CN114692693A - Distributed optical fiber signal identification method, device and storage medium based on fractal theory - Google Patents

Distributed optical fiber signal identification method, device and storage medium based on fractal theory Download PDF

Info

Publication number
CN114692693A
CN114692693A CN202210354540.4A CN202210354540A CN114692693A CN 114692693 A CN114692693 A CN 114692693A CN 202210354540 A CN202210354540 A CN 202210354540A CN 114692693 A CN114692693 A CN 114692693A
Authority
CN
China
Prior art keywords
optical fiber
distributed optical
fractal
signal
signal identification
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210354540.4A
Other languages
Chinese (zh)
Inventor
吴雪竹
李珣
宓雪沁
应昶昊
张乐
陈佳雷
杨旭光
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hangzhou Dinglan Technology Co ltd
Original Assignee
Hangzhou Dinglan Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hangzhou Dinglan Technology Co ltd filed Critical Hangzhou Dinglan Technology Co ltd
Priority to CN202210354540.4A priority Critical patent/CN114692693A/en
Publication of CN114692693A publication Critical patent/CN114692693A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/08—Feature extraction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12—Classification; Matching

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Signal Processing (AREA)
  • Complex Calculations (AREA)

Abstract

The invention provides a fractal theory-based distributed optical fiber signal identification method, a fractal theory-based distributed optical fiber signal identification device and a storage medium, and belongs to the technical field of optical fiber signal identification. The method solves the problem that the accuracy is not credible due to extreme unbalance of the class quantity of the data sample of the existing distributed optical fiber. The distributed optical fiber signal identification method based on the fractal theory comprises the following steps: step S1: obtaining global characteristics of distributed optical fiber time sequence signal data; step S2: obtaining local characteristics of distributed optical fiber time sequence signal data; step S3: the model is trained using the parameters of the global and local features obtained in steps S1 and S2, and a prediction class is output. The invention has the advantages of utilizing the overall characteristics of the single fractal dimension reaction signal and the local characteristics of the multi-fractal spectrum reaction signal to obtain more comprehensive and accurate characteristic information, reducing the false alarm rate caused by the high sensitivity of the optical fiber signal and overcoming the singleness of signal description.

