Cable temperature monitoring method and system based on optical fiber

Abstract

The invention relates to an optical fiber-based cable temperature monitoring method and system, and the method comprises the steps: achieving the positioning of a cable temperature measurement point through employing a backscattering signal and an optical time domain reflection signal generated when light is transmitted in an optical fiber, obtaining the temperature data of a corresponding temperature measurement point, and storing the temperature data in the form of an electric signal; the temperature data of the plurality of temperature measurement points form the temperature of the curved surface of the cable, a rectangular coordinate system is established by taking the cable laying direction as an x axis, a geometric model based on the rectangular coordinate system is formed by using the temperature data of the curved surface of the cable, and the direction of a normal vector pointing to the rectangular coordinate system of the curved surface of the cable under different sampling rates is determined through a geometric hyperboloid vector method. Establishing a data flow temperature sliding window model; a preset standard cable landmark model is compared with a data structure of a data flow temperature sliding window model established by a measured cable, error data on the curved surface of the cable is extracted, namely a prediction error, and when the prediction error exceeds the preset standard cable landmark model, temperature data acquired at a temperature measurement point is corrected.

Description

Technical Field

[0001] The present invention relates to the field of cable technology, and in particular to an optical fiber-based cable temperature monitoring method and system. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] For online monitoring of the entire cable line and the temperature of the environment in which it is laid, the most commonly used technology is distributed optical fiber temperature measurement. The main principle is Raman scattering. When light propagates in an optical fiber, it will scatter with the optical fiber material and its surrounding environment. The Raman scattering spectrum will change due to changes in the ambient temperature of the optical fiber. By injecting pulsed lasers into the optical fiber and demodulating the Raman scattering signal using spectral analysis technology, the temperature distribution information along the length of the optical fiber can be obtained.

[0004] Distributed optical fiber temperature measurement technology obtains temperature data that changes over time, and the temperature data at each time point exists in the form of a data stream. Under normal circumstances, the temperature of the cable will not suddenly increase or decrease at a certain point in time, but when measuring temperature with optical fiber, it is affected by the loss and noise of the entire system, which is prone to omissions and false alarms, causing the associated temperature alarm function to erroneously issue warnings. Summary of the invention

[0005] In order to solve the technical problems existing in the above-mentioned background technology, the present invention provides an optical fiber-based cable temperature monitoring method and system. Through the prediction model of the sensor data stream, when the prediction error exceeds a preset threshold, the prediction model is automatically adjusted, thereby improving the quality of distributed optical fiber sensor information and being able to make effective corrections when abnormal data occurs, thereby reducing the defect rate of the optical fiber detection system to a certain extent and avoiding a large amount of on-site work.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] A first aspect of the present invention provides a cable temperature monitoring method based on optical fiber, comprising the following steps:

[0008] The backscattering signal and optical time domain reflection signal generated when light is transmitted in the optical fiber are used to locate the temperature measurement point of the cable and obtain the temperature data of the corresponding temperature measurement point. The temperature data is saved in the form of electrical signals.

[0009] The temperature data of multiple temperature measurement points form the temperature of the cable surface. The temperature data of the cable surface is used to form a geometric model. The normal vector pointing to the rectangular coordinate system of the cable surface under different sampling rates is determined by the geometric hyperboloid vector method, and a data stream temperature sliding window model is established.

[0010] Through the pre-set standard cable landmark model, the data structure of the data stream temperature sliding window model established for the tested cable is compared to extract the error data on the cable surface, which is the prediction error. The projection area of ​​the prediction error on each coordinate plane in the rectangular coordinate system is used as the input of the wavelet transform, and the corrected temperature data is obtained after denoising and reconstruction.

[0011] Furthermore, the temperature data of the cable surface is used to form a geometric model based on a rectangular coordinate system, including: establishing a rectangular coordinate system with the cable laying direction as the x direction, the vertical direction as the y direction, and the direction perpendicular to the xy plane as the z direction.

[0012] Furthermore, the geometric model based on the rectangular coordinate system includes upper and lower side models, left and right side models, and front and rear side models of the cable.

[0013] Furthermore, the upper and lower side models of the cable are shown as follows:

[0014] If ∑:z=z(x,y);

[0015] Upper side model: It is an acute angle, COSγ>0;

[0016] Lower side model: It is an obtuse angle, COSγ<0;

[0017] The left and right side models of the cable are shown below:

[0018] If ∑:y=y(x,z);

[0019] Right model: It is an acute angle, COSβ>0;

[0020] Model on the left: It is an obtuse angle, COSβ<0;

[0021] The front and rear side models of the cable are shown in the following equations:

[0022] If ∑:x=x(y,z);

[0023] Front Model: It is an acute angle, COSα>0;

[0024] Back side model: It is an obtuse angle, COSα<0;

[0025] Where, ∑ is the cable surface, is the normal vector of ∑.

[0026] Furthermore, the projection of the prediction error (ΔS) xy , as shown below:

[0027]

[0028] Among them, (Δσ) xy Represents the area of ​​the projection area, and γ is the angle between the normal vector and the z-axis.

[0029] Furthermore, in the landmark model, when the standard cable is set to run, the query range of the data flow model composed of the data flow information generated is: from a known initial time point to the current time point, that is, {a s , …, a n}.

[0030] Furthermore, the sliding window model is a data model established based on the temperature information measured by the current cable. In the data stream, the latest W data, its query range is {a ,max(c-w+1,0 , ..., a n}, as data continues to arrive, the data in the window continues to shift.

[0031] Furthermore, the backscattered signal and optical time domain reflection signal generated when light is transmitted in the optical fiber are used to locate the temperature measurement point of the cable and obtain the temperature data of the corresponding temperature measurement point, as shown in the following formula:

[0032] R(T)=I as / I s =(λ as / λ s ) 4 exp[-hcu0 / (kT)](1)

[0033] In the formula, I as I is the light intensity of the anti-bilateral curved surface; s is the light intensity of the double-sided curved surface; u as u is the frequency of the inverse double-sided curved surface; s is the double-sided surface spectrum frequency; u0 is the Rayleigh frequency; c is the speed of light; h is the Planck constant; k is the Boltzmann constant; T is the absolute temperature.

[0034] Furthermore, by using the Haar wavelet function method, when the prediction error exceeds the preset standard cable landmark model, the temperature data obtained from the temperature measuring point is corrected.

[0035] A second aspect of the present invention provides an optical fiber-based cable temperature monitoring system, comprising:

[0036] The data acquisition unit is configured to: utilize the backscattering signal and the optical time domain reflection signal generated when the light is transmitted in the optical fiber to realize the positioning of the temperature measuring point of the cable, and obtain the temperature data of the corresponding temperature measuring point, and the temperature data is stored in the form of electrical signals;

[0037] The data correction unit is configured to: form the temperature of the cable surface using the temperature data of multiple temperature measurement points, form a geometric model using the temperature data of the cable surface, determine the rectangular coordinate system of the normal vector pointing to the cable surface under different sampling rates through the geometric hyperboloid vector method, and establish a data stream temperature sliding window model;

[0038] The data correction unit is also configured to: compare the data structure of the data stream temperature sliding window model established with the measured cable through a pre-set standard cable landmark model, extract the error data on the cable surface, that is, the prediction error, use the projection area of ​​the prediction error on each coordinate plane in the rectangular coordinate system as the input of the wavelet transform, and obtain the corrected temperature data after denoising and reconstruction.

