Event identification device, event identification method, and program

Abstract

To obtain highly accurate identification results for each of multiple identification items related to the speed of a moving object.SOLUTION: An event identification device is configured as follows. On the basis of optical signals output from optical fibers arranged corresponding to a moving surface, the event identification device is capable of generating, as identification-use information, time-series vibration information indicating changes in vibration over time at specific positions, time-series vibration position information indicating changes in vibration positions over time, and frequency characteristic information indicating the frequency characteristics of vibrations within unit time at specific positions. Using the input identification-use information, the device identifies a target to be recognized. In relation to events such as the speed of a moving object on the moving surface, identification items to be recognized are set. One or more pieces of identification-use information selected from the time-series vibration information, time-series vibration position information, and frequency characteristic information are input to an event identification unit for identifying the set identification item.SELECTED DRAWING: Figure 11

Description

[Technical Field]

[0001] The present invention relates to an event identification device, an event identification method, and a program. [Background technology]

[0002] A technology is known in which a pattern corresponding to the traveling state of a vehicle on a road is detected based on an optical signal received from an optical communication fiber laid on the road, and the traveling state of a vehicle on the road is detected based on the detected pattern (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] International Publication No. 2020 / 116030 Summary of the Invention [Problem to be solved by the invention]

[0004] Based on the optical signal received from the optical fiber, multiple identification items can be set according to, for example, the distance traveled, direction of travel, etc. of the moving object, such as a vehicle moving on a moving surface such as a road. It is preferable to obtain as accurate a classification result as possible for each of these multiple classification items.

[0005] In consideration of the above-mentioned problems, an object of the present invention is to obtain highly accurate classification results for each of a plurality of classification items related to the speed of a moving object. [Means for solving the problem]

[0006] (1) One aspect of the present invention that solves the above-mentioned problems is an event identification device that includes an information generation unit that is capable of generating, as identification use information used to identify an event, at least time-series vibration information that indicates changes in vibration over time for each specific position, time-series vibration position information that indicates changes in vibration position over time, and frequency characteristic information that indicates the frequency characteristics of vibration in unit time for each specific position, based on an optical signal output from an optical fiber arranged corresponding to the moving plane on which the moving body moves; an event identification unit that identifies an identification target using the input identification use information; an identification target setting unit that sets an identification item to be identified in relation to an event as the speed of the moving body moving on the moving plane; and an input unit that inputs one or more identification use information from the time-series vibration information, time-series vibration position information, and frequency characteristic information to the event identification unit, to be used to identify the identification item set as the target by the identification target setting unit.

[0007] (2) One aspect of the present invention is the event identification device described in (1), wherein the identification target setting unit may set a specified identification item from among a plurality of identification items defined in relation to the event as the speed of a moving body as the identification target.

[0008] (3) One aspect of the present invention is an event identification device described in (1) or (2), wherein the input unit may identify identification use information to be used to identify the identification item set as the target by the identification target setting unit based on an identification use information table that indicates identification use information defined as suitable for identification for each of a plurality of identification items, and input the identified identification use information to the event identification unit.

[0009] (4) One aspect of the present invention is an event identification device described in (1) or (2), wherein the input unit selects one or more pieces of identification use information suitable for identifying an identification item set as a target by the identification target setting unit from among the identification use information of the time series vibration information, the time series vibration position information, and the frequency characteristic information, and inputs the selected identification use information to the event identification unit.

[0010] (5) One aspect of the present invention is the event identification device described in (4), wherein the input unit may determine one or more pieces of identification use information to be input to the event identification unit in correspondence with the identification item set as the target by the identification target setting unit, using a machine learning model obtained by performing machine learning to estimate a combination pattern of one or more pieces of identification use information that will produce the best identification result for the identification item.

[0011] (6) One aspect of the present invention is an event identification device according to any one of (1) to (5), wherein the event identification unit may include a plurality of primary classifiers, each of which outputs a primary identification result in response to input of any one of the plurality of pieces of identification-use information, and a secondary classifier, which outputs a final identification result in response to input of the primary identification results output by each of the primary classifiers.

[0012] (7) One aspect of the present invention is an event identification device described in any one of (1) to (5), wherein the event identification unit may include a single identifier that outputs a final identification result in response to input of all of the identification utilization information by the input unit.

[0013] (8) One aspect of the present invention is an event identification device described in any one of (1) to (5), wherein the event identification unit may include a predetermined number of primary classifiers that output a primary identification result in response to input of a portion of a plurality of pieces of identification-use information, and a secondary classifier that outputs a final identification result in response to input of the primary identification result and the identification-use information that is not input to the primary classifiers.

