An OTDR event detection method

Abstract

The application discloses an OTDR event detection method and relates to the technical field of data analysis. The method comprises the following steps: acquiring an OTDR response curve of a communication optical fiber to be detected and performing wavelet denoising processing to obtain wavelet coefficients decomposed to each layer; an improved simulated annealing algorithm is used to acquire optimal threshold values of the wavelet coefficients of each layer, and threshold value processing is performed on the wavelet coefficients of each layer, a signal is reconstructed, and a denoised OTDR response curve is obtained; a middle data section of the denoised response curve is extracted, suspected reflection events and non-suspected reflection events are found out in the middle data section, false points caused by noise in the suspected events are removed according to event interval distances and event duration distances, and finally, reflection event position information, non-reflection event position information and the attenuation loss of each event are determined. The method can obtain a response curve with a higher signal-to-noise ratio, intelligently acquires the sizes of each threshold value through the improved simulated annealing algorithm, and reduces the error probability of reported events.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data analysis, and particularly relates to an OTDR event detection method. BACKGROUND

[0002] With the rapid development of optical communication networks in China, the communication optical fiber can often reach dozens or even hundreds of kilometers, and the communication optical fiber not only exists on the land surface, but also often distributes underwater. If the optical fiber is bent or broken during communication, the communication quality will be affected or even communication failure. In order to ensure the quality of communication services, a device needs to be designed to detect the working condition of the communication optical fiber, and quickly detect and accurately locate the events that may occur on the optical fiber.

[0003] At present, the detection of communication optical fiber at home and abroad generally adopts an optical time domain reflectometer (OTDR). The OTDR is a device for detecting the working condition of the communication optical fiber in an optical communication network, and detects through the power of backscattered Rayleigh scattering light and Fresnel reflection light generated when light is transmitted in the optical fiber. The OTDR receives an optical power signal during the working process. According to the optical power signal, an OTDR response curve graph can be drawn. In the existing technical solution, the dynamic range size and the start-end blind area position are usually obtained by manually comparing the horizontal and vertical coordinates in the response curve graph. In addition, by analyzing the sudden rising or falling trend in the response curve graph, it can be judged whether there is an event in the communication optical fiber, and the occurrence position corresponding to each event on the horizontal coordinate can be found, and the event attenuation loss can be calculated.

[0004] In actual detection, the OTDR system is affected by noise. The accumulated average method is usually used for denoising processing of the OTDR response curve. The received data is stored and accumulated and averaged to reduce the noise level. Therefore, researchers need to spend a lot of time and effort on data acquisition and accumulated average to improve the signal-to-noise ratio of the response curve. The accuracy of the dynamic range, the start-end blind area and the event information accuracy of the OTDR system determined by manual visual inspection and slope method is not high, and events are easily missed or over-reported. Finally, the manual recording of the working condition also consumes a certain amount of time. SUMMARY

[0005] The present inventors propose an OTDR event detection method based on a wavelet transform algorithm and an improved simulated annealing algorithm to solve the above problems and technical needs.The wavelet transform algorithm is used to denoise OTDR response curves, calculate dynamic range and initial blind area, and find signal singular points and mutation points to calculate event location information and event attenuation loss.The improved simulated annealing algorithm is used to select a threshold during denoising and finding signal singular points and mutation points, and select an appropriate threshold according to different signal lengths and signal range information.The technical solution of the present application is as follows:

[0006] An OTDR event detection method includes the following steps:

[0007] Obtain the OTDR response curve of the communication optical fiber to be measured and perform wavelet denoising to obtain wavelet coefficients decomposed to each layer; the horizontal coordinate of the OTDR response curve is the measurement distance, and the vertical coordinate is the relative power intensity;

[0008] An improved simulated annealing algorithm is used to obtain the optimal threshold of each layer of wavelet coefficients;

[0009] According to the optimal threshold, the threshold of each layer of wavelet coefficients is processed, and the wavelet coefficients after threshold processing are used to reconstruct the signal to obtain the denoised OTDR response curve;

[0010] Extract the middle data segment of the denoised OTDR response curve, find out suspected reflection events and non-suspected reflection events in the middle data segment, and remove false positive points caused by noise according to event interval distance and event duration distance to finally determine reflection event location information, non-reflection event location information and attenuation loss of each event;

[0011] The event interval distance and the event duration distance are fixed values set according to the resolution space, and the resolution space is the propagation distance corresponding to one pulse width.