Description

Distributed optical fiber signal identification method, device and storage medium based on fractal theory
Technical Field
The invention belongs to the technical field of optical fiber signal identification, and particularly relates to a fractal theory-based distributed optical fiber signal identification method, a fractal theory-based distributed optical fiber signal identification device and a storage medium.
Background
The distributed optical fiber sensing system has the characteristics of high sensitivity and real-time performance, and physical parameters such as temperature, acoustics, vibration and the like can be measured in real time along the whole length of the optical fiber by utilizing the optical fiber. The vibration signal has the characteristics of nonlinearity, non-stationarity, complexity and the like.
Most of the existing optical fiber signal feature extraction methods are based on frequency domain features and time domain features. The frequency domain characteristics are generally realized by a Fourier transform method, and the method has certain limitation and is easily influenced by disturbance signals such as noise and the like, so that the false alarm rate of an intrusion signal is high; although the time domain features are simple to calculate, a threshold needs to be manually set, and the size of the threshold determines the identification accuracy.
Fractal theory was first proposed in 1975 by the mathematician Benoit b. The theory is not mature as a new concept and method in all aspects, and most of the application is in the fields of image processing, financial signals and biomedical signals. At present, in the application of fractal theory in the extraction of the distributed optical fiber signal characteristics, the existing method is only based on multi-fractal spectral parameters. However, the method has low applicability and cannot distinguish more interference signal types and intrusion signal types; and the recognition rate is low because the feature vectors are too few.
Methods for calculating the multi-fractal spectrum are various, and a partition function method, a multi-fractal detrended fluctuation analysis Method (MFDFA) and a wavelet film maximum method (WTMM) are common. However, the partition function method does not give correct results when dealing with non-stationary time series that are affected by the trend term or cannot be normalized; the WTMM algorithm is more complex to calculate than the MFFA algorithm; MFDFAs are not only computationally simple, but also have advantages in computing and processing short time series for negative q-moments.
Fractal techniques can be divided into two categories, single fractal and multi-fractal. Fractal dimension is the most important parameter in single fractal theory, and common one-dimensional fractal dimension calculation methods include a box method and a Hausdorff method (Hausdorff). Although the box method is simple in calculation, a high signal identification rate can be obtained only when the side length is small enough, but the calculation time is long; the Hausdorff method is the most standard calculation method, but has high calculation complexity and is difficult to implement in practice, so that the Hausdorff dimension is the most common method for obtaining the Hausdorff dimension by using the Higuchi algorithm.
In machine learning modeling, data is divided into a training set and a test set, a common method does not consider whether the class quantity proportion of the training set and the test set is consistent or not, and model performance is judged by using model accuracy. This approach can lead to unreliable accuracy due to the extreme imbalance of the data sample class size of the distributed fiber.
Disclosure of Invention
The invention aims to solve the problems in the prior art, adjust the classification method, ensure the consistent proportion of the classification quantity of samples in a training set and a test set, reduce the error caused by the unbalanced sample based on the roc-auc condition when judging the performance of the model, and finally give the final prediction category by utilizing the classification probability returned by various models, thereby achieving the purpose of improving the identification accuracy rate and providing a distributed optical fiber signal identification method, a device and a storage medium based on a fractal theory.
The first object of the present invention can be achieved by the following technical solutions: a fractal theory-based distributed optical fiber signal identification method is characterized by comprising the following steps:
step S1: obtaining global characteristics of distributed optical fiber time sequence signal data;
step S2: obtaining local characteristics of distributed optical fiber time sequence signal data;
step S3: the model is trained using the parameters of the global and local features obtained in steps S1 and S2, and a prediction class is output.