[0014] (9) One aspect of the present invention is an event identification method in an event identification device, comprising: an information generation step in which an information generation unit is capable of generating, as identification use information to be used for identifying an event, at least time-series vibration information indicating changes in vibration over time for each specific position, time-series vibration position information indicating changes in vibration position over time, and frequency characteristic information indicating frequency characteristics of vibration in unit time for each specific position, based on an optical signal output from an optical fiber arranged corresponding to a moving surface on which a moving body moves; an event identification step in which the event identification unit identifies an identification target using the input identification use information; an identification target setting step in which an identification target setting unit sets an identification item to be identified in relation to an event as the speed of a moving body moving on the moving surface; and an input step in which an input unit inputs, to the event identification step, one or more pieces of identification use information to be used for identifying the identification item set as the target by the identification target setting step from among the time-series vibration information, time-series vibration position information, and frequency characteristic information.

[0015] (10) One aspect of the present invention is a program for causing a computer as an event identification device to function as an information generation unit that is capable of generating, as identification use information used to identify an event, at least time-series vibration information indicating changes in vibration over time for each specific position, time-series vibration position information indicating changes in vibration position over time, and frequency characteristic information indicating the frequency characteristics of vibration in unit time for each specific position, based on an optical signal output from an optical fiber arranged corresponding to the moving surface on which the moving body moves; an event identification unit that identifies an identification target using the input identification use information; an identification target setting unit that sets an identification item to be identified in relation to an event as the speed of the moving body moving on the moving surface; and an input unit that inputs one or more identification use information from the time-series vibration information, time-series vibration position information, and frequency characteristic information to the event identification unit to be used to identify the identification item set as the target by the identification target setting unit. [Effects of the Invention]

[0016] According to the present invention, it is possible to obtain an effect that highly accurate classification results can be obtained for each of a plurality of classification items related to the speed of a moving object. [Brief explanation of the drawings]

[0017] [Figure 1] 1 is a diagram illustrating an example of the overall functional configuration of an event identification system according to a first embodiment. [Figure 2] 1 is a diagram illustrating a configuration example of an optical fiber-held sensor according to a first embodiment. FIG. [Figure 3] 5A and 5B are diagrams illustrating another configuration example of the optical fiber sensor according to the first embodiment. [Figure 4] FIG. 2 is a diagram illustrating an example of the functional configuration of an event identification device according to the first embodiment. [Figure 5] 4 is a diagram showing an example of identification use information generated by an information generation unit in the first embodiment. FIG. [Figure 6] FIG. 2 is a diagram illustrating a first example of the configuration of an event identification unit in the first embodiment. [Figure 7] FIG. 10 is a diagram illustrating a second example of the configuration of the event identification unit in the first embodiment. [Figure 8] FIG. 10 is a diagram illustrating a third example of the configuration of the event identification unit in the first embodiment. [Figure 9] FIG. 10 is a diagram illustrating a specific application example of the event identification system according to the second embodiment. [Figure 10] FIG. 11 is a diagram showing an example of the identification accuracy of identification use information for each identification item in the second embodiment. [Figure 11] FIG. 10 is a diagram showing a configuration example (first example) of an event identification device according to a second embodiment. [Figure 12] FIG. 10 is a diagram illustrating a configuration example (second example) of an event identification device according to the second embodiment. [Figure 13] FIG. 10 is a diagram illustrating a configuration example (third example) of an event identification device according to the second embodiment. [Figure 14] 10A and 10B are diagrams illustrating an example of the structure of an optical fiber sensor as a modified example of the embodiment. [Figure 15]10A and 10B are diagrams illustrating another example of the structure of the optical fiber sensor as a modified example of the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0018] First Embodiment [Overall configuration example of an event identification system] 1 shows an example of the overall configuration of an event discrimination system according to the first embodiment. The event discrimination system in the figure includes an optical fiber 22 and an event discrimination device 300.

[0019] In this embodiment, the optical fiber 22 is arranged in a predetermined manner in the event recognition environment ENV. Specifically, the optical fiber 22 is included in the structure of the optical fiber-carrying sensor 20, and the optical fiber-carrying sensor 20 is then arranged in the event recognition environment ENV. The event identification environment ENV is an environment in which an event to be identified by the event identification system occurs.

[0020] In this embodiment, the optical fiber sensor 20 is a sensor that detects vibrations and sounds generated in response to an event to be identified as a change in an optical signal. The optical fiber sensor 20, which is disposed in the event identification environment ENV, is connected to the event identification device 300 via the end of an optical fiber 22.

[0021] The event identification device 300 is a device that identifies an event to be identified by utilizing an optical signal input from the optical fiber sensor 20. In the figure, an example in which the event identification device 300 is configured as a single device is shown, but it may also be configured, for example, by multiple devices in which predetermined functions are distributed.

[0022] [Configuration example of an optical fiber sensor] An example of the configuration of the optical fiber held sensor 20 of this embodiment will be described with reference to Fig. 2. The optical fiber held sensor 20 shown in the figure has a cylindrical resin tube 21 and an optical fiber 22 extending in the axial direction (pipe axis) O1 of the resin tube 21 within a cylindrical wall 21A of the resin tube 21. In this embodiment, the resin tube 21 is a core material.

