A Demodulation Method for a Distributed Fiber Optic Acoustic Sensing System Based on FPGA Timing Segmentation

Abstract

This invention belongs to the field of acoustic sensing demodulation, specifically a demodulation method for a distributed fiber optic acoustic wave sensing system based on FPGA time-sequence segmentation. This invention uses an FPGA as the processing module for signal demodulation. On the FPGA, three signals with a 120-degree phase difference output from the distributed fiber optic system are acquired. Then, time-sequence segmentation is achieved by FIFO buffering of the spatial point data, ensuring data alignment at each spatial point during signal demodulation. Finally, moving average and other signal demodulation processes are performed on the data at each spatial point in the time domain. This invention requires only a small amount of data buffering to achieve demodulation of the entire system, reaching the theoretical maximum distance for signal demodulation. Ultimately, it achieves signal demodulation of a long-distance distributed fiber optic acoustic wave sensing system with high repetition rate under low resource consumption, providing a new approach and method for hardware-based signal demodulation of distributed fiber optic sensing systems.

Description

Technical Field

[0001] This invention belongs to the field of acoustic sensing demodulation, specifically a demodulation method for a distributed fiber optic acoustic wave sensing system based on FPGA time-sequence segmentation, which can be applied to structural health monitoring, perimeter security, railway tracks, oil pipelines and other fields. Background Technology

[0002] Distributed fiber optic acoustic wave sensing systems monitor vibration signals from long-distance fiber optic cables and locate the vibration source. When an external sound source is present, the sound waves acting on the sensing fiber cause a phase change in the backscattered Rayleigh light within the fiber. This phase change information includes the frequency and phase of the disturbance. Different demodulation algorithms can be used to reconstruct, enhance, and locate the vibration source signal. Distributed fiber optic acoustic wave sensing systems have wide applications in structural health monitoring, perimeter security, railway tracks, and oil pipelines. They are used to monitor the structural safety and health status of systems in real time. For railway tracks and oil pipelines where damage or leakage may occur, they primarily monitor stress wave changes or abnormal sound signals generated by pipe wall vibration. In these complex working environments, where the monitoring range and distance are both long, traditional electronic or mechanical sensors are insufficient for monitoring in such challenging conditions.

[0003] Most currently used distributed fiber optic acoustic wave sensing systems are PC-based signal demodulation systems. However, these systems require demodulation of long-distance data, and the high repetition rates present significant challenges in terms of data processing capabilities, demanding even higher hardware performance. Furthermore, the serial processing of data on the PC leads to substantial delays during demodulation, posing a challenge to real-time monitoring in applications such as perimeter security and oil pipelines.

[0004] Field-Programmable Gate Arrays (FPGAs) offer advantages in parallel data processing. For systems like distributed fiber optic acoustic wave sensing systems (FAAS) with large amounts of data processing, hardware acceleration via FPGAs is crucial for demodulation. However, traditional demodulation methods for FAS require separating spatial point data, necessitating more hardware resources for data buffering and signal demodulation, thus failing to fully leverage the parallel data processing advantages of FPGAs. Furthermore, the large buffer size and spatial point data separation increase demodulation latency in FPGA-based FAS implementations, challenging the real-time performance of the demodulation system.

[0005] For the demodulation schemes of commonly used distributed fiber optic acoustic wave sensing systems, high-performance signal acquisition cards are typically used, and data signal demodulation processing is performed on a PC. However, this places very high demands on the signal acquisition card and the data processing capabilities of the PC, and the huge amount of data makes the real-time performance of data processing still a serious challenge, especially the demodulation distance at high repetition frequencies is severely limited. Summary of the Invention

[0006] To address the aforementioned problems and shortcomings, and to resolve the demodulation distance issues and hardware implementation resource limitations of existing distributed fiber optic acoustic wave sensing systems at high repetition frequencies, this invention provides a demodulation method for distributed fiber optic acoustic wave sensing systems based on FPGA time-series segmentation. Based on the demodulation principle of distributed fiber optic acoustic wave sensing and leveraging the advantages of FPGA data parallel processing, this method achieves real-time signal demodulation for long-distance distributed fiber optic acoustic wave sensing systems with high repetition frequencies under low resource consumption.

[0007] A demodulation method for a distributed fiber optic acoustic wave sensing system based on FPGA time-sequence segmentation includes the following steps:

[0008] (1) The distributed fiber optic acoustic wave sensing system is triggered by the FPGA at a repetition frequency of f (e.g., 10KHz). After each trigger, the distributed fiber optic acoustic wave sensing system will output three detection signals with a phase difference of 120 degrees. The AD acquisition module samples the detection signals (analog signals) output by the system at a sampling frequency of F (e.g., 10MHz). The acquired analog signals are converted into k-bit (e.g., 16-bit) signed digital signals by the AD module.

