JPH06201488A - Temperature distribution detector - Google Patents

Abstract

PURPOSE:To improve total accuracy without narrowing light pulse width by preventing solution from diverging due to influence of noise during inverse transformation. CONSTITUTION:Each time Fresnel reflection light is emitted from an optical fiber 1, the light is converted into photocurrent while being led by a photo diode 11 via an optical filter 8 and a light switch 9, and an average output is supplied through a photocurrent amplifying circuit 12 and a digital averager circuit 13 to a control storage calculation circuit 14 and stored in it. The control storage calculation circuit 14 stores a number of times wherein waveforms become linear as an optimum number of repetitions while repeating deconvolution with respect to a stored average output. When temperature is measured, deconvolution is repeated by the number of repetitions with respect to waveform data of two kinds of Raman scattering light to calculate a temperature distribution along the optical fiber.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a temperature distribution detecting device utilizing backscattered light in an optical fiber.

[0002]

2. Description of the Related Art In a sensor utilizing backscattered light from an optical fiber, the scattered light in the optical fiber 100 is excited by a laser pulse incident on the optical fiber 100 as shown in FIG. Along the optical fiber 100, the backscattered light propagating to the incident side of the excitation laser pulse in the rear of the optical fiber 100 including the temperature information of the scattered place in the state of polarization is detected as a time series signal. Measure the one-dimensional distribution of temperature.

In this case, scattered light generated when a laser pulse is incident on the optical fiber 100 includes Rayleigh scattered light due to density fluctuation, Brillouin scattered light due to propagating fluctuation, and Raman scattered light due to vibration and rotation of molecules. .

Of these, the Brillouin scattered light and the Raman scattered light are inelastic scattered light, and have different spectra from the excitation light. The temperature information is contained in all three scattered lights, but Raman scattered light has the highest sensitivity to temperature, and its intensity changes depending on the temperature.

When the temperature is measured using Raman scattered light, the Stokes Raman scattered light whose wavelength shifts to the longer wavelength side than the excitation light and the anti-Stokes Raman scattered light whose wavelength shifts to the shorter wavelength side are filtered by a filter. It is selected and the temperature distribution is calculated from the value based on the ratio of the two scattered lights. Further, even if the ratio is not used, it is necessary to select one of the scattered lights with a filter.

[0006]

By the way, in such a distributed sensor, not only the accuracy of the physical quantity to be measured but also its position resolution is an important quantity.

In this case, the resolution is generally determined by the incident pulse width, and when the backscattered signal has a discrete value system, it is further influenced by the sampling frequency.

However, in the method of increasing the sampling period, if the sampling period is sufficiently shorter than the half value width, the backscattered light measured at a certain time is the total of the scattered light over the half value width of the incident pulse. Therefore, the measured temperature is an average temperature over the half-value width, and thus the position resolution cannot be made higher than the width of the incident pulse.

Therefore, a method of narrowing the width of the incident pulse is effective for increasing the position resolution. As a method of realizing this method, a method of extremely narrowing the width of the incident pulse itself and a finite method are available. A method is conceivable in which the response to a light pulse having a width is regarded as one conversion process and inverse conversion is performed to obtain an impulse response.

However, a method of narrowing the width of the incident pulse itself is currently being developed because the width of the optical pulse cannot be narrowed beyond the pulse width determined by the rising characteristic and the falling characteristic of the light emitting system. It is difficult to make the width of the incident pulse narrower than the performance of the device, which is unsuitable as a method for improving the position resolution.

On the other hand, the method for obtaining the impulse response is considered to be effective as a method for improving the position resolution.

Therefore, the method of obtaining the impulse response will be examined in detail here.

First, since the transformation used in the method for obtaining the impulse response is a convolution process within the range of linear approximation, assuming that the impulse response is h (t) and the input light pulse is P (t), backscattering is performed. The optical signal g (t) can be expressed by the following equation.

[0014]

[Equation 1] Then, the impulse response h (t) can be obtained by measuring the input optical pulse P (t) as an input signal in advance and performing deconvolution on the measurement result.

