Fiber grating sensor wavelength demodulation method and system based on FPGA

Abstract

The invention provides a fiber grating sensor wavelength demodulation method and system based on an FPGA, belongs to the field of fiber sensing technology and hardware acceleration, can realize high-speed, high-precision and low-power consumption multichannel FBG wavelength demodulation, and meets the reliability requirement of spaceflight embedded deployment. By utilizing the parallel processing capability of the FPGA, simultaneous processing of multichannel spectral data is realized, and the demodulation speed is greatly improved; a pipeline architecture is adopted, so that data acquisition, processing and output can be continuously carried out, and the system throughput is remarkably improved; the innovative adaptive threshold algorithm improves the accuracy and reliability of peak detection; a centroid algorithm realized by hardware ensures high precision and stability of wavelength calculation; the whole scheme has the characteristics of low power consumption and high integration level, and is very suitable for embedded application.

Description

TECHNICAL FIELD

[0001] The application relates to an FPGA-based fiber grating sensor wavelength demodulation method and system, and belongs to the field of optical fiber sensing technology and hardware acceleration. BACKGROUND

[0002] Key structures such as aircraft and spacecraft shells are prone to structural fatigue and deformation in complex and harsh service environments, which threatens mission safety and causes damage. Long-term and real-time health monitoring of the structures is of great significance to safety protection. Fiber Bragg grating (FBG) sensors have become an ideal technology to achieve this demand due to their advantages such as anti-interference and easy networking. The sensors perceive strain and temperature through the drift of the reflection center wavelength, so fast and accurate demodulation of the FBG wavelength is the core prerequisite for obtaining state information and assessing structural safety.

[0003] Existing demodulation systems are mostly based on commercial demodulators or general-purpose computers, which have obvious limitations: the data processing speed is limited by the serial architecture, making it difficult to meet the requirements of real-time processing of multiple signals; the system is large in size and high in power consumption, and low in integration, which is difficult to meet the harsh requirements of space applications for lightweight, low power consumption and high reliability. SUMMARY

[0004] Therefore, the application provides an FPGA-based fiber grating sensor wavelength demodulation method and system, which can realize high-speed, high-precision and low-power multi-channel FBG wavelength demodulation and meet the reliability requirements of space embedded deployment.

[0005] To achieve the purpose of the application, the following technical solutions are provided.

[0006] The application provides an FPGA-based fiber grating sensor wavelength demodulation method, which comprises the following steps: Step S1: Collecting original spectrum data of a fiber grating sensor and inputting the data to an FPGA; Step S2: Real-time preprocessing of the original spectrum data in the FPGA to obtain smoothed spectrum data; Step S3: Based on the smoothed spectrum data, using a sliding window to calculate local statistical features in real time and generating an adaptive dynamic threshold; Step S4: Comparing the smoothed spectrum data with the corresponding adaptive dynamic threshold and combining a neighborhood comparison algorithm to identify candidate peak data segments; Step S5: Based on a pre-stored wavelength-pixel calibration lookup table, using a centroid positioning algorithm to perform parallel hardware calculation on the candidate peak data segments to calculate the center wavelength value with sub-pixel accuracy; Step S6: Outputting the center wavelength value.

[0007] A wavelength demodulation system for fiber grating sensor based on FPGA, comprising: a spectrum acquisition module for acquiring original spectrum data; an FPGA processing module connected with the spectrum acquisition module for receiving and processing the original spectrum data; an output module connected with the FPGA processing module for outputting demodulation results; wherein the FPGA processing module further comprises: a data preprocessing unit for buffering and smoothing filtering the original spectrum data; an adaptive threshold calculation unit for calculating adaptive dynamic threshold in real time; a peak detection unit for identifying candidate peak data segments; a high-precision wavelength calculation unit for calculating central wavelength values.

[0008] The method according to claim 1, wherein the real-time preprocessing of the original spectrum data in step S2 comprises buffering the original spectrum data using a ping-pong FIFO or RAM data storage structure, and smoothing filtering the buffered data using one of a rectangular filter, a triangular filter or a Gaussian filter.

[0009] The method according to claim 1, wherein the local statistical features in step S3 comprise an arithmetic mean μ and a standard deviation σ within a sliding window, and the adaptive dynamic threshold is calculated by the formula Threshold = μ + K * σ, wherein K is a configurable sensitivity coefficient.

[0010] The method according to claim 1, wherein the neighborhood comparison algorithm in step S4 is to determine whether a current data point is greater than M adjacent data points on the left and right sides of the current data point and greater than the adaptive dynamic threshold of the current point, wherein M is a predetermined positive integer.