[0012] The beneficial technical effects of the present application are:

[0013] The method is suitable for detecting the working condition of an optical fiber in an optical communication network by using an OTDR system. The method can save the time for storing multiple received data, obtain an OTDR response curve with a higher signal-to-noise ratio, calculate the modulus maximum value of the curve by using a wavelet transform algorithm, obtain the position information of each singular point and abrupt point according to the modulus maximum value, further analyze the curve information, and obtain more detailed information such as the size of a dynamic range, the range of a blind area at a starting end, the position information of a reflection event / non-reflection event, and event attenuation loss. In the process of analyzing the response curve, the improved simulated annealing algorithm can be used to intelligently obtain the size of each optimal threshold value, the annealing function can be selected according to the change rate of the objective function, the disturbance range of the threshold value can be folded, the annealing temperature can be more effectively controlled, and the algorithm has the opportunity to jump out of a local optimal solution. The method is convenient and has high precision, greatly reduces the error probability of a reported event, provides strong technical support for detecting the working condition of a communication optical fiber, and improves the intelligent detection level of a communication optical fiber. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 FIG. 1 is a flowchart of an OTDR event detection method provided by the present application.

[0015] Figure 2 FIG. 2 is an OTDR response curve without denoising provided by the present application.

[0016] Figure 3 FIG. 3 is a flowchart of an improved simulated annealing algorithm for obtaining the optimal threshold value of a wavelet coefficient provided by the present application.

[0017] Figure 4 FIG. 4 is an OTDR response curve after wavelet denoising and removal of a tail-end noise part provided by the present application.

[0018] Figure 5 FIG. 5 is an algorithm code diagram for obtaining a modulus maximum value provided by the present application.

[0019] Figure 6 FIG. 6 is a modulus maximum value diagram of a useful signal provided by the present application.

[0020] Figure 7 FIG. 7 is an algorithm code diagram for positioning the starting position of a reflection event provided by the present application.

[0021] Figure 8 FIG. 8 is a slope diagram of a middle data segment after removal of a blind area at a starting end and a terminal position provided by the present application.

[0022] Figure 9 FIG. 9 is a document screenshot for automatically recording the working condition of a communication optical fiber to be detected provided by the present application. DETAILED DESCRIPTION

[0023] The specific embodiments of the present application will be further described in conjunction with the accompanying drawings.

[0024] As Figure 1 shown, the embodiment provides an OTDR event detection method, comprising the following steps:

[0025] Step 1: Obtain the OTDR response curve of the communication optical fiber to be detected, comprising:

[0026] Step 1.1: Detect the communication optical fiber to be detected to obtain the original received data, including the collection time (unit: μs) and the corresponding collection voltage (unit: v).

[0027] Since in the original received data, the data with the collection time less than zero does not contain the communication optical fiber signal and belongs to the OTDR system empty collection, the data of this time period can be regarded as the background noise part before the detection of the communication optical fiber, and step 1.2 is executed.

[0028] Step 1.2: Extract the data segment with the collection time less than zero, and perform average processing on the data segment to obtain the bias voltage.

[0029] Step 1.3: Identify the data segment with the collection time greater than zero as the signal generation interval, extract the data of this time period, convert the collection time into the measurement distance (unit: km), and take it as the horizontal coordinate of the OTDR response curve, convert the collection voltage into the relative power intensity (unit: dB), and take it as the vertical coordinate of the OTDR response curve, and obtain the OTDR response curve of the communication optical fiber to be detected as shown in Figure 2 . The expression of the relative power intensity is:

[0030]

[0031] Wherein, S represents the relative power intensity, V represents the collection voltage, V0 represents the bias voltage, and V1 represents the collection voltage of the first measurement signal.