[0023] A plurality of optical fibers 22 may be provided inside the cylindrical wall 21A. In the figure, an example in which four optical fibers 22 are provided is shown. Fig. 2 shows an example in which four optical fibers 22 are positioned at 90° intervals around the tube axis 2 in a cross section perpendicular to the axis O1. Fig. 2 also shows an example in which the four optical fibers 22 are arranged in a spiral shape with the axis O1 as the center. The spiral pitch P of the optical fibers 22 may be set according to the required measurement accuracy, etc.

[0024] The optical fiber held sensor 20 of this embodiment has a folded portion 222 and a folded portion 223 that protrude from one end edge of the resin tube 21 in the longitudinal direction (direction of the axis O1). The folded portion 222 is a member in which the optical fiber 22a and the optical fiber 22b protruding from one longitudinal end edge of the resin pipe 21 are connected to each other outside the resin pipe 21 (in the direction of the axis O1 away from the resin pipe 21). The optical fiber 22a and the optical fiber 22b are adjacent to each other in the circumferential direction of the resin pipe 21. The folded portion 223 is a member in which the optical fiber 22c and the optical fiber 22d protruding from one end edge of the resin pipe 21 are connected to each other outside the resin pipe 21. The optical fiber 22c and the optical fiber 22d are adjacent to each other in the circumferential direction of the resin pipe 21. That is, in this embodiment, the optical fiber 22 protrudes from the resin tube 21 at one end in the direction of the axis O1, and has folded portions 222, 223 outside the resin tube 21 in which an arbitrary optical fiber 22 is connected to another arbitrary optical fiber 22 adjacent to the arbitrary optical fiber 22 in the circumferential direction of the resin tube 21, which is the core material.

[0025] The optical fiber sensor 20 has a folded portion 229 that protrudes from the other end edge of the resin tube 21 in the longitudinal direction. The folded portion 229 is a member in which optical fibers 22g and 22h protruding from the other longitudinal edge of the resin wall portion 24 are connected to the outside of the resin tube 21 (in the direction of axis O1 away from the resin tube 21). Optical fibers 22g and 22h are adjacent to each other in the circumferential direction of the resin tube 21. Optical fibers 22g and 22h protrude from the other longitudinal edge of the resin tube 21. Optical fiber end 22e and optical fiber end 22f do not form a folded portion. Optical fiber end 22e is connected to the event identification device 300.

[0026] In this embodiment, one optical fiber is folded back to form folded portions 222, 223, and 229, and optical fiber ends 22e and 22f are protruded from the other edge, thereby positioning four optical fibers 22 within the resin tube 21.

[0027] The optical fiber 22 may have a core, a cladding, and a coating layer, and may generate scattered light, such as Brillouin scattering or Raman scattering, resulting from distortion or temperature of the core when discontinuous pump light, such as laser light, is incident on the core. For example, an optical fiber consisting of a core and a cladding may be suitably used as this type of optical fiber 22. Examples of materials for the core and cladding include plastic and quartz glass.

[0028] Examples of optical fibers include an optical fiber strand having a primary coating around the cladding, an optical fiber core having a secondary coating around the primary coating, and an optical fiber cord having a reinforcing material around the secondary coating and an outer jacket covering the reinforcing material.

[0029] The type of optical fiber 22 is not particularly limited and can be selected depending on the detection target (such as strain, vibration, or sound), the detection method, the type of scattered light used during detection, etc. For example, at least one type of optical fiber selected from the group consisting of a single-mode optical fiber, a multi-mode optical fiber, and a polarization-maintaining optical fiber can be used.

[0030] When a plurality of optical fibers 22 are arranged in the optical fiber-carrying sensor 20, the plurality of optical fibers 22 may be the same type of optical fiber or different types of optical fibers.

[0031] 2, the optical fiber is located inside the cylindrical wall of a cylindrical resin tube. However, the present invention is not limited to this, and the optical fiber may be located outside the cylindrical wall (i.e., on the outer circumferential surface of the resin tube). In the above-described embodiment, the core material is a cylindrical resin tube, but the present invention is not limited to this. The core material may not be cylindrical (i.e., hollow) but may be solid. When the core material is solid, the optical fiber may be located within the core material or may be located on the outer periphery of the core material.

[0032] 3, a configuration example of an optical fiber sensor 20A in which the core material is solid and the optical fiber is located on the outer periphery of the core material will be described. The optical fiber sensor 20A can be used in place of the optical fiber sensor 20 in the event identification system of this embodiment. Alternatively, when the event identification system of this embodiment is provided with multiple optical fiber sensors, the optical fiber sensor 20 and the optical fiber sensor 20A may be mixed among the multiple optical fiber sensors. 3, the same parts as those in FIG. 2 are designated by the same reference numerals and the explanation thereof will be omitted, and the differences from the optical fiber sensor 20 in FIG. 1 will be mainly explained.