The invention utilizes the integral characteristic of the single fractal dimension reaction signal and the local characteristic of the multi-fractal spectrum reaction signal to obtain more comprehensive and accurate characteristic information, thereby realizing the reduction of false alarm rate caused by high sensitivity of the optical fiber signal and overcoming the singleness of signal description.
In the foregoing method for identifying a distributed optical fiber signal based on a fractal theory, the step S1 includes the following steps:
-step s 1.a.: defining a maximum value t of adjacent distributed optical fiber time sequence intervals, and reconstructing original distributed optical fiber time sequence data s (1), s (2),.. multidot.s (N) as
Figure BDA0003582309820000031
Calculate each
Figure BDA0003582309820000032
Obtaining the total average length l (i) of the time sequence;
step s1. b: and obtaining a relation expression related to the fractal dimension through logarithmic transformation, and calculating the slope to obtain the global characteristic Hausdorff fractal dimension D of different signals of the distributed optical fiber.
In the foregoing method for identifying a distributed optical fiber signal based on a fractal theory, the step S2 includes the following steps:
step s2. a: reconstructing the sequence s (1), s (2),.. s, (N) as y (l), l ═ 1, 2.. N;
step s2. b: defining the value of s, dividing the reconstructed sequence y (l) into segments which are not intersected with each other and have the length of s from the beginning to the end, and dividing the reconstructed sequence y (l) into segments which are not intersected with each other and have the length of s from the end to obtain the total number of 2NsA line segment;
-step s2. c: defining m value, calculating 2N by least square fittingsThe m-order polynomial of each line segment and the label l in the line segment correspond to residual errors;
step s2. d: calculating a local root mean square value according to the size condition of v;
step s2. e: defining the size range of q, thereby obtaining a global q-order fluctuation function Fq(s)；
-step s2. f: repeating the steps from the step S2.b to the step S2.e, calculating the fluctuation functions corresponding to different s, and obtaining s and Fq(s);
step s2. g: by s and Fq(s) obtaining a relation of a linear regression slope h (q) and obtaining a quality parameter value by utilizing the relation of h (q) and the quality index tau (q); using Legendre transformation to obtain a holder index alpha and a multi-fractal spectrum f (alpha);
step s2. h: and calculating the spectrum width delta alpha, the fractal dimension difference delta f and the inclination rate r of the local characteristic parameters of different signal categories of the distributed optical fiber based on the holder index alpha and the multi-fractal spectrum f (alpha) obtained in the step S2. g.
In the foregoing method for identifying a distributed optical fiber signal based on a fractal theory, the step S3 includes the following steps:
step S3. a: combining the local features and the global features of all the signal classes into a feature vector, and defining the feature vector as X; defining the category corresponding to each signal data as a label, representing by y, and dividing the characteristic vector and the label into a training set and a test set with the same signal category proportion;
step S3. b: the training set is brought into an XGboost model, the best parameter value of the model is found according to roc-auc conditions of the test set by using grid search of a five-fold cross-validation method, signal data needing to be predicted is brought into the trained model, the prediction class probability of each signal is output,
step S3.c: repeating the step S3.b by using an SVC model;
step S3. d: and (4) calculating the average value of the prediction class probabilities output in the step S3.b and the step S3.c, wherein the signal class with the larger probability average value is obtained.
In the fractal theory-based distributed optical fiber signal identification method, the step s1.a includes the following steps:
step S1. a.1: the time sequence is reconstructed from time sequences s (1), s (2),. gtoreq.s (N), where N is the total length of data, and the interval between two adjacent time sequences is t, i ═ 1, 2,. gtoreq.t:
Figure BDA0003582309820000041
step S1. a.2: for each one
Figure BDA0003582309820000042
The average length is:
Figure BDA0003582309820000043
step S1. a.3: the total average length of the time series was obtained as:
Figure BDA0003582309820000044
in the fractal theory-based distributed optical fiber signal identification method, the step s1.b includes the following steps:
step S1. b.1: through logarithmic transformation, the relation can be obtained, and the slope D is the fractal dimension:
Figure BDA0003582309820000051
in the fractal theory-based distributed optical fiber signal identification method, in step s2.a, the calculation formula of y (l) is as follows:
Figure BDA0003582309820000052
in step S2.b, the calculation formula of s is as follows:
Figure BDA0003582309820000053
in step s2.c, l ═ 1 (v-1) s + i, i ═ 1, 2.., s;
Figure BDA0003582309820000054
step S2. d: the calculation formula of the local root mean square value is as follows:
Figure BDA0003582309820000055
Figure BDA0003582309820000056
step S2. e: the calculation formula of the q-order fluctuation function is as follows:
Figure BDA0003582309820000057
Figure BDA0003582309820000058
step S2. f: obtaining s and Fq(s) is as follows:
Fq(s)∝sh(q)；
step S2. g: the slope h (q) of the linear regression is calculated by the formula:
Figure BDA0003582309820000061
the quality parameter value of the quality index tau (q) is calculated by the formula:
τ(q)＝qh(q)-1；
the calculation formula of the holder index α is:
Figure BDA0003582309820000062
the calculation formula of the multi-fractal spectrum f (alpha) is as follows:
f(α)＝qα-τ(q)；
step S2. h: the formula for calculating the spectral width Δ α is:
Δα＝αmax-αmin；
the fractal difference Δ f is calculated by the following formula:
Δf＝fmax-fmin；
the formula for calculating the rate of inclination r is:
r＝(αmax-α0)/Δα。
the second object of the present invention can be achieved by the following technical solutions: a fractal theory based distributed optical fiber signal identification apparatus, comprising an input terminal, an output terminal, one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs comprising instructions for performing a fractal theory based distributed optical fiber signal identification method as described above.
The third object of the present invention can be achieved by the following technical solutions: a computer-readable storage medium storing a computer program for use with an input terminal and an output terminal, wherein the computer program is executable by a processor to perform a fractal theory based distributed optical fiber signal identification method as described above.
Compared with the prior art, the method adjusts the classification method, ensures the consistent proportion of the sample classification quantities of the training set and the test set, reduces the error caused by the unbalanced sample on the basis of the roc-auc condition when judging the performance of the model, and finally gives the final prediction category by utilizing the classification probability returned by various models, thereby having the advantage of improving the identification accuracy.
Drawings
FIG. 1 is a block diagram of the present invention.
FIG. 2 is a design flow diagram of the present invention.
Detailed Description
The following are specific embodiments of the present invention and are further described with reference to the drawings, but the present invention is not limited to these embodiments.
The first embodiment is as follows:
a fractal theory-based distributed optical fiber signal identification method is characterized by comprising the following steps:
step S1: obtaining global characteristics of distributed optical fiber time sequence signal data;
-step s 1.a.: defining a maximum value t of adjacent distributed optical fiber time sequence intervals, and reconstructing original distributed optical fiber time sequence data s (1), s (2),.. multidot.s (N) as
Figure BDA0003582309820000071
Calculate each of
Figure BDA0003582309820000072
Obtaining a total average length l (i) of the time sequence;
step s1. a.1: the time sequence is reconstructed from time sequences s (1), s (2),. so., s (N), where N is the total length of data, and the interval between two adjacent time sequences is t, i ═ 1, 2,. so., t:
Figure BDA0003582309820000073
step s1. a.2: for each one
Figure BDA0003582309820000074
The average length is:
Figure BDA0003582309820000075
step s1. a.3: the total average length of the time series was obtained as:
Figure BDA0003582309820000076
step s1. b: obtaining a relational expression related to fractal dimension through logarithmic transformation, and calculating a slope to obtain a global characteristic Hausdorff fractal dimension D of different signals of the distributed optical fiber;
step s1. b.1: through logarithmic transformation, the relation can be obtained, and the slope D is the fractal dimension:
Figure BDA0003582309820000081
step S2: obtaining local characteristics of distributed optical fiber time sequence signal data;
step s2. a: the reconstruction sequence s (1), s (2),. so., s (N) is y (l), and the calculation formula of l 1, 2.. so, N, y (l) is:
Figure BDA0003582309820000082
step s2. b: defining the value of s, dividing the reconstructed sequence y (l) into segments which are not intersected with each other and have the length of s from the beginning to the end, and dividing the reconstructed sequence y (l) into segments which are not intersected with each other and have the length of s from the end to obtain the total number of 2NsThe calculation formula of the line segment s is as follows:
Figure BDA0003582309820000083
-step s2. c: defining m value, calculating 2N by least square fittingsThe m-th order polynomial of each line segment and the label l in the line segment correspond to the residual error,
