The present invention is based upon and claims the benefit of priority from Japanese patent application No. 2023-102155, filed on Jun. 22, 2023, the disclosure of which is incorporated herein in its entirety by reference.
The present disclosure relates to an information processing device, an information processing method, and a storage medium.
With a technology called optical fiber sensing, it is possible to analyze surrounding environments of a plurality of points along an optical fiber from vibration and sounds applied to the optical fiber. Patent Literature 1 discloses art of specifying the position of a utility pole by sensing vibration and sounds at a plurality of points along an optical fiber by using optical fiber sensing. Specifically, in Patent Literature 1, vibration data for each distance is measured from an optical fiber cable, and the position where the vibration data has a unique pattern is identified as the position of a utility pole.
However, the art described in Patent Literature 1 merely detects whether or not vibration data has a unique pattern and specifies only the position of a utility pole. Therefore, in the case where vibration data has an unknown pattern, it is impossible to identify the position of a utility pole, and also impossible to specify the position of any object. Moreover, this art causes a problem that it is impossible to accurately recognize the surrounding environment at a given position, without being limited to recognizing existence of an object such as a utility pole at a position around the laid optical fiber cable.
Therefore, an example object of the present invention is to solve the problem described above, that is, a problem that it is impossible to accurately recognize the surrounding environment of a laid optical fiber cable.
An information processing device, according to one aspect of the present disclosure, is configured to include
Further, an information processing method, according to one aspect of the present disclosure, is configured to include
Further, a program, according to one aspect of the present disclosure, is configured to cause a computer to execute processing to
With the configurations described above, the present disclosure is capable of accurately recognizing the surrounding environment of a laid optical fiber cable.
The present disclosure will be described with reference to the drawings. Note that the drawings may relate to any embodiments.
An information processing system according to the present embodiment is used to recognize the surrounding environment of each of a plurality of points along a laid optical fiber cable from measurement signals measured from the optical fiber cable, by using a technology called optical fiber sensing. For example, the information processing system first acquires information such as acoustic and vibration in a time-series manner, by using back-scattered light with respect to incident light as measurement signals, at measurement points along an optical fiber cable. Hereinafter, the measurement signals will be described as time-series signals of acoustic signals or vibration signals. Then, from the acquired time-series signals, the information processing system recognizes an event occurring at a measurement point as the environment of the measurement point. For example, examples of recognizing an event occurring at a measurement point include specifying a signal source such as dog's bark or human footfalls and specifying a place such as a city area, a forest, or a park. Hereinafter, a configuration for realizing such a configuration will be described.
As illustrated in
The signal measurement device 20 measures measurement signals at a plurality of points (measurement points, hereinafter referred to as “channels”) along the optical fiber cable 30. The signal measurement device 20 is, for example, a distributed acoustic sensor (DAS), and measures acoustic signals and vibration signals transmitted to the optical fiber cable 30 as measurement signals.
The signal processing device 10 is configured of one or a plurality of information processing devices each having an arithmetic device and a storage device. As illustrated in
The acquisition unit 11 acquires measurement signals at respective measurement points measured by the signal measurement device 20 as described above, and stores them in the signal storage unit 16. For example, the acquisition unit 11 acquires the measurement signals D1, D2, D3, and D4 at four places, namely, the measurement points C1, C2, C3, and C4 along the optical fiber cable 30 as illustrated in
x
c∈: time−sereis signal of series length T observed at channel c [Expression 1]
The prediction unit 12 predicts an event (predictive environment) representing the environment of the corresponding measurement point, on the basis of the measurement signal acquired as described above. Specifically, the prediction unit 12 of the present embodiment first extracts a feature value from the measurement signal D at a predetermined measurement point C by using a feature extractor 12a as illustrated in
Then, the prediction unit 12 inputs the feature value extracted from the measurement signal into a predetermined prediction model to predict a predictive event y at the measurement point C. Here, a prediction model 12b is a machine learning model generated by machine-learning the training data in advance. For example, it is generated by performing machine learning to optimize the parameter of the machine learning model by using, as training data, a pair of a feature value extracted from a measurement signal consisting of an acoustic signa or a vibration signal measured from each measurement point in advance, and a teacher signal that is an event occurring at the measurement point, and is stored in the model storage unit 17. As illustrated in
ŷ
c∈[0,1]K: event type set predicted at chennel c [Expression 2]
Here, K represents the total number of event types defined in advance, and it is assumed that the number of predictive values to be output is equal to the total number K of events. Therefore, the predictive event set y is represented as a vector as expressed by Expression 3, when the channel c=1 and the total number K of event types=3.