[0011] The method according to claim 1, wherein the centroid positioning algorithm in step S5 calculates the central wavelength value according to the formula: λo = Σ(Ii * λi) / ΣIi, wherein Ii is the light intensity value of the i-th pixel point in the candidate peak data segment, and λi is the wavelength value corresponding to the i-th pixel point obtained through the wavelength-pixel calibration lookup table.

[0012] The system according to claim 2, wherein the data preprocessing unit realizes data buffering using a ping-pong RAM structure.

[0013] The system according to claim 2, wherein the high-precision wavelength calculation unit comprises: Block ROM pre-storing wavelength-pixel calibration data; a product-accumulator tree for calculating the product of "Ii * λi" and "ΣIi"; a hardware divider for performing the division operation "λo = Σ(Ii * λi) / ΣIi".

[0014] The system according to claim 2, wherein each unit in the FPGA processing module is connected through a pipeline architecture, the spectral acquisition module is connected with the FPGA through an SPI data interface, and the output module is an Ethernet interface or a UART interface.

[0015] The system according to claim 2, wherein the system is applied to aviation, spacecraft or other structural health monitoring, and is used for real-time demodulation of strain and temperature signals of a plurality of fiber grating sensors arranged thereon.

[0016] Advantages (1) The present application utilizes the parallel processing capability of FPGA to realize simultaneous processing of multi-channel spectral data, greatly improving the demodulation speed; the pipeline architecture is adopted to enable continuous data acquisition, processing and output, significantly improving the system throughput; the innovative adaptive threshold algorithm improves the accuracy and reliability of peak detection; the centroid algorithm realized by hardware ensures the high precision and stability of wavelength calculation; the overall scheme has the characteristics of low power consumption and high integration, and is very suitable for embedded applications.

[0017] (2) The system of the present application comprises a continuous spectral acquisition module, an FPGA processing module and an output module; the FPGA processing module adopts a parallel pipeline architecture, and sequentially integrates a spectral preprocessing module, an adaptive peak detection module and a centroid method positioning module; The preprocessing module performs real-time baseline correction and filtering on the spectrum; the adaptive peak detection module generates a dynamic threshold to robustly identify candidate peaks by real-time calculation of local noise statistical characteristics; the centroid method positioning module quickly calculates the sub-pixel precision center wavelength through a pre-stored wavelength lookup table by a hardware pipeline.

[0018] (3) The present application is a fiber grating wavelength demodulation scheme based on FPGA parallel pipeline architecture, which solves the problems of poor real-time performance, difficulty in processing multi-channel data and inability to meet the requirements of embedded deployment of existing demodulation schemes, realizes high speed, high precision and strong anti-noise capability of wavelength demodulation, and is suitable for multi-channel high-speed grating demodulation applications.

[0019] (4) The present application is particularly suitable for fields such as spacecraft structural health monitoring, and can meet the demand for real-time and high-precision demodulation of multiple fiber grating sensors. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 The application provides a wavelength demodulation method for a fiber grating sensor based on an FPGA.

[0021] Figure 2 The application provides a wavelength demodulation method for a fiber grating sensor based on an FPGA. DETAILED DESCRIPTION

[0022] The application will be further described below in combination with the drawings and specific embodiments.

[0023] The application provides a wavelength demodulation method for a fiber grating sensor based on an FPGA, and a method flowchart is shown in the figure. Figure 1 The flowchart and its description completely embody the parallel and pipeline processing architecture of the application, how to convert original spectrum data into high-precision wavelength information step by step, and highlight the high speed and real-time technical advantages. The steps of the application include: Step S1: collecting original spectrum data of a fiber grating sensor and inputting the data into an FPGA; Step S2: performing real-time preprocessing on the original spectrum data in the FPGA to obtain smoothed spectrum data; Step S3: based on the smoothed spectrum data, using a sliding window to calculate local statistical features in real time and generating an adaptive dynamic threshold; Step S4: comparing the smoothed spectrum data with the corresponding adaptive dynamic threshold and combining a neighborhood comparison algorithm to identify candidate peak data segments; Step S5: based on a pre-stored wavelength-pixel calibration lookup table, using a centroid positioning algorithm to perform parallel hardware calculation on the candidate peak data segments to calculate a center wavelength value with sub-pixel accuracy; Step S6: outputting the center wavelength value.