l＝(v-1)s+i，i＝1，2，...，s；
Figure BDA0003582309820000084
step s2. d: and calculating a local root mean square value according to the magnitude condition of v, wherein the calculation formula of the local root mean square value is as follows:
Figure BDA0003582309820000085
Figure BDA0003582309820000086
step s2. e: defining the size range of q, thereby obtaining a global q-order fluctuation function Fq(s), the calculation formula of the q-order fluctuation function is as follows:
Figure BDA0003582309820000091
Figure BDA0003582309820000092
-step s2. f: repeating the steps from the step S2.b to the step S2.e, calculating the fluctuation functions corresponding to different s, and obtaining s and Fq(s) obtaining s and Fq(s) is as follows:
Fq(s)∝sh(q)；
step s2. g: by s and Fq(s) obtaining a relation of a linear regression slope h (q) and obtaining a quality parameter value by utilizing the relation of h (q) and the quality index tau (q); using Legendre transformation to obtain a holder index alpha and a multi-fractal spectrum f (alpha);
the slope h (q) of the linear regression is calculated by the formula:
Figure BDA0003582309820000093
the mass parameter value of the holder mass index tau (q) is calculated by the formula:
τ(q)＝qh(q)-1；
the formula for the index α is:
Figure BDA0003582309820000094
the calculation formula of the multi-fractal spectrum f (alpha) is as follows:
f(α)＝qα-τ(q)；
step s2. h: based on the holder index alpha and the multi-fractal spectrum f (alpha) obtained in the step S2.g, calculating the spectrum width delta alpha, the fractal dimension difference delta f and the inclination rate r of local characteristic parameters of different signal categories of the distributed optical fiber:
the formula for calculating the spectral width Δ α is:
Δα＝αmax-αmin；
the fractal difference Δ f is calculated by the following formula:
Δf＝fmax-fmin；
the formula for calculating the rate of inclination r is:
r＝(αmax-α0)/Δα；
step S3: training a model by using the parameters of the global features and the local features obtained in the steps S1 and S2, and outputting prediction categories;
-step s3. a: combining the local features and the global features of all the signal classes into a feature vector, and defining the feature vector as X; defining the category corresponding to each signal data as a label, expressing the label by y, and dividing the characteristic vector and the label into a training set and a test set with the same signal category proportion;
step s3. b: the training set is brought into an XGboost model, the best parameter value of the model is found according to roc-auc conditions of the test set by using grid search of a five-fold cross-validation method, signal data needing to be predicted is brought into the trained model, the prediction class probability of each signal is output,
-step s3. c: repeating the step S3.b by using an SVC model;
step s3. d: and (4) calculating the average value of the prediction class probabilities output in the step S3.b and the step S3.c, wherein the signal class with the larger probability average value is obtained.
The invention relates to a fractal theory-based XGboost and SVM mixed model machine learning method, which is used for realizing the class prediction of signal data acquired by a distributed optical fiber in the next time period.
The prediction categories are intrusion signals and interference signals.
According to the invention, the identification of the interference signal and the intrusion signal is realized by processing the vibration parameter, the time sequence signal category of the next time period is predicted, and further the real-time early warning of the intrusion signal is provided.
The second embodiment:
a fractal theory based distributed optical fiber signal identification apparatus, comprising an input terminal, an output terminal, one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs comprising instructions for performing a fractal theory based distributed optical fiber signal identification method as described above.
Example three:
a computer-readable storage medium storing a computer program for use with an input terminal and an output terminal, wherein the computer program is executable by a processor to perform a fractal theory based distributed optical fiber signal identification method as described above.
The specific embodiments described herein are merely illustrative of the spirit of the invention. Various modifications or additions may be made to the described embodiments or alternatives may be employed by those skilled in the art without departing from the spirit or ambit of the invention as defined in the appended claims.
Although a large number of terms are used here more, the possibility of using other terms is not excluded. These terms are used merely to more conveniently describe and explain the nature of the present invention; they are to be construed as being without limitation to any additional limitations that may be imposed by the spirit of the present invention.