[0039] Compared with the prior art, one or more of the above technical solutions have the following beneficial effects:

[0040] The cable temperature data obtained by the distributed optical fiber is a data stream, that is, the temperature data obtained in sequence at a set time interval forms a data stream. The data stream can be converted into a geometric model reflecting the cable surface temperature. The data stream temperature value of the geometric hyperboloid vector method is used to determine the normal vector pointing to the inner or outer rectangular coordinate system of the cable surface at different sampling rates, thereby establishing a data stream temperature sliding window model. The data structure obtained by the data threshold of the landmark model established by the pre-set standard cable (referring to the temperature data collected when the cable is not powered on) is compared with the data structure obtained by the data stream temperature sliding window model collected when the cable is running, and the prediction error is extracted and converted to the projection on each coordinate plane, and then denoised and reconstructed, so that it can be automatically adjusted when the prediction error exceeds or decreases the threshold of the pre-set landmark model data, thereby improving the quality of the distributed optical fiber sensor information and being able to make effective corrections when abnormal data occurs. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0042] Figure 1 A schematic diagram of a cable temperature monitoring process based on optical fiber provided for one or more embodiments of the present invention;

[0043] Figure 2 A schematic diagram of establishing a data flow temperature model through a rectangular coordinate system after cable temperature measurement is provided for one or more embodiments of the present invention;

[0044] Figure 3 A schematic diagram of the structure of an optical fiber temperature sensor provided by one or more embodiments of the present invention;

[0045] Figure 4 A schematic diagram of the working process of the optical fiber sensor provided by one or more embodiments of the present invention;

[0046] Figure 5 A schematic diagram of a cable structure provided by one or more embodiments of the present invention;

[0047] Figure 6 A schematic diagram of the principle of a device for measuring the performance of an erbium-doped fiber amplifier provided in one or more embodiments of the present invention.

[0048] In the figure, 101 is the cable core; 102 is the sensing optical fiber; 103 is the outer sheath. DETAILED DESCRIPTION

[0049] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0050] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0051] Embodiment 1:

[0052] like Figure 1 As shown, the optical fiber-based cable temperature monitoring method comprises the following steps:

[0053] The backscattering signal and optical time domain reflection signal generated when light is transmitted in the optical fiber are used to locate the temperature measurement point of the cable and obtain the temperature data of the corresponding temperature measurement point. The temperature data is saved in the form of electrical signals.

[0054] The temperature data of multiple temperature measurement points form the temperature of the cable surface. The temperature data of the cable surface is used to form a geometric model. The normal vector pointing to the rectangular coordinate system of the cable surface under different sampling rates is determined by the geometric hyperboloid vector method, and a data stream temperature sliding window model is established.

[0055] Through the pre-set standard cable landmark model, the data structure of the data stream temperature sliding window model established for the tested cable is compared to extract the error data on the cable surface, which is the prediction error. The projection area of ​​the prediction error on each coordinate plane in the rectangular coordinate system is used as the input of the wavelet transform, and the corrected temperature data is obtained after denoising and reconstruction.

[0056] In this embodiment, the pre-set standard cable landmark model refers to a data model formed by measuring the temperature of the cable before it is powered on (mainly affected by the external environment temperature, so this temperature is extracted). The data stream temperature sliding window model is a data model formed by measuring the temperature of the cable after it is powered on.

[0057] The distributed fiber Bragg grating temperature measurement technology for power cables monitors the cable temperature online by utilizing the backscattering signal and optical time domain reflection generated when light is transmitted in the optical fiber to locate the cable temperature measurement point and obtain the temperature perception and measurement. The temperature function algorithm of the cable to be tested is:

[0058] R(T)=I as / I s =(λ as / λ s ) 4 exp[-hcu0 / (kT)](1)

[0059] In the formula, I as I is the light intensity of the anti-bilateral curved surface; s is the light intensity of the double-sided curved surface; u as u is the frequency of the inverse double-sided curved surface; s is the double-sided surface spectrum frequency; u0 is the Rayleigh frequency; c is the speed of light; h is the Planck constant; k is the Boltzmann constant; T is the absolute temperature.

[0060] According to formula (1), we can get:

[0061]

[0062] For a fixed temperature

[0063]

[0064] From equation (2) and equation (3), we can get:

[0065]

[0066] It can be seen from formula (4) that after the temperature measurement system is calibrated, by measuring R(T), using R(T0) at a known temperature T0, and then according to the propagation time of the backward light wave, the temperature value of each measurement point along the optical fiber can be determined. However, power cables will be affected by rain, snow, sunshine, electromagnetic induction and human factors during long-term operation, as well as the loss and noise of the entire system, causing the authenticity of the measured power cable data to be damaged during the transmission of light waves in the cable and when the data is transmitted to the data center, resulting in missed reports and false reports.

[0067] The temperature data obtained by the distributed optical fiber temperature measurement system on the cable is a data stream, that is, the temperature data is obtained in sequence at a set time interval to form a data stream. Let t represent any timestamp, a i Represents the data arriving at this timestamp. Stream data can be represented as {..., a i-1 , a i , a i+1 , ...}.

[0068] This embodiment adopts the double-sided surface integral method to give the measured temperature of the sensor data stream according to the temperature generated by the cable surface. The measured data is sampled by the cable sensing data and calculated at different sampling rates. The data constitutes a geometric model, and then the data stream temperature value of the geometric hyperboloid vector method is used to determine the normal vector at different sampling rates. Point to the cable∑inside or outside rectangular coordinate system and establish a data stream temperature sliding window model.

[0069] like Figure 2 As shown, a sensing optical fiber (102) is arranged on a cable core (101), an outer sheath (103) is coated on the sensing optical fiber (102), the sensing optical fiber and the cable core are laid in the same direction, and a rectangular coordinate system is established with the laying direction of the cable core as the x direction, the vertical direction as the y direction, and the direction perpendicular to the xy plane as the z direction.

[0070] 1) Establishment of the upper and lower side models of the cable:

[0071] If ∑:z=z(x,y);

[0072] Upper side model: is an acute angle, COSγ>0

[0073] Lower side model: Obtuse angle, COSγ<0

[0074] 2) Establishment of the left and right side models of the cable:

[0075] If ∑:y=y(x,z);

[0076] Right model: is an acute angle, COSβ>0

[0077] Model on the left: Obtuse angle, COSβ<0

[0078] 3) Establishment of the front and rear side models of the cable:

[0079] If ∑:x=x(y,z);

[0080] Front Model: is an acute angle, COSα>0

[0081] Back side model: Obtuse angle, COSα<0

[0082] in Determine a certain point threshold for the sensor temperature.