[0033] The optical fiber sensor 20A has a long rod-shaped core material 121 and an optical fiber 22 located on the outer circumferential surface of the core material 121. The optical fiber 22 forms a spiral around an axis O2. The spiral pitch P2 of the spiral formed by the optical fiber 22 is the same as the spiral pitch P.

[0034] The core material 121 is a straight rod-shaped solid member. By making the core material 121 a solid member, the rigidity of the optical fiber held sensor 20A can be further increased, and the displacement detection accuracy can be further improved. The outer diameter R2 of the core material 121 is the same as the outer diameter R of the resin pipe 21. Examples of materials for the core material 121 include a resin composition, ceramic, metal, and glass.

[0035] The optical fiber 22 is fixed in close contact with the outer circumferential surface of the core material 121. Examples of methods for fixing the optical fiber 22 to the outer circumferential surface of the core material 121 include adhesion, fusion, and tape attachment.

[0036] In the optical fiber sensor 20A, the optical fiber 22 may be fixed to the core material 121 by, for example, winding the optical fiber 22 around the core material 121 and bonding it with an adhesive.

[0037] [Example of functional configuration of event identification device] An example of the functional configuration of the event identification device 300 of this embodiment will be described with reference to FIG. In the following explanation, an example will be given in which the event identification device 300 is connected to one optical fiber sensor 20. Also, an example will be given in which one optical fiber sensor 20 has four optical fibers 22 each individually arranged in a spiral.

[0038] The event identification device 300 in the figure includes a receiving unit 301, an information generating unit 302, an event identification unit 303, and a storage unit 304. The functions of the event identification device 300 in the figure may be realized by executing a program using a CPU (Central Processing Unit) included in the event identification device 300 as hardware.

[0039] The receiving unit 301 receives optical signals from the optical fibers 22. In this case, the receiving unit 301 is connected to the optical fiber ends 22e (22e-1, 22e-2, 22e-3, 22e-4) of each of the four optical fibers 22. In other words, the receiving unit 301 is connected to each of the four optical fibers 22 provided in one optical fiber-carrying sensor 20.

[0040] The receiving unit 301 inputs pulsed light from the optical fiber end 22e into each of the four connected optical fibers 22. Scattered light generated in the optical fiber 22 as the input pulsed light is transmitted returns to the input side as returned light via the same optical fiber 22. The receiving unit 301 receives the returned light as an optical signal.

[0041] For example, in the event identification environment ENV, vibrations and sounds occur in response to the occurrence of an event to be identified. The occurrence of such vibrations and sounds affects the scattered light transmitted through the optical fiber 22 via the ground or air, and also affects the returning light received by the receiving unit 301. In other words, the returning light changes in response to the vibrations and sounds occurring in the event identification environment ENV.

[0042] The optical signal Sopt received by the receiving unit is input to the information generating unit 302. The information generating unit 302 uses the input optical signal Sopt to generate a plurality of pieces of identification information. The identification information is information used by the event identifying unit 303 to identify a target event (event identification).

[0043] The event identification unit 303 performs event identification using the multiple pieces of identification-use information generated by the information generation unit 302, and outputs the identification result (event identification result). For example, the event identification unit 303 may store the event identification result in the storage unit 304. The event identification unit 303 may also output the event identification result via an external display or printer. Furthermore, data that associates the event identification result obtained by the event identification unit 303 with the corresponding identification-use information may be used as training data for building a trained model used by the event identification unit 303.

[0044] The storage unit 304 stores various information related to the event identifier 300 .

[0045] [Generation of Identification and Use Information by the Information Generation Unit] The generation of identification-use information by the information generating unit 302 of this embodiment will be described with reference to FIG. The optical signal Sopt received by the receiving unit 301 is input to the information generating unit 302. In Fig. 4, the receiving unit 301 receives four optical signals. For ease of explanation, an example will be given in which the receiving unit 301 inputs one optical signal Sopt, which is a combination of the four received optical signals, to the information generating unit 302.

[0046] The information generating unit 302 generates three pieces of identification and use information based on the input optical signal Sopt. That is, as shown in Fig. 5, the information generating unit 302 is capable of generating time-series waveform information D1, time-series vibration position information D2, and frequency characteristic information D3 as the identification and use information.

[0047] The time-series waveform information D1 is information (an example of time-series vibration information) that indicates a change over time in a waveform corresponding to a vibration occurring at a specific position in the event identification environment ENV. The information generating unit 302 may generate a plurality of pieces of time-series waveform information D1 corresponding to each of a plurality of specific positions defined in the event identification environment ENV.