[0032] Optionally, in step 1, it further comprises: defining other fixed values required by the method, such as the speed of light, the refractive index of the communication optical fiber, the pulse width of the optical signal, the propagation distance corresponding to one pulse width (hereinafter referred to as the resolution space), etc.

[0033] Step 2: Perform wavelet denoising processing on the OTDR response curve to obtain the wavelet coefficients decomposed to each layer.

[0034] After obtaining the OTDR measurement signal, wavelet denoising processing is needed to be performed on the signal, and selection of the wavelet base function and the decomposition layer number is particularly important. In the process of selecting the wavelet base function, the following factors need to be considered: symmetry, tight support orthogonality, vanishing moments and support width. For the OTDR system, a wavelet base function with good symmetry and tight support orthogonality is generally used. Good symmetry ensures that the singular points and abrupt points of the signal do not shift after wavelet transformation, and high tight support orthogonality reduces the complexity of the algorithm and improves the calculation speed. After the wavelet base function is determined, the decomposition layer number of the wavelet denoising needs to be determined. If the decomposition layer number is too small, the elimination of high-frequency noise is not thorough enough, and if the decomposition layer number is too large, some useful signals will also be removed as noise. In the OTDR system, 3 to 5 layers are generally selected.

[0035] The embodiment comprehensively considers the denoising effect, algorithm complexity and calculation speed of the wavelet denoising algorithm in the OTDR system, selects db4 wavelet in Daubechies, Symlets and Biorthogonal as the wavelet base function, and selects 3 layers as the signal decomposition layer number. The measurement signal with noise is decomposed into different scales to obtain the wavelet coefficients of each layer.

[0036] Step 3: Obtain the optimal threshold value of each layer of wavelet coefficients by using the improved simulated annealing algorithm.

[0037] At this time, the measurement signal with noise is decomposed into multiple layers, and the size of each layer threshold needs to be determined to process each layer of wavelet coefficients. The embodiment uses the improved simulated annealing algorithm to optimize the wavelet coefficient optimal threshold value. The method of optimizing the threshold value of each layer of wavelet coefficients is the same, as shown in FIG. 3, and specifically includes the following steps: Figure 3

[0038] Step 3.1: Set the initial parameters of the algorithm, including the initial temperature T0 and the termination temperature T k , and set the control parameters, including the number of cycles N S at each temperature step.

[0039] In the process of setting the algorithm parameters, the initial temperature T0 and the termination temperature T k should be set according to the actual situation. If the initial temperature T0 is too high or the termination temperature T k is too low, the algorithm program will waste unnecessary time. If the initial temperature T0 is too low or the termination temperature T k is too high, the algorithm program will be premature and fall into a local optimal solution. The number of cycles N S at each temperature step is generally set to 20 times.

[0040] ​Step 3.2: give the wavelet coefficients of the layer a threshold value ρ i (where i represents the layer number), and calculate the corresponding objective function f(ρ i ) according to the initial threshold value.

[0041] Step 3.3: set a suitable threshold generation function to perturb the threshold value in the neighborhood of the initial threshold value to generate a new threshold value ρ i ', and calculate the objective function f(ρ i ') of the new threshold value, and let n = n + 1.

[0042] Step 3.4: compare the increments Δf of the objective functions before and after to determine whether to accept the new threshold value, if Δf ≤ 0, then accept the new threshold value ρ i ', otherwise, according to the Metropolis criterion, determine whether to accept the new threshold value. Where Δf = f(ρ i ') - f(ρ i ).

[0043] Step 3.5: if the number of cycles n at the current temperature T reaches the set number of cycles N S , further determine whether the termination condition is met, if not, go to step 3.6, if yes, go to step 3.9; if the number of cycles n at the current temperature T does not reach the set number of cycles N S , then generate a new threshold value by perturbing the threshold value in the neighborhood again, i.e. return to step 3.3.