[0083] The data threshold of the landmark model established by the pre-set standard cable is compared with the data structure of the sliding window model established by the tested cable to extract the error data on the directed surface ∑, which is the prediction error model ΔS. The projection (ΔS) on the prediction error model ΔS is xy for:

[0084]

[0085] Where (Δσ) xy Represents the area of ​​the projection area, and γ is the angle between the normal vector and the z-axis.

[0086] Similarly, the projection of a directed surface on other coordinate planes can be given, and then the Haar wavelet function method can be used to automatically adjust when the prediction error exceeds or decreases the threshold of the pre-set landmark model data, thereby improving the quality of distributed fiber optic sensor information and being able to make effective corrections when abnormal data occurs. It can also display the cable temperature in real time or issue an alarm, which reduces the defect rate of the fiber optic detection system to a certain extent and avoids a lot of on-site work.

[0087] The characteristics of the fiber optic temperature detection data stream determine that the core of the data stream prediction model algorithm is to design a single-pass scanning algorithm, that is, in a memory space that is much smaller than the data size, a structure representing the data set - the prediction model data structure is continuously updated, so that approximate query results can be quickly obtained at any time based on this structure. The scale of the prediction model data structure should be at most sublinear, that is, if the length of the data stream model is N, the size of the prediction model data structure does not exceed 0 (polylog (N)), and the time to process each set of data on the stream does not exceed 0 (polylog (N)), that is, (ΔS) = 0.

[0088] xy

[0089] The data stream prediction model can be divided into a landmark model and a sliding window model according to different time series ranges. Let n represent the current timestamp, and s and e are two known timestamps respectively.

[0090] In the landmark model, the query range of the data flow model consisting of the data flow information generated during the operation of the set standard cable is from a known initial time point to the current time point, that is, {a s , …, a n}.

[0091] The sliding window model is only for the temperature information of the current cable measurement. The data model established is the latest W (W is also called the sliding window size) data in the data stream. Its query range is {a ,max(c-w+1,0 , ..., a n}, as data continues to arrive, the data in the window continues to shift.

[0092] The landmark model and sliding window model are closer to the predictive model to automatically adjust the sensing fiber to detect defects because they have to continuously process new data.

[0093] Due to the objective existence of factors such as node environment changes, electromagnetic interference, measurement errors, sampling errors and network delays, the acquired perception data is often uncertain. It is obviously inappropriate to use traditional data cleaning methods to improve the data quality in the perception database. This is because in the distributed optical fiber sensing network of power cables, although some abnormal data has a low probability of occurrence, it is an important data source for fault detection. If low-quality perception data is used directly, it will have serious consequences for the detection and management of power cables. Therefore, how to improve the quality of these perception data and perform effective calculations has brought new challenges to the data management of optical fiber detection, and is also the key to the success of distributed optical fiber sensing technology in power cable detection applications.

[0094] This embodiment adopts the Haar wavelet function method, provides a signal processing method, and a fast algorithm brings great convenience to the analysis and solution of practical problems, and provides a prediction model and prediction mechanism of the sensor data stream.

[0095] Haar wavelet transform. In order to reduce the prediction error, a prediction model automatic adjustment strategy is proposed, so that when the prediction error exceeds the preset threshold, the prediction model is automatically adjusted. The function value is as follows:

[0096]

[0097] For positive (negative) k, The graph is The result after translating to the right (left) by unit k. Assume v0 is all linear combinations:

[0098] (set of real numbers)

[0099] The space composed of The discontinuity points are x=k, x=(k+1), so the above linear combination is essentially a piecewise continuous function, and when the value of k is a finite number, the obtained function is a finite or compactly supported function.

[0100] Let v1 be the following linear combination:

[0101] (set of real numbers)

[0102] From a geometric point of view, the linear space composed of v0 can only describe the signal with a duration of more than 1 unit, while v1 can describe the signal with a duration of 1 / 2 unit. Based on the Haar function, the general space sequence is defined as follows.

[0103] Assume that n is an arbitrary integer, and the step function space v with level n is n The following collection:

[0104] {...,φ(2 n x+1),φ(2 n x),φ(2 n x-1),φ(2 n x-2)}

[0105] Composition. It can be deduced that the spatial sequence v n It has the following increasing property.

[0106]

[0107] because The above inclusion relationship is strictly established. Using space v n The definition of leads to the following conclusions.

[0108] (1) For a function f(x), if and only if f(2 n x)∈v n When f(x)∈v0.

[0109] (2) For a function f(x), if and only if f(2 -n x)∈v n When f(x)∈v n .

[0110] (3) Function set Composition n The standard orthogonal basis of .

[0111] In denoising applications, noise often appears as the high-frequency part of the signal. If we take v n-1 In v n The orthogonal complement w n , then w n This means that if and only if the length is at most The space where the constant is taken on the interval of , which represents the high-frequency components in the signal. In wavelet analysis, the function φ(x) is stretched and translated Get space w n .w n The definitions are as follows.

[0112] Assume w n It is formed by:

[0113]

[0114] The linear space composed of functions of k Contains a finite number of non-zero terms, then w n Composition V n In V n+1 The orthogonal complement of n+1 =V n +W n .

[0115] Function φ(x) and function φ(x) can also establish a direct relationship as follows.

[0116]

[0117] φ(x)=φ(2x)-φ(2x-1)

[0118] φ(x)=φ(2x)+φ(2x-1)

[0119] For any energy-limited space, there is the following algorithm:

[0120] Algorithm 1: Energy-limited space L 2 (R) can be decomposed into the following sum:

[0121] L 2 (R) = V0 + W0 + W1 + W2

[0122] So f = L 2 (R) can be uniquely expressed as where f0∈V0, w k ∈W k .

[0123] where w k It means that if and only if the length is A constant function on the interval. When k is sufficiently large, the function can be used to represent noise. For example, assuming that a signal with a duration less than 0.02 is regarded as noise, then 2 -6 <0.02<2 -5 , so the components with k ≥ 6 are set to W k =0, and the remaining signal part is approximately regarded as the actual signal content.

[0124] Algorithm 2: Wavelet orthogonal signal decomposition and reconstruction algorithm: Assumptions:

[0125]

[0126] Then f1(x) can be decomposed into the following form:

[0127] f1(x)=w l-1 (x)+f l-1 (x)

[0128] in,

[0129]

[0130] satisfy:

[0131]

[0132] Among them, l-1 is called resolution, is called scale. The above decomposition process can be continued until the following decomposition formula is obtained:

[0133] f n =w n-1 +w n-2 +...+w1+f0

[0134] Algorithm 3: Assumptions:

[0135] f n =w n-1 +w n-2 +...+w1+f0,w1∈W1

[0136] satisfy:

[0137]

[0138] and

[0139]

[0140] but

[0141]

[0142] The coefficient a can be recursively realized according to the following expression:

[0143]

[0144] The process of signal processing is to reconstruct the coefficient b before the wavelet signal is reconstructed according to the needs of the actual problem. 1 k Make appropriate corrections and then reconstruct the wavelet.