[0048] The time-series vibration position information D2 is information indicating a change in position of a vibration occurring in the event identification environment ENV over time. The time-series waveform information D1 may generate a plurality of pieces of time-series vibration position information D2 for each occurring vibration.

[0049] The frequency characteristic information D3 is information indicating changes over time in frequency characteristics corresponding to vibrations occurring at specific positions in the event identification environment ENV. The information generator 302 may generate multiple pieces of frequency characteristic information D3 corresponding to each of multiple specific positions defined in the event identification environment ENV. The frequency characteristic information D3 may be a spectrogram.

[0050] The optical signal Sopt input to the information generator 302 is a signal that changes in response to distortion of the optical fiber 22 caused by vibrations occurring in the event identification environment ENV, for example. In other words, the optical signal Sopt is information indicating distortion of the optical fiber 22. Therefore, the information generating unit 302 performs a predetermined calculation on the input optical signal Sopt to convert it into vibration velocity information indicating the vibration velocity. The information generating unit 302 then performs differentiation (first-order differentiation) on the vibration velocity information to convert it into vibration acceleration information indicating the acceleration of the vibration (vibration acceleration).

[0051] The information generating unit 302 detects vibration waveforms that occur at each time corresponding to each specific position from the vibration acceleration indicated by the vibration acceleration information obtained as described above, and generates time-series waveform information D1 that indicates vibrations over time at each specific position. Furthermore, the information generating unit 302 generates time-series vibration position information D2 by detecting a position that changes with time for a specific vibration from the vibration acceleration indicated by the vibration acceleration information. Furthermore, the information generating unit 302 performs a process of detecting frequency characteristics corresponding to a unit time by performing frequency analysis (e.g., FFT) on a time window set for the vibration acceleration indicated by the vibration acceleration information, while shifting the time window, and generates frequency characteristic information D3 by arranging the frequency characteristics obtained corresponding to each time window over time.

[0052] [Example of event recognition unit configuration: Example 1] Below, three examples of the configuration of the event identification section 303 of this embodiment will be described, and the first example will be described first. 6 shows a first example of the configuration of the event discriminator 303. The event discriminator 303 in the figure includes one discriminator 331. The discriminator 331 receives as input the time-series waveform information D1, the time-series vibration position information D2, and the frequency characteristic information D3 generated by the information generating unit 302 (not shown in the figure). The classifier 331 performs event classification in response to inputs of the time-series waveform information D1, the time-series vibration position information D2, and the frequency characteristic information D3, and outputs a final classification result D10.

[0053] The classifier 331 may be configured to use artificial intelligence to output the final classification result D10. Specifically, the classifier 331 may output the final classification result D10 using a machine learning model based on machine learning. In this case, the machine learning model used by the classifier 331 may be constructed by causing the learner to perform learning using learning data in which the data sets of the time-series waveform information D1, the time-series vibration position information D2, and the frequency characteristic information D3 correspond to the classification result (final classification result). Note that the machine learning model used by the classifier 331 may be constructed by, for example, deep learning.

[0054] In this manner, in this embodiment, a predetermined event is identified using a plurality of pieces of identification-use information. The identification-use information differs in factors such as the events and various conditions for which identification accuracy is high. In other words, each piece of identification-use information has different advantages. For this reason, for example, event identification using a single piece of identification-use information may not provide a sufficiently accurate identification result. Therefore, by performing event identification using multiple pieces of identification-use information as in this embodiment, it is possible to compensate for elements for which the identification accuracy is low using some identification-use information by using the advantages of other identification-use information, thereby improving the accuracy of the final identification result D10.

[0055] Furthermore, in this example, a final identification result D10 is output by inputting multiple pieces of identification-use information to one classifier 331. In this configuration, since there is only one classifier 331, the processing load is reduced and high-speed processing can be expected.

[0056] In this example, the event identification unit 303 may be configured to output the final identification result D10 in response to input of a plurality of (two in this case) pieces of the time-series waveform information D1, the time-series vibration position information D2, and the frequency characteristic information D3. In this case, when the identification-use information input to the identifier 331 is fixed, the information generating unit 302 may be configured to generate, from the identification-use information that can be generated, the identification-use information that is input to the identifier 331, and not generate the identification-use information that is not input to the identifier 331. Furthermore, some of the identification use information to be input to the identifier 331 may be changed in response to changes in the event to be identified or changes in various conditions in the event identification environment ENV.

[0057] [Example of event recognition unit configuration: Example 2] 7 shows a second example of the configuration of the event discrimination unit 303. The event discrimination unit 303 in the figure includes three primary discriminators 332 (332-1, 332-2, 332-3) corresponding to the time-series waveform information D1, the time-series vibration position information D2, and the frequency characteristic information D3, respectively. The event discrimination unit 303 also includes one secondary discriminator 333.