[0009] (2) The AD acquisition clock is used as the read / write clock for the First-Input-First-Output (FIFO) queue. By delaying the FIFO's read / write clock, the write enable of the FIFO is effectively 1 / f seconds earlier than the read enable, ensuring that the signal sequence stored in the FIFO is the data of all spatial points within one trigger period of 1 / f seconds. Simultaneously, it ensures that the data S written to the FIFO at time n is... n+1 With the read data S n The data is generated by two consecutive triggers of a pulse signal corresponding to the same spatial point, where n represents the trigger time of the nth clock signal, thereby achieving time-series segmentation of the acquired data stream.

[0010] (3) The three signals acquired by the AD are connected to a FIFO, and then the read ports of each FIFO are connected in series with the write ports of a new FIFO. After each channel is connected in four stages, the data of the detection signals of the same spatial point triggered by the pulse signal five times in a row are obtained.

[0011] The five detection data points from one of the three signals can be represented as S. n+4 S n+3 S n+2 S n+1 S n Then the FIFO read channel of this signal will receive signal X. t+4 X t+3 X t+2 X t+1 X t Here, X represents the continuous signal sequence output by the FIFO, and t indicates that the signal becomes valid after time t. By weighting and averaging the five signals from this path, the moving average output signal XL of the distributed fiber optic sensing system is obtained. t+2 Among them, the processing of the three signals is the same, and L corresponds to signals 1, 2, and 3.

[0012] (4) The three signals after the moving average are connected in series through a read / write controlled FIFO to obtain a time-divided data stream X1. t+3 and X1 t+2 X2 t+3 and X2 t+2 X3 t+3 and X3 t+2 Then, the sum of the squares of the three signals is used to obtain signal P, which is then differentiated and cross-multiplied to obtain signal Q.

[0013] (5) Using signal P as the divisor and signal Q as the dividend, perform a division operation to obtain signal Y. Then, perform time-sequential partitioning on signal Y to obtain data stream Y. t0 and Yt1 Y t0 and Y t1 The sums are used to obtain the demodulated signal M.

[0014] Furthermore, the pulse optical signal frequency f of the FPGA is a pulse signal with a frequency of 1kHz to 10kHz and a pulse width greater than 100ns, and the sampling rate F of the AD is an integer multiple of f.

[0015] Furthermore, the FIFO's read / write clock is only activated when enabled, and the read / write clock maintains a fixed delay of 1 / f to keep the data depth in the FIFO constant. This ensures that the data read from and written to the FIFO represents two consecutive samples of a specific spatial point, achieving time-series segmentation of the acquired signal. If the AD sampling clock is F and the trigger repetition frequency is f, then the data depth in the FIFO is F / f data samples. Therefore, the actual depth of the FIFO must be greater than F / f. This invention achieves a time-series segmentation-based moving average by cascading FIFOs. The read data channel of the previous FIFO is connected to the write data channel of the next FIFO. The read / write control enable of each FIFO is consistent, ensuring that the data at the read port of each FIFO represents five consecutive samples of the same spatial point in time, thus achieving time-series segmentation control of the acquired sequence.

[0016] Furthermore, the five consecutive samples are assigned weights of 1 / 8, 1 / 8, 1 / 2, 1 / 8 and 1 / 8 respectively, which makes the acquired signal have good stability and removes noise from the signal.

[0017] Furthermore, step 4 specifically involves: dividing the moving average signal into two sampling sequences using a time-series method, then subtracting these two data points to obtain the differentiated signal. Both the averaged and differentiated data are k bits. Simultaneously, the sum of squares and cross-multiplication will expand the data bit width to 2k bits. Thus, the data obtained after the sum of squares and the cross-multiplication of the differential are both 2k bits. Since the results of the sum of squares and the cross-multiplication of the differential may have different delays, a delay module is added to ensure that the results of the sum of squares and the cross-multiplication of the differential correspond to the same spatial point data.

[0018] Furthermore, in step 5, the division operation involves dividing the result of the sum of squares multiplied by the differential cross-multiplication. During digital signal quantization and acquisition by the distributed fiber optic acoustic sensing system, some distortion points may exist, leading to inaccurate demodulation results. Simultaneously, due to quantization errors in digital signals, the data after the sum of squares is shifted right by d1 bits, and the data after the differential cross-multiplication is shifted left by d2 bits before performing the division operation. Here, d1 and d2 are less than 2k, reducing errors caused by fading and quantization, and improving demodulation stability. This invention, through the design of a combination of an integrator and a half-band filter, performs time-series segmentation on the data stream obtained from the divider, resulting in two segmented data streams. By adding these data streams, integration and half-band filtering of the signal are achieved, yielding the demodulated signal. This replaces integration and simultaneously filters the signal in the frequency domain, increasing the sensitivity of the demodulation system.