[0145] Wavelet signal denoising. Using modulus maximum reconstruction filtering, assume that θ(t) is a smooth function, and wavelet φ(t) is the first-order derivative of θ(t), that is, φ(t) = dθ(t) / dt, then the continuous wavelet transform of f is:

[0146]

[0147] The modulus maximum Wf(s,u) of the wavelet transform is the maximum value of the first-order derivative of the polished function f, which corresponds exactly to the mutation point of the signal f.

[0148] Signal modulus maximum reconstruction refers to the reconstruction of the signal using the modulus maximum of the wavelet coefficients at all scales. The modulus maximum of the signal wavelet coefficients contains the peak variability and singularity of the signal. By reconstructing the signal from these maxima, the singularity of the signal can be modified by processing the modulus maximum of the wavelet coefficients. The singularity intensity can be modified by changing the modulus maximum, or the corresponding singularity can be removed by suppressing certain maximum points.

[0149] The following are the specific algorithm steps of wavelet modulus maximum denoising.

[0150] ① Perform discrete binary wavelet transform on the noisy signal. The selected decomposition scale number should make the number of signal modulus maxima dominate at the maximum decomposition scale and the important singular points of the signal are not easily lost.

[0151] ② Find the modulus maximum point corresponding to the wavelet transform coefficient at each scale.

[0152] ③ At the maximum decomposition scale J, the wavelet transform modulus maximum is almost completely controlled by the signal. A threshold is selected so that points with modulus maximum values ​​less than the threshold are removed, thereby obtaining a new modulus maximum point on the maximum scale.

[0153] ④ Starting from each modulus maximum point on scale J, search upward for its corresponding modulus maximum curve. Specifically, find the propagation point corresponding to each modulus maximum point on scale j on scale j-1 (j=J, ...4,3), retain the modulus maximum points generated by the signal, eliminate the modulus maximum points caused by noise, and remove the extreme points on each scale j that are not on any modulus maximum curve. Search step by step in this way until scale j=2.

[0154] ⑤ For scale j=1, the corresponding extreme points when j=1 are retained at the positions where extreme points exist in j=2, and the extreme points at other positions are set to zero.

[0155] ⑥ The denoised signal is obtained by using the reconstruction method based on the modulus maximum and its extreme point positions retained at each scale.

[0156] The cable sensing data is sampled at different sampling rates, with the signal reconstruction error and reconstruction time as the perception. The algorithm is as follows:

[0157] Sampling rate (S R )S R =M / N

[0158] Relative error (R E )

[0159] Root mean square error (RMSE)

[0160] Among them: x is the original signal, To reconstruct the signal, M is the number of measurements and N is the length of the original signal.

[0161] The reconstruction comparison of temperature signals at different sampling rates is shown in Table 1.

[0162] Table 1: Comparison of reconstruction of temperature signals at different sampling rates

[0163]

[0164]

[0165] As can be seen from the table above, the relative error, root mean square error and reconstruction algorithm time of this algorithm are significantly better than those of traditional algorithms. In this algorithm, when M = 0.1 × N, although the trend of the reconstructed signal is similar to that of the original signal, the relative error and root mean square error of the reconstructed signal are greatly increased, the reconstruction quality is significantly reduced, and the operation speed is also greatly reduced. The results show that the prediction model automatically adjusts the sensing fiber detection and senses the monitoring signal of the temperature sensor in the optical fiber network in the wavelet domain. This algorithm is more effective than the traditional CoSaMP algorithm.

[0166] Kalman filtering is used for filtering, which is convenient for superposition. The Gaussian distribution model is used to describe the expected value and error of the measured value and the estimated value. Then, the formula properties of the Gaussian distribution model are used to superimpose the estimated value and the measured value to output, and the error range of the optimal value is calculated: the filtering process of the next cycle is consistent with the current process, and it can be continuously iterated; the weight configuration of the measured value and the estimated value finally outputs the optimal value formula:

[0167]

[0168] Among them, x- represents the optimal output value, is the estimated value, z is the measured value, k g It represents the Kalman gain. It can be seen from the formula that the Kalman filter is consistent with the first-order low-pass filter and complementary filter, relying on the weight coefficient k gTo affect the final output value. The source of the predicted value depends on the control system model. For example, if the current temperature is 30°, heating with a power of 10° per second can predict that the temperature will be 50° in 2 seconds (ignoring the effects of heat dissipation, etc.), with an error of 5 degrees. Of course, in the case of simple first-order filtering, the control model can be simplified, and the predicted value after 1 second is always assumed to be the actual optimal output value before. The measured value is obtained from the sensor, and the size of the error is determined by the sensor accuracy.

[0169] The optimal output value is the superposition of the estimated value and the measured value, but the estimated value and the measured value are not accurate values ​​and there are certain errors. For example, when the sensor measures the temperature, the measured value is 30° and the error is 2°. The estimated value will also have process errors in the calculation process due to the accuracy of the model, as well as external interference errors. The Gaussian distribution expectation and variance form are consistent with this value and error form, which is very suitable for describing this model. Therefore, Kalman used the Gaussian probability distribution model to specifically functionalize the description of each component value and its error, and indirectly introduced the amount of error value. Gaussian distribution is also called normal distribution, and its probability density function is shown below:

[0170]

[0171] e is the natural base; σ 2 is the variance, corresponding to the error; μ is the expectation, corresponding to the estimated value / measured value

[0172] (1) Assume that the Gaussian distribution probability density functions of the estimated value and the measured value are:

[0173]

[0174] (2) According to the properties of the Gaussian formula, multiplication is performed. This part is the core part of the Kalman filter. The superposition principle formula is the Bayesian formula: posterior probability = k × likelihood probability × prior estimated probability. Correspondingly, the optimal output probability density function = k × measured value probability density function × estimated value probability density function.

[0175]

[0176] (3) Simplify and construct a new Gaussian formula

[0177]

[0178] Among them, S g It is called the compression factor and determines the size of the total probability area.

[0179] Variance σ after transformation superposition 2 With the expected μ, extract the Kalman gain coefficient k g

[0180]

[0181] μ=μ1+k g (μ2-μ1);

[0182] σ 2 =(1-k g ).σ1 2 ;

[0183] So far, the description and superposition of the Gaussian distribution of the estimated value and the measured value have been completed. The optimal estimated value is obtained based on the expected value μ after superposition, and the variance after superposition is used for the variance calculation of the next estimated value.

[0184] During the filtering process, the filtering method of superimposing estimated values ​​and measured values ​​has the following effects and benefits on improving the quality of collected temperature data:

[0185] Reduce the impact of random errors. Random errors often exist in the temperature measurement process. These errors may be caused by factors such as noise in the measuring instrument and slight fluctuations in the environment.

[0186] The measured value itself will be disturbed by these random factors and fluctuate. When the estimated value and the measured value are superimposed, the estimated value can smooth these random fluctuations to a certain extent. For example, if the measured value has an abnormally high value due to momentary environmental interference, the estimated value can pull it back to a more reasonable range, thereby reducing the impact of this abnormal value on the final result.