[0024] The original spectrum data is buffered by using a ping-pong FIFO or a RAM data storage structure, and the buffered data is smoothed by using a rectangular filter, a triangular filter or a Gaussian filter suitable for peak features. The local statistical features in step S3 include an arithmetic mean (μ) and a standard deviation (σ) in the sliding window, and the adaptive dynamic threshold is calculated by the formula Threshold = μ + K * σ, wherein K is a configurable sensitivity coefficient. The neighborhood comparison algorithm in step S4 is to determine whether the current data point is greater than M adjacent data points on the left and right and greater than the adaptive dynamic threshold of the current point, wherein M is a preset positive integer. The centroid positioning algorithm in step S5 calculates the center wavelength value according to the following formula: λo = Σ(Ii * λi) / ΣIi wherein Ii is the light intensity value of the i-th pixel point in the candidate peak data segment, and λi is the wavelength value corresponding to the i-th pixel point obtained through the wavelength-pixel calibration lookup table.

[0025] A piece of Xilinx A7 series FPGA is used as the core processor to construct the demodulation system. The FPGA is connected with the spectrum signal acquisition end through the SPI data interface or other interfaces to collect the original spectrum data at a rate of 1Msps. The wavelength and peak value are calculated. The system is externally equipped with an independent clock management chip and a DDR3 memory to provide accurate clock and additional data buffer space for the FPGA.

[0026] As Figure 1 The core data processing pipeline of the fiber grating wavelength demodulation system based on FPGA is clearly shown, wherein the data input is specifically as follows: the original spectrum data stream collected by the acquisition module of the demodulation system is input to the FPGA through the data interface.

[0027] The data preprocessing is specifically as follows: Input buffer: the data is first buffered in the "ping-pong FIFO" structure to ensure the continuity of the data stream.

[0028] Smooth filtering: then, the data is subjected to smooth processing in the Gaussian filter window to effectively suppress noise.

[0029] The specific steps of adaptive threshold calculation are as follows: The smoothed data stream enters the sliding window calculation module to calculate the average value (mu) and the standard deviation (sigma) of the local data in real time. According to the formula Threshold = mu + K * sigma, the detection threshold of each data point is dynamically generated.

[0030] The specific steps of peak detection are as follows: The processed data stream is simultaneously sent to the neighborhood comparison logic and the threshold comparison logic.

[0031] The neighborhood comparison logic judges whether the current point is greater than its left and right neighborhood points.

[0032] The threshold comparison logic judges whether the current point is greater than the dynamic threshold.

[0033] Only when all the above conditions are met, the data point is determined as a valid candidate peak, and the data segment where the data point is located is output to the next stage.

[0034] The specific steps of high-precision wavelength calculation are as follows: Lookup table (LUT): according to the pixel coordinates in the candidate peak segment, the corresponding wavelength value (lambda i) is indexed from the pre-calibrated wavelength lookup table.

[0035] Centroid method calculation: through a dedicated hardware multiply-accumulate tree, Σ(Ii * λi) and ΣIi are calculated in parallel.

[0036] Division operation: finally, the operation of λo= Σ(Ii * λi) / ΣIi is completed through a divider, and the sub-pixel accuracy center wavelength result is obtained.

[0037] The application provides a fiber grating wavelength demodulation system based on FPGA, comprising a spectrum acquisition module for acquiring original spectrum data; an FPGA processing module connected with the spectrum acquisition module for receiving and processing the original spectrum data; and an output module connected with the FPGA processing module for outputting demodulation results; wherein the FPGA processing module further comprises: a data preprocessing unit for buffering and smoothing filtering the original spectrum data; an adaptive threshold calculation unit for calculating adaptive dynamic threshold in real time; a peak detection unit for identifying candidate peak data segments; and a high-precision wavelength calculation unit for calculating center wavelength values. Specifically, the data preprocessing unit is responsible for buffering and smoothing filtering the input spectrum data, adopts a suitable filter for noise suppression of the original spectrum data according to signal characteristics, and retains effective peak characteristics; the adaptive threshold calculation unit calculates dynamic threshold of each data point in real time based on local statistical analysis, and provides a basis for subsequent peak detection; the peak detection unit preliminarily identifies peak segments in the spectrum through a neighborhood data comparison algorithm combined with dynamic threshold and peak width parameters; and the high-precision wavelength calculation unit adopts a centroid positioning algorithm, performs pixel coordinate, light intensity value and wavelength value multiply-accumulate operation in a hardware pipeline manner through a pre-stored wavelength-pixel calibration lookup table, and finally realizes center wavelength calculation of sub-pixel accuracy through a divider.