However, in practice, when this inverse transformation process is applied to the actually measured data, it is often impossible to obtain a correct impulse response because of the noise component contained in the measurement signal. The reason for this is easy to understand in the following way.

First, when the noise component N (ω) is added to the backscattered light by Fourier transforming the equation (1), G (ω) + N (ω) = H (ω) .P (ω) (2) ), And both sides of this equation (2) are P (
When divided by ω), H (ω) = G (ω) / P (ω) + N (ω) / P (ω) (3)

Here, the problem is the part of the second item on the right side, where the spectrum of the input optical pulse component P (ω) is finite, whereas the spectrum of the noise component N (ω) is considerably large. , The division result diverges in a high frequency region (a region where the angular velocity ω is large).

For this reason, there is a problem that the overall position measurement accuracy is deteriorated by this divergence.

In view of the above situation, the present invention prevents the divergence of the solution due to the influence of noise in the inverse conversion process, and thereby the optimum approximation which can most improve the overall accuracy without narrowing the width of the optical pulse. It is possible to improve the position resolution without decreasing the S / N ratio that occurs when the optical pulse width is narrowed, the measurement time accompanying this, and the difficulty of the semiconductor laser drive circuit. It is an object of the present invention to provide a temperature distribution detecting device that can easily detect an abnormal heating portion such as a hot spot.

[0020]

In order to achieve the above-mentioned object, the present invention provides a light source for generating an optical pulse and two kinds of Raman scattered light from the backscattered light of the light pulse incident on the optical fiber. An optical system for selecting, a photoelectric conversion unit that converts and amplifies the Raman scattered light into an electric signal, a digital averager unit that converts the output signal of the photoelectric conversion unit into a digital signal and performs averaging, and the digital average. In a temperature distribution detecting device having a calculation unit for calculating the temperature distribution along the optical fiber based on the processing result of the optical fiber unit, deconvolution processing is performed on the waveform data of the Fresnel reflected light from the end face of the optical fiber. When the temperature is measured by calculating the number of repetitions until the waveform of the waveform data has a predetermined shape, the waveform data of two types of Raman scattered light is compared. It is characterized by calculating the temperature distribution along the optical fiber by repeating the repetition number of times deconvolution process.

[0021]

In the above structure, the waveform data of the Fresnel reflected light from the end face of the optical fiber is subjected to deconvolution processing to obtain the number of repetitions until the waveform of the waveform data has a predetermined shape, and temperature measurement is performed. When doing 2
By calculating the temperature distribution along the optical fiber by repeating the deconvolution process for the number of times of the above-mentioned Raman scattered light waveform data, the solution will diverge due to the noise in the inverse conversion process. To provide the optimum approximation that can most improve the overall accuracy without narrowing the width of the optical pulse, and reduce the S / N ratio that occurs when the width of the optical pulse is narrowed, and the accompanying measurement. The position resolution is improved and the abnormal heating portion such as a hot spot is easily detected without increasing the time and making the driving circuit of the semiconductor laser difficult.

[0022]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS First, the measurement principle of the temperature distribution detecting device according to the present invention will be described prior to the detailed description of the temperature distribution detecting device according to the present invention.

Now, in an optical fiber for temperature measurement, the backscattering coefficient per unit length is S (z), the attenuation rate received by the incident light pulse is Rf (z), and the attenuation received by the backscattered light up to the incident end. When the rate is Rb (z) and the group velocity of the optical pulse in the optical fiber is Vg, the power h (t) of the impulse response of the backscattered light obtained by making the optical pulse incident on this optical fiber is expressed by the following equation. be able to.

H (t) = Rf (z) .S (z) .Rb (z) .Vg / 2 (4) And the convolution with the incident optical pulse P is the anti-Stokes Raman. Since the respective coefficients S of the scattered light and the Stokes Raman scattered light are used, the inverse conversion is performed on the respective backscattered signals of the anti-Stokes Raman scattered light and the Stokes Raman scattered light, and then the attenuation effect of the optical fiber is compensated. Therefore, the procedure is such that each of these backscattered signals is divided and the temperature distribution is calculated.