[0019] In summary, this invention fully leverages the parallel data processing advantages of FPGAs. It implements time-series segmentation by using FIFO buffering for spatial data points, ensuring data alignment at each spatial point during signal demodulation. Only a small amount of data buffering is needed to demodulate the entire system, achieving the theoretical maximum distance for signal demodulation. For example, at a repetition frequency of 10 kHz, it can demodulate signals over a 10-kilometer fiber optic cable. This invention meets the real-time signal demodulation requirements of distributed fiber optic sensing systems, solving the problem of difficult real-time demodulation of signals over long distances at high repetition frequencies. Furthermore, it exhibits good demodulation performance at different repetition frequencies, making it significant for the application and development of distributed fiber optic acoustic sensing systems. Attached Figure Description

[0020] Figure 1 This is a flowchart of the demodulation method of the present invention;

[0021] Figure 2 This is a timing diagram of signal acquisition for an embodiment;

[0022] Figure 3 This is a signal control design diagram for an embodiment;

[0023] Figure 4 The moving average design diagram is shown in the example.

[0024] Figure 5 This is a demodulation design diagram for an embodiment;

[0025] Figure 6 This is a design diagram of the signal integration and filtering module for an embodiment;

[0026] Figure 7 This is the overall design drawing of the present invention.

[0027] Figure 8 The measured signal time-frequency diagram is shown in the example. Detailed Implementation

[0028] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0029] Figure 1 This is a flowchart of the demodulation method based on FPGA timing segmentation according to the present invention. After the acquired signal passes through the timing segmentation module, it is divided into multiple spatially aligned signals with temporal delays. Then, a weighted moving average filtering module performs a time-domain moving average on the acquired signal. The signal is then used as the input for cross-multiplication of the sum of squares and the derivative by the timing segmentation module, and finally demodulated and output after passing through an integral filtering module. This embodiment is applied to belt conveyor idler roller fault testing, with an actual laying length of 5 kilometers, using a 10kHz repetition frequency triggering system and a 10MHz system sampling frequency.

[0030] A demodulation algorithm for a distributed fiber optic sensing acoustic system based on FPGA time-sequence segmentation is described below:

[0031] (1) By designing the time delay of the sampling clock, a read / write enable signal with a fixed time delay is obtained as the read / write control of the FIFO, ensuring that the data written and read by the FIFO corresponds to the same spatial point of the distributed optical fiber sensing system.

[0032] (2) The FPGA will process the signal X currently acquired from the distributed fiber optic sensing system. t0 After being connected via the FIFO controlled by the read / write mechanism in step 1, the read and write data of the FIFO can always correspond to the same spatial point S due to the read / write control of the FIFO. n The data at t0 and t1 are obtained by connecting the read data of one FIFO with the write data of another FIFO to obtain the same spatial point S. n t0, t1, t2...t n Signals sampled multiple times consecutively.

[0033] (3) Data from the same spatial point are compared with the intermediate sampling time t. n / 2 Weight determination is performed to achieve sliding weight filtering of signals at each spatial point.

[0034] (4) The signal after moving average is connected in series through a read / write controlled FIFO to obtain a time-divided data stream X1. t0 and X1 t1 The processed signals P and Q are then obtained by cross-multiplying the sum of squares and the derivative. t0 and X1 t1By subtracting two sampled data points from the same spatial point, the result of signal differential cross-multiplication can be obtained without separating the collected data from each spatial point.

[0035] (5) The data obtained by dividing signals P and Q are concatenated through a FIFO to obtain a time-divided data stream Y. t0 and Y t1 Y t0 and Y t1 The sums are used to obtain the output signal M.

[0036] Figure 2 This is a timing diagram for signal acquisition based on FPGA timing segmentation in an embodiment. The AD acquisition card outputs a trigger acquisition signal with a frequency of 10kHz and a pulse width of 100ns. The optical distributed fiber optic acoustic wave sensing system outputs three signals with a phase difference of 120 degrees. The AD acquires the signals at a rate of 10MHz, acquiring 1000 data points after each trigger, corresponding to 1000 consecutive locations in space. The actual interval between each data point is 10m, and the sampling clock for each spatial point is 10kHz, with k = 16bit.