ŷ
1∈(0.0,0.3,0.9) [Expression 3]
Here, as expressed by Expression 3, the predictive value may be represented as a continuous value representing the probability that it may correspond to each predictive event, or may be represented as a binary in which “0” represents nonoccurrence of an event and “1” represents occurrence.
As an example, events in the case where the total number K of types=3 in the above example include “place: city area”, “place: forest”, and “place: park”. In this case, for each of the four measurement points C1, C2, C3, and C4 illustrated in
Then, the present embodiment is configured to correct the predictive event set that is a set of predictive events predicted by the prediction unit 12, by the space information extraction unit 13, the similarity calculation unit 14, and the correction unit 15.
The space information extraction unit 13 generates an event space correlation matrix RP (first relationship information) expressed by Expression 4 described below, on the basis of the predictive event set y expressed by Expression 3 calculated as described above.
The event space correlation matrix RP is defined by a matrix having a size of channel by channel, as expressed by Expression 4. Here, an element γnm of the event space correlation matrix RP is a value representing the relationship of the predictive event sets y between channels (between measurement points). As an example, the element γnm is obtained by calculating a Pearson's correlation coefficient (corr(•)) of a predictive event set between two channels with respect to every combination. The elements γnm are aligned to form the event space correlation matrix RP. As corr(•) for calculating each element, distance, pseudometric, or a similarity function such as dispersion, Euclidean distance, Kullback-Leibler divergence, cosine similarity, or the like may be used.
Moreover, the space information extraction unit 13 generates a measurement signal space correlation matrix RA (second relationship information) as expressed by Expression 5 provided below, on the basis of the measurement signal.
As expressed by Expression 5, the measurement signal space correlation matrix RA is defined by a matrix having a size of channel by channel. Here, the element ρnm of the measurement signal space correlation matrix RA is a value representing the relationship between measurement signals of the channels (measurement points), and is calculated similarly to the element of the event signal space matrix described above. As an example, the element ρnm is obtained by calculating a Pearson's correlation coefficient (corr(•)) of measurement signals between two channels with respect to every combination. The element ρnm are aligned to form the measurement signal space correlation matrix RA. Note that a measurement signal x used in corr(•) calculating each element is an amplitude spectrum obtained by taking an absolute value after the discrete Fourier transform with respect to a measurement signal of a channel. However, the measurement signal x may be power, phase spectrum, or a time region signal.
The similarity calculation unit 14 calculates similarity between the event space correlation matrix RP (first relationship information) and the measurement signal space correlation matrix RA (second relationship information) calculated as described above. Specifically, as expressed by Expression 6, the similarity calculation unit 14 calculates the similarity between corresponding elements (γnm, ρnm) of the event space correlation matrix RP and the measurement signal space correlation matrix RA, that is, similarity anm between elements corresponding to the same channels, and takes the total sum am in the row direction or column direction of the matrix to thereby calculate a vector “a” of the number of dimensions that is equal to the number of channels.
Here, it is considered that when prediction of an event at each measurement point by the prediction model 12b is spatially correct, it conforms to spatial information of a measurement signal measured at each measurement point. Therefore, when the prediction by the prediction model 12b is correct, the similarity between the elements of the event space correlation matrix RP and the measurement signal space correlation matrix RA is high, and the similarity between the elements of the vector “a” may be high. As a result, it can be said that the similarity corresponding to each channel that is each element of the vector “a” calculated as described above represents the reliability of the prediction result by the prediction model at each channel.
The correction unit 15 corrects the prediction result by the prediction model by using the vector “a” representing the reliability of each channel calculated as described above. For example, as expressed by Expression 7 provided below, with respect to a predictive event set yc of each channel by the prediction model, a predictive value included in the predictive event set is corrected by giving a weight to the reliability of each channel by the vector “a”.
For the correction of a predictive event as described above, correction can be made by giving a weight using a softmax function or by using only some channels having higher reliability, rather than simple multiplication. For example, it is possible to make a correction in which an importance is not placed on a prediction result of a channel having lower reliability, as compared with other channels.
Next, operation of the signal processing device 10 will be described.
First, the signal processing device 10 acquires measurement signals at respective measurement points along the optical fiber cable 30 measured by the signal measurement device 20 (step S1 in
Then, the signal processing device 10 predicts an event representing the environment of the corresponding measurement point, on the basis of the acquired measurement signal (step S2 in
Then, on the basis of the calculated predictive event set y, the signal processing device 10 generates the event space correlation matrix RP expressed as Expression 4 (step S3 in
Then, the signal processing device 10 calculates the similarity between the calculated event space correlation matrix RP and the measurement signal space correlation matrix RA (step S4 in
Then, the signal processing device 10 corrects the predictive event set by the prediction model, by using the vector “a” representing the calculated reliability of the prediction result of each channel. For example, as expressed by Expression 7 provided above, the signal processing device 10 corrects the prediction result by giving a weight to the reliability of each channel by the vector “a” with respect to the predictive event set yc of each channel by the prediction model.