[0038] Specifically, the data preprocessing unit realizes data buffering in a ping-pong RAM structure. The high-precision wavelength calculation unit comprises: a Block ROM pre-stored with wavelength-pixel calibration data; a multiply-accumulator tree for calculating Σ(Ii * λi) and ΣIi; and a hardware divider for performing division operation λo= Σ(Ii * λi) / ΣIi. i * λ i ” and “ΣI i ” o = Σ(I i * λ i ) / ΣI i ”

[0039] The units in the FPGA processing module are connected through a pipeline architecture, the spectrum acquisition module is connected with the FPGA through a data interface, and the output module is a common Ethernet interface or a UART interface or other communication interfaces.

[0040] The system is applied to the health monitoring of an aircraft, a spacecraft and other shells or structures, and is used for real-time demodulation of strain and temperature signals of a plurality of fiber grating sensors arranged thereon.

[0041] The present application includes but is not limited to the above embodiments, any equivalent replacement or partial improvement made under the principle of the present application shall be considered within the protection scope of the present application.

Claims

1. A wavelength demodulation method for a fiber grating sensor based on FPGA, characterized in that, The method comprises the following steps: Step S1: collecting original spectrum data of the fiber grating sensor and inputting the original spectrum data into the FPGA; Step S2: performing real-time preprocessing on the original spectrum data in the FPGA to obtain smoothed spectrum data; Step S3: based on the smoothed spectrum data, using a sliding window to calculate local statistical features in real time and generating an adaptive dynamic threshold value; Step S4: comparing the smoothed spectrum data with the corresponding adaptive dynamic threshold value and combining a neighborhood comparison algorithm to identify candidate peak data segments; Step S5: based on a pre-stored wavelength-pixel calibration lookup table, using a centroid positioning algorithm to perform parallel hardware calculation on the candidate peak data segments to calculate a center wavelength value with sub-pixel accuracy; Step S6: outputting the center wavelength value.

2. The method of claim 1, wherein, The real-time preprocessing of the original spectrum data in the step S2 comprises: using a ping-pong FIFO or RAM data storage structure to cache the original spectrum data; and using one of rectangular filtering, triangular filtering or Gaussian filtering to perform smoothing filtering on the cached data.

3. The method according to claim 1 or 2, characterized in that, The local statistical features in the step S3 comprise an arithmetic mean μ and a standard deviation σ in the sliding window, and the adaptive dynamic threshold value is calculated by the formula Threshold = μ + K * σ, wherein K is a configurable sensitivity coefficient.

4. The method of claim 3, wherein, The neighborhood comparison algorithm in the step S4 is to determine whether the current data point is greater than M adjacent data points on the left and right and greater than the adaptive dynamic threshold value of the current point, wherein M is a pre-set positive integer.

5. The method of claim 4, wherein, The centroid positioning algorithm in the step S5 calculates the center wavelength value according to the following formula: λo = Σ(Ii * λi) / ΣIi wherein Ii is the light intensity value of the i-th pixel point in the candidate peak data segment, and λi is the wavelength value corresponding to the i-th pixel point obtained through the wavelength-pixel calibration lookup table.

6. A wavelength demodulation system for fiber grating sensors based on FPGA, characterized in that, The method comprises: a spectrum acquisition module configured to acquire original spectrum data; an FPGA processing module connected with the spectrum acquisition module and configured to receive and process the original spectrum data; an output module connected with the FPGA processing module and configured to output demodulation results; The FPGA processing module further comprises: a data preprocessing unit configured to cache and smoothing filter the original spectrum data; an adaptive threshold calculation unit configured to calculate an adaptive dynamic threshold value in real time; a peak detection unit configured to identify candidate peak data segments; a high-precision wavelength calculation unit configured to calculate a center wavelength value.

7. The system of claim 6, wherein, The data preprocessing unit uses a ping-pong RAM structure to realize data caching.

8. The system of claim 6 or 7, wherein, The high-precision wavelength calculation unit comprises: a Block ROM pre-stored with wavelength-pixel calibration data; a product accumulator tree configured to calculate "Ii * λi" and "ΣIi"; a hardware divider configured to perform a division operation "λo = Σ(Ii * λi) / ΣIi".

9. The system of claim 6 or 7, wherein, The units in the FPGA processing module are connected through a pipeline architecture, the spectrum acquisition module is connected with the FPGA through an SPI data interface, and the output module is an Ethernet interface or a UART interface.

10. The system of claim 6 or 7, wherein, The system is applied to the structural health monitoring of an aircraft, a spacecraft or other structures, and is used for real-time demodulation of strain and temperature signals of a plurality of fiber grating sensors arranged thereon.

Images