[0044] Step 3.6: if the current temperature T does not reach the set termination temperature T k , calculate the change rate ΔC of the objective function to select a suitable annealing function. Wherein the calculation formula of the change rate ΔC of the objective function is:

[0045]

[0046] The annealing function is generally a linear function T' = k1-T, where the slope k1 is adjusted according to the size of ΔC. That is, when ΔC is greater than a set value, the value of the slope k1 is adjusted to be greater than 1 to achieve heating; otherwise, the value of the slope k1 is adjusted to be less than 1 to achieve cooling. In this embodiment, the slope k1 is set to three optional fixed values: 0.80, 0.96 and 1.025, corresponding to rapid cooling, normal cooling and heating, respectively, and the size of ΔC is selected according to the size of ΔC, so that the annealing temperature can be more effectively controlled, and the algorithm has the opportunity to jump out of the local optimal solution.

[0047] Step 3.7: perform annealing operation according to the selected annealing function, and reset the value of n.

[0048] Step 3.8: Fold the perturbation range of the threshold value, i.e. reduce the upper and lower limits of the threshold value neighborhood range by a certain proportion, and then perform the operation of generating a new threshold value by perturbing in the threshold value neighborhood range again, i.e. return to step 3.3.

[0049] Adding this step can further determine the new threshold value, shorten the algorithm running time, and speed up the calculation. Optionally, the folding proportion of the perturbation range is set to 0.92 in the embodiment.

[0050] Step 3.9: If the current temperature T reaches the set termination temperature T k , output the optimal threshold value of the wavelet coefficient of the layer, and the algorithm ends.

[0051] Step 4: Perform threshold value processing on each layer of wavelet coefficients according to the optimal threshold value, and use the wavelet coefficients after threshold value processing to reconstruct the signal to obtain the OTDR response curve after denoising, remove the unnecessary noise interval at the tail end, and retain the measurement part to obtain the OTDR response curve as shown in Figure 4 .

[0052] Traditional threshold value judgment functions are divided into hard threshold value functions and soft threshold value functions, but the forced zeroing method in the hard threshold value function will cause additional oscillation of the signal and thus discontinuity, and the soft threshold value function is easy to lose signal details, thereby affecting the authenticity of the reconstructed signal. Based on this, the embodiment redefines a threshold value judgment function between the hard and soft threshold value functions, which reduces the deviation while ensuring continuity. Based on the improved threshold value judgment function, by comparing each layer of wavelet coefficients d and the optimal threshold value p of each layer of wavelet coefficients, the updated each layer of wavelet coefficients d' is obtained, and the expression is:

[0053]

[0054] Step 5: Extract the middle data segment of the OTDR response curve after denoising, find out the suspected reflection events and non-suspected reflection events in the middle data segment, and remove the misjudgment points caused by noise in the suspected events according to the event interval distance and event duration distance, and finally determine the reflection event position information, non-reflection event position information and the attenuation loss of each event.

[0055] Among them, the extraction of the middle data segment of the OTDR response curve after denoising includes the following steps:

[0056] Step 5.1: Perform multi-segment least squares fitting on the beginning data segment of the extracted OTDR response curve after denoising to locate the beginning blind area range of the measured communication optical fiber measurement section, including:

[0057] The initial end data in the OTDR response curve after denoising is acquired, the optical fiber initial end data point is j, the total data length is L, one least square fitting is performed every w data points, it is assumed that w=50, the fitting data segment corresponding to the data points is (j, j+49), a total of L-49 data segments are obtained, such as the fitting data segment includes: (1, 50), (2, 51), (3, 52)…, the least square fitting is performed on each data segment, the basic principle of the fitting method is to minimize the variance. Taking the first fitting data segment as an example, the data segment contains 50 data {(t1, s1), (t2, s2), …, (t 50 ,s 50 )}, wherein t and s respectively represent the measured distance and the relative power intensity of the measured signal, the OTDR measurement segment signal is assumed to be a linear function form s=k2t+b, then the least square fitting is to solve the optimization problem:

[0058]

[0059] The slope k2 of the fitting data segment is needed to locate the range of the initial end blind area, the derivative of formula (3) is used to obtain the calculation formula of the slope k2 of each fitting data segment as follows:

[0060]

[0061] Wherein, t j represents the measured distance of the jth data point in a fitting data segment, s j represents the relative power intensity of the jth data point in a fitting data segment.