However, in an actual device, after the backscattered light is converted into an electric signal by a photodiode or the like,
Since it is A / D converted and treated as a discrete value digital signal, the convolution expressed in the integral form has a matrix form as shown in FIG.

The optical pulse input is an n × n matrix, which has only a diagonal neighborhood where the position of the incident pulse moves with the passage of time. Then, for the elements in each row, the ones to which the actually measured values of the optical pulse waveform have moved are substituted.

After that, in order to obtain the impulse response h,
Although the inverse matrix is multiplied from both sides, in order to obtain the impulse response h while monitoring the influence of noise, the method of obtaining the impulse response h while converging the solution by the iterative method (Jacobi Gauss-Seidel method) is suitable.

In this iterative method, if the number of iterations is k,
The k + 1-th i-th solution hi is given by the following equation.

[0029]

[Equation 2] On the other hand, as shown in FIG. 3, a signal processing device 101, a thermostatic chamber 102 at 80 ° C., and an optical fiber 103 of 1 km are prepared, and a predetermined portion of the optical fiber 103 is set to 2 m, 5 m, 1
The signal processing device 1 was put into the constant temperature bath 102 for each length of 0 m, 20 m, and 50 m, while keeping the temperature at 80 ° C.
When the anti-Stokes Raman scattered light and the Stokes Raman scattered light are measured by 01 by the OTDR method and the ratio of these is obtained, the waveform shown in FIG. 4 is obtained.

In this case, the full width at half maximum of the optical pulse used for measurement corresponds to 10 m in length of the optical fiber per 100 nsec.
The calculation results of the part put in 02 should all have the same value, but as is clear from FIG. 4, due to convolution, in practice, sufficient amplitude is obtained in the part of 5 m or 2 m of 10 m or less. Can't get

Therefore, in order to eliminate the influence of such convolution, deconvolution is performed by applying the above-mentioned iterative method to the data obtained by the above-mentioned measurement operation to perform deconvolution. When the waveform of the data before the execution has the shape shown in FIG. 5A, when the data is subjected to deconvolution with 5 iterations, the waveform shown in FIG. When the deconvolution is performed 10 times, the waveform shown in FIG. 5C is obtained.

As is apparent from this figure, it is found that the deconvolution improves the location where the amplitude was insufficient and improves the rising edge.

However, it can be seen that the influence of noise becomes more remarkable as the number of iterations increases along with such improvement.

Therefore, as a total accuracy, an error in which the amplitude does not reach a predetermined value due to convolution is a systematic error, and a variation due to the influence of noise is a random error, and the number of times of deconvolution is repeated, the systematic error, and the random error. In order to investigate the relationship of the above, the above-described deconvolution processing was performed on the portion having a heating length of 5 m, using the data immediately before heating as random noise. At this time, the table shown in FIG. 6 was obtained.

As is apparent from FIG. 6, as the number of iterations is increased, the random error increases at a constant rate, but the systematic error initially decreases at a constant rate and increases at a certain number of times. After that, it increases at a constant rate.

That is, there is the number of iterations that minimizes the total error obtained by adding the systematic error and the random error.

The optimum number of iterations is calculated by calculating the total error for each iterative calculation when the temperature distribution is accurately known in advance as in the inverse conversion reference temperature part given here. The number of times it becomes a value can be obtained.

However, in such an inverse transform method, random noise increases with the number of calculations. Therefore, when there is almost no systematic error, for example, only random noise increases in a portion where the temperature distribution is almost uniform. Become.

That is, this inverse conversion method is advantageous in the case where the temperature change is spatially intense, particularly when the temperature distribution is to be measured when the temperature is locally high or low.

Therefore, as a plant system to be used for such an inverse conversion method, there is a system in which a system abnormality is manifested as a local temperature rise. When rising, this can be detected efficiently by using the optimal number of iterations.