[0058] The primary classifier 332-1 receives time-series waveform information D1 as input and outputs a primary classification result according to the input time-series waveform information D1. The primary classifier 332-2 receives the time-series vibration position information D2 as input and outputs a primary classification result according to the input of the time-series vibration position information D2. The primary classifier 332-3 receives frequency characteristic information D3 as input and outputs a primary classification result according to the input frequency characteristic information D3. That is, in the second example event identification unit 303, as the first stage of event identification, three primary identification results are obtained corresponding to the individual identification use information of time series waveform information D1, time series vibration position information D2, and frequency characteristic information D3.

[0059] The primary classification results from the primary classifiers 332-1, 332-2, and 332-3 are input to the secondary classifier 333. The secondary classifier 333 outputs a classification result (final classification result) based on the three input primary classification results.

[0060] In this example, each piece of identification-use information is input to a corresponding primary classifier 332, thereby obtaining a primary classification result corresponding to the respective piece of identification-use information. In this case, each primary classifier 332 can be configured to be adapted to the corresponding identification-use information, thereby improving the accuracy of the primary classification result corresponding to the respective piece of identification-use information. Furthermore, the configuration of this example makes it easy to weight the primary classification result of each primary classifier 332 corresponding to each piece of identification-use information, for example by assigning a coefficient to it. Then, the secondary classifier 333 integrates these primary classification results to obtain the final classification result D10, thereby improving the accuracy of the final classification result D10 as well.

[0061] [Example of event recognition unit configuration: Example 3] 8 shows a third example of the configuration of the event discriminator 303. The event discriminator 303 in the figure includes one primary discriminator 334 and one secondary discriminator 335. The primary classifier 334 outputs a primary classification result according to the input of the time-series waveform information D1. The secondary classifier 335 outputs a final classification result D10 according to the input of the primary classification result output by the primary classifier 334, the time-series vibration position information D2, and the frequency characteristic information D3.

[0062] That is, in the event identification unit 303 of this example, the primary classifier 334 outputs a primary identification result in response to input of some of all the plurality of pieces of identification-use information, and the secondary classifier 335 outputs a final identification result D10 in response to input of the primary identification result output by the primary classifier 334 and the remaining identification-use information out of all the plurality of pieces of identification-use information that has not been input to the primary classifier 334.

[0063] In this example, the identification-use information input to the primary classifier 334 may be one or more pieces of identification-use information as long as it is a part of all of the plurality of pieces of identification-use information. Furthermore, the one or more pieces of identification-use information input to the primary classifier 334 and the remaining identification-use information input to the secondary classifier 335 may be changed as appropriate depending on conditions such as the phenomenon to be identified.

[0064] In this example, for example, a primary classification result may be obtained by inputting one or more pieces of basic identification-use information that are most suitable for identifying the phenomenon to be identified from among the plurality of pieces of identification-use information into the primary classifier 334. Then, in this example, the secondary classifier 335 is constructed so that elements with low classification accuracy in the primary classification result of the primary classifier 334 can be appropriately reinforced with other identification-use information, thereby making it possible to obtain a highly accurate final classification result D10.

[0065] Second Embodiment [Example of an event identification system configuration] Next, a second embodiment will be described. In the second embodiment, an event identification system based on the first embodiment is used to identify a predetermined identification item related to an event such as the speed of a moving body such as a vehicle moving on a road (an example of a moving surface).

[0066] A specific example of the configuration of the event identification system of this embodiment will be described with reference to Fig. 9. In Fig. 9, the same parts as in Fig. 1 are given the same reference numerals and their description will be omitted. In the event identification system shown in the figure, an optical fiber sensor 20 is buried under a road RD on which a vehicle CR such as an automobile moves as a moving body. The figure shows an example in which the optical fiber sensor 20 is buried so that its extension direction is aligned with the direction in which the vehicle CR travels on the road RD. The optical fiber 22 in the optical fiber sensor 20 thus provided changes the return light in response to vibrations and sounds generated as a moving object moves along the road RD. In the event identification device 300, the information generation unit 302 generates identification use information (time-series waveform information D1, time-series vibration position information D2, and frequency characteristic information D3) using the returned light received by the receiving unit 301. The event identification unit 303 outputs a final identification result D10 according to the input of the identification use information.

[0067] In such an event identification system, the identification utilization information (time series waveform information D1, time series vibration position information D2, and frequency characteristic information D3) generated by the information generation unit 302 of the event identification device 300 can be treated as information detected regarding vibrations that occur in response to a moving object moving on a road RD, for example. In addition, the event identification unit 303 is configured to be able to identify an event as the speed of a moving object moving on the road RD in response to the input of the identification use information.

[0068] The event identification unit 303 of this embodiment is configured to be able to identify the speed of a moving object by focusing on one of a plurality of identification items.