Claims (9)

1.A fractal theory-based distributed optical fiber signal identification method is characterized by comprising the following steps:
step S1: obtaining global characteristics of distributed optical fiber time sequence signal data;
step S2: obtaining local characteristics of distributed optical fiber time sequence signal data;
step S3: the model is trained using the parameters of the global and local features obtained in steps S1 and S2, and a prediction class is output.
2. The fractal theory-based distributed optical fiber signal identification method according to claim 1, wherein the step S1 includes the following steps:
-step s 1.a.: defining a maximum value t of adjacent distributed optical fiber time sequence intervals, and reconstructing original distributed optical fiber time sequence data s (1), s (2),. multidot.s (N) as
Figure FDA0003582309810000012
1, 2,. t; calculate each
Figure FDA0003582309810000011
Obtaining a total average length l (i) of the time sequence;
step s1. b: and obtaining a relation of related fractal dimensions through logarithmic transformation, and calculating a slope to obtain a global feature Hausdorff fractal dimension D of different signals of the distributed optical fiber.
3. The fractal theory-based distributed optical fiber signal identification method according to claim 1, wherein the step S2 includes the following steps:
step s2. a: reconstructing the sequence s (1), s (2),.. s, (N) as y (l), l ═ 1, 2.. N;
step s2. b: defining the value of s, dividing the reconstructed sequence y (l) into segments which are not intersected with each other and have the length of s from the beginning to the end, and dividing the reconstructed sequence y (l) into segments which are not intersected with each other and have the length of s from the end to obtain the total number of 2NsA line segment;
-step s2. c: defining m value, calculating 2N by least square fittingsThe m-order polynomial of each line segment and the label l in the line segment correspond to residual errors;
step s2. d: calculating a local root mean square value according to the size condition of v;
step s2. e: defining the size range of q, thereby obtaining a global q-order fluctuation function Fq(s)；
-step s2. f: repeating the steps from the step S2.b to the step S2.e, calculating the fluctuation functions corresponding to different s, and obtaining s and Fq(s);
step s2. g: by s and Fq(s) obtaining a relation of a linear regression slope h (q) and obtaining a quality parameter value by utilizing the relation of h (q) and the quality index tau (q); using Legendre transformation to obtain a holder index alpha and a multi-fractal spectrum f (alpha);
step s2. h: and calculating the spectrum width delta alpha, the fractal dimension difference delta f and the inclination rate r of the local characteristic parameters of different signal categories of the distributed optical fiber based on the holder index alpha and the multi-fractal spectrum f (alpha) obtained in the step S2. g.
4. The fractal theory-based distributed optical fiber signal identification method according to claim 1, wherein the step of S3 includes the following steps:
step S3. a: combining the local features and the global features of all the signal classes into a feature vector, and defining the feature vector as X; defining the category corresponding to each signal data as a label, expressing the label by y, and dividing the characteristic vector and the label into a training set and a test set with the same signal category proportion;
step S3. b: bringing the training set into an XGboost model, searching for the best parameter value of the model according to the roc auc condition of the test set by using a grid search of a five-fold cross-validation method, bringing signal data to be predicted into the trained model, outputting the prediction class probability of each signal,
step S3. c: repeating the step S3.b by using an SVC model;
step S3. d: and (4) calculating the average value of the prediction class probabilities output in the step S3.b and the step S3.c, wherein the signal class with the larger probability average value is obtained.
5. The fractal theory-based distributed optical fiber signal identification method according to claim 2, wherein the step S1.a comprises the following steps:
step S1. a.1: the time sequence is reconstructed from time sequences s (1), s (2),. gtoreq.s (N), where N is the total length of data, and the interval between two adjacent time sequences is t, i ═ 1, 2,. gtoreq.t:
Figure FDA0003582309810000021
step S1. a.2: for each one
Figure FDA0003582309810000031
The average length is:
Figure FDA0003582309810000032
step S1. a.3: the total average length of the time series was obtained as:
Figure FDA0003582309810000033
6. the fractal theory-based distributed optical fiber signal identification method according to claim 2, wherein the step S1.b comprises the following steps:
step S1. b.1: through a logarithmic transformation, the relationship can be obtained, and the slope D is the fractal dimension:
Figure FDA0003582309810000034
7. the fractal theory-based distributed optical fiber signal identification method according to claim 3, wherein in the step S2.a, the calculation formula of y (l) is as follows:
Figure FDA0003582309810000035
in step S2.b, the calculation formula of s is as follows:
Figure FDA0003582309810000036
in step s2.c, l ═ 1 (v-1) s + i, i ═ 1, 2.., s;
Figure FDA0003582309810000037
step S2. d: the calculation formula of the local root mean square value is as follows:
Figure FDA0003582309810000038
Figure FDA0003582309810000039
step S2. e: the calculation formula of the q-order fluctuation function is as follows:
Figure FDA00035823098100000310
Figure FDA00035823098100000311
step S2. f: obtaining s and Fq(s) is as follows:
Fq(s)∝sh(q）；
step S2. g: the slope h (q) of the linear regression is calculated by the formula:
Figure FDA0003582309810000041
the quality parameter value of the quality index tau (q) is calculated by the formula:
τ(q)＝qh((q)-1；
the calculation formula of the holder index α is:
Figure FDA0003582309810000042
the calculation formula of the multi-fractal spectrum f (alpha) is as follows:
f(α)＝qα-τ(q)；
step S2. h: the formula for calculating the spectral width Δ α is:
Δα＝αmax-αmin；
the fractal difference Δ f is calculated by the following formula:
Δf＝fmax-fmin；
the formula for calculating the rate of inclination r is:
r＝(αmax-α0)/Δα。
8. a fractal theory-based distributed optical fiber signal identification apparatus comprising an input terminal, an output terminal, one or more processors, a memory, and one or more programs stored in the memory and configured to be executed by the one or more processors, wherein the programs include instructions for performing a fractal theory-based distributed optical fiber signal identification method as claimed in any one of claims 1 to 7.
9. A computer-readable storage medium storing a computer program for use with an input and an output, wherein the computer program is executable by a processor to perform a method for fractal theory based distributed optical fiber signal identification as claimed in any one of claims 1 to 7.
CN202210354540.4A 2022-04-06 2022-04-06 Distributed optical fiber signal identification method, device and storage medium based on fractal theory Pending CN114692693A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210354540.4A CN114692693A (en) 2022-04-06 2022-04-06 Distributed optical fiber signal identification method, device and storage medium based on fractal theory