In the present invention, in order to obtain the optimum number of iterations in the inverse conversion process, Fresnel reflected light from the incident surface or the terminal surface of the optical fiber is taken in as shown in FIG. The deconvolution process for light is repeated, and the optimum calculation process for obtaining the optimum approximate solution is selected by calculating the number of times the pulse waveform becomes linear as shown in FIG. 7B, and this optimum calculation process is selected. It is applied over the entire area of the optical fiber to perform the inverse transformation,
As a result, when performing temperature distribution measurement using an optical fiber, the position resolution is greatly improved while suppressing the influence of noise to a minimum.

FIG. 1 is a block diagram showing an embodiment of a temperature distribution detecting device according to the present invention.

The temperature distribution detecting device shown in this drawing comprises an optical fiber 1 and a signal processing device 2, and generates a very narrow optical pulse by the signal processing device 2 at initialization and periodically. Then, the light is incident on the optical fiber 1, the Fresnel reflected light from the end surface of the optical fiber 1 is taken in, and the signal corresponding to the Fresnel reflected light is obtained by performing deconvolution by the iterative method based on the Fresnel reflected light. Find the optimal number of iterations for which the waveform of becomes linear. When performing normal measurement, the signal processing device 2
Generates a very narrow light pulse, makes it enter the optical fiber 1, takes in this backscattered light, and based on the anti-Stokes Raman scattered light and the Stokes Raman scattered light, the optimum number of repetitions. The deconvolution is performed based on the above to measure the temperature distribution of each part where the optical fiber 1 is laid.

The optical fiber 1 is laid so as to pass through each portion whose temperature is to be measured. When an optical pulse is incident from the incident end, the optical fiber 1 is taken in and the backscattered light corresponding to the temperature of each portion is taken. And returning the Fresnel reflected light at the end surface to the incident end.

The signal processing device 2 also includes a timing control circuit 4, a laser drive circuit 5, a laser light source 6, a coupler 7, an optical filter 8, an optical switch 9, a bias control circuit 10, and a photodiode. 11, a photocurrent amplifier circuit 12, a digital averager circuit 13, and a control memory arithmetic circuit 14, which generate an extremely narrow optical pulse at initialization and periodically to generate an optical pulse. The Fresnel reflected light from the end face of the optical fiber 1 is incident on the fiber 1 and the convolution of the Fresnel reflected light is performed by an iterative method based on the Fresnel reflected light so that the signal waveform corresponding to the Fresnel reflected light is linear. Find the optimal number of iterations. Then, during normal measurement, a light pulse having a very narrow width is generated and is made incident on the optical fiber 1, and this backscattered light is taken in to obtain the anti-Stokes Raman scattered light and the Stokes Raman scattered light. Based on the light, deconvolution is performed based on the optimum number of times of repetition, and the temperature distribution of each part where the optical fiber 1 is laid is measured.

The timing control circuit 4 is a portion for generating a light emission timing signal, a synchronous addition timing signal and the like while exchanging signals with the control storage arithmetic circuit 14, and a laser emission timing signal obtained by the generation operation. It is supplied to the drive circuit 5 to control it and
Synchronous addition timing signal to digital averager circuit 1
3 to control this.

The laser drive circuit 5 is a portion for generating a very narrow drive signal based on the light emission timing signal output from the timing control circuit 4, and supplies this drive signal to the laser light source 6. This is made to emit light.

The laser light source 6 is a portion that emits light to generate an optical pulse having a preset wavelength while the laser drive circuit 5 outputs the drive signal. The generated optical pulse is transmitted to the coupler 7. Supply.

The coupler 7 guides the optical pulse output from the laser light source 6 to the incident end of the optical fiber 1,
This is a part that guides the backscattered light emitted from this incident end and the Fresnel reflected light at the end face to the optical filter 8, which takes in the optical pulse output from the laser light source 6 and uses it. Is introduced into the optical fiber 1 and is introduced into the optical fiber 1, and the light pulse is generated when the optical pulse propagates in the optical fiber 1, and the backscattered light emitted from the incident end or the Fresnel reflected light generated at the termination surface is generated. Is taken in and guided to the optical filter 8.

The optical filter 8 takes in the backscattered light and the Fresnel reflected light emitted from the coupler 7, and discriminates the wavelengths of the backscattered light and the Fresnel reflected light into anti-Stokes Raman scattered light and Stokes Raman scattered light. The anti-Stokes Raman scattered light and the Stokes Raman scattered light obtained by this separating operation are guided to the optical switch 9.