[0069] Specifically, an example of an identification item is the speed (short distance speed) of a moving object when moving over a short distance (for example, within about 20 m). Furthermore, an identification item paired with the short-distance speed can be the speed (long-distance speed) of a moving object moving over a long distance (for example, about 200 m). The long-distance speed may be the average speed over the section to be identified.

[0070] Furthermore, the speed of a moving object moving along the traffic direction of the road RD (traffic direction corresponding speed) can be cited as an identification item. Furthermore, an identification item that is paired with the travel direction speed is the speed of a moving object moving in the direction crossing the road RD (crossing direction speed).

[0071] Here, when identifying the above-mentioned various identification items, if the identification use information of this embodiment (time-series waveform information D1, time-series vibration position information D2, and frequency characteristic information D3) is used alone, the accuracy of the identification result will differ between the identification use information. In addition, the accuracy of the identification result (identification accuracy) will differ between the identification use information depending on the identification item.

[0072] 10(A), 10(B), and 10(C) show examples of the identification accuracy of the identification use information for each identification item. FIG. 10(A) shows a comparison of the identification accuracy of identification-usable information for short-distance speeds and long-distance speeds. In the figure, the identification accuracy is shown in three levels: "1," "2," and "3." Level "1" indicates that the identification accuracy is high above a certain level. Level "2" indicates that the identification accuracy is medium. Level "3" indicates that the identification accuracy is low below a certain level. These identification accuracy levels may be classified, for example, based on the probability that a correct identification result is obtained.

[0073] 10(A) shows that the identification accuracy of short-distance speed is medium at level "2" for the time-series waveform information D1 and frequency characteristic information D3, and low at level "3" for the time-series vibration position information D2. In this case, it can be said that the time-series vibration position information D2 has low superiority in identifying short-distance speed. As for the time-series vibration position information D2, the detected positions are scattered over time, particularly for large vehicles such as trucks, and therefore there are cases where sufficiently high accuracy cannot be obtained, and therefore the level is set to "3". That is, in terms of identifying short-distance speed, the time-series waveform information D1 and the frequency characteristic information D3 are superior to the time-series vibration position information D2.

[0074] 10(A), the identification accuracy of long-distance speed is high for the time-series vibration position information D2 at level "1," while it is medium for the time-series waveform information D1 and the frequency characteristic information D3 at level "2." In other words, it can be said that the time-series vibration position information D2 has a higher advantage over the time-series waveform information D1 and the frequency characteristic information D3 in terms of identification accuracy of long-distance speed.

[0075] FIG. 10(B) shows a comparison of the identification accuracy of the identification use information for the speed along the traffic direction of road RD (traffic direction speed) and the speed along the crossing direction of road RD (crossing direction speed). According to FIG. 10(B), the identification accuracy of the travel direction speed is level "1" for the time-series vibration position information D2, and level "3" for both the time-series waveform information D1 and the frequency characteristic information D3. That is, in terms of identifying the speed in the travel direction, the time-series vibration position information D2 is superior to the time-series waveform information D1 and the frequency characteristic information D3.

[0076] 10(B), the identification accuracy of the transverse direction velocity is level "2" for the time-series waveform information D1 and the frequency characteristic information D3, and level "3" for the time-series vibration position information D2. In other words, it can be said that the time-series waveform information D1 and the frequency characteristic information D3 are superior to the time-series vibration position information D2 in terms of the identification accuracy of the transverse direction velocity.

[0077] Furthermore, Figure 10(C) shows the identification accuracy of the identification utilization information for each of the following cases: "detection of extremely high-speed objects," "detection of multiple objects in a mixed state," "application to discrete vibrations," and "use as wheelbase information." The classification accuracy for "detection of an extremely high-speed object" is level "1" for the time-series waveform information D1, level "3" for the time-series vibration position information D2, and level "2" for the frequency characteristic information D3. In other words, the classification accuracy for "detection of an extremely high-speed object" is highest in the order of time-series waveform information D1, highest in frequency characteristic information D3, and lowest in time-series vibration position information D2. Furthermore, the classification accuracy for "detection of multiple objects in a mixed state" is level "2" for the time-series waveform information D1, level "3" for the time-series vibration position information D2, and level "1" for the frequency characteristic information D3. In other words, the classification accuracy for "detection of multiple objects in a mixed state" is highest in the frequency characteristic information D3, lowest in the time-series vibration position information D2, and lowest in the time-series waveform information D1. Furthermore, the identification accuracy for "application to discrete vibration" is level "2" to "3" for the time-series waveform information D1, level "2" for the time-series vibration position information D2, and level "2" to "3" for the frequency characteristic information D3. In other words, the identification accuracy for "application to discrete vibration" is considered to be approximately equal for the time-series waveform information D1, the time-series vibration position information D2, and the frequency characteristic information D3, and it can be said that the time-series vibration position information D2 has a higher advantage than the time-series waveform information D1 and the frequency characteristic information D3 depending on the conditions of the measurement environment, etc. Furthermore, the identification accuracy for "use as wheelbase information" is level "1" to "2" for the time-series waveform information D1, level "2" to "3" for the time-series vibration position information D2, and level "1" to "2" for the frequency characteristic information D3. In other words, the identification accuracy for "use as wheelbase information" is higher for the time-series waveform information D1 and the frequency characteristic information D3 than for the time-series vibration position information D2, and depending on the circumstances of the measurement environment, etc., it can be said that there are cases where the identification accuracy can be approximately equal among the time-series waveform information D1, the time-series vibration position information D2, and the frequency characteristic information D3.