[0037] Figure 3 The example is a signal control design diagram based on FPGA timing segmentation. The 10MHz clock of the AD is used as the standard clock to generate a 10KHz repetitive trigger signal as the trigger for the distributed fiber optic acoustic wave sensing system. When the sensing system is triggered, the FIFO performs a read operation. After the next trigger, the FIFO performs a write operation, ensuring that the effective difference between the read and write enable of the FIFO is 1 trigger cycle, that is, ensuring that there are 1000 data in the FIFO clock, and at the same time ensuring that the data written to the FIFO and the data read out are data from the same spatial point.

[0038] The write enable signal of the FIFO starts valid one repetition frequency cycle earlier than the read enable signal. Taking a repetition frequency of 10kHz as an example, the write enable signal starts valid 1ms earlier than the read enable signal, ensuring that the FIFO is not empty when reading. The depth of the FIFO must be greater than the number of sampling clock cycles within one repetition frequency cycle. In this embodiment, the depth of the FIFO should be greater than 1000 to ensure that data buffering does not cause FIFO overflow and thus data loss, and that the data in the FIFO is always maintained at the level of data collected in one trigger cycle, i.e., 1000.

[0039] Figure 4This is an example of a moving average design based on FPGA time-series segmentation. To achieve a lower noise floor in the demodulated signal, four FIFOs are connected in series to obtain five consecutive samples of the acquired signal at the same spatial point. These five samples are assigned weights of 1 / 8, 1 / 8, 1 / 2, 1 / 8, and 1 / 8, respectively. By averaging these five samples, a moving average of the data at each spatial point in the time domain is achieved.

[0040] Figure 5 This is a signal demodulation design diagram based on FPGA timing segmentation, as shown in the example. After the moving average of signals X1, X2, and X3 is processed through a FIFO timing segmentation control, the signals at time t0 and t1 are obtained. Then, the sum of squares and the derivative of these two signals are cross-multiplied to obtain signal P. t1 and Q t1 .

[0041] Figure 6 This is a design diagram of a signal integration and filtering module based on FPGA timing segmentation, as shown in the example. The signal P... t1 and Q t1 The signal Y is obtained after division. t1 Then signal Y t1 After time-series segmentation using a FIFO, and then passing through a designed filter and integrator, the output signal M is obtained by adding the signals at time t1 and t0.

[0042] Figure 7 This is a design diagram of the FPGA-based time-sequence segmentation demodulation method for this invention. In this embodiment, the entire system outputs a 10kHz trigger signal, while the AD converter samples the entire distributed fiber optic sensing system at 10MHz. The operation of all FIFOs is controlled by a clock read / write control module. This module, through synchronization control logic with sampling and triggering, achieves time-sequence segmentation of the acquired signal. By cascading the FIFOs, the system completes time-sequence segmentation-based moving average and signal demodulation. Finally, the demodulated signal output M is obtained through an integral filtering module. This embodiment illustrates the specific implementation flow of the FPGA-based time-sequence segmentation demodulation method for a distributed fiber optic sensing system. The acquisition card speed, trigger clock frequency, FIFO depth, data delay during differential cross-multiplication, and data shifting used in this invention can all be adjusted according to actual conditions. Figure 8 This is the measured signal time-frequency diagram of this embodiment. In the actual belt conveyor system test, this example can correctly restore the time-domain waveform of the signal and see the frequency of the signal in the frequency domain.

[0043] As can be seen from the above embodiments, this invention uses an FPGA as the processing module for the signal demodulation method. On the FPGA, three signals with a 120-degree phase difference output from the distributed optical fiber system are acquired. Then, in terms of timing, the data at each spatial point is buffered using a FIFO to achieve time-series segmentation, ensuring data alignment at each spatial point during signal demodulation. Finally, in the time domain, the data at each spatial point is processed by moving average and signal demodulation. This invention requires only a small amount of data buffering to achieve demodulation of the entire system, reaching the theoretical maximum distance for signal demodulation. Ultimately, it achieves signal demodulation of a high-repetition-rate, long-distance distributed optical fiber acoustic wave sensing system with low resource consumption, providing a new approach and method for hardware-based signal demodulation in distributed optical fiber sensing systems.

Claims

1. A demodulation method for a distributed fiber optic acoustic wave sensing system based on FPGA time-sequence partitioning, characterized in that, Includes the following steps: Step 1: The distributed fiber optic acoustic wave sensing system is triggered by the FPGA at a repetition frequency of f. After each trigger, the distributed fiber optic acoustic wave sensing system outputs three detection signals with a phase difference of 120 degrees. The AD acquisition module samples the detection signals output by the system at a sampling frequency of F. The acquired detection signals are analog signals, which are converted into k-bit signed digital signals by the AD module. Step 2, the collection clock of AD is taken as the read-write clock of the first-in first-out queue FIFO, the write enable of FIFO is made valid faster than the read enable by 1 / f time interval through the read-write clock delay of FIFO, the signal sequence stored in FIFO is guaranteed to be the data of all spatial points in a trigger period 1 / f; at the same time, the data S n+1 corresponding to the same spatial point is written into FIFO at time n, and the data S n corresponding to the same spatial point is read out, wherein n represents the time when the nth clock signal triggers, so as to realize the time sequence segmentation of the collected data stream; Step 3: Connect the three signals acquired by the AD converter to a FIFO, and then connect the read port of each FIFO to the write port of a new FIFO in series. After each FIFO is connected in four stages, the data of the detection signal at the same spatial point triggered by the pulse signal five times in a row are obtained. The five detection data points for each of the three signals can be represented as S. n+4 S n+3 S n+2 S n+1 S n Then the data stream obtained by the FIFO read channel of this signal is X. t X t+1 X t+2 X t+3 X t+4 Where X represents the continuous signal sequence output by the FIFO, and time t indicates that the data from that time onwards is valid; by performing a moving average on the five signals of this path, the moving average output signal XL of the distributed fiber optic sensing system is obtained. t+2 Among them, the processing of the three signals is the same, and L corresponds to signals 1, 2, and 3. Step 4: Connect the three signals after the moving average to a read / write controlled FIFO in series to obtain a time-divided data stream X1. t+3 and X1 t+2 X2 t+3 and X2 t+2 X3 t+3 and X3 t+2 Then, the sum of the squares of the three signals is used to obtain signal P, which is then differentiated and cross-multiplied to obtain signal Q. Step 5: Using signal P as the divisor and signal Q as the dividend, perform a division operation to obtain signal Y; then, perform time-series partitioning on signal Y to obtain data stream Y. t0 and Y t1 Y t0 and Y t1 The sums are used to obtain the demodulated signal M.

2. The demodulation method for a distributed fiber optic acoustic wave sensing system based on FPGA time-sequence partitioning as described in claim 1, characterized in that: The frequency f of the pulsed optical signal of the FPGA is 1kHz-10KHz, and the pulse width is greater than 100ns; the sampling rate F of the AD is an integer multiple of f.

3. The demodulation method for a distributed fiber optic acoustic wave sensing system based on FPGA time-sequence partitioning as described in claim 1, characterized in that: The FIFO's read / write clock is only used when enabled, and the read / write clock always maintains a fixed delay of 1 / f to keep the data depth in the FIFO constant. It also ensures that the data read from and written to the FIFO are two consecutive samples of a certain spatial point, thus achieving time-series segmentation of the acquired signal. The data depth in the FIFO is F / f data samples, and the actual depth of the FIFO is greater than F / f. By cascading FIFOs, a time-series segmentation-based moving average is achieved. The read data channel of the upper-level FIFO is connected to the write data channel of the lower-level FIFO. The read and write control enable of each FIFO is consistent, ensuring that the data of the read port of each FIFO is sampled continuously five times in time from the same spatial point, thus realizing time-series segmentation control of the acquisition sequence.

4. The demodulation method for a distributed fiber optic acoustic wave sensing system based on FPGA time-sequence partitioning as described in claim 3, characterized in that: The five consecutive samples are assigned weights of 1 / 8, 1 / 8, 1 / 2, 1 / 8 and 1 / 8 respectively, which makes the acquired signal have good stability and removes noise from the signal.

5. The demodulation method for a distributed fiber optic acoustic wave sensing system based on FPGA time-sequence partitioning as described in claim 3, characterized in that, Step 4 specifically involves: dividing the time sequence of the signal after the moving average into two sampling sequences, then subtracting the two data to obtain the differentiated signal. Both the averaged and differentiated data are k bits. Simultaneously, the sum of squares and cross-multiplication will expand the data bit width to 2k bits. Thus, the data obtained after the sum of squares and the cross-multiplication of the derivative are both 2k bits. Furthermore, by adding a delay module, it is ensured that the results of the sum of squares and the cross-multiplication of the derivative correspond to the same spatial point data.

6. The demodulation method for a distributed fiber optic acoustic wave sensing system based on FPGA time-sequence partitioning as described in claim 3, characterized in that: In step 5, the division operation involves shifting the squared data to the right by d1 bits, and the differential cross-multiplication data to the left by d2 bits, and then performing the division operation again. d1 and d2 are less than 2k.

Images