[0187] From a statistical point of view, taking the average of multiple measurements can reduce random errors, and the estimated value here can be regarded as a "pre-average" based on historical data or models. When superimposed with the measured value, it further reduces the impact of random errors on the quality of temperature data.

[0188] Improve data stability. For temperature data acquisition systems, if the acquisition frequency is high, the data may fluctuate rapidly and slightly. The filtering method of superimposing estimated values ​​and measured values ​​can make the data more stable. Estimated values ​​are often calculated based on certain rules or trends, and they can reflect the general trend of temperature data. When superimposed with measured values, it is like adding a "stabilizer" to rapidly fluctuating data.

[0189] For example, in a temperature control system, stable temperature data is very important for accurate control. This filtering method can avoid frequent and unnecessary adjustments of the control system due to excessive fluctuations in data, thereby improving the stability and reliability of the entire system.

[0190] Enhance data accuracy. The estimated value can use existing knowledge or models to predict the temperature. In some cases, the measured value may deviate due to local failure of the sensor or local interference. The estimated value can make a reasonable estimate of the temperature based on the surrounding environment information, historical data, etc., and can correct the deviation of the measured value after being superimposed with the measured value, thereby improving the accuracy of the data.

[0191] For example, if there is a temporary heat source near the cable fiber optic sensor that interferes with the measurement, and the estimated value is based on the average temperature of the entire environment or the historical temperature trend, the superimposed result will be closer to the actual temperature value.

[0192] To determine the superposition of estimated and measured values, it is necessary to consider the accuracy of the measuring instrument, the stability of the environment, the frequency of data collection, etc., so that the quality of the filtered temperature data is optimized.

[0193] In the calculation of superposition ratio, the error can be controlled in the following ways: First, ensure the accuracy and reliability of the measured data and reduce the error of the original data. Second, choose a suitable calculation method, such as optimizing the parameters of the weighted average method or the least square method. Third, perform multiple calculations and take the average value to reduce the impact of random errors.

[0194] The modulation signal is realized by encoding the digital baseband signal consistent with the device communication protocol. In the test system of communication equipment, the standard computer platform is generally used as the core, and the D / A conversion card is used to connect to the device under test, and the access to these devices is realized through software. However, in order for the system to work normally and reliably, the digital baseband signal encoding generated by the computer must comply with the communication protocol of the device. The communication protocol is an agreement that both parties of the communication abide by. It involves information format, sequence control, flow control, information generation and unlocking, error control and other aspects.

[0195] The generation of digital baseband signal coding can be divided into two steps:

[0196] ① Generate initial data that conforms to the communication information format. The asynchronous communication information format of the device generally includes a start bit, a data bit, a parity bit, an end bit, etc., which are the basis for generating the initial data. In this embodiment, the asynchronous communication information format of the communication device is shown in Table 2:

[0197] Table 2: Asynchronous communication information format of communication equipment

[0198] 0 B0 B1 B2 B3 B4 B5 B6 B7 1

[0199] From the format, we know that sending 8b data requires sending 1b start bit and 1b end bit at the same time. It can be understood that the communication data is a number between 100000000-111111110, which is converted into a decimal number between 512 (2 to the 9th power) and 1023. The initial data is generated by multiplying the valid data by 2 plus 512.

[0200] ② Convert the initial data into a binary sequence and store it in an array, and you will get a digital baseband signal encoding that is consistent with the device asynchronous communication information format. Since the binary number corresponding to the 1B data is 10b, sending 28 numbers requires generating a binary number of 280b.

[0201] Each code element of the digital baseband signal encoding is converted into a certain number of data describing the signal waveform (frequency f, and cosine waveform of f2), and stored in an array. These data are sent out through the D / A card at a certain time interval to form the required modulation signal. The obtained modulation signal must ensure the continuity of phase and the accuracy of frequency.

[0202] The signal can be represented as:

[0203] Phase continuity means that the phases of the two frequencies f1 and f2 are continuous at the moment of conversion; to achieve this in the modulated signal generated by software simulation, it is necessary to make the initial phase of the first data of the subsequent code element the same as or differ by 2π from the phase of the last data of the previous code element when the transmitted code elements are converted to each other (1 to 0 or 0 to 1). Otherwise, the phase of the obtained FM signal will be discontinuous.

[0204] Frequency accuracy. Using a D / A card to send data inevitably involves timing issues, that is, the time interval for data transmission must be set. If the time interval is not set properly, the frequency modulation signal with accurate frequency cannot be obtained. The data transmission rate of the communication equipment tested by the author is 1200b / s. If each code element is to be converted into 100 data describing the signal waveform and sent out, the data transmission interval must be set to 8.3μs (120k data are sent per second). It is worth noting that the setting of the transmission interval must ensure that the number of data converted by each code element is an integer, otherwise the obtained signal frequency will be unstable.

[0205] The software implements the D / A card's sending timing method by calling API functions under the C++Builder platform to implement timing. The WIN32API provides two related functions that can achieve μs-level timing: QueryPerformanceFrequency() and QueryPerformance Counter(). Their definitions are as follows:

[0206] ①BOOL Query PerformanceFrequency(LARGE-INTEGER*IpFrequency): This function obtains the frequency of the high-resolution performance counter (if it exists); lpFre quency represents a variable pointer, and the count per second of the variable value is set by the function as the frequency of the current performance counter; if the installed hardware supports high-resolution performance counters, the function returns a non-zero value, otherwise the return value is zero.

[0207] ②BOOL QueryPerformanceCounter(LARGE-INTEGER*lpPerformance-Count); This function obtains the current value of the high-resolution performance counter (if it exists) (with the boot time as the base value 0); lpPerformanceCount represents a variable pointer, the value of which is set by the function to the current performance counter value; if the installed hardware supports high-resolution performance counters, the function returns a non-zero value, otherwise the return value is zero.

[0208] like Figure 6 As shown in the figure, the basic device for measuring the performance of erbium-doped fiber amplifiers consists of a 980nm pump DC light, a 1540nm DC laser, an isolator EDF, a WDM, and a section of erbium-doped fiber to be tested. The 1540nm DC light is used as the signal light, and the 980nm DC light is used as the pump light. They are input into the WDM, and the insertion loss of the WDM is set to 1dB. If the length of the erbium fiber is 10.29m. After passing through an isolator (insertion loss is measured to be 0.5dB) to ensure unidirectional transmission of the light, it is input into the erbium-doped fiber through the WDM together with the 980nm pump light. The solder joint loss of the erbium fiber is 1.6dB. The model of the erbium-doped fiber is EDF502-10720, the absorption of 1530nm is about 6dB / m, and the length of the erbium fiber is 10.29m. The output is filtered out of the 1540nm band through a filter, and the insertion loss of the filter is about 1dB. Then the signal light and pump light are filtered out by the demultiplexer. It is calculated that when the 1540nm signal light power is 160uW,

[0209] The fiber pulse amplifier is a laser without feedback. Its physical mechanism is to amplify the incident light through stimulated radiation. It transitions from the ground state to the laser emission, and then quickly relaxes to a metastable state with lower energy, causing a population inversion between energy levels. When the signal light is incident, the erbium ions are stimulated to radiate and amplify the signal light. The gain coefficient of the uniformly broadened gain medium is calculated as follows:

[0210]

[0211] Where g0 is the peak value of the optical gain, which is determined by the pump power of the amplifier, ω ais the atomic transition frequency, and P is the signal light power. s is the saturation power, which is related to the fluorescence time and transition cross section. Equation (29) shows that when the signal light frequency is the same as the atomic transition frequency, the gain is maximum.