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210354540.4A CN114692693A (en) 2022-04-06 2022-04-06 Distributed optical fiber signal identification method, device and storage medium based on fractal theory

Publications (1)

Publication Number Publication Date
CN114692693A true CN114692693A (en) 2022-07-01

Family

ID=82142185

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210354540.4A Pending CN114692693A (en) 2022-04-06 2022-04-06 Distributed optical fiber signal identification method, device and storage medium based on fractal theory

Country Status (1)

Country Link
CN (1) CN114692693A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2024102233A1 (en) * 2022-11-09 2024-05-16 Microsoft Technology Licensing, Llc Machine learning training duration control

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2024102233A1 (en) * 2022-11-09 2024-05-16 Microsoft Technology Licensing, Llc Machine learning training duration control

Similar Documents

Publication Publication Date Title
CN108742697B (en) Heart sound signal classification method and terminal equipment
CN111680642B (en) Terrain classification method and device
Kershenbaum et al. An image processing based paradigm for the extraction of tonal sounds in cetacean communications
CN115795282B (en) Shock tube dynamic pressure reconstruction method and device, electronic equipment and storage medium
CN117116290B (en) Method and related equipment for positioning defects of numerical control machine tool parts based on multidimensional characteristics
CN111933185A (en) Lung sound classification method, system, terminal and storage medium based on knowledge distillation
Gómez-García et al. On the design of automatic voice condition analysis systems. Part III: Review of acoustic modelling strategies
CN117853827B (en) Sampling pump working state operation monitoring system and method for atmospheric chamber gas monitoring
CN115290596A (en) FCN-ACGAN data enhancement-based hidden dangerous goods identification method and equipment
CN114692693A (en) Distributed optical fiber signal identification method, device and storage medium based on fractal theory
CN117949821B (en) Asynchronous motor fault diagnosis method and related equipment
CN115273904A (en) Angry emotion recognition method and device based on multi-feature fusion
US20190331721A1 (en) Noise spectrum analysis for electronic device
CN114168586A (en) Abnormal point detection method and device
CN114582325A (en) Audio detection method and device, computer equipment and storage medium
Yadav et al. Breast cancer prediction using SVM with PCA feature selection method
CN112133289B (en) Voiceprint identification model training method, voiceprint identification device, voiceprint identification equipment and voiceprint identification medium
CN117493961A (en) Partial discharge signal separation and identification method, device and storage medium
CN107894837A (en) Dynamic sentiment analysis model sample processing method and processing device
CN114844796B (en) Method, device and medium for abnormality detection of time sequence KPI
CN115526882A (en) Medical image classification method, device, equipment and storage medium
CN115355166A (en) Air compressor fault diagnosis method and system based on short-time Fourier transform
Malfante et al. Machine learning for automatic classification of volcano-seismic signatures
CN113925495A (en) Arteriovenous fistula abnormal tremor signal identification system and method combining statistical learning and time-frequency analysis
Li et al. A denosing method of frequency spectrum for recognition of dashboard sounds

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination