Unsupervised distributed aerial optical cable anomaly detection method

Abstract

The invention discloses an unsupervised distributed aerial optical cable anomaly detection method, and relates to the field of aerial optical cable state monitoring based on distributed optical fiber sensing, and the method comprises the steps: collecting and demodulating to obtain phase signals returned by an optical fiber under different disturbance conditions, dividing the original phase signals into subsequences through a sliding window, calculating the norm of each sub-sequence, and calculating the norm difference between each sub-sequence and the previous adjacent sub-sequence; calculating a mean value, a minimum value, a maximum value and a standard deviation of each sub-sequence on each dimension to obtain a feature sequence, and inputting the obtained feature sequence into an LSTM-AE model; and calculating a root mean square error RMSE between the input sequence and the reconstructed sequence, and judging the sample condition according to the RMSE value. According to the method, the LSTM-AE is trained only by using the normal sample, so that abnormal event detection based on the unlabeled data set is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of overhead optical cable status monitoring based on distributed optical fiber sensing, and in particular to an unsupervised distributed overhead optical cable anomaly detection method. Background Art

[0002] As one of the main transmission media connecting urban facilities, overhead optical cables play a vital role in contemporary optical fiber communication networks. However, overhead optical cables are exposed to outdoor environments for a long time. External factors such as storms, tower collapse and human damage may cause damage to the cables, resulting in service interruptions and economic losses. Distributed optical fiber sensing technology based on phase-sensitive optical time domain reflectometer (Φ-OTDR) can directly use redundant optical fibers in overhead optical cables for sensing. It has the advantages of long measurement range, high vibration sensitivity and multi-point positioning, and is very suitable for full-line blind-spot monitoring of optical cable status.

[0003] In order to reduce the false alarm rate of Φ-OTDR in practical applications, efficient pattern recognition methods have received widespread attention. At present, research in this technical field is mainly based on supervised machine learning and deep learning methods. The model is trained by collecting and establishing a data set of target events to distinguish the alarm types of Φ-OTDR. However, in complex overhead optical cable scenarios, due to the wide variety of abnormal events, it is very difficult to build a complete data set, resulting in poor recognition performance of supervised learning methods. In addition, collecting enough signals and annotating them requires huge manpower and material costs. How to use unlabeled signals to achieve unsupervised detection of abnormal events has become a key issue in the development of this field. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides an unsupervised distributed aerial optical cable anomaly detection method. Based on the time series characteristics of the Φ-OTDR phase signal, an unsupervised learning method based on feature enhancement preprocessing and LSTM-AE is proposed to realize the detection of the abnormal vibration state of the aerial optical cable by Φ-OTDR. In the preprocessing method, two sliding window operations are used to capture the temporal dynamic characteristics of the Φ-OTDR phase signal; then, the obtained multidimensional feature sequence is input into an autoencoder (AE) with a long short-term memory neural network (LSTM) as the basic component for encoding and reconstruction; finally, the error between the input sequence and the reconstructed sequence is calculated, and a suitable threshold is set to realize abnormal judgment.

[0006] In order to solve the above technical problems, the present invention provides the following technical solution, an unsupervised distributed overhead optical cable anomaly detection method, comprising:

[0007] The phase signal returned by the optical fiber under different disturbance conditions is collected and demodulated, and the original phase signal is divided into subsequences using a sliding window. The norm of each subsequence is calculated, and the norm difference between each subsequence and the previous adjacent subsequence is calculated; the mean, minimum, maximum and standard deviation of each subsequence in each dimension are calculated to obtain the feature sequence, and the obtained feature sequence is input into the LSTM-AE model; the root mean square error RMSE between the input sequence and the reconstructed sequence is calculated, and the sample situation is judged based on the RMSE value.