[0212] The amplifier gain G is defined as G = P out / P in , where P out is the output signal optical power, P in is the input signal optical power. The amplifier gain can be obtained by solving the following equation:

[0213]

[0214] The above formula shows that the amplifier gain G and the gain factor g have the following relationship:

[0215] G(ω)=exp[g(ω)*L];

[0216] In the above formula, L is the length of the gain fiber.

[0217] Equation (29) also illustrates the phenomenon of gain saturation, that is, when the power of the signal light is comparable to the saturation power, the gain coefficient becomes smaller, so the amplifier gain becomes smaller. After considering the gain saturation, equation (30) becomes:

[0218]

[0219] Generally speaking, the higher the doping concentration, the higher the gain of the fiber. However, when the doping concentration is too high, the erbium ions will not only absorb all the 980-nanometer optical pump power, but also absorb the signal light, and the gain will decrease. If the pump is input from one end of the gain fiber, due to the absorption of erbium ions, the 980-nanometer optical pump power will decrease with the fiber length, and the gain coefficient g will decrease. When the fiber is too long, the 980-nanometer optical pump light is exhausted, and the erbium ions begin to absorb the signal light, which will also cause the gain to decrease. Therefore, there is an optimal length for the gain fiber. The relationship between the too small doping radius and the intensity division of the 980-nanometer optical pump light mode in the fiber will also affect the gain.

[0220] Real-time display and alarm of cable temperature. In this embodiment, rectangles are used to represent FBG temperature sensors. When the user observes all sensors in a certain area, the sensor measurement data will be displayed on the rectangle. When the sensor measurement data is below the dangerous temperature threshold, the sensor will be represented by a white rectangle; when the sensor measurement data is greater than the threshold, it indicates that the cable temperature at that location is abnormal, which should attract the user's attention and remind the user to take relevant measures.

[0221] When the user needs to observe a specific sensor, move the mouse to the sensor and click it, and the system will draw the temperature data flow of the sensor in real time with a dynamic curve graph. At the same time, considering the strict safety requirements of the power system, the system will display the highest temperature of other monitoring areas in real time with a bar graph.

[0222] When the temperature of a sensor exceeds the threshold, the corresponding SVG rectangular basic element turns red, and the user is prompted to take measures in the lower right corner of the interface, which can effectively prevent dangerous accidents caused by high temperature of power lines. After the user takes emergency measures for the sensor, the sensor can return to normal. At the same time, when a sensor is in a dangerous alarm state, other sensors are still in normal working state.

[0223] Principle of short-time autocorrelation method. By calculating the short-time autocorrelation function value of the signal, the possible half-cycle and double-cycle errors can be verified. Since the autocorrelation function can determine multiple sensor data stream periods by comparing the similarities between multiple signal segments, it can complement the previous wavelet transform method and more effectively verify the true position of the next extreme point and the true length of the next sensor data stream period.

[0224] The calculation formula of the short-time autocorrelation function is as follows:

[0225]

[0226] Where: N is the window length. Since the short-time autocorrelation function R w (l) There are large peaks at points that are integer multiples of the sensor data stream period. Generally, the length of the sensor data stream period can be determined based on the position of the first peak.

[0227] The main function of short-time autocorrelation verification is to predict the situation where the periodic change of the front sensor data stream is too large, and to establish multiple sensor data streams with the maximum center error absolute value reaching the minimum and the maximum length error absolute value reaching the minimum. The autocorrelation verification is performed through the autocorrelation algorithm. The verification method is: determine the next extreme point from the two candidate extreme points. So first, in order to avoid too much computational complexity, it is only necessary to calculate the short-time autocorrelation function value at the position of the two candidate extreme points for preliminary comparison: if one of the two function values ​​is more than 3 times the other, the function value with the larger value is determined as the next extreme point to be determined.

[0228] If one of the two function values ​​is between 1.5 and 3 times the other, it is necessary to calculate the sequence function value according to the short-time autocorrelation method, find the position of the first short-time autocorrelation function peak point, and then take the one of the two extreme points that is closer to the peak point as the next extreme point. If the two function values ​​are close, then directly determine the extreme point, and the multiple sensor data stream cycles formed by it are close to the current sensor data stream cycle as the next extreme point.

[0229] The established multiple sensor data streams were then autocorrelated and verified using an autocorrelation algorithm. The waveform period was determined by short-time autocorrelation verification, and the verified correction data was fed back to the automatic adjustment prediction model for data correction, which improved the quality of power cable perception data, ensured the calculation speed, and reduced the defect rate of optical fiber detection.

[0230] Embodiment 2:

[0231] Fiber optic based cable temperature monitoring system, including:

[0232] The data acquisition unit is configured to: utilize the backscattering signal and the optical time domain reflection signal generated when the light is transmitted in the optical fiber to realize the positioning of the temperature measuring point of the cable, and obtain the temperature data of the corresponding temperature measuring point, and the temperature data is stored in the form of electrical signals;

[0233] The data correction unit is configured to: form the temperature of the cable surface using the temperature data of multiple temperature measurement points, form a geometric model using the temperature data of the cable surface, determine the rectangular coordinate system of the normal vector pointing to the cable surface under different sampling rates through the geometric hyperboloid vector method, and establish a data stream temperature sliding window model;

[0234] The data correction unit is also configured to: compare the data structure of the data stream temperature sliding window model established with the measured cable through a pre-set standard cable landmark model, extract the error data on the cable surface, that is, the prediction error, use the projection area of ​​the prediction error on each coordinate plane in the rectangular coordinate system as the input of the wavelet transform, and obtain the corrected temperature data after denoising and reconstruction.

[0235] like Figure 4 As shown, the light source system sends the test signal to the sensing unit or array through the light transmission and conversion part, and also sends the modulated signal to the user interface through the light transmission and conversion part.

[0236] Fiber optic cable structure Figure 5 As shown, it comprises a cable core (101); a sensing optical fiber (102); and an outer sheath (103); wherein the sensing optical fiber (102) is arranged on the cable core (101), and the outer sheath (103) is coated on the sensing optical fiber (102).

[0237] The sensing unit or array may be composed of a light source (S), a light detector (D), two 3dB couplers (DC1, DC2), a mode filter (P), a phase modulator (M) and a sensitive ring L connected by optical fibers.