The optical switch 9 corresponds to the control memory arithmetic circuit 1
4 is a part that selects either the anti-Stokes Raman scattered light or the Stokes Raman scattered light emitted from the optical filter 8 based on the selection signal from the anti-Stokes Raman scattered light obtained by this selection processing. Alternatively, either one of the Stokes Raman scattered light is guided to the photodiode 11.

The bias control circuit 10 is a part for generating a bias voltage based on the bias command signal output from the control storage arithmetic circuit 14, and when the bias command signal is a high amplification factor command, it is previously set. A preset high voltage is generated and supplied to the photodiode 11, and when the bias command signal is a low amplification factor command, a preset low voltage is generated and supplied to the photodiode. Supply to 11.

The photodiode 11 is an avalanche diode made of silicon or the like, and when the bias voltage output from the bias control circuit 10 is a high voltage, the anti-Stokes Raman scattered light emitted from the optical switch 9 is emitted. Alternatively, the Stokes Raman scattered light is converted into a photocurrent with a high amplification factor and supplied to the photocurrent amplification circuit 12, and when the bias voltage output from the bias control circuit 10 is a low voltage, the optical switch 9 Anti-Stokes Raman scattered light or Stokes Raman scattered light emitted from the low amplification factor (for example,
It is converted into a photocurrent with an amplification factor “1”) and supplied to the photocurrent amplifier circuit 12.

The photocurrent amplifier circuit 12 amplifies the photocurrent of the photodiode 11 with an amplification factor according to the amplification factor designating signal output from the control storage arithmetic circuit 14 to generate a scattered signal or a Fresnel reflection signal. This is a part, and supplies the scattered signal and the Fresnel reflection signal obtained by the amplification operation to the digital averager circuit 13.

The digital averager circuit 13 takes in the scattered signal and Fresnel reflection signal outputted from the photocurrent amplifier circuit 12 based on the synchronous addition timing signal outputted from the timing control circuit 4, and synchronously adds them. Average output of anti-Stokes Raman scattering signal or average output of Stokes Raman scattering signal,
An average output of the Fresnel reflection signal is generated and supplied to the control storage arithmetic circuit 14.

The control storage arithmetic circuit 14 is a part which controls the entire temperature distribution detecting device, and controls the timing control circuit 4 and the bias control circuit 10 to adjust the light emission timing and the synchronous addition timing, and the processing described above. The process of taking in the average output of the anti-Stokes Raman scattering signal, the average output of the Stokes Raman scattering signal, or the average output of the Fresnel reflected light and storing this, which is output at the time of initialization or is preset. The optimal number of iterations (optimal number of iterations) by performing the above-mentioned deconvolution on the stored average output of Fresnel reflected light for each period
When the processing for obtaining the value and the measurement command are set, the deconvolution described above is performed for the optimum number of iterations for the stored average output of the anti-Stokes Raman scattering signal and the average output of the Stokes Raman scattering signal. A process of generating a signal, a process of outputting a temperature signal obtained by this process to another device through a temperature distribution output interface, and the like are performed.

Next, with reference to the block diagram shown in FIG. 1, the optimum number of iterations detecting operation and the temperature measuring operation of this embodiment will be sequentially described.

<< Optimum Repeat Count Detection Operation >> First, when the power of this temperature distribution detecting device is turned on to initialize the circuit, or whenever a preset calibration timing is reached.
The control storage arithmetic circuit 14 of the signal processing device controls the bias control circuit 10 to generate a low voltage and set the amplification factor of the photodiode 11 to a low value (for example, amplification factor “1”), and then the timing control circuit. 4 is driven to drive the laser drive circuit 5, the laser light source 6 connected to the laser drive circuit 5 is caused to continuously generate a very narrow optical pulse at a predetermined timing, and this is applied to the coupler 7. To supply.

