates the waveform burrs of the output signal by adopting the Kalman filtering method based on the central value of data points, making the output curve obtained by 100 kHz demodulation smoother, thereby improving the demodulation accuracy; and the Kalman filtering based on the central value of data points, as an estimation algorithm integrating numerical calculation and recursive thinking, can use numerical methods to solve the central points of a set of multi-channel acquisition data, and then perform Kalman filtering processing on the new data composed of each central point. Since the problem of finding the central points occupies less computing resources than solving the Kalman filtering equation, the processing speed of the waveform can be accelerated, and the stability and real-time performance of clutter processing are enhanced.

[0047] (2) The multi-channel sampling clutter processing method, system and device of an optical fiber F-P sensor of the present invention are not only applicable to the demodulation of optical fiber F-P sensors, but also can be applied to other optical sensor systems that require high-speed and high-precision multi-channel sampling, and have broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is a flowchart of a multi-channel sampling clutter processing method for an optical fiber F-P sensor of the present invention.

[0049] Figure 2 is a four-channel sampling signal diagram of the present invention.

[0050] Figure 3 is a waveform diagram after processing of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0051] The embodiments of the present invention will be described in detail below in conjunction with the drawings, so as to more clearly understand the purpose, features and advantages of the present invention. It should be understood that the embodiments shown in the drawings are not limitations on the scope of the present invention, but only to illustrate the essential spirit of the technical solution of the present invention. Embodiment 1

[0052] As Figure 1As shown in the figure. Embodiment 1 provides a multi-channel sampling clutter processing method for an optical fiber F-P sensor, and the method includes the following steps:

[0053] Step 1: Collect the reflected light signals of the optical fiber F-P sensor through multi-channel parallel acquisition to obtain multi-channel reflected spectral data.

[0054] Specifically, perform time-sharing exposure and parallel sampling through a spectrometer. Input the broadband light of the SLD light source from the first port of the optical fiber circulator, enter the optical fiber F-P sensor through the second port of the optical fiber circulator. The reflected light of the optical fiber F-P sensor enters from the second port of the optical fiber circulator, passes through the third port of the optical fiber circulator and then enters the input arm of the optical fiber splitter. The reflected light of the optical fiber F-P sensor is divided into n beams of light signals by the optical fiber splitter, and respectively enter n spectrometers. The electrical signals of the n-channel reflected spectral data of the optical fiber F-P sensor are obtained simultaneously by the n spectrometers. The electrical signal of each channel is converted into a digital signal through an AD sampling circuit, and finally the converted digital signal is sent to the FPGA controller for signal processing to obtain n-channel reflected spectral data.

[0055] As a preference, the present invention uses 4 spectrometers, and the sampling frequency of each spectrometer is 25 kHz.

[0056] The number of sampling channels is 4, and the sampling rate is 100k.

[0057] Step 2: Take the data points corresponding to the same moment of each channel of reflected spectral data as a set of first data sets, and obtain multiple sets of first data sets at consecutive moments.

[0058] Specifically, since the sampling frequency of the spectrometer is 25 kHz, that is, 25k data are collected in 1 s, n spectrometers will generate n data points in 1 / 25k seconds. Take the n data points corresponding to the same moment of the n-channel reflected spectral data as the first data set, and obtain multiple sets of first data sets at consecutive moments.

[0059] Step 3: Sequentially obtain the center points of each set of first data sets, and use the center points as the second data set.

[0060] Specifically, the center point is obtained by using the numerical solution method and is processed by the FPGA controller. Set the center point of the first data set at the same moment , according to the distance formula between two points, write the distances between the n data points and the center point , and sum them up to obtain a binary function , find the minimum value point of the binary function , which is the coordinate of the center point .

[0061] The specific numerical solution method is as follows: It is set that the first data set includes n data points , which are P1(x1,y1), P2(x2,y2), P3(x3,y3)... P n-1 (x n-1 ,y n-1 ), P n (x n ,y n ), such that the difference in the sum of the squares of the distances from the center point to these n data points is minimized.

[0062] Define the square of the distance from the center point to each data point as:

[0063] ; (1)

[0064] In formula (1), represents the distance from the th point to the center point , , are the coordinates of the center point , , are the coordinates of the th data point;

[0065] Set the function of the distance between the center point and each data point as:

[0066] ; (2)

[0067] In formula (2), ∑ represents summation, that is, summing the distances from n data points to the center point ;

[0068] Use the gradient descent method to solve the minimum value of the function , and the iterative formula of the gradient descent method is:

[0069] ; (3)

[0070] In formula (3), is the learning rate (step size), and are the partial derivatives of the function with respect to the coordinate and the coordinate respectively; then the partial derivative of is:

[0071] ; (4)

[0072] Substitute the partial derivative of 4 (formula) into the iterative formula (3) to obtain the coordinates of the center point of , .

[0073] Step 4: After performing Kalman filtering on the center point of the second data set, obtain the wavelength signal data, and plot the wavelength signal data as a waveform curve for output.

[0074] Specifically, the center point is processed by the FPGA controller by establishing a Kalman state prediction equation. When establishing the Kalman state prediction equation, the parameters of the Kalman filter can be configured, such as the state transition matrix, observation matrix, noise covariance, etc. The parameters are adjusted according to the shape of the actual generated curve. The center point obtains the wavelength signal data through processing by the FPGA controller. The wavelength signal data is sent to the host computer software through a high-speed communication interface, and the wavelength signal data is plotted as a waveform curve through the plotting library in the host computer software, and the waveform curve is displayed through the host computer.

[0075] Among them, the Kalman state prediction equation is:[[]]

[0076] ; (5)

[0077] In formula (5), is the state prediction value at time is the state transition matrix, is the optimal state estimate value at time is the control input matrix, is the control input at time

[0078] Among them, the prediction error covariance is:[[]]

[0079] ; (6)

[0080] In formula (6), is the prediction error covariance at time is the optimal estimation error covariance at time is the transpose matrix of is the process noise covariance.

[0081] Further, in order to effectively perform state estimation and prediction on the collected spectral data, the central point is processed by the Kalman state prediction equation and then updated by the FPGA controller through the state update equation to obtain accurate wavelength signal data.

[0082] Among them, the state update equation is:

[0083] ; (7)

[0084] In formula (7), is the Kalman gain at time is the observation matrix, is the observation noise covariance, is the optimal state estimate value at time is the observed value at time is the optimal estimation error covariance at time is the identity matrix.

[0085] The simulation verification steps of Example 1 with 4 sampling channels are as follows:

[0086] First, use matlab software to construct a mathematical model of the fiber optic F-P sensor to simulate its interference phenomenon and the generation of spectral signals. Set the simulation parameters as follows: the number of sampling channels is 4, the sampling rate is 100k, the frame rate of the spectrometer is 25kHz, the simulation time is 1s, and a simulated spectral signal is generated using a 50Hz simplified sine wave signal.

[0087] Second, add time shift and random bias to the sine wave signal to automatically generate three other signals with time shift and bias, and obtain 4-channel sampling signals, as Figure 2 shown.

[0088] Third, introduce the 4-channel sampling signals into the mathematical model and simulate the time shift and bias between the 4-channel sampling channels to generate a reflected spectral data signal with clutter.

[0089] Fourth, discretize the reflected spectral data signal into data points, find the central point for a set of 4 data points collected by 4 sampling channels as a set, perform Kalman filtering on each central point, and then plot the waveform diagram after processing, as Figure 3 shown. Through Figure 2 , Figure 3 it can be seen that after Kalman filtering, the previous four waveform curves with time shift and bias have been processed into smooth waveform curves. Example Two

[0090] Embodiment 2 provides a multi-channel sampling clutter processing system for an optical fiber F-P sensor. Based on the multi-channel sampling clutter processing method of an optical fiber F-P sensor in Embodiment 1, the system includes:

[0091] A reflected spectrum data acquisition module, configured to collect the reflected spectrum signals of the optical fiber F-P sensor in parallel and obtain multiplexed reflected spectrum data;

[0092] A first data set module, configured to use the data points corresponding to each reflected spectrum data at the same moment as a group of first data sets, and obtain multiple groups of first data sets at consecutive moments;

[0093] A second data set module, configured to sequentially obtain the center points of each group of first data sets, and use the center points as the second data set;

[0094] A waveform curve output module, configured to perform Kalman filtering on the center points of the second data set to obtain wavelength signal data, and plot the wavelength signal data as a waveform curve for output. Embodiment Three

[0095] Embodiment 3 provides an electronic device, which includes at least one processor; and a memory communicatively connected to the processor; wherein, the memory stores instructions executable by the processor, and when the instructions are executed by the processor, the processor is enabled to execute the multi-channel sampling clutter processing method of an optical fiber F-P sensor as in Embodiment 1.

Claims

1. A multi-channel sampling clutter processing method for an optical fiber F-P sensor, characterized in that, The method includes the following steps: Step S1, parallelly collect the reflected spectral signals of the fiber optic F-P sensor to obtain multiplexed reflected spectral data; Step S2, take the data points corresponding to each path of the reflected spectral data at the same moment as a set of first data sets, and obtain multiple sets of the first data sets at consecutive moments; Step S3, sequentially obtain the center points of each set of the first data sets, and use the center points as the second data set; Obtaining the center points of each set of the first data sets specifically includes: Define the center point to each data point in the first data set The square of the distance is: ； Among them, represents the distance from the nth point to the center point , and are the coordinates of the center point , and are the coordinates of the nth data point; Set the center point The function of the distance from each data point is as follows: ； Solve the minimum value of the function using the iterative formula The iterative formula is as follows: ； Among them, is the learning rate, and are the partial derivatives of the function with respect to the coordinate and the coordinate respectively. Then, the partial derivative of ； Substitute the partial derivatives into the iterative formula to obtain the coordinates of the center point ; , ; Step S4, perform Kalman filtering on the center points of the second data set to obtain wavelength signal data, and plot the wavelength signal data as a waveform curve for output.

2. The multi-channel sampling clutter processing method of an optical fiber F-P sensor according to claim 1, characterized in that In step S1, parallelly collecting the reflected spectral signals of the fiber optic F-P sensor to obtain multiplexed reflected spectral data specifically includes: Input broadband light from the first port of the fiber optic circulator into the fiber optic F-P sensor. The reflected light of the fiber optic F-P sensor is output from the third port of the fiber optic circulator. The output reflected light is split and then enters multiple spectrometers respectively. Each spectrometer simultaneously obtains the electrical signals of the reflected spectral data of the fiber optic F-P sensor. After converting the electrical signals of each path into digital signals, they are sent to the FPGA controller for signal processing to obtain multiplexed reflected spectral data.

3. A multi-channel sampling clutter processing method for an optical fiber F-P sensor according to claim 1, characterized in that In the step S4, the Kalman filtering specifically includes: using the center point to process through establishing a Kalman state prediction equation in sequence. The Kalman state prediction equation is: ； Among them, is the predicted value of the state at the moment, is the state transition matrix, is the optimal state estimate value at the moment, is the control input matrix, is the control input at the moment; The prediction error covariance is: ； Among them, is the predicted error covariance at time is the optimal estimation error covariance at time is the transpose matrix of is the process noise covariance.

4. A multi-channel sampling clutter processing method for an optical fiber F-P sensor according to claim 3, characterized in that The Kalman filter further includes: after processing by establishing a Kalman state prediction equation, updating the center point by establishing a state update equation, where the state update equation is: ； ； ； Among them, is the Kalman gain at time is the observation matrix, and is the observation noise covariance; is the optimal state estimate value at time is the observed value at time is the optimal estimation error covariance at time is the identity matrix.

5. A multi-channel sampling clutter processing method for an optical fiber F-P sensor according to claim 3, characterized in that The Kalman state prediction equation is established by the FPGA controller.

6. A multi-channel sampling clutter processing method for an optical fiber F-P sensor according to claim 4, characterized in that, The state update equation is established by the FPGA controller.

7. A multi-channel sampling clutter processing method for an optical fiber F-P sensor according to claim 1, characterized in that In step S4, plotting the wavelength signal data as a waveform curve for output specifically includes: The FPGA controller sends the wavelength signal data to the host computer software, and the host computer software plots the wavelength signal data as a waveform curve and displays it.

8. A system for multi-channel sampling clutter processing method of an optical fiber F-P sensor according to any one of claims 1-7, characterized in that, It includes: A reflected spectral data acquisition module, configured to parallelly collect the reflected spectral signals of the fiber optic F-P sensor to obtain multiplexed reflected spectral data; A first data set module, configured to take the data points corresponding to each path of the reflected spectral data at the same moment as a set of first data sets, and obtain multiple sets of the first data sets at consecutive moments; A second data set module, configured to sequentially obtain the center points of each set of the first data sets, and use the center points as the second data set; A waveform curve output module, configured to perform Kalman filtering on the center points of the second data set to obtain wavelength signal data, and plot the wavelength signal data as a waveform curve for output.

9. An electronic device, characterized in that, It includes at least one processor; and a memory communicatively connected to the processor; wherein, the memory stores instructions executable by the processor, and when the instructions are executed by the processor, the processor is enabled to execute a multi-channel sampling clutter processing method for a fiber optic F-P sensor according to any one of claims 1-7.

Images