[0238] The distributed optical fiber temperature sensing system is composed of an optical fiber core containing one or more coatings of the sensing optical fiber and then an outer sheath of a reinforcing member; the reinforcing members of the distributed optical fiber temperature sensing include metal heavy reinforcing members, non-metallic reinforcing members, and non-metallic heavy reinforcing members; the metal heavy reinforcing members generally use stainless steel pipes or copper pipes, and the non-metallic reinforcing members generally use polyethylene protective tubes, polyethylene sleeves, polyethylene coated with nylon sleeves or polyvinyl chloride sleeves and other materials as reinforcing members and outer sheaths;

[0239] The optical fiber core of the sensing fiber in the optical sensing part is usually coated with a layer of acrylate polymer, imide or fiber jacket to prevent light leakage from one fiber from entering another fiber, reduce crosstalk between fibers, or reduce the spot in fiber bundle imaging applications;

[0240] There are two forms of distribution of the sensing optical fiber of the optical sensing part in the cable: the first one uses winding to connect the waveguide offset and winds it at a certain pitch. When the pitch is too small, the additional loss increases and the optical fiber length increases, which will cause signal attenuation and transmission delay; the second one uses rising sinusoidal S-shaped bends to connect the waveguide offset, which can obtain excellent splitting ratio, extremely low additional loss and shorter connection length.

[0241] The sensing fiber is contained within deployment cables suitable for different environmental protections. Each end of the cable can be terminated with a dedicated fiber optic connector to make it easy to connect and disconnect to the sending and receiving equipment, and different types of cables are used for different applications; such as long-distance telecommunications or providing high-speed data connections between different parts of a building.

[0242] The sensing optical fiber adopts non-dispersion-shifted optical fiber, cut-off wavelength-shifted optical fiber, dispersion-shifted optical fiber or non-zero dispersion-shifted optical fiber.

[0243] The sensing fiber is a fiber Bragg grating temperature sensor, which is used to transmit information data during the operation of the cable. The fiber Bragg grating temperature sensor adopts Raman fiber. The Raman scattering technology uses the function of the fiber Bragg grating temperature sensor to measure the temperature distribution along the cable. It has certain advantages in the online monitoring of long-distance cable temperature, but it has particularly high requirements for the light source.

[0244] The function of the fiber grating temperature sensor is similar to that of a point-shaped electrical signal temperature sensor. It is encapsulated in a stainless steel shell and has a short thermal response time. Multiple probes can be connected in series in the same detection circuit to achieve distributed and rapid measurement of multiple target temperatures.

[0245] The fiber grating temperature sensor is specially packaged and laid on the surface of several 1kV~500kV power cables to realize online temperature computer monitoring.

[0246] Optical fiber is both a transmission medium and a sensor. The high-speed driving circuit drives the laser to emit a narrow pulse width laser pulse. The laser pulse is transmitted forward along the sensing optical fiber after passing through the wavelength division multiplexer. The laser pulse interacts with the optical fiber molecules to produce a variety of weak backscattering. Among them, Raman scattering is due to the thermal vibration of the optical fiber molecules, which produces temperature-insensitive double-sided surface integral (Stokes) light and temperature-sensitive anti-double-sided surface integral (Anti-Stokes) light. The wavelengths of the two are different. After being separated by the wavelength division multiplexer, they are detected by a highly sensitive detector. The Anti-Stokes light intensity in the optical fiber is modulated by the external temperature, and the ratio of the Anti-Stokes to Stokes light intensity accurately reflects the temperature information; the time for Raman scattering signals at different positions to return to the detector is different. By measuring the echo time, the optical fiber position corresponding to the scattering signal can be determined; combined with high-speed signal acquisition and data processing technology, the temperature distribution information of any point on the entire sensing optical fiber can be accurately and quickly obtained.

[0247] DWDM is a dense wavelength division multiplexer that uses the wavelength of lasers to transmit data in optical fibers in bit-by-bit parallel or string-by-string mode. It carries 8 to 160 different wavelength channels on one optical fiber. The different wavelengths of each channel are separated by 1.6 / 0.8 / 0.4nm (200GHz / 100GHz / 50GHz) wavelengths, making it suitable for long-distance signal transmission.

[0248] The most direct application of fiber pulse lasers is as ultrafast light sources, forming a variety of time-resolved spectral all-optical sampling / electro-optical sampling measurement techniques, as seed light sources for femtosecond solid-state laser amplifiers, which can range from nanometers (for example, measuring the displacement of scanning detection microscopes) to kilometers, based on the correlation measurement method, which is dispersion interferometry. According to spectroscopy or cable characterization, it is achieved by encoding a series of femtosecond pulse sequences from the "detection" laser, and then using a local oscillator laser signal to generate a beat signal with the detection laser signal. The repetition frequency of the local oscillator laser signal is slightly different from the repetition frequency of the detection signal. In this case, when the detection signal is reflected back by the target plane and the reference plane, the encoding is completed, and two colinear propagating pulses with different time delays are generated. For each consecutive pair of detection and local oscillator signals, the relative pulse timing between them is slightly shifted, and when the local oscillator signal passes through the detection signal, a cross-correlation is formed in the correlation signal. For sufficiently stable femtosecond lasers, this cross-correlation contains precise information about the timing signal between the target and reference pulses. The envelope of the pulse provides a time-of-flight range measurement while the carrier phase of the pulse provides correlation range monitoring and analysis.

[0249] The hardware platform mainly consists of three parts: subsystems, clients and integrated software platforms. Among them, each subsystem is responsible for the collection, analysis, processing and storage of its own monitoring parameters, and uploads the data to the integrated software platform through TCP / IP (Transmission Control Protocol / Internet Protocol). The data uploaded by each subsystem is received by the interface module of the integrated software platform and uniformly converted into a standard format and stored in the database for upper-level analysis and application of the integrated software platform. Each client views the data on the integrated software platform through the network and makes corresponding business function requests. The integrated software platform adopts a networked architecture to realize data collection, conversion, processing, analysis and display, and exchanges data with other software systems in the power department through TCP / IP.

[0250] The structure of distributed optical fiber temperature sensing system is as follows: Figure 3 As shown. The principle is: the driving circuit sends out a driving signal to modulate the high-power laser, and at the same time sends out a synchronous pulse to control the A / D (analog / digital) acquisition card. The pulse light emitted by the modulated laser is injected into the reference optical fiber after CWDM (coarse wavelength division multiplexing), and the end of the reference optical fiber is connected to the sensing optical fiber. When the pulse light is transmitted in the optical fiber, Rayleigh scattering will occur. The backscattered light with a scattering direction opposite to the forward direction of the light wave contains double-sided curved surface light and anti-double-sided curved surface light. After CWDM branching, it is received by the APD (avalanche photodiode) for optical / electrical conversion. The converted electrical signal is collected by the A / D acquisition card after filtering and optical fiber pulse amplifier, and input into the computer for processing.

[0251] For example, the extinction ratio of the ZnO / A1 thin film polarizer at 1.3 μm wavelength reaches 45 dB, and the ZnO is made by high-frequency sputtering;

[0252] The outer diameter of the single-mode optical fiber required for coupling of the coupler (DC1, DC2) device is usually 125μm, and the optical fiber is fixed in a reinforcement member that can be precisely positioned. The position of the optical fiber in the groove should meet the following requirements:

[0253] Where: ω0 is the groove width of the optical fiber in the V-groove in the reinforcing member; Δ is the distance from the center of the optical fiber to the surface of the V-groove; r is the core diameter of the optical fiber.

[0254] When LiNbO3 waveguide is coupled with SiO2 fiber, there is about 4% Fresnel reflection, which forms two unstable Michelson interferences in the optical path and superimposes on the Sagnac interference to produce harmful bias offset. To eliminate this reflection, the end faces of the optical fiber and waveguide are ground into oblique coupling at different angles to the axis. The angles should satisfy the law of refraction and are generally ground into 15° and 10°. In this oblique coupling, neither the optical fiber nor the waveguide will receive any reflected light, and the reflected light is coupled into the substrate. As a result of this combined effect, the reflection suppression can reach more than 60dB. The oblique grinding method uses a waveguide end face that is not perpendicular to the waveguide surface and forms a certain angle. This method is incompatible with flip-chip coupling and can only use a method similar to the coupling of optical fiber and laser, but the reflected light is not transmitted on the waveguide surface, and there is no re-coupling of the reflected light.

[0255] The hardware platform mainly consists of three parts: subsystems, clients and integrated software platforms. Among them, each subsystem is responsible for the collection, analysis, processing and storage of its own monitoring parameters, and uploads the data to the integrated software platform through TCP / IP (Transmission Control Protocol / Internet Protocol). The data uploaded by each subsystem is received by the interface module of the integrated software platform and uniformly converted into a standard format and stored in the database for upper-level analysis and application of the integrated software platform. Each client views the data on the integrated software platform through the network and makes corresponding business function requests. The integrated software platform adopts a networked architecture to realize data collection, conversion, processing, analysis and display, and exchanges data with other software systems in the power department through TCP / IP.

[0256] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A cable temperature monitoring method based on optical fiber, characterized in that: The following steps are involved: The backscattering signal and optical time domain reflection signal generated when light is transmitted in the optical fiber are used to locate the temperature measurement point of the cable and obtain the temperature data of the corresponding temperature measurement point. The temperature data is saved in the form of electrical signals. The temperature data of multiple temperature measurement points form the temperature of the cable surface. The temperature data of the cable surface is used to form a geometric model. The normal vector pointing to the rectangular coordinate system of the cable surface under different sampling rates is determined by the geometric hyperboloid vector method, and a data stream temperature sliding window model is established. Through the pre-set standard cable landmark model, the data structure of the data stream temperature sliding window model established for the tested cable is compared to extract the error data on the cable surface, which is the prediction error. The projection area of ​​the prediction error on each coordinate plane in the rectangular coordinate system is used as the input of the wavelet transform, and the corrected temperature data is obtained after denoising and reconstruction.

2. The optical fiber-based cable temperature monitoring method according to claim 1, characterized in that: The temperature data of the cable surface is used to form a geometric model based on a rectangular coordinate system, including: establishing a rectangular coordinate system with the cable laying direction as the x direction, the vertical direction as the y direction, and the direction perpendicular to the xy plane as the z direction.

3. The optical fiber-based cable temperature monitoring method according to claim 1, characterized in that: The geometric model based on the rectangular coordinate system includes the upper and lower side models, the left and right side models, and the front and rear side models of the cable.

4. The optical fiber-based cable temperature monitoring method according to claim 3, characterized in that: The upper and lower side models of the cable are shown in the following formula: If ∑:z=z(x,y); Upper side model: It is an acute angle, COSγ>0; Lower side model: It is an obtuse angle, COSγ<0; The left and right side models of the cable are shown below: If ∑:y=y(x,z); Right model: It is an acute angle, COSβ>0; Model on the left: It is an obtuse angle, COSβ<0; The front and rear side models of the cable are shown below: If ∑:x=x(y,z); Front Model: It is an acute angle, COSα>0; Back side model: It is an obtuse angle, COSα<0; Where ∑ is the cable surface.

5. The optical fiber-based cable temperature monitoring method according to claim 1, characterized in that: Projection of prediction error (ΔS) xy , as shown below: Among them, (Δσ) xy Represents the area of ​​the projection area, and γ is the angle between the normal vector and the z-axis.

6. The optical fiber-based cable temperature monitoring method according to claim 1, characterized in that: In the landmark model, when the standard cable is set to run, the query range of the data flow model composed of the data flow information generated is: from a known initial time point to the current time point, that is, {a s , …, a n }.

7. The optical fiber-based cable temperature monitoring method according to claim 1, characterized in that: The sliding window model is a data model established based on the temperature information measured by the current cable. In the data stream, the latest W data, its query range is {a ,max(c-w+1,0 , ..., a n }, as data continues to arrive, the data in the window continues to shift.

8. The optical fiber-based cable temperature monitoring method according to claim 1, characterized in that: By utilizing the backscattered signal and optical time domain reflection signal generated when light is transmitted in the optical fiber, the cable temperature measurement point is located and the temperature data of the corresponding temperature measurement point is obtained, as shown in the following formula: R(T)=I as / I s =(λ as / λ s ) 4 exp[-kcu0 / (kT)] (1) In the formula, I as I is the light intensity of the anti-bilateral curved surface; s is the light intensity of the double-sided curved surface; u as u is the frequency of the inverse double-sided curved surface; s is the double-sided surface spectrum frequency; u0 is the Rayleigh frequency; c is the speed of light; h is the Planck constant; k is the Boltzmann constant; T is the absolute temperature.

9. The optical fiber-based cable temperature monitoring method according to claim 1, characterized in that: By using the Haar wavelet function method, the temperature data obtained from the temperature measuring point is corrected when the prediction error exceeds the preset standard cable landmark model.

10. A system for implementing the optical fiber-based cable temperature monitoring method according to any one of claims 1 to 9, characterized in that: include: The data acquisition unit is configured to: utilize the backscattering signal and the optical time domain reflection signal generated when the light is transmitted in the optical fiber to realize the positioning of the temperature measuring point of the cable, and obtain the temperature data of the corresponding temperature measuring point, and the temperature data is stored in the form of electrical signals; The data correction unit is configured to: form the temperature of the cable surface using the temperature data of multiple temperature measurement points, form a geometric model using the temperature data of the cable surface, determine the rectangular coordinate system of the normal vector pointing to the cable surface under different sampling rates through the geometric hyperboloid vector method, and establish a data stream temperature sliding window model; The data correction unit is also configured to: compare the data structure of the data stream temperature sliding window model established with the measured cable through a pre-set standard cable landmark model, extract the error data on the cable surface, that is, the prediction error, use the projection area of ​​the prediction error on each coordinate plane in the rectangular coordinate system as the input of the wavelet transform, and obtain the corrected temperature data after denoising and reconstruction.

Images