Then, each time the optical pulse is guided to the optical fiber 1 by the coupler 7 and Fresnel reflected light is emitted from the optical fiber 1, this is sent to the photodiode 11 via the optical filter 8 and the optical switch 9. While being guided and converted into photocurrent, the photocurrent amplifier circuit 12
After being amplified by and converted into a Fresnel reflection signal, the digital averager circuit 13 synchronously adds the designated number of times to generate an average output of the Fresnel reflection signal, which is supplied to and stored in the control storage arithmetic circuit 14. .

When this processing is completed, the control storage operation circuit 14 repeats the deconvolution processing with respect to the stored average output of the Fresnel reflection signal and determines the number of times the waveform becomes linear to the optimum number of iterations. Memorize as.

<< Temperature Measuring Operation >> Then, when the above-described optimum iteration number detecting operation is completed, the control storage arithmetic circuit 14 controls the optical switch 9 to select the anti-Stokes Raman scattered light, and then the timing control circuit 4 is operated. The laser drive circuit 5 is controlled and driven, the laser light source 6 connected to the laser drive circuit 5 is made to continuously generate an extremely narrow optical pulse at a predetermined timing, and this is supplied to the coupler 7. .

Then, every time the optical pulse is guided to the optical fiber 1 by the coupler 7 and the backscattered light is emitted from the optical fiber 1, the wavelength is discriminated by the optical filter 8 and the anti-Stokes Raman scattered light is emitted. When,
It is separated into Stokes Raman scattered light.

Then, the optical switch 9 selects the anti-Stokes Raman scattered light obtained by the wavelength discrimination processing of the optical filter 8 and guides it to the photodiode 11 where it is converted into a photocurrent and a photocurrent amplifier circuit. After being amplified by 12 and made into an anti-Stokes Raman scattering signal, it is synchronously added by the number of times specified by the digital averaging circuit 13 to generate an average output of the anti-Stokes Raman scattering signal, which is supplied to the control memory arithmetic circuit 14. Are stored.

After this, when this processing is completed, the control storage arithmetic circuit 14 controls the optical switch 9 to select the Stokes Raman scattered light, and then controls the timing control circuit 4 to drive the laser drive circuit 5. The laser light source 6 connected to the laser driving circuit 5 is caused to continuously generate a light pulse having a very narrow width at a predetermined timing and to supply it to the coupler 7.

Then, every time the optical pulse is guided to the optical fiber 1 by the coupler 7 and the backscattered light is emitted from the optical fiber 1, the wavelength is discriminated by the optical filter 8 and the anti-Stokes Raman scattered light is emitted. When,
It is separated into Stokes Raman scattered light.

Next, the Stokes Raman scattered light obtained by the wavelength discrimination processing of the optical filter 8 is selected by the optical switch 9 and is guided to the photodiode 11 to be converted into a photocurrent, and at the same time, the photocurrent amplifier circuit 12 is selected. After being amplified by the Stokes Raman scattering signal to be a Stokes Raman scattering signal, the digital averager circuit 13 synchronously adds the number of times specified to generate an average output of the Stokes Raman scattering signal.
Are stored and stored in.

When this processing is completed, the control storage arithmetic circuit 14 performs the above-described optimum iteration number detection operation on the stored average output of the anti-Stokes Raman scattering signal and the stored average output of the Stokes Raman scattering signal. While repeating the deconvolution processing for the obtained optimum number of iterations, the ratio between the two is calculated, and the finally obtained temperature signal is output to another device via the temperature distribution output interface.

As described above, in this embodiment, the optical pulse having a very narrow width is generated by the signal processing device 2 at the time of initialization and is made incident on the optical fiber 1, and the optical pulse of this optical fiber 1 is generated. The Fresnel reflected light from the terminal surface is taken in, and the deconvolution is performed by the iterative method based on this Fresnel reflected light to find the optimum number of repetitions when the signal waveform corresponding to this Fresnel reflected light becomes linear. At the time of measurement, a very narrow optical pulse is generated by the signal processing device 2 and is made incident on the optical fiber 1. This backscattered light is taken in, and its anti-Stokes Raman scattered light and Stokes Raman scattering are taken. The temperature distribution of each part where the optical fiber 1 is laid is measured by performing deconvolution based on the optimum number of repetitions based on the light. Therefore, it is possible to prevent the solution from diverging due to the influence of noise in the inverse transformation process, and to give an optimal approximation that can improve the overall accuracy without narrowing the width of the optical pulse. S / N generated when the optical pulse width is narrowed
It is possible to improve the position resolution and facilitate the detection of abnormal heating portions such as hot spots without lowering the ratio, increasing the measurement time associated therewith, and making the semiconductor laser drive circuit difficult.