As described above, the signal processing device 10 of the present embodiment first calculates an event space correlation matrix representing the spatial relationship of prediction results by the prediction model and a measurement signal space correlation matrix representing the spatial relationship of measurement signals. Then, the signal processing device 10 calculates the similarity by comparing respective spatial correlation matrixes, whereby calculates the reliability of the prediction result by the prediction model, and corrects the prediction result by the prediction model by using the reliability. Therefore, even in the case of a measurement signal that largely deviates from the training data used at the time of generating the prediction model, it is possible to correct the prediction in consideration of spatial relationship of the measurement signal, and to improve the prediction accuracy of the event.
For example, in the measurement data D2 and D4 measured at the measurement points C2 and C4 along the optical fiber cable 30 as illustrated in
Next, a second example embodiment of the present disclosure will be described with reference to the drawings. The present embodiment shows the outlines of the configuration of the signal processing device explained in the above-described embodiment. Note that
First, a hardware configuration of an information processing device 100 will be described with reference to
Note that
The information processing device 100 can construct, and can be equipped with, an acquisition unit 121, a prediction unit 122, and a correction unit 123 illustrated in
The acquisition unit 121 acquires measurement signals measured at a plurality of measurement points on a laid optical fiber cable.
The prediction unit 122 predicts a predictive environment representing the environment of each measurement point from each measurement signal by using a predetermined prediction model.
The correction unit 123 corrects the predictive environment on the basis of a comparison between the predictive environment of each measurement point and the measurement signal measured as the measurement point. For example, the correction unit generates first relationship information representing the relationship of the predictive environments between the measurement points and second relationship information representing the relationship of the measurement signals between the measurement points, and corrects the predictive environment on the basis of the comparison between the first relationship information and the second relationship information.
Since the present disclosure is configured as described above, a prediction result by the prediction model is corrected on the basis of the comparison between the prediction result by the prediction model and the measurement signal. Therefore, even in the case of a measurement signal that largely deviates from the training data used at the time of generating the prediction model, it is possible to correct the prediction from the comparison result, and to improve the prediction accuracy.
Further, at least one of the functions of the acquisition unit 121, the prediction unit 122, and the correction unit 123 described above may be carried out by an information processing device provided and connected to any location on the network, that is, may be carried out by so-called cloud computing.
The program described above can be supplied to a computer by being stored on a non-transitory computer-readable medium of any type. Non-transitory computer-readable media include tangible storage media of various types. Examples of non-transitory computer-readable media include magnetic storage media (for example, flexible disk, magnetic tape, and hard disk drive), magneto-optical storage media (for example, magneto-optical disk), a CD-ROM (Read Only Memory), a CD-R, a CD-R/W, and semiconductor memories (for example, mask ROM, PROM (Programmable ROM), EPROM (Erasable PROM), flash ROM, and RAM (Random Access Memory)). The program may be supplied to a computer by a transitory computer-readable medium of any type. Examples of transitory computer-readable media include electric signals, optical signals, and electromagnetic waves. A transitory computer-readable medium can supply a program to a computer via a wired communication channel such as a wire and an optical fiber, or a wireless communication channel.
While the present disclosure has been described with reference to the example embodiments described above, the present disclosure is not limited to the above-described embodiments. The form and details of the present disclosure can be changed within the scope of the present disclosure in various manners that can be understood by those skilled in the art. Moreover, the example embodiments may be combined with other example embodiments as appropriate.
The whole or part of the example embodiments disclosed above can be described as the following supplementary notes. Hereinafter, outlines of the configurations of an information processing device, an information processing method, and a program, according to the present disclosure, will be described. However, the present disclosure is not limited to the configurations described below.
An information processing device comprising:
The information processing device according to supplementary note 1, wherein
The information processing device according to supplementary note 2, wherein
The information processing device according to supplementary note 2 or 3, wherein
The information processing device according to supplementary note 4, wherein
The information processing device according to supplementary note 5, wherein
The information processing device according to supplementary note 6, wherein
An information processing method comprising:
The information processing method according to supplementary note 8, further comprising
A program for causing cause a computer to execute processing to:
Number | Date | Country | Kind |
---|---|---|---|
2023-102155 | Jun 2023 | JP | national |