CN111337160A

CN111337160A - Distributed optical fiber temperature measurement system based on double-end demodulation

Info

Publication number: CN111337160A
Application number: CN202010371547.8A
Authority: CN
Inventors: 卫欢; 蒋俊
Original assignee: Hangzhou Sensys Photoelectric Co ltd
Current assignee: Hangzhou Sensys Photoelectric Co ltd
Priority date: 2020-05-06
Filing date: 2020-05-06
Publication date: 2020-06-26

Abstract

The invention discloses a distributed optical fiber temperature measurement system based on double-end demodulation, which comprises an optical structure, a processing module and an optical fiber, wherein double-end demodulation is utilized to overcome the influence of wavelength-dependent loss in practical application of RDTS, meanwhile, the problem that two ends of a temperature curve signal-to-noise ratio obtained by double-end demodulation are much lower than a central part is effectively solved by utilizing an optical fiber partition-dependent noise reduction technology, and the temperature measurement precision of RDTS is further improved. The method is beneficial to improving the overall performance of the distributed optical fiber temperature measurement system in special environments such as fire detection, power cable detection, nuclear power station detection and the like.

Description

Distributed optical fiber temperature measurement system based on double-end demodulation

Technical Field

The invention relates to a distributed optical fiber sensing system, in particular to a distributed optical fiber temperature measuring system based on double-end demodulation.

Background

Raman Distributed Temperature Sensors (RDTS) have been studied for many years and, due to their well-known advantages over electrically equivalent devices, have been successfully used in many different fields, such as fire detection, power cable monitoring and leakage detection. Most RDTS systems are based on the Optical Time Domain Reflectometry (OTDR) principle for backscattered Raman scattered anti-Stokes (AS) and Stokes (S) optical signals. Where single-ended RDTS represents the most common solution in long distances, but it is inherently affected by AS and S Wavelength Dependent Loss (WDL); the slow variation of the WDL over time actually causes slow and undetectable distortion in the demodulation temperature profile. This is typically due to fiber aging, but is strongly enhanced in certain harsh application environments, such as nuclear power plant monitoring, where the presence of ionizing radiation can significantly increase fiber loss over time and result in significant WDL. Studies have also shown that degradation of the fibers in high temperature and humid environments can greatly increase fiber loss. Furthermore, in geothermal well applications, the WDL of the optical fiber typically varies over time due to the effects of high temperature and hydrogen concentration.

For use in a wider field, it is preferable to improve the measurement accuracy by post-processing without expensive light sources, filters, circuits, and the like. A band pass filter that cuts off unnecessary signal bands or an adaptive filter that extracts an effective signal band based on a designed noise model may be used as post-processing for noise reduction. But the frequency band of the noise overlaps with the frequency band of the signal component, that is, in any filtering process, the signal component is attenuated with the suppression of the noise, and even if the filtering process is performed in order to suppress the reduction of the measurement accuracy, the signal component itself is attenuated, and the noise of the anti-stokes optical signal becomes particularly large at the far end.

In practical RDTS systems both WDL and local losses must be effectively accounted for and eliminated, which can be achieved by using an alternate double-ended demodulation scheme (also known as loop demodulation). In this scheme, the AS and S optical signals are obtained in the forward and backward directions, and then averaged appropriately. But double-ended demodulation also has one such problem: the signal to noise ratio at the two ends of the demodulation temperature curve is much lower than in the central part.

Disclosure of Invention

The invention aims to solve the problems that: how to overcome the influence of wavelength-dependent loss of RDTS in practical application, solve the problem that two ends of a temperature curve signal-to-noise ratio obtained by double-end demodulation are much lower than a central part, and further improve the temperature measurement precision of RDTS so as to meet the application requirement of RDTS in more complex environments.

The problem proposed by the invention is solved as follows: a double-ended demodulation scheme based on fiber partition correlation noise reduction is provided.

The specific technical scheme is as follows:

the utility model provides a distributed optical fiber temperature measurement system based on bi-polar demodulation, includes optical structure, processing module and optic fibre, its characterized in that: the optical structure is configured to detect a first stokes light signal and a first anti-stokes light signal from backscattered light generated when light is input to the first end of the optical fiber, and to detect a second stokes light signal and a second anti-stokes light signal from backscattered light generated when light is input to the second end of the optical fiber; the processing module is configured to calculate, within a region including the first end of the optical fiber, a first region length based on a first correlation between the second stokes light signal and at least one of the first stokes light signal and the first anti-stokes light signal, calculate a second region length based on a second correlation between the second anti-stokes light signal and at least one of the first stokes light signal and the first anti-stokes light signal, smooth the second stokes light signal in the first region, smooth the second anti-stokes light signal in the second region, calculate a temperature of the sampling point using the smoothed second stokes light signal, the smoothed second anti-stokes light signal, the first stokes light signal, and the first anti-stokes light signal.

Further, pulsed light is alternately input to the first end and the second end of the optical fiber through the optical switch.

Further, the first region is elongated as the first correlation becomes smaller, and the second region is elongated as the second correlation becomes smaller.

Further, the length of the first region and the second region sets an upper limit.

Further, when the first correlation is equal to or greater than the threshold, the second stokes light signal is not smooth; when the second correlation is equal to or greater than another threshold, the second anti-stokes light signal is not smooth.

Further, a pearson product-moment correlation coefficient is applied to the first correlation and the second correlation.

Further, a spearman rank correlation coefficient is applied to the first correlation and the second correlation.

Further, the fiber has a constant temperature in the region around the sampling point; the region is larger than a zeroth-order width of a temperature distribution obtained when another constant temperature different from the constant temperature is given to a minimum heating length portion centered on a sampling point, and is smaller than a principal component width of the temperature distribution.

The invention has the beneficial technical effects that the influence of wavelength dependent loss in practical application of RDTS is overcome by utilizing double-end demodulation, and the problem that the two ends of the signal-to-noise ratio of a temperature curve obtained by double-end demodulation are much lower than the central part is effectively solved by utilizing the optical fiber partition dependent noise reduction technology, thereby further improving the temperature measurement precision of RDTS. The method is beneficial to improving the overall performance of the distributed optical fiber temperature measurement system in special environments such as fire detection, power cable detection, nuclear power station detection and the like.

Drawings

Fig. 1 shows the general structure of a distributed optical fiber temperature measurement system 1 based on double-end demodulation;

fig. 2 shows a case where a portion of the optical fiber 3 is immersed in hot water of 55 c when the room temperature is 24 c;

FIG. 3 shows the results obtained from FIG. 2 and equation (3);

FIG. 4 shows a typical example of a calculated impulse response;

fig. 5 to 7 show a comparison between an output waveform estimated from an impulse response with respect to each immersion length in hot water and an actually obtained output waveform;

fig. 8 shows an output waveform in the case where a center portion to which no high temperature is applied is provided between two high temperature application portions of 0.2m, and the width of the center portion is gradually changed;

fig. 9 shows a flow chart executed when the distributed optical fiber thermometry system 1 based on double-ended demodulation measures temperature;

FIG. 10 shows a comparison between Pearson product-moment correlation coefficients and spearman rank correlation coefficients for a set of experimental data;

fig. 11 shows another example of a flowchart executed when the temperature measured by the temperature detector 42 is corrected by the corrector 43;

FIG. 12 shows another flowchart executed when the temperature measured by the temperature detector 42 is corrected by the corrector 43;

FIG. 13 shows another example of a flow chart performed when the distributed fiber optic thermometry system 1 measures temperature;

fig. 14 shows equation (5).

Detailed Description

Fig. 1 shows the overall structure of a distributed fiber optic thermometry system 1 based on double-ended demodulation, comprising an optical structure 2, an optical fiber 3 and a processing module 4. The optical structure 2 comprises a laser 21, a beam splitter 22, an optical switch 23, a filter 24, a detector 25a and a detector 25 b. The processing module 4 comprises an instructor 41, a temperature detector 42 and a corrector 43. The laser 21 emits laser pulses of a predetermined wavelength range at predetermined time intervals in accordance with the instruction from the commander 41. The optical splitter 22 inputs the pulsed light emitted from the laser 21 into the optical switch 23, and the optical switch 23 alternately inputs the pulsed light into the first end and the second end of the optical fiber 3 in a predetermined period as instructed by the commander 41. In this embodiment, the length of the optical fiber 3 is L meters, the position of the first end is 0 meters, and the position of the second end is L meters.

When the pulsed light propagates through the optical fiber 3, raman scattering occurs, and forward scattered light advancing in the propagation direction and backward scattered light advancing in the return direction are generated, and the backward scattered light is input again to the optical splitter 22 through the optical switch 23. The backscattered light input into the spectroscope 22 is emitted to the filter 24, and long-wavelength light (stokes light) and short-wavelength light (anti-stokes light) are extracted from the backscattered light. The

detectors

25a and 25b convert the anti-stokes light and the stokes light into electrical signals, respectively, and send the electrical signals to the temperature detector 42 and the corrector 43. The corrector 43 corrects the anti-stokes light signal and the temperature measurer 42 measures the temperature using the stokes light signal and the anti-stokes light signal.

The temperature detector 42 measures the temperature of each position of the optical fiber 3 according to the following formula (1), and when the ratio of the two components is used, the difference between the two weak components is enhanced and comes to practical value. The gain and compensation depend on the design of the optical fiber 3 and are therefore pre-calibrated.

Temperature = gain/[ offset-2 x ln (anti-stokes/stokes intensity) ] (1)

When the incident position of the optical switch 23 to the optical fiber 3 is fixed at one of the first end or the second end, the temperature measurement can be achieved using the above formula (1). When the incident position is alternately switched to the first end and the second end with a constant period, the anti-stokes optical signal and the stokes optical signal are averaged (average value is calculated) with respect to the position of the optical fiber 3, and thus this method is called double-ended demodulation. The double-ended demodulation changes the above equation (1) to the following equation (2), i.e., averages the anti-stokes optical signal and the stokes optical signal at each position of the optical fiber 3 using the above equation (1).

Temperature = gain/[ offset-2 x ln (average anti-stokes light intensity/average stokes light intensity) ] (2)

If 01S denotes a stokes light signal in the case where pulsed light is input to the first end (0 to L meters), 01A denotes an anti-stokes light signal in the case where pulsed light is input to the first end (0 to L meters), 02S 'denotes a stokes light signal in the case where pulsed light is input to the second end (L to 0 meters), 02A' denotes an anti-stokes light signal in the case where pulsed light is input to the second end (L to 0 meters), and 02S 'and 02A' are inverted with respect to the elapsed time, 02S and 02A are obtained. The positions can be unified through inversion, and the influence of the optical fiber loss on the measured temperature can be eliminated by using double-end demodulation.

Next, the relationship between the length of the temperature measuring optical fiber and the demodulation temperature is explained, and FIG. 2 shows a case where a part of the optical fiber 3 is immersed in hot water of 55 ℃ when the room temperature is 24 ℃. When the length of immersion in hot water is extended from 0.5m to 10.5m, the maximum temperature becomes 55 ℃, which is the same as the temperature when the length of immersion in hot water is 2m or more. Therefore, in order to accurately measure the temperature, it is preferable to extend the length of the temperature measuring fiber. The sensitivity of the thermometric system at this time can be expressed by the following equation (3).

Sensitivity = (peak temperature of hot water soak position-room temperature measured with fiber before and after soak position)/applied temperature 100% (3)

Fig. 3 shows the results obtained from fig. 2 and equation (3), where the curves show a slight overshoot, because the impulse response of the system is not gaussian, but the waveform of the impulse response has a negative component close to the sinc function and a high order peak. The minimum length for which the sensitivity is 100% or considered 100% is called the minimum heating length.

As can be seen from fig. 2, the temperature profile of the fiber immersed in the hot water section can be seen as a single square wave into which the impulse response is convoluted, and fig. 4 shows a typical example of the calculated impulse response. When using backward raman scattered light for temperature measurement, the impulse response can be viewed as a wave form in which a window function is applied to the sinc function in order to smoothly attenuate the optical signal away from the center. When the impulse response is convolved into a temperature distribution along the fiber 3, an approximately accurate output prediction can be achieved.

Fig. 5 to 7 show that the output waveform can be predicted approximately accurately as found from the comparison between the estimated output waveform and the actually obtained output waveform with respect to the impulse response for each immersion length in the hot water. When the immersion length in hot water is 3.25m, the peaks are smoothed due to interference of convolutions of the impulse response.

Fig. 8 shows an output waveform in the case where a center portion to which no high temperature is applied is provided between two high temperature application portions of 0.2m, and the width of the center portion is gradually changed. The peak temperature is normalized to 1 and the reference temperature is normalized to 0. When the length of the central portion is 1.2 m to 1.6 m, two high temperature application portions can be considered. This is because of the interference caused by the amplification of the impulse response waveform. When the distance between the two high temperature application portions is the half width of the impulse response of fig. 8 or more, it can be considered that there are two high temperature application portions. Preferably, the distance is equal to half the value of the zeroth order width at which the gradient reverses or is greater to ensure that the two portions are significantly spaced from each other. As can be seen from fig. 8, when the lowest temperature of the central non-heating portion is equal to the reference temperature, the distance between the two high temperature-applied portions is greater than the main peak width and approximately equal to the main component width, i.e., the disturbance of the impulse response waveform can be ignored in fig. 8.

In order to accurately measure the temperature, it is preferable to focus on the temperature variation in a range in which the center position is the current processing position and the width is equal to or greater than the zero-order width and equal to or less than the width of the principal component. The pulsed light propagates while being gradually broadened and gradually attenuated due to influences such as broadening of wavelength, incident angle, scattering, and the like, and therefore, it is preferable to average values of the near end, the center, and the far end.

For measuring the temperature with higher accuracy, it is preferable that the plurality of portions are determined so that the range of difference between the convolution of the impulse response, which is calculated and stored at the center position of each region, and the output data shown in fig. 5 to 7 is not a problem, and is considered to be the same. The form of the impulse response wave slightly changes over time due to the degradation of the laser light or the like. Therefore, in a constant cycle, the impulse response is preferably calibrated at the same position as the initially obtained position for a more accurate measurement of the temperature.

For use in a wider field, it is preferable to improve the measurement accuracy by post-processing without expensive light sources, filters, circuits, and the like. A band pass filter that cuts off unnecessary signal bands or an adaptive filter that extracts an effective signal band based on a designed noise model may be used as post-processing for noise reduction. But the frequency band of the noise overlaps with the frequency band of the signal component, that is, in any filtering process, the signal component is attenuated with the suppression of the noise, and even if the filtering process is performed in order to suppress the reduction of the measurement accuracy, the signal component itself is attenuated, and the noise of the anti-stokes optical signal becomes particularly large at the far end. That is, it is preferable to focus on a method of reducing the noise of the far-end anti-stokes optical signal in order to reduce the noise. Therefore, the present embodiment uses double-ended demodulation and the relationship between the fluctuation of the stokes light signal and the fluctuation of the anti-stokes light signal.

When the temperature changes, the stokes light signal and the anti-stokes light signal also synchronously change along with the change of the temperature. That is, when the synchronization range is specified, it is considered that the temperatures of the other ranges do not change with respect to the next position of the optical fiber on the light source side, or only the temperature gradient changes smoothly. Thus, the minimum heating length described with reference to fig. 2 to 4 can be focused on. It is believed that the minimum heating length response is approximately the same within a given fiber segment as measured by temperature measurements taken by detecting backscattered raman light. When a portion of the optical fiber of the minimum heating length is heated beyond the region where the temperature remains constant, a waveform substantially identical to the impulse response of fig. 2 is obtained. As described above, as the range having an influence on the surroundings, it is preferable to focus on a range having a width equal to or larger than the zeroth-order width and equal to or smaller than the width of the principal component having an amplitude that is approximately attenuated to zero.

Fig. 9 shows a flow chart executed when the distributed optical fiber temperature measurement system 1 based on double-ended demodulation measures temperature. According to the measurement accuracy distribution of the optical fiber 3 in the double-ended demodulation method, the optical fiber 3 is equally divided into a first region, a central region, and a third region. In the first region, the measurement accuracy is low near 0 meter (first end); the second region is a central portion; in the third region, the measurement accuracy near L meters (second end) is low. The corrector 43 executes the flowchart of fig. 7 for three regions.

First, the corrector 43 determines whether the currently processed region of the optical fiber 3 is the first region of 1/3 equal to or smaller than the total length of the optical fiber 3 (step S1). When the determination in step S1 is yes, the corrector 43 calculates, for each sampling point, a large value of the correlation in a predetermined range having a width equal to or larger than the zeroth-order width of the response waveform of the minimum heating length and equal to or smaller than the principal component width of the response waveform of the minimum heating length.

Next, the corrector 43 targets 02S and 02A as objects to be processed, 01S and 01A as targets, and calculates large values of the four correlations, for example, the corrector 43 selects, as the large value α _02S, the smaller one of the correlation between 02S and 01S and the correlation between 02S and 01A (step S2). next, the corrector 43 selects, as the large value α _02A, the smaller one of the correlation between 02A and 01S and the correlation between 02A and 01A (step S3).

Next, the corrector 43 sets the average region so that the larger each correlation is, the smaller the average range of the center position of its current processing region, for example, the corrector 43 determines each integer of the samples to one side of each average range by rounding 1/α _02A and 1/α _02S to the nearest integer (step S4). next, after the corrector 43 averages 02S and 02A within the average range determined for each sampling point, the corrector 43 averages 01S and 02S at each sampling point, uses the average value as a stokes component, averages 02A and 01A at each sampling point, uses the average value as an anti-stokes component, and calculates the temperature (step S5). 01S and 01A are targets.

When the determination is "no" in step S1, the corrector 43 determines whether the currently processed region of the optical fiber is a third region which is equal to or greater than 2/3 of the total length of the optical fiber 3 (step S6). in the third region, 02S corresponds to 01S, 02A corresponds to 01A. therefore, when the determination is "yes" in step S6, the corrector 43 replaces 02S with 01S, 01S with 02S, 02A with 01A, and 01A with 02A. the corrector 43 performs the same procedure as in steps S2 to S4 (step S7). accordingly, the corrector 43 selects the smaller of the two correlations calculated with respect to 01S as a large value α _01S, and selects the smaller of the two correlations calculated with respect to 01A as a large value α _ 01A. the corrector 43 calculates the average value of the two correlations with 1/α _01A and 1/α _01S as a rounded to the nearest whole, and determines that the average value of the two correlation values of the samples calculated by rounding off 1/α _01A to 1/α _01S to the nearest to the average value of the sampling point S, and the average value of the sampling point S43, and determines whether the average value of the sampling point of each of the sampling point S02, and the sampling point of the sampling point S02, so that the average value of the sampling point S02, and the sampling point of the sampling point, so that the sampling point S01A 02S 01A, and the sampling point of the sampling point, and the sampling point of the sampling point.

The method carries out relative weighting on the values with small noise and high reliability, and eliminates the noise. Although one of the smaller values in step S2 and step S3 is used, a larger value may be used. Alternatively, an average of the values of step S2 and step S3 may be used. When the smaller one is used, the data other than the portion regarded as the temperature change is equalized. When a larger one is used, even data that is hidden in a slight change in noise is not deleted as much as possible.

There are many ways to determine the correlation. For example, Pearson product-moment correlation coefficients may be used. The pearson product-

moment correlation coefficients

02A and 01S are represented by the following formula (4).

Correlation coefficient α = (covariance of 02A, the range of which is the same as the specified range of 01S covariance)/(standard deviation of 01S of the same range)/(standard deviation of 02A of the same range) (4)

The Pearson product-moment correlation coefficient centered at the sampling point k of the optical fiber 3 is α [ k ], one array of 01S is 01S [ k ], one array of 02A is 02A [ k ], the number of samples in a specified range is n, the average 01S [ k ] in the specified range is 01save, the average 02A [ k ] in the specified range is 02Aave, and the above equation (4) can be expressed by equation (5), as shown in FIG. 14.

As another example, when the modified spearman rank correlation coefficient is used, the n numbers of 01S and 02A (n in the above equation (5)) in the specified range are sorted, and the pearson product-moment correlation coefficient is used for sorting. When there are two or more of the same ranks, a compensation formula is used. However, for the stokes component and the anti-stokes component, there are generally few cases where two or more have the same rank. Thus, the one that appears previously may be considered a higher level.

For example, in the portion shown in the impulse response of fig. 2, 3.6 m is set as a range satisfying the above condition. Fig. 10 shows a comparison between pearson product-moment correlation coefficients and spearman rank correlation coefficients for a set of experimental data. In general, a Pearson product-moment correlation coefficient of 1 or-1 indicates complete correlation. The absolute value of the pearson product-moment correlation coefficient is 0.4 or more than 0.4 and less than 0.7, indicating that the correlation is high. The absolute value of the pearson product-moment correlation coefficient is 0.2 or more and less than 0.4, indicating that the correlation is low. A pearson product-moment correlation coefficient with an absolute value less than 0.2 indicates no correlation. However, although the gradient of spearman is much greater than that of pearson, an approximate ratio of 1:1 is achieved in a range less than 0.2 indicating no correlation and in a range 0.3 or more indicating low correlation.

The inverse of the large value of the correlation coefficient in fig. 9 is an index of the average range, however, the inverse is not always required to be used. The larger the correlation coefficient, the narrower the average range is. And the smaller the correlation coefficient, the wider the average range is.

Fig. 11 shows another example of a flowchart executed when the corrector 43 corrects the temperature measured by the temperature detector 42, as shown in fig. 11, the corrector 43 executes steps S11 to S13 which are the same as steps S1 to S3 of fig. 9 next, the corrector 43 executes steps S14 to S18 for the correlation large value α _02S and the correlation large value α _02A detailed description will be given of steps S14 to S18, the correlation large value α _02S and the correlation large value α _02A are shortened to "α _".

The corrector 43 determines whether the correlation large value α _ is equal to or smaller than a first threshold value (0.2 or smaller) (step S14). when it is determined "yes" in step S14, the temperature measurer 42 expands the average range of the sample points currently being processed to an upper limit value (step S15). for example, the upper limit value may be 6 samples as one side or 11 samples as a total number. when it is determined "no" in step S14, the corrector 43 determines whether the correlation coefficient α _ is equal to or smaller than a second threshold value (e.g., 0.55) which is larger than the first threshold value of step S14 (step S16). when it is determined "yes" in step S16, the corrector 43 determines the average range as "1" (step S17). when it is determined "no" in step S16, the temperature measurer 42 calculates an integer by rounding 1/α _ and takes the nearest integer as the average range of the average coefficient, and executes the average coefficient of the average range of step S9619 or the average range of the same step S3619, when it is determined "no" S9619 ", and the average range of step S3619 is equal to or the average coefficient of step S9619, S9685, which is executed after this step S3619.

When it is determined as "no" in step S11, steps S20 to S22, which are the same as steps S6 to S8 of fig. 9, are performed, that is, when it is determined as "yes" in step S20, an average range with respect to the correlation large value α _01S and the correlation large value α _01A is determined using the first threshold value and the second threshold value, and the temperature is calculated, when it is determined as "no" in step S20, the temperature is calculated using a conventional double-ended method.

Fig. 12 shows another flowchart executed when the temperature measured by the temperature detector 42 is corrected by the corrector 43. Differences between fig. 12 and fig. 11 will be described. First, the optical fiber 3 is divided into a first portion near 0 m (first end) where the measurement accuracy is low, a central third portion where the measurement accuracy is low, a fifth portion near L m (second end), a second portion between the first portion and the third portion, and a fourth portion between the third portion and the fifth portion, on the average, based on the measurement accuracy distribution in the longitudinal direction of the optical fiber 3. The corrector 43 executes the flowchart of fig. 12 for five sections.

The corrector 43 determines whether the currently processed portion of the optical fiber 3 is the first portion of 1/5 equal to or less than the total length of the optical fiber 30 (step S31). When the determination in step S31 is yes, the corrector 43 performs steps S32 to S39, which are the same as steps S12 to S19.

When the determination in step S31 is "no", the corrector 43 determines whether the currently processed portion of the optical fiber 3 is equal to or greater than 4/5 of the total length of the optical fiber 3 (step S40). When it is determined as "yes" in step S40, after replacing 02S with 01S, 01S with 02S, 02A with 01A, and 01A with 02A, the same processing as in steps S32 to S39 is performed (step S41). However, as for the upper limit values of the correlation coefficient and the average range, other values may be used as the threshold values.

When the determination in step S40 is "no", the corrector 43 determines whether the currently processed section is equal to or smaller than 2/5 of the total length of the optical fiber 3 or equal to or larger than 3/5 of the total length of the optical fiber 3 (step S42). When it is determined as "yes" in step S42 and the currently processed section is equal to or smaller than 2/5 of the total length of the optical fiber 3, the same processing as steps S32 to S39 is performed. However, the threshold values of the correlation coefficient and the upper limit value with respect to the average range are different from the threshold values of step S32 to step S39. When it is determined as "yes" in step S42 and the currently focused section is equal to or larger than 3/5 of the total length of the optical fiber 3, the same processing as step S41 is performed. However, the threshold value and the upper limit of the average range regarding the correlation coefficient are different from the threshold value of step S41. Therefore, the minimum heating length according to the sectional change can be corrected. When the determination in step S42 is no, the same step S44 as step S22 is performed.

In fig. 9, 11 and 12, the average range is equal to or less than the principal component width of the minimum heating length. This is because when the averaging range exceeds the principal component width of the minimum heating length, the probability that crosstalk adjacent to another signal affects the averaged signal becomes higher. For example, when the obtained correlation coefficient is-1, it is determined that complete correlation has occurred. However, in the present embodiment, the correlation is regarded as noise. The reason is as follows: the Stokes and anti-Stokes curves are convex upward as the temperature increases. The Stokes and anti-Stokes curves are convex downward as the temperature decreases. In this case, the anti-stokes curve never has an upside-down shape of the stokes curve. The anti-stokes curve may have a shape in which the stokes curve is inverted upside down only when noise occurs or a connector with poor connection or poor fusion, or optical fibers having a large difference in refractive index are fused to each other.

In fig. 9, 11 and 12, the optical fiber is divided into three or five parts. However, the optical fiber may be divided into two portions at one center, or four portions obtained by dividing the two portions, or eight portions obtained by dividing the four portions. In these cases, only the central portion where the temperature is calculated without processing 01S, 01A, 02S, and 02A is deleted.

In an embodiment, double ended demodulation is used. When pulsed light is input to the second end, the 02S is averaged in accordance with at least one of the correlation coefficients of 02S and 01S and the correlation coefficients of 02S and 01A within a predetermined range of sampling points including a partial range of the first end side. In this case, the noise of 02S is reduced. Further, the 02A is averaged within an averaging range according to at least one of the correlation coefficients of 02A and 01S and the correlation coefficients of 02A and 01A, in which case the noise of 02A is reduced. When the average 02S and the average 02A are used, the measured temperature can be corrected. When pulsed light is input to the first end, 01S is averaged within an average range within a predetermined range including sampling points of a partial range of the second end side according to at least one of correlation coefficients of 01S and 02S and correlation coefficients of 01S and 02A. In this case, the noise of 01S is reduced. Further, 01A is averaged within an averaging range according to at least one of the correlation coefficients of 01A and 02S and the correlation coefficients of 01A and 02A. In this case, the noise of 01A is reduced. When the average 01S and the average 01A are used, the measured temperature can be corrected. When averaging 02A, it is preferred that 01S and 01A are the components just before 02A, relative to the switching of the optical switch 23. When averaging 01A, it is preferable that 02S and 02A are the components before 01A in terms of switching of the optical switch 23.

In the present embodiment, the average value of the Stokes optical signal and the anti-Stokes optical signal is calculated within the average range. However, when the variability of the data within the average range is suppressed, it is not necessary to calculate the average value. Thus, another average may be used, for example an arithmetic average, a geometric average or a harmonic average taking into account the w-weights. When calculating the average values of 01S and 02S and the average values of 01A and 02A, an arithmetic average value considering the weights may be used.

The averaging may be performed after smoothing the Stokes light signal and the anti-Stokes light signal within a smoothing range according to a correlation large value of the Stokes light signal and the anti-Stokes light signal within a predetermined range including predetermined sampling points. In this case, the measured temperature can be corrected. For example, when the correlation is small, the noise becomes large, and therefore it is preferable to smooth the two optical signals. In this case, noise can be reduced. Preferably, the smaller the correlation, the longer the smoothing range. In this case, the noise is reduced more. Preferably, the upper limit of the smoothing range is determined. In this case, redundancy of the smoothing range is suppressed, suppressing a decrease in temperature measurement accuracy. When the correlation is small, the temperature variation around the sampling point is small, and therefore, both the smoothing processing and the reduction in the accuracy of measuring the temperature are suppressed. On the other hand, when the correlation is large, the smoothing range is shortened, or no correction is performed. When the correlation is large, the temperature around the sampling point changes greatly, so the influence of noise is small, and the accuracy of measuring the temperature can be maintained. However, in the processes of fig. 9, 11, and 12, when the correlation coefficient between anti-stokes optical signals is high, it is preferable not to perform smoothing. This is because the anti-stokes optical signal is small and detection of the anti-stokes component becomes difficult due to the smoothing process. This process is explained based on fig. 13.

Fig. 13 shows another example of a flowchart executed when the distributed optical fiber thermometry system 1 measures temperature. In step S3 of fig. 9, the corrector 43 determines whether the correlation coefficient between 01A and 02A is equal to or smaller than a threshold (S51). When the determination in step S51 is "no", the correlation between 01A and 02A is high. Thus, the flowchart terminates, the smoothing of the anti-stokes light signal is not performed and the accuracy of fig. 9, 11 and 12 is maintained.

When the determination in step S51 is yes, the corrector 43 calculates a correlation large value α between the Stokes optical signal and the anti-Stokes optical signal within a predetermined range including the sampling points, the width of which is equal to or larger than the zeroth-order width of the minimum heating length response waveform and equal to or smaller than the principal component width of the minimum heating length response waveform with respect to each sampling point (step S52).

Next, the corrector 43 determines whether the correlation coefficient α is equal to or smaller than a threshold value (for example, 0.2 or smaller) (step S53) — when it is determined "yes" in step S53, the corrector 43 expands the smoothing range of the currently processed sampling point to an upper limit (step S54) — the upper limit may be 11 samples, one side of which may be 6 samples with respect to the sample point — when it is determined "no" in step S53, the corrector 43 calculates an integer by rounding 1/α to the nearest integer, and determines the number of samples on the smoothing range side to the calculated integer (step S55) — after step S54 or step S55 is performed, the corrector 43 smoothes the Stokes optical signal and the anti-Stokes optical signal respectively within the determined smoothing range (step S56) — specifically, both 01A and 01S are within the smoothing range determined according to the large value of the correlation coefficient of 01A and 01S, both 02 and 02 are within the large value of the correlation coefficient of 02 and the anti-Stokes optical signal α, when it is determined as an average value of the correlation coefficient of 1, or when it is not necessarily equal to the average value of fig. 1, or fig. 1, and when it is not necessarily performed as an average value of the graph 13.

The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention in any way, and any simple modification, equivalent change and modification made by the technical essence of the present invention to the above embodiment by those skilled in the art can be made within the technical scope of the present invention.

Claims

1. The utility model provides a distributed optical fiber temperature measurement system based on bi-polar demodulation, includes optical structure, processing module and optic fibre, its characterized in that: the optical structure is configured to detect a first stokes light signal and a first anti-stokes light signal from backscattered light generated when light is input to the first end of the optical fiber, and to detect a second stokes light signal and a second anti-stokes light signal from backscattered light generated when light is input to the second end of the optical fiber; the processing module is configured to calculate, within a region including the first end of the optical fiber, a first region length based on a first correlation between the second stokes light signal and at least one of the first stokes light signal and the first anti-stokes light signal, calculate a second region length based on a second correlation between the second anti-stokes light signal and at least one of the first stokes light signal and the first anti-stokes light signal, smooth the second stokes light signal in the first region, smooth the second anti-stokes light signal in the second region, calculate a temperature of the sampling point using the smoothed second stokes light signal, the smoothed second anti-stokes light signal, the first stokes light signal, and the first anti-stokes light signal.

2. The distributed optical fiber temperature measurement system based on double-end demodulation as claimed in claim 1, wherein: pulsed light is alternately input to the first and second ends of the optical fiber through the optical switch.

3. The distributed optical fiber temperature measurement system based on double-end demodulation as claimed in claim 1, wherein: the first region is elongated as the first correlation becomes smaller, and the second region is elongated as the second correlation becomes smaller.

4. The distributed optical fiber temperature measurement system based on double-end demodulation as claimed in claim 3, wherein: the length of the first region and the second region sets an upper limit.

5. The distributed optical fiber temperature measurement system based on double-end demodulation as claimed in claim 1, wherein: when the first correlation is equal to or greater than the threshold, the second stokes light signal is not smooth; when the second correlation is equal to or greater than another threshold, the second anti-stokes light signal is not smooth.

6. The distributed optical fiber temperature measurement system based on double-end demodulation as claimed in claim 1, wherein: the pearson product-moment correlation coefficient is applied to the first correlation and the second correlation.

7. The distributed optical fiber temperature measurement system based on double-end demodulation as claimed in claim 1, wherein: a spearman rank correlation coefficient is applied to the first correlation and the second correlation.

8. The distributed optical fiber temperature measurement system based on double-end demodulation as claimed in claim 1, wherein: the fiber has a constant temperature in a region around the sampling point; the region is larger than a zeroth-order width of a temperature distribution obtained when another constant temperature different from the constant temperature is given to a minimum heating length portion centered on a sampling point, and is smaller than a principal component width of the temperature distribution.