[0008] As a preferred solution of the unsupervised distributed overhead optical cable anomaly detection method of the present invention, wherein: the calculation of the norm of each subsequence includes, for the original phase signal of length T Use a sliding window to divide it into K subsequences, the sliding window length is w1, the step size is w1 / 2, and the norm of each subsequence is calculated. The i-th norm N i The formula is expressed as:

[0009]

[0010] As a preferred solution of the unsupervised distributed overhead optical cable anomaly detection method of the present invention, wherein: the calculation of the norm difference between each subsequence and the previous adjacent subsequence includes calculating the norm difference between each subsequence and the previous adjacent subsequence, the i-th norm difference D i The formula is expressed as:

[0011]

[0012] As a preferred embodiment of the unsupervised distributed overhead optical cable anomaly detection method of the present invention, wherein: the calculation of the mean, minimum, maximum and standard deviation of each subsequence in each dimension includes, after the first sliding window processing, obtaining a new sequence G containing two-dimensional features = [g1, g2, ..., g K ]; Set the sliding window with length w2 and step size w2 / 2 to divide the new sequence G into K' subsequences, calculate the mean, minimum, maximum and standard deviation of each subsequence in each dimension, and get the feature sequence F = [f1,f2,…,f K' ], f at each time step i They are all 8-dimensional feature vectors.

[0013] As a preferred solution of the unsupervised distributed overhead optical cable anomaly detection method described in the present invention, the LSTM-AE model includes inputting the obtained feature sequence F into the LSTM-AE model; the LSTM-AE model consists of three parts, namely an input layer, a hidden layer and a reconstruction layer, wherein the input layer and the hidden layer constitute an encoder, the hidden layer and the reconstruction layer constitute a decoder, and each layer uses LSTM units as components.

[0014] As a preferred solution of the unsupervised distributed overhead optical cable anomaly detection method described in the present invention, wherein: the LSTM-AE model also includes, in the LSTM-AE model, the encoder receives the input feature vector of the current time step each time, and the hidden state of the time step is calculated by the LSTM forward propagation until the entire sequence is processed;

[0015] The decoder receives the hidden state of the last LSTM output in the encoder and reconstructs the feature vector at the first time step until the complete reconstructed sequence is output. The jth element value of the i-th vector in the reconstructed sequence is recorded as

[0016] As a preferred solution of the unsupervised distributed overhead optical cable anomaly detection method described in the present invention, wherein: the calculation of the root mean square error RMSE between the input sequence and the reconstructed sequence includes calculating the root mean square error RMSE between the input sequence and the reconstructed sequence, and the formula is expressed as:

[0017]

[0018] Among them, f ij is the i-th vector f in the input sequence F i The j-th element value of is the corresponding reconstructed value.

[0019] As a preferred solution of the unsupervised distributed overhead optical cable anomaly detection method described in the present invention, the judgment of the sample situation based on the RMSE value includes: if the obtained RMSE value does not exceed the set threshold, the corresponding input sample is judged as a normal sample, otherwise it is judged as an abnormal sample.

[0020] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of an unsupervised distributed overhead optical cable anomaly detection method are implemented.

[0021] A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of an unsupervised distributed overhead optical cable anomaly detection method are implemented.

[0022] Beneficial effects of the present invention: The present invention effectively solves the problem of high data set construction cost in the current Φ-OTDR for abnormal detection of overhead optical cables, and realizes abnormal event detection based on unlabeled data sets by training LSTM-AE using only normal samples;

[0023] In addition, the present invention effectively improves the anomaly detection performance of the autoencoder when processing Φ-OTDR phase signals. Considering the advantages of LSTM network in time series data analysis, it is applied to process Φ-OTDR phase signals to better capture the long-term dependencies in the time domain; in addition, by performing two sliding window operations on the original phase signal, the temporal dynamic information in the sequence is effectively enhanced, further enriching the time series features in the original signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0025] Figure 1 A schematic structural diagram of an unsupervised distributed overhead optical cable anomaly detection method provided by an embodiment of the present invention.

[0026] Figure 2 A schematic diagram of an overhead optical cable experimental device for an unsupervised distributed overhead optical cable anomaly detection method provided by an embodiment of the present invention.

[0027] Figure 3 A schematic diagram of a feature enhancement preprocessing method for an unsupervised distributed overhead optical cable anomaly detection method provided by an embodiment of the present invention.

[0028] Figure 4 A schematic diagram of the LSTM-AE model structure of an unsupervised distributed overhead optical cable anomaly detection method provided for an embodiment of the present invention.