[0062] The slope k2 of all fitting data segments is obtained according to formula (4), considering that the initial end blind area contains the rising edge, in order to prevent misjudgment, the method for locating the end point of the initial end blind area includes: if the slope of the previous fitting data segment and the slope of the next fitting data segment of the current fitting data segment are both less than zero, and the slope of the next fitting data segment and the theoretical optical fiber attenuation rate (-0.21) differ by not more than a preset value (such as 0.02), then the horizontal coordinate of the end data point of the current fitting data segment is selected as the initial end blind area end point of the measured communication optical fiber measurement segment.

[0063] Step 5.2: Extract the end data segment in the OTDR response curve after denoising, and set the first threshold according to the modulus maximum value corresponding to the end data segment, and consider the horizontal coordinate of the first data point exceeding the first threshold in the end data segment as the end position of the measured communication optical fiber measurement segment.

[0064] Specifically, after the range of the blind area at the start end of the optical fiber measurement section is located, the end position of the optical fiber measurement is located. Here, the db1 wavelet base function in Daubechies, that is, the haar wavelet, is used to perform one layer of decomposition on the denoised signal and directly reconstruct the high-frequency coefficients, with the purpose of retaining only the information of the event point and ignoring the normal attenuation that does not need to be concerned. In the field of signal processing, the singular point and the mutation point of a signal represent the discontinuity or turning point of the signal at the point. In general, the singular point and the mutation point of a signal carry a large amount of important information of the signal and present the local characteristics of the signal. In the wavelet transform algorithm, the modulus maximum point corresponds to the singular point and the mutation point of the signal, and therefore the modulus maximum value can be used as a means to detect the singularity of the signal. The modulus maximum point can be described by the following principle: for t belonging to a certain field of t0, if Wf(t,s) satisfies |Wf(t,s)|≤|Wf(t0,s)| and there exists |Wf(t,s)|<|Wf(t0,s)| in the left and right neighborhoods of t0, then (t0,s) is called the modulus maximum point at the scale s, |Wf(t0,s)| is the modulus maximum value corresponding to the point (t0,s), and the algorithm code is as shown in Figure 5 The modulus maximum value method is used to analyze the reconstructed signal, and the signal and the noise are separated by the signal singularity, and the modulus maximum value of the useful signal is retained, as shown in Figure 6 Then, the last 3 kilometers of the optical fiber measurement section signal (that is, the end data section in the denoised OTDR response curve) is intercepted as the end space signal, 1 / 2 of the maximum value of the modulus maximum value corresponding to the end space signal is selected as the first threshold value, and the first data point whose horizontal coordinate exceeds the first threshold value is considered as the end position of the communication optical fiber measurement section to be measured.

[0065] Step 5.3: After the blind area at the start end of the optical fiber and the end position are determined, the data section between the blind area at the start end and the end position is intercepted as the middle data section of the denoised OTDR response curve, which is used to locate the event position information and calculate the event loss.

[0066] Among them, the suspected reflection events are found in the middle data section, and the misjudgment points caused by noise in the suspected events are removed according to the event interval distance and the event duration distance, and finally the reflection event position information is determined, including the following steps:

[0067] Step 5.4: A second threshold value is set according to the modulus maximum value corresponding to the middle data section, and the data point whose horizontal coordinate exceeds the second threshold value in the middle data section is initially determined as the suspected reflection event start position. Optionally, the second threshold value set in this embodiment is 1 / 3 of the maximum value of the modulus maximum value corresponding to the middle data section.

[0068] Step 5.5: Starting from the first suspected reflection event start position, the distance between two adjacent suspected reflection event start positions is determined in sequence, and the relationship between the distance and the set event interval distance is determined. If the distance is greater than the event interval distance, the two adjacent suspected reflection events are considered to be different suspected reflection events, and the two suspected reflection event start positions are retained. Otherwise, the suspected reflection event start position of the earlier suspected reflection event is taken as the start position of the same suspected reflection event.

[0069] Step 5.6: After the first screening, the first suspected reflection event start position is taken as the starting point again, and the duration of each suspected reflection event is determined in sequence, and the relationship between the duration and the set event duration is determined. If the duration is less than the event duration, the suspected reflection event start position is considered to be a misjudgment point, and it is removed. The algorithm code of step 5.5 and step 5.6 is shown in Figure 7 . Optionally, the event interval distance and the event duration set in the embodiment are fixed values set according to the resolution space, wherein the event interval distance is 1.2 times the resolution space, and the event duration is 0.8 times the resolution space.

[0070] Step 5.7: After the second screening, the start position of each reflection event is obtained, the end position of the reflection event is determined according to the start position of each reflection event, and the attenuation loss of each event is calculated according to the relative power intensity corresponding to the start position and the relative power intensity corresponding to the end position.

[0071] Among them, the suspected non-reflection event is found in the intermediate data segment, and the misjudgment points caused by noise in the suspected event are removed according to the event interval distance and the event duration, and the non-reflection event position information is finally determined, including the following steps:

[0072] Step 5.8: The data in the intermediate data segment is subjected to multi-segment least square fitting to obtain the slope of all fitting data segments, as shown in Figure 8 The implementation steps of multi-segment least square fitting are the same as step 5.1, and will not be repeated here. When the slope of the fitting data segment satisfies less than the third threshold value and greater than the fourth threshold value, the end data abscissa of the fitting data segment is initially determined as the suspected non-reflection event start position. Among them, the third threshold value is set to 1.15 times the average slope k0 of all fitting data segments, and the fourth threshold value is set to a theoretical fiber attenuation within one resolution space.

[0073] Similarly, the non-reflection event start position is determined according to the event interval distance and the event duration, which is the same as the two screening steps for determining the reflection event start position. For specific content, please refer to step 5.5 and step 5.6, which will not be repeated here.

[0074] Step 5.9: After the secondary screening, the start position of each non-reflection event is obtained, the end position of each non-reflection event is determined according to the start position of each non-reflection event, and the attenuation loss of each event is calculated according to the relative power intensity corresponding to the start position and the relative power intensity corresponding to the end position.

[0075] In the above steps 5.7 and 5.9, the method for determining the end position of a reflection event or a non-reflection event is the same, and the following is specifically included for an event:

[0076] The position obtained by adding a resolution space to the start position of the event is defined as the preliminary event end position. Starting from the preliminary event end position, the data points are fitted by multiple least squares to obtain the slope of each fitting data segment, that is, least squares fitting is performed every 50 data points, and the fitting method is the same as that in step 5.1, which will not be repeated here. When the slope differs from the theoretical fiber attenuation rate (-0.21) by no more than a preset value (such as 0.02), the horizontal coordinate of the end data point of the fitting data segment is selected as the end position of the event.

[0077] Optionally, the OTDR system dynamic range (i.e., the total fiber attenuation) is calculated according to the relative power intensity corresponding to the fiber start end blind area and the relative power intensity corresponding to the fiber end position.

[0078] Optionally, the detection method further includes step 6: combining the information obtained by the above methods, finally merging the event misjudgment, and outputting a document, at this time the document automatically records the OTDR system dynamic range, the start end blind area range, the fiber end position, the positions of various events and the attenuation loss, as shown in Figure 9

[0079] After generating the document, researchers can directly download the document, reducing the workload of manual discrimination and recording; reducing the missed or multiple events generated by the prior art, that is, the event detection method provided in the embodiment is more efficient and intelligent.

[0080] The above is only a preferred embodiment of the present application, and the present application is not limited to the above embodiments. It can be understood that other improvements and changes directly derived or thought of by those skilled in the art without departing from the spirit and concept of the present application should be considered to be included in the protection scope of the present application.​

Claims

1. A method of OTDR event detection, characterized by, The method comprises: Obtaining an OTDR response curve of a communication optical fiber to be measured and performing wavelet denoising processing to obtain wavelet coefficients decomposed to each layer; wherein the horizontal coordinate of the OTDR response curve is a measurement distance, and the vertical coordinate is a relative power intensity; Obtaining an optimal threshold value of each layer wavelet coefficient by using an improved simulated annealing algorithm; Performing threshold processing on each layer wavelet coefficient according to the optimal threshold value, and using the wavelet coefficient after threshold processing to reconstruct a signal to obtain an OTDR response curve after denoising; Extracting a middle data segment of the OTDR response curve after denoising, finding out suspected reflection events and non-suspected reflection events in the middle data segment, and removing misjudgment points caused by noise in the suspected events according to an event interval distance and an event duration distance to finally determine reflection event position information, non-reflection event position information and the attenuation loss of each event; The event interval distance and the event duration distance are fixed values set according to a resolution space, and the resolution space is a propagation distance corresponding to one pulse width; The method of obtaining the optimal threshold value of each layer wavelet coefficient by using the improved simulated annealing algorithm comprises, for each layer wavelet coefficient: Giving each layer wavelet coefficient a threshold value a certain initial value and a corresponding objective function; Setting a suitable threshold value generating function to generate a new threshold value by perturbation in the threshold value initial value neighborhood range, and calculating the objective function of the new threshold value; Comparing the increments of the objective functions before and after to determine whether to accept the new threshold value; If the current temperature T If the number of cycles has reached the set number of cycles, an annealing operation is performed, otherwise the operation of perturbing the threshold value in the threshold value neighborhood range to generate a new threshold value is performed again. After annealing a certain number of times, if the current temperature T reaches the set termination temperature, output the optimal threshold value of the layer wavelet coefficient, otherwise re-perform the annealing operation; Before each execution of the annealing operation, calculating the change rate of the objective function to select a suitable annealing function; after each annealing operation, reducing the upper and lower limits of the threshold value neighborhood range by a certain proportion; wherein the change rate of the objective function is the ratio of the increments of the objective functions before and after to the objective function of the previous time.

2. The OTDR event detection method of claim 1, wherein, The method for obtaining the OTDR response curve of the communication optical fiber to be measured comprises: Detecting the communication optical fiber to be measured to obtain original received data, including collection time and corresponding collection voltage; Extracting a data segment with collection time less than zero, and performing average processing on the data segment to obtain a bias voltage; Extracting a data segment with collection time greater than zero, and converting the collection time into a measurement distance and converting the collection voltage into a relative power intensity, with the expression being: ； wherein denotes the relative power intensity, denotes the acquisition voltage, denotes the bias voltage, denotes the acquisition voltage of the first measurement signal.

3. The OTDR event detection method of claim 1, wherein, The method for selecting a suitable annealing function according to the change rate of the objective function comprises: The annealing function is defined as a linear function: where the slope is adjusted according to the magnitude of the change rate of the objective function; when the change rate is greater than a set value, the slope is adjusted to be greater than 1 to achieve temperature rise; otherwise, the slope is adjusted to be less than 1 to achieve temperature drop.

4. The OTDR event detection method of claim 1, wherein, The method for performing threshold processing on each layer wavelet coefficient according to the optimal threshold value comprises: Based on the improved threshold judgment function, by comparing each layer wavelet coefficient with the optimal threshold of each layer wavelet coefficient , the updated wavelet coefficient of each layer is obtained , the expression is: 。 5. The OTDR event detection method of claim 1, wherein, The method for extracting the middle data segment of the OTDR response curve after denoising comprises: Performing multi-segment least square fitting on the initial end data segment of the extracted OTDR response curve after denoising to locate the initial end blind area range of the measurement section of the communication optical fiber to be measured; Extracting the terminal data segment of the OTDR response curve after denoising, and setting a first threshold value according to the modulus maximum value corresponding to the terminal data segment, and regarding the horizontal coordinate of the first data point exceeding the first threshold value in the terminal data segment as the terminal position of the measurement section of the communication optical fiber to be measured; Taking the data segment between the initial end blind area and the terminal position as the middle data segment of the OTDR response curve after denoising.

6. The OTDR event detection method of claim 1, wherein, The method for finding out suspected reflection events in the intermediate data section and removing false points caused by noise in the suspected events according to event interval distance and event duration distance to finally determine the position information of the reflection events comprises: Setting a second threshold according to the modulus maximum value corresponding to the intermediate data section, and initially determining the abscissa of the data point exceeding the second threshold in the intermediate data section as the start position of the suspected reflection event; Taking the start position of the first suspected reflection event as the starting point, judging the distance between the start positions of two adjacent suspected reflection events and the set event interval distance in sequence, if the distance is greater than the event interval distance, considering that the two adjacent suspected reflection events are different suspected reflection events; otherwise, taking the start position of the suspected reflection event appearing earlier as the start position of the same suspected reflection event; After the first screening, taking the start position of the first suspected reflection event as the starting point again, judging the duration distance of each suspected reflection event and the set event duration distance in sequence, if the duration distance is less than the event duration distance, considering that the start position of the suspected reflection event is a false point, and removing it; After the second screening, obtaining the start positions of various reflection events, and determining the end position of each reflection event according to the start position of the reflection event.

7. The OTDR event detection method of claim 1, wherein, The method for finding out suspected non-reflection events in the intermediate data section and removing false points caused by noise in the suspected events according to event interval distance and event duration distance to finally determine the position information of the non-reflection events comprises: Performing multi-segment least square fitting on the data in the intermediate data section, and initially determining the abscissa of the end data of the fitting data section as the start position of the suspected non-reflection event when the slope of the fitting data section satisfies less than a third threshold and greater than a fourth threshold; wherein the third threshold is set according to the average slope of all fitting data sections, and the theoretical fiber attenuation in a resolution space is set as the fourth threshold; Taking the start position of the first suspected non-reflection event as the starting point, judging the distance between the start positions of two adjacent suspected non-reflection events and the set event interval distance in sequence, if the distance is greater than the event interval distance, considering that the two adjacent suspected non-reflection events are different suspected non-reflection events; otherwise, taking the start position of the suspected non-reflection event appearing earlier as the start position of the same suspected non-reflection event; After the first screening, taking the start position of the first suspected non-reflection event as the starting point again, judging the duration distance of each suspected non-reflection event and the set event duration distance in sequence, if the duration distance is less than the event duration distance, considering that the start position of the suspected non-reflection event is a false point, and removing it; After the second screening, obtaining the start positions of various non-reflection events, and determining the end position of each non-reflection event according to the start position of the non-reflection event.

8. The OTDR event detection method of claim 5, wherein, The method for positioning the start blind area range of the measured communication fiber section by performing multi-segment least square fitting on the start data section in the extracted denoised OTDR response curve comprises: Obtain the initial data from the denoised OTDR response curve, and set each w Perform a least-squares fit on each data point to obtain the slope of each fitted data segment. The expression is: ； in, t j Represents the first element in a fitted data segment. j The measured distance of each data point s j Represents the first element in a fitted data segment. j The relative power intensity of each data point; If the slope of the previous fitting data segment and the slope of the next fitting data segment of the current fitting data segment are both less than zero, and the slope of the next fitting data segment differs from the theoretical fiber attenuation rate by not more than a preset value, then the horizontal coordinate of the end data point of the current fitting data segment is selected as the end point of the blind area of the start end of the measured communication fiber measuring segment.

9. The OTDR event detection method of claim 6 or 7, wherein, The method for determining the end position of the event comprises, for each event: defining the position after the start position of the event plus a resolution space as the initial end position of the event; starting from the initial end position of the event, performing multi-segment least square fitting on the data points to obtain the slope of each fitting data segment; when the slope differs from the theoretical fiber attenuation rate by not more than a preset value, then the horizontal coordinate of the end data point of the fitting data segment is selected as the end position of the event.

Images