Further, in the above-mentioned embodiment, the value of the bias voltage applied to the photodiode 11 is switched to reduce the amplification factor so that the Fresnel reflected light is converted into photocurrent. Without changing the bias voltage of the photodiode 11, the Fresnel reflected light is made to have the same size as the anti-Stokes Raman scattered light or the Stokes Raman scattered light by applying a treatment such as coating to the incident surface and the end surface of No. 1. Even if the Fresnel reflected light is converted into a photocurrent while preventing the photodiode 11 from being saturated, the same effect as that of the above-described embodiment can be obtained.

[0070]

As described above, according to the present invention, the solution divergence due to the influence of noise is prevented in the inverse conversion process,
This makes it possible to provide an optimum approximation that can most improve the overall accuracy without narrowing the width of the light pulse, and reduce the S / N ratio that occurs when the width of the light pulse is narrowed, and the accompanying measurement. It is possible to improve the position resolution and facilitate detection of abnormal heating portions such as hot spots without increasing the time and making the semiconductor laser drive circuit difficult.

[Brief description of drawings]

FIG. 1 is a configuration diagram showing an embodiment of a temperature distribution detection device according to the present invention.

FIG. 2 is a schematic diagram showing a procedure example of deconvolution processing used in the basic principle of the temperature distribution detection device according to the present invention.

FIG. 3 is a configuration diagram showing a device configuration example for explaining a basic principle of a temperature distribution detection device according to the present invention.

4 is a waveform diagram showing an example of a temperature signal obtained by the temperature distribution detecting device shown in FIG.

5 is a waveform diagram showing an example of deconvolution processing for the temperature signal shown in FIG.

FIG. 6 is a table showing an example of a relationship between a systematic error of deconvolution processing and a random error used in the basic principle of the temperature distribution detecting device according to the present invention.

FIG. 7 is a waveform diagram for explaining the relationship between the number of repetitions of deconvolution processing used in the basic principle of the temperature distribution detecting device according to the present invention and the signal of Fresnel reflected light.

FIG. 8 is a schematic diagram showing a temperature measurement principle of a sensor using backscattered light of an optical fiber.

[Explanation of symbols]

1 Optical Fiber 2 Signal Processing Device 4 Timing Control Circuit 5 Laser Driving Circuit 6 Laser Light Source (Light Source) 7 Coupler 8 Optical Filter (Optical System) 9 Optical Switch (Optical System) 10 Bias Control Circuit 11 Photodiode (Photoelectric Conversion Section) 12 Photocurrent amplifier circuit (photoelectric conversion unit) 13 Digital averager circuit (digital averager unit) 14 Control memory arithmetic circuit (arithmetic unit)

Claims (1)

[Claims] 1. A light source that generates an optical pulse, an optical system that inputs the optical pulse into an optical fiber, and selects two types of Raman scattered light from the backscattered light thereof, and the Raman scattered light into an electrical signal. A photoelectric conversion unit for amplification, a digital averager unit for converting the output signal of the photoelectric conversion unit into a digital signal for averaging, and a temperature distribution along the optical fiber based on the processing result of the digital averager unit. In a temperature distribution detection device having a calculation unit for calculating, the waveform data of the Fresnel reflected light from the end face of the optical fiber is subjected to deconvolution processing and repeated until the waveform of the waveform data has a predetermined shape. When the number of times is calculated and the temperature is measured, the deconvolution process is repeated for the number of times of the Raman scattered light waveform data for the number of times described above. Calculating the temperature distribution along the optical fiber Te, the temperature distribution detection device according to claim.