[0029] Figure 5 A schematic diagram of the relationship between different thresholds and LSTM-AE anomaly detection accuracy of an unsupervised distributed overhead optical cable anomaly detection method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0030] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0031] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0032] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0033] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0034] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0035] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0036] Example 1, reference Figure 1-Figure 5 , which is the first embodiment of the present invention, and provides an unsupervised distributed overhead optical cable anomaly detection method, comprising:

[0037] S1: Collect and demodulate the phase signal returned by the optical fiber under different disturbance conditions, use a sliding window to divide the original phase signal into subsequences, calculate the norm of each subsequence, and calculate the norm difference between each subsequence and the previous adjacent subsequence.

[0038] S2: Calculate the mean, minimum, maximum, and standard deviation of each subsequence in each dimension to obtain a feature sequence, and input the obtained feature sequence into the LSTM-AE model.

[0039] S3: Calculate the root mean square error (RMSE) between the input sequence and the reconstructed sequence, and make a judgment on the sample situation based on the RMSE value.

[0040] The calculation of the norm of each subsequence includes, for the original phase signal of length T Use a sliding window to divide it into K subsequences, the sliding window length is w1, the step size is w1 / 2, and the norm of each subsequence is calculated. The i-th norm N i The formula is expressed as:

[0041]

[0042] The calculation of the norm difference between each subsequence and the previous adjacent subsequence includes calculating the norm difference between each subsequence and the previous adjacent subsequence, the i-th norm difference D i The formula is expressed as:

[0043]

[0044] The calculation of the mean, minimum, maximum and standard deviation of each subsequence in each dimension includes: after the first sliding window processing, a new sequence G=[g1, g2, ..., g K ]; Set the sliding window with length w2 and step size w2 / 2 to divide the new sequence G into K' subsequences, calculate the mean, minimum, maximum and standard deviation of each subsequence in each dimension, and get the feature sequence F = [f1,f2,…,f K' ], f at each time step i They are all 8-dimensional feature vectors.

[0045] The LSTM-AE model includes inputting the obtained feature sequence F into the LSTM-AE model; the LSTM-AE model consists of three parts, namely an input layer, a hidden layer and a reconstruction layer, wherein the input layer and the hidden layer constitute an encoder, the hidden layer and the reconstruction layer constitute a decoder, and each layer uses an LSTM unit as a component.

[0046] The LSTM-AE model also includes that in the LSTM-AE model, the encoder receives the input feature vector of the current time step each time, and the hidden state of the time step is calculated by LSTM forward propagation until the entire sequence is processed.

[0047] The decoder receives the hidden state of the last LSTM output in the encoder and reconstructs the feature vector at the first time step until the complete reconstructed sequence is output. The jth element value of the i-th vector in the reconstructed sequence is recorded as

[0048] The calculating of the root mean square error RMSE between the input sequence and the reconstructed sequence includes calculating the root mean square error RMSE between the input sequence and the reconstructed sequence, and the formula is expressed as:

[0049]

[0050] Among them, f ij is the i-th vector f in the input sequence F i The j-th element value of is the corresponding reconstructed value.

[0051] The determination of the sample condition according to the RMSE value includes: if the obtained RMSE value does not exceed the set threshold, the corresponding input sample is determined as a normal sample, otherwise it is determined as an abnormal sample.

[0052] Example 2, reference Figure 2-Figure 5 , which is an embodiment of the present invention, provides an unsupervised distributed overhead optical cable anomaly detection method. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.

[0053] Step 1: Connect one core of the overhead optical cable to the Φ-OTDR system, collect and demodulate the phase signal returned by the optical fiber under different disturbance conditions;

[0054] In this example, the schematic diagram of the overhead optical cable experimental device is as follows: Figure 2 shown.

[0055] In this example, a 6-core single-mode finished armored optical cable is suspended between two equal-height brackets with a span of about 17.2m. Both ends of the cable are pulled to the ground winding drum by pulleys and fixed. The mid-span sag of the overhead optical cable is controlled by adjusting the pay-off length of the winding drum. The initial sag under normal conditions is about 25cm.

[0056] In this example, one core of the optical cable is selected as the sensing optical fiber, and 4 km and 1 km of single-mode tight-buffered optical fiber are fused at its front and rear ends respectively, and the other end of the 4 km is connected to the Φ-OTDR system.

[0057] In this example, the phase signal collected when no disturbance is applied is the normal state of the optical cable, which is mainly affected by the environmental noise in the experiment; the phase signal collected when disturbance is applied is the abnormal state of the optical cable. The disturbance types include but are not limited to optical cable vibration caused by external activities such as wind blowing, ice shedding, human pulling, object knocking, shaking of pole towers, etc.

[0058] In this example, a binary classification data set is constructed based on normal state samples and abnormal state samples, and is divided into a training set, a validation set, and a test set. The training set only contains normal state samples, while the validation set and the test set both contain normal state samples and abnormal state samples. The data set composition is shown in Table 1:

[0059] Table 1 Dataset composition table

[0060]

[0061] In this example, the training set is used to train the model, the validation set is used to set the appropriate threshold, and the test set is used to evaluate the anomaly detection performance of the model. In order to meet the needs of experimental result analysis, for the validation set and the test set containing mixed samples, labels are annotated for each sample during the data processing stage. In actual applications, labeling is not required, and only normal samples need to be provided for training.

[0062] Step 2: For the original phase signal of length T Use a sliding window to divide it into K subsequences, the sliding window length is w1, the step size is w1 / 2, and the norm of each subsequence is calculated. The i-th norm N i The formula is expressed as:

[0063]

[0064] Step 3: Calculate the norm difference between each subsequence and the previous adjacent subsequence, the i-th norm difference D i The formula is expressed as:

[0065]

[0066] Step 4: After the first sliding window processing, a new sequence G = [g1, g2, ..., g K ]. Set a sliding window with a length of w2 and a step size of w2 / 2 to divide the new sequence G into K' subsequences, calculate the mean, minimum, maximum and standard deviation of each subsequence in each dimension, and obtain the feature sequence F = [f1, f2, ..., f K' ], f at each time step i They are all 8-dimensional feature vectors;

[0067] In this example, the feature enhancement preprocessing method is shown in the figure Figure 3 shown.

[0068] In this example, the phase signal length of each sample is 600 points, corresponding to a time of 600ms. The length of the two sliding windows is set to 6, and the final feature sequence F obtained by preprocessing has a total of 66 time steps. At this time, the reconstruction error of LSTM-AE on the validation set is about 3.74, and the training time of each cycle is about 193ms.

[0069] Step 5: Input the feature sequence F obtained above into the LSTM-AE model;

[0070] In this example, the LSTM-AE model structure diagram is as follows Figure 4 As shown in the figure, the model consists of three parts: input layer, hidden layer and reconstruction layer. The input layer and hidden layer form the encoder, and the hidden layer and reconstruction layer form the decoder. Each layer uses LSTM units as components.

[0071] Step 6: In the LSTM-AE model, the input of LSTM-AE is usually a sequence S = [s1, s2, …, s L ], where s at each time step t are all 8-dimensional feature vectors, i.e., f t , the encoder receives the input feature vector of the current time step t each time, combined with the hidden state h of the previous time step t-1 , the hidden state h of this time step is calculated by LSTM forward propagation t , until the entire sequence is processed, the update formula for each time step is as follows:

[0072] i t =σ(W ix f t +W ir h t-1 +b i )

[0073] r t =σ(W rx f t +W rh h t-1 +b r )

[0074]

[0075]

[0076] o t =σ(W ox f t +Woh h t-1 +b o )

[0077] h t =o t ⊙tanh(c t )

[0078] Among them, i t 、r t and t are the outputs of the input gate, forget gate, and output gate respectively. t is the cell state at time step t, is the candidate unit state. ⊙ represents the Hadamard product. All W represents the weight matrix in the relevant calculation, b represents the bias vector, and σ(·) is the activation function, i.e., the sigmoid function.

[0079] After that, the decoder receives the hidden state of the last LSTM output in the encoder, linearly transforms it through the weight matrix and bias vector of the fully connected layer, and reconstructs the feature vector at the first time step. At each subsequent time step, the decoder takes the feature output and hidden state of the previous time step as input, calculates the hidden state of the current time step through the LSTM layer, and thus obtains the reconstructed features. This process is repeated until the complete reconstructed sequence is output;

[0080] Step 7: Calculate the root mean square error (RMSE) between the input sequence and the reconstructed sequence. The formula is:

[0081]

[0082] Among them, f ij is the i-th vector f in the input sequence F i The j-th element value of is the corresponding reconstructed value.

[0083] Step 8: If the obtained RMSE value does not exceed the set threshold, the corresponding input sample is judged as a normal sample, otherwise it is judged as an abnormal sample.

[0084] In this example, based on the validation set, we observe the anomaly detection accuracy of the LSTM-AE model when different thresholds are set. Figure 5 As shown, when the threshold is set to 4, the test accuracy is the highest, reaching 92.73%.

[0085] In this example, the results of LSTM-AE on the test set are evaluated using accuracy, precision, recall, and F1 score, and compared with the other three control experimental groups. The results are shown in Table 2:

[0086] Table 2

[0087]

[0088] The LSTM-AE proposed in the present invention using preprocessed features as input achieves an anomaly recognition accuracy of 90.91%, and its precision, recall and F1 score are better than those of the other three control experimental groups.

[0089] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

[0090] Example 3

[0091] The third embodiment of the present invention is different from the first two embodiments in that:

[0092] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0093] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0094] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0095] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0096] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0097] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. An unsupervised distributed overhead optical cable anomaly detection method, characterized in that: include, The phase signal returned by the optical fiber under different disturbance conditions is collected and demodulated, and the original phase signal is divided into subsequences using a sliding window. The norm of each subsequence is calculated, and the norm difference between each subsequence and the previous adjacent subsequence is calculated. Calculate the mean, minimum, maximum and standard deviation of each subsequence in each dimension to obtain a feature sequence, and input the obtained feature sequence into the LSTM-AE model; Calculate the root mean square error (RMSE) between the input sequence and the reconstructed sequence, and judge the sample situation based on the RMSE value.

2. The unsupervised distributed overhead optical cable anomaly detection method according to claim 1, characterized in that: The calculation of the norm of each subsequence includes, for the original phase signal of length T Use a sliding window to divide it into K subsequences, the sliding window length is w1, the step size is w1 / 2, and the norm of each subsequence is calculated. The i-th norm N i The formula is expressed as:

3. The unsupervised distributed overhead optical cable anomaly detection method according to claim 2, characterized in that: The calculation of the norm difference between each subsequence and the previous adjacent subsequence includes calculating the norm difference between each subsequence and the previous adjacent subsequence, the i-th norm difference D i The formula is expressed as:

4. The unsupervised distributed overhead optical cable anomaly detection method according to claim 3, characterized in that: The calculation of the mean, minimum, maximum and standard deviation of each subsequence in each dimension includes: after the first sliding window processing, a new sequence G=[g1, g2, ..., g K ]; Set the sliding window with length w2 and step size w2 / 2 to divide the new sequence G into K' subsequences, calculate the mean, minimum, maximum and standard deviation of each subsequence in each dimension, and get the feature sequence F = [f1,f2,…,f K' ], f at each time step i They are all 8-dimensional feature vectors.

5. The unsupervised distributed overhead optical cable anomaly detection method according to claim 4, characterized in that: The LSTM-AE model includes inputting the obtained feature sequence F into the LSTM-AE model; the LSTM-AE model consists of three parts, namely an input layer, a hidden layer and a reconstruction layer, wherein the input layer and the hidden layer constitute an encoder, the hidden layer and the reconstruction layer constitute a decoder, and each layer uses an LSTM unit as a component.

6. The unsupervised distributed overhead optical cable anomaly detection method according to claim 5, characterized in that: The LSTM-AE model also includes: in the LSTM-AE model, each time the encoder receives the input feature vector of the current time step, the hidden state of the time step is calculated by the LSTM forward propagation until the entire sequence is processed; The decoder receives the hidden state of the last LSTM output in the encoder and reconstructs the feature vector at the first time step until the complete reconstructed sequence is output. The jth element value of the i-th vector in the reconstructed sequence is recorded as 7. The unsupervised distributed overhead optical cable anomaly detection method according to claim 6, characterized in that: The calculating of the root mean square error RMSE between the input sequence and the reconstructed sequence includes calculating the root mean square error RMSE between the input sequence and the reconstructed sequence, and the formula is expressed as: Among them, f ij is the i-th vector f in the input sequence F i The j-th element value of is the corresponding reconstructed value.

8. The unsupervised distributed overhead optical cable anomaly detection method according to claim 7, characterized in that: The determination of the sample condition according to the RMSE value includes: if the obtained RMSE value does not exceed the set threshold, the corresponding input sample is determined as a normal sample, otherwise it is determined as an abnormal sample.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images