Method for enhancing DTS spatial resolution based on inverse filtering

Abstract

The invention discloses a method for enhancing DTS spatial resolution based on inverse filtering. The method comprises the following steps that firstly, observation signals are collected and preprocessed; establishing an observation model, representing the preprocessed observation signal as convolution of real temperature distribution and system detection pulse in a spatial domain, and superposing noise; then discrete sequence representation of the detection pulse is obtained or constructed, and normalization processing is carried out; constructing a regularization inverse filter in a frequency domain based on the detection pulse and the preprocessed observation signal; performing frequency domain transformation on the preprocessed observation signal, and multiplying the frequency spectrum of the observation signal by a regularization inverse filter in a frequency domain to obtain a recovered frequency domain signal; and carrying out inverse frequency domain transformation on the recovered frequency domain signal, and obtaining a time domain recovered signal after taking a real part. The method can effectively recover the fuzzy temperature signal characteristics of the detector, improves the reduction precision of the peak value and the edge, introduces the regularization term to suppress the high-frequency noise amplification, and gives consideration to the recovery performance and the stability.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of optical fiber sensing and signal processing, and particularly to a method for enhancing spatial resolution of DTS based on inverse filtering. BACKGROUND

[0002] Many application scenarios require distributed temperature sensors (DTS) to accurately locate temperature gradients or hotspots at a finer spatial scale, in order to quickly locate leaks, local overheating or structural damage. These requirements have driven the comprehensive improvement of higher spatial resolution, while taking into account the range, time response and reliability.

[0003] DTS usually reverses the temperature distribution along the fiber based on the optical signal of Raman scattering in the fiber changing with temperature. The physical upper limit of spatial resolution is mainly determined by the pulse width of the laser: the propagation distance of light corresponding to the pulse length is the minimum distinguishable interval between two points; the shorter the pulse, the higher the theoretical resolution, but short pulses carry less energy, resulting in a decrease in signal-to-noise ratio (SNR). On the other hand, time-domain measurement is also limited by the bandwidth of the detector, fiber dispersion and system response function; in long-distance applications, signal attenuation and noise accumulation further reduce the available resolution.

[0004] In order to break through the contradiction of SNR reduction caused by short pulses, multiple technical routes have been developed in research and engineering practice: 1) Pulse and cable design: shorten the pulse width, increase the peak power, use low-loss or functional (such as Bragg grating array, multi-core / multi-mode) optical cables to improve the distinguishable ability at the source.

[0005] 2) Transmission / reception strategy: use pulse coding (e.g. PRBS, Golay code), pulse compression, coherent detection or frequency domain technology (such as OFDR) to improve the equivalent resolution and SNR.

[0006] 3) Signal processing and inversion algorithm: through deconvolution, super-resolution reconstruction, compressed sensing, sparse representation, Bayesian estimation and deep learning, etc. Methods recover finer temperature structure from measurements blurred by noise and system response.

[0007] 4) Multi-modal and fusion scheme: combine multi-channel, multi-wavelength, Bragg fiber, other sensing (such as acoustics or strain) and prior models for joint inversion to enhance local resolution information.

[0008] 5) System integration and calibration: reduce system errors through accurate instrument response modeling, calibration and field calibration, so that algorithm-based super-resolution or deconvolution methods are more effective.

[0009] In practical engineering, shorter pulse or higher peak power will reduce SNR or increase the risk of nonlinearity and fiber damage; encoding and coherent technology brings complex hardware and synchronization requirements; algorithm method is sensitive to noise, model error and calculation amount; and the requirements of long distance, large dynamic range, real-time and cost limit the applicability of a single scheme. SUMMARY

[0010] The purpose of the present application is to provide a method for enhancing the spatial resolution of DTS based on inverse filtering, which can effectively improve the spatial resolution of distributed optical fiber temperature measurement system under noise conditions. The method is based on the convolution model of observation signal and detection pulse, and realizes deconvolution recovery through frequency domain regularization inverse filtering, and significantly improves the detection ability of the system to spatial temperature change under the premise of ensuring the suppression of noise amplification.

[0011] To achieve the above purpose, the present application provides a method for enhancing the spatial resolution of DTS based on inverse filtering, comprising the following steps: S1, collecting observation signals and performing pretreatment operation; S2, establishing an observation model, expressing the pretreated observation signal as the convolution of real temperature distribution and system detection pulse in spatial domain and superimposing noise; S3, obtaining or constructing a discrete sequence representation of the detection pulse, and performing normalization processing; S4, based on the detection pulse of S2 and the pretreated observation signal, constructing a regularization inverse filter in the frequency domain; S5, performing frequency domain transformation on the pretreated observation signal, and multiplying its frequency spectrum with the regularization inverse filter in the frequency domain to obtain a recovered frequency domain signal; S6, performing inverse frequency domain transformation on the recovered frequency domain signal, and taking the real part to obtain a time domain recovery signal.

[0012] Preferably, S1 is specifically: The original backscattering signal is collected by a distributed temperature sensor DTS, the original temperature data sequence along the length of the optical fiber is obtained by demodulation, and the original temperature data sequence is oversampled.

[0013] Preferably, in S2, the along-line observation signal of the DTS system is expressed as the convolution of real temperature distribution and system detection pulse in spatial domain and superimposed noise, and the expression is as follows: ; Wherein, is the observation signal, is the detection pulse, is the real temperature distribution, is the observation noise.

[0014] Preferably, in S3, the step of obtaining the probe pulse is specifically: Through experimental calibration, the short pulse light emitted by the DTS is directly received, the signal waveform is measured and stored, and the signal waveform is taken as the probe pulse response.

[0015] Preferably, in S3, the step of constructing the probe pulse is specifically: The probe pulse response is fitted by using a parameterized function, the parameterized function is a Gaussian window function , the window width and shape are controlled by the ProbeWidth and alpha parameters, and the probe pulse response is normalized to make .

[0016] Preferably, the original observation signal is also resampled, baseline removed, windowed or zero-padded.

[0017] Preferably, S4 is specifically: The probe pulse and the preprocessed observation signal are subjected to FFT Fourier transform, the length of FFT is set as N, and and are obtained, and a regularization inverse filter is constructed in the frequency domain , and the expression is as follows: ; Wherein, is the conjugate complex of , is the power spectrum, is a regularization parameter, and is used to suppress noise amplification at near zero frequency.

[0018] Preferably, in S5, the calculation formula of the recovered frequency domain signal is as follows: .

[0019] Preferably, in S6, the inverse Fourier transform is performed on to obtain the time domain recovered signal .

[0020] Therefore, the method for enhancing the spatial resolution of the DTS based on the inverse filter in the application has the following beneficial effects: (1) In the presence of noise, the spatial broadening effect caused by convolution is effectively suppressed by the frequency domain regularization inverse filter, thereby significantly improving the spatial resolution of the DTS system, and making the observed temperature jump or local hot spot more sharp and narrow in space.​​

[0021] (2) The present invention has a certain tolerance for the form of the detection pulse, and can be approximated by the experimentally measured pulse or parameterized Gaussian window. It is highly adaptable and simple to implement, and is easy to deploy in existing DTS systems.

[0022] (3) The present invention has a simple structure and moderate computational overhead, making it suitable for implementation on embedded platforms and PCs, and has good engineering application value.

[0023] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0024] Figure 1 This is an overall flowchart of an embodiment of the method for enhancing DTS spatial resolution based on inverse filtering according to the present invention; Figure 2 This is a hardware connection diagram of a first embodiment of the method for enhancing DTS spatial resolution based on inverse filtering according to the present invention; Figure 3 This is an example diagram comparing the original signal and the reconstructed signal on the distance axis in Embodiment 1 of the method for enhancing DTS spatial resolution based on inverse filtering according to the present invention, where (a) is the original signal and (b) is the reconstructed signal. Detailed Implementation

[0025] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0026] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0027] like Figure 1 As shown, a method for enhancing DTS spatial resolution based on inverse filtering includes the following steps: S1. Acquire and preprocess the observed signals, specifically: The original backscattering signal is collected by a distributed temperature sensor (DTS), a raw temperature data sequence along the length of the optical fiber is obtained by demodulation, and the raw temperature data sequence is oversampled.

[0028] S2, the observation model is established, the preprocessed observation signal is represented as the convolution of the real temperature distribution and the system detection pulse (point spread function) in the spatial domain and the superposition of noise, and the expression is as follows: ; Among them, is the observation signal, is the detection pulse, that is, the system response or the point spread function, is the real temperature distribution, is the observation noise.

[0029] S3, a discrete sequence representation of the detection pulse is obtained or constructed, and normalization processing is performed.

[0030] The detection pulse can be obtained or constructed by experimental measurement, system modeling or fitting with a parameterized function (for example, a Gaussian window function) , wherein the step of obtaining the detection pulse is specifically: In the form of experimental calibration, the short pulse light emitted by the DTS is directly received, the signal waveform is measured and stored, and the signal waveform is taken as the detection pulse response.

[0031] The step of constructing the detection pulse is specifically: The parameterized function is used to fit the detection pulse response, and the Gaussian window function is selected as the parameterized function , wherein the window width and shape are controlled by the ProbeWidth and alpha parameters, and the detection pulse response is normalized to make .

[0032] The present application can also resample, baseline (average value), window processing or zero padding operation on the original observation signal , so as to facilitate frequency domain processing and reduce boundary effect. Resampling is only used for numerical processing accuracy, and is not considered as a physical resolution improvement means.

[0033] S4, based on the detection pulse of S2 and the preprocessed observation signal, a regularization inverse filter is constructed in the frequency domain, and S4 is specifically: The detection pulse and the preprocessed observation signal are subjected to FFT Fourier transform, the length of FFT is set to N, and and are obtained, and a regularization inverse filter is constructed in the frequency domain , whose expression is as follows: ; wherein, is a conjugate complex of is a power spectrum, is a regularization parameter for suppressing noise amplification at near zero frequencies.

[0034] S5, the pre-processed observation signal is subjected to frequency domain transformation, and its spectrum is multiplied by a regularized inverse filter in the frequency domain to obtain a recovered frequency domain signal , and the calculation formula is as follows: .

[0035] S6, the recovered frequency domain signal is subjected to inverse Fourier transformation in the frequency domain, and the real part is taken to obtain a time domain recovered signal .

[0036] Embodiment One A hardware environment is built, including a water bath, a distributed fiber temperature sensing device DTS, and a sensing optical fiber. In this embodiment, the water bath is used to provide a stable constant temperature environment, and the temperature of the water bath is set to be constant at 37 The sensing optical fiber serves as a temperature sensing carrier and is used to conduct laser signals along the length direction of the optical fiber, and simultaneously generates temperature-related optical signals through Raman scattering as original physical signal sources.

[0037] The distributed fiber temperature sensing device DTS includes a DTS host, an ADC module, a DSP module, a display module, and a storage module. The DTS host includes a laser emission module for emitting a probe pulse light of a specific width, i.e., a physical light source corresponding to ProbePulse; an optical receiving module for receiving scattered optical signals returned by the sensing optical fiber and converting them into analog electrical signals such as current / voltage signals. The ADC module is used to receive analog electrical signals output by the DTS host and convert them into digital signals, i.e., original sequences Ta, to provide a calculable data source for subsequent digital signal processing.

[0038] The DSP module is the core processing unit of the present application and is used for data preprocessing, ProbePulse construction, frequency domain inverse filtering, signal recovery, and the implementation of other algorithms, and is the hardware core for improving spatial resolution.

[0039] The display module is used for visual display of the processing results, including comparison curve graphs of original signals and recovered signals, temperature distribution maps, etc., and supports real-time observation of resolution improvement effects.

[0040] ​The storage module is used for persistently storing the original collected data, the constructed probe pulse data ProbePulse, the restored signal data inverseRestored and the visualized image file, and supports subsequent data tracing and secondary analysis.

[0041] The signal transmission process is as follows: The sensing and collecting stage: the laser emission module of the DTS host emits probe pulse light to the sensing optical fiber, and the optical signal produces Raman scattered light related to temperature when propagating in the optical fiber, and the scattered light is returned to the optical receiving module (sensor) of the DTS host along the optical fiber.

[0042] The signal conversion stage: the sensor converts the returned optical signal into an analog electrical signal, which is transmitted to the ADC module, and the ADC quantizes the analog electrical signal into a digital sequence Ta.

[0043] The data processing stage: the DSP module reads the digital sequence Ta output by the ADC module, executes the subsequent software process (intercepts the effective points Sm0, pre-processes, constructs ProbePulse, inverse filtering in frequency domain, signal restoration), and obtains the restored signal with improved spatial resolution.

[0044] The result output stage: the DSP module transmits the restored signal and the contrast image to the display module for real-time display, and writes various data into the storage module for persistent storage.

[0045] Dynamic adaptability: when the temperature of the sensing optical fiber changes or the DTS host collects a new frame of data, the above process is executed in a loop, the DSP module continuously performs inverse filtering processing on the newly input digital sequence, realizes real-time response to temperature changes, and supports the dynamic demand of repeatedly performing inverse filtering in frequency domain on each frame / each segment of data.

[0046] The software process is as follows: First, data reading and preprocessing are performed: The original sequence Ta is read from the collection system, and 12000 effective points near the position of the wave peak are intercepted as the initial sequence Sm0.

[0047] If it is necessary to improve the processing resolution, a resampling operation can be performed on Sm0 to obtain the pre-processed signal Sm = resample(Sm0, 10, 1), and the resampling ratio can be adjusted according to actual needs. To further optimize the signal quality, Sm can also be selectively de-meaned or band-pass filtered, which removes low-frequency drift and high-frequency noise in the signal and lays a foundation for subsequent processing.

[0048] Then, the construction of the light pulse, i.e. the probe pulse ProbePulse, is performed, such as Figure 2As shown, the oscilloscope is used to directly measure the probe pulse light of the DTS system to obtain the real pulse signal. If the conditions are limited, the pulse ProbePulse is constructed based on the characteristics of the actual pulse light using a Gaussian window. After the construction of the pulse, the ProbePulse is normalized to ensure the stability and accuracy of the subsequent filtering process.

[0049] Subsequently, the construction of the frequency domain inverse filter is entered. The length of FFT is first set to N. To ensure the processing effect, N needs to be no less than the length of the preprocessed signal Sm. In actual application, it is recommended that N be no less than the sum of the length of Sm and the length of ProbePulse, so as to avoid time domain wraparound artifacts. Then, the Fourier transform is performed on the constructed and normalized ProbePulse to obtain its frequency domain representation H = fft(ProbePulse, N). When the frequency domain inverse filter inverseFilter is constructed based on H, the regularization processing is added to avoid the problem of numerical explosion of H when it is close to 0.

[0050] The signal recovery process starts with the preprocessed signal Sm. First, the FFT transform is performed on Sm to obtain the corresponding frequency domain representation The Sm_fft is multiplied point by point with the constructed inverseFilter in the frequency domain to obtain the recovered frequency domain signal . Then, the inverse Fourier transform is performed on and the real part is taken to obtain the recovered time domain signal . Combined with the zero padding operation before, the effective segment with the same length as Sm is cut out from as the final recovered signal .

[0051] After the signal recovery, the visualization and saving of the results are performed. The comparison curve of the original signal and the reconstructed signal in the corresponding distance range is drawn as shown in Figure 3 . The difference between the two signals is intuitively presented to complete the analysis. At the same time, the generated comparison image and the recovered signal data are saved to the file system, which is convenient for subsequent tracing and in-depth study.

[0052] In the implementation process, attention needs to be paid to multiple key parameters and implementation considerations. The regularization constant has a significant impact on the recovery effect, the smaller the value, the more obvious the high-frequency enhancement effect, but the risk of noise amplification also increases. In actual selection, the noise variance or cross-validation can be used to determine it.

[0053] The selection of ProbePulse preferentially adopts the real probe pulse response obtained by experiment calibration, and if a Gaussian window fitting is adopted, key parameters such as window width ProbeWidth and shape parameter alpha need to be accurately set. The setting of FFT length N in combination with zero padding operation, after frequency domain multiplication, must be clipped to the required length. When the probe pulse response is unknown, blind deconvolution or iterative method can be used as a deformation embodiment.

[0054] From the perspective of module relationship, the static level functional modules follow the connection logic from sensor to ADC module, then to DSP module, and finally to display and storage. The data flow between software modules starts from Sm and ProbePulse, and after FFT transformation, inverseFilter processing, and IFFT inverse transformation, inverseRestored is output.

[0055] The input Sm at the dynamic level changes over time or distance, and the system needs to repeatedly perform frequency domain inverse filtering operation for each frame or each segment of data. The regularization term controls the gain of different frequency components in the frequency domain, and realizes the balance between signal recovery quality and noise amplification suppression when the signal appears transient change.

[0056] Therefore, the present application adopts the above-mentioned method for enhancing the spatial resolution of DTS based on inverse filtering, performs inverse filtering processing through known or estimable probe pulse response, can significantly restore the temperature characteristics blurred by the system response, makes the signal peak more sharp and the edge more clear, and the regularization processing effectively avoids the amplification problem of high-frequency noise when noise exists, and is particularly suitable for the data post-processing work of optical and thermal detection systems with known or calibrated probe pulse response.

[0057] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application and not to limit it, although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that: it can still modify or equivalently replace the technical solutions of the present application, and these modifications or equivalent replacements also cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present application.

Claims

1. A method for enhancing DTS spatial resolution based on inverse filtering, characterized in that, Includes the following steps: S1. Acquire observation signals and perform preprocessing operations; S2. Establish an observation model, representing the preprocessed observation signal as the convolution of the real temperature distribution and the system detection pulse in the spatial domain, with noise superimposed. S3. Obtain or construct a discrete sequence representation of the probe pulse and perform normalization processing; S4. Based on the probe pulse of S2 and the preprocessed observation signal, a regularized inverse filter is constructed in the frequency domain; S5. Perform frequency domain transformation on the preprocessed observation signal, and multiply its spectrum with the regularized inverse filter in the frequency domain to obtain the recovered frequency domain signal; S6. Perform an inverse frequency domain transformation on the recovered frequency domain signal and take the real part to obtain the recovered time domain signal.

2. The method for enhancing DTS spatial resolution based on inverse filtering according to claim 1, characterized in that, S1 specifically refers to: The raw backscattered signal is acquired by a distributed temperature sensor (DTS), demodulated to obtain the raw temperature data sequence distributed along the fiber length, and then oversampled.

3. The method for enhancing DTS spatial resolution based on inverse filtering according to claim 2, characterized in that, In S2, the observation signal along the DTS system is represented as the convolution of the real temperature distribution and the system's probe pulses in the spatial domain, with noise added, as shown in the following expression: ； in, For observing signals, To detect pulses, For the true temperature distribution, To observe noise.

4. The method for enhancing DTS spatial resolution based on inverse filtering according to claim 3, characterized in that, In S3, the detection pulse is acquired. The specific steps are as follows: By experimentally calibrating, short pulse light emitted by DTS is directly received, its signal waveform is measured and stored, and this signal waveform is used as the probe pulse response.

5. The method for enhancing DTS spatial resolution based on inverse filtering according to claim 3, characterized in that, In S3, a probe pulse is constructed. The specific steps are as follows: The probe impulse response is fitted using a parameterized function, specifically a Gaussian window function. The window width and shape are controlled by the ProbeWidth and alpha parameters, and the probe impulse response is normalized to make... .

6. The method for enhancing DTS spatial resolution based on inverse filtering according to claim 3, characterized in that, It also includes the raw observation signal Perform resampling, baseline removal, windowing, or zero-filling operations.

7. The method for enhancing DTS spatial resolution based on inverse filtering according to claim 6, characterized in that, S4 specifically refers to: For the detection pulse and preprocessed observation signals Perform an FFT Fourier transform, setting the length of the FFT to N, to obtain and Construct a regularized inverse filter in the frequency domain Its expression is as follows: ； in, for The conjugate of complex numbers, Power spectrum, This is a regularization parameter used to suppress... Noise amplification at near-zero frequencies.

8. The method for enhancing DTS spatial resolution based on inverse filtering according to claim 7, characterized in that, In S5, the frequency domain signal is recovered. The calculation formula is as follows: 。 9. A method for enhancing DTS spatial resolution based on inverse filtering according to claim 8, characterized in that, In S6, for The time-domain recovered signal is obtained by performing an inverse Fourier transform. .

Images