[0078] In this way, which of the time-series waveform information D1, the time-series vibration position information D2, and the frequency characteristic information D3 is the identification use information that is superior for identification varies depending on the identification item. Therefore, the event identification device 300 of this embodiment is configured to be able to use, for identification, one or more pieces of identification use information that are advantageous for identifying the identification item designated as the identification target from among the time-series waveform information D1, the time-series vibration position information D2, and the frequency characteristic information D3.

[0079] An example of the configuration of the event identification device 300 in this embodiment will be described with reference to Figures 11 to 13. In each of Figures 11 to 13, the receiving unit 301 and the information generating unit 302 in the event identification device 300 are not shown. In Figures 11 to 13, the same parts as in Figures 6 to 8 are given the same reference numerals and their description will be omitted as appropriate.

[0080] 11 shows a first example of the configuration of an event identification device 300 in this embodiment. The event identification device 300 in the figure includes an identification target setting unit 305. The identification target setting unit 305 sets an identification item to be identified by the event identification unit 303 from among a plurality of identification items.

[0081] The storage unit 304 in the event identification device 300 in the figure also stores an identification item / identification-use information table 341. The identification item / identification-use information table 341 is a table in which identification-use information designated to be used for identification is associated with each identification item. The identification-use information associated with each identification item in the identification item / identification-use information table 341 may be determined based on the superiority of the identification accuracy obtained for each identification item, as exemplified in FIG. 10 (FIG. 10(A), FIG. 10(B), FIG. 10(C)).

[0082] The event identification device 300 in this embodiment also includes an input unit 306. The input unit 306 inputs, to the event identification unit 303, identification use information that has been specified as information to be used for identification in accordance with the identification item set by the identification target setting unit 305, from among the time-series waveform information D1, the time-series vibration position information D2, and the frequency characteristic information D3 generated by the information generating unit 302. At this time, the input unit 306 refers to the identification item / identification use information table 341 to identify the identification use information associated with the identification item set by the identification target setting unit 305. The input unit 306 may cause the event identification unit 303 to input the identified identification use information. The identification and use information input by the input unit 306 to the event identification unit 303 in this way is one or more of the time-series waveform information D1, the time-series vibration position information D2, and the frequency characteristic information D3. The input unit 306 may take in identification use information specified from the time-series waveform information D1, the time-series vibration position information D2, and the frequency characteristic information D3 output by the information generation unit 302, and input it to the event identification unit 303. Alternatively, the input unit 306 may cause the information generation unit 302 to generate identification use information that is recognized as being used for identification from the time-series waveform information D1, the time-series vibration position information D2, and the frequency characteristic information D3, and input the generated identification use information to the event identification unit 303.

[0083] In the configuration of the event identifying section 303 shown in FIG. 11, the input section 306 inputs the identification use information identified as information to be used for identification to one identifier 331 provided in the event identifying section 303.

[0084] 12 shows a second example of the configuration of the event identification device 300 in this embodiment. In this figure, the same parts as in FIG. 11 are given the same reference numerals and the description thereof will be omitted as appropriate. 11, the input unit 306 may specify identification-use information to be input to the event identification unit 303, as in the first example of FIG. 11. Then, when the identification-use information specified to be used for identification includes time-series waveform information D1, the input unit 306 inputs the time-series waveform information D1 to the primary classifier 332-1. Furthermore, when the identification-use information specified to be used for identification includes time-series vibration position information D2, the input unit 306 inputs the time-series vibration position information D2 to the primary classifier 332-2. Furthermore, when the identification-use information specified to be used for identification includes frequency characteristic information D3, the input unit 306 inputs the frequency characteristic information D3 to the primary classifier 332-3.

[0085] Fig. 13 shows a third example of the configuration of the event identification device 300 in this embodiment. In this figure, the same parts as in Fig. 11 and Fig. 12 are given the same reference numerals and the description thereof will be omitted as appropriate. 11, the input unit 306 may also specify the identification use information to be input to the event identification unit 303. Then, if the identification use information specified to be used for identification includes time-series waveform information D1, the input unit 306 inputs the time-series waveform information D1 to the primary classifier 334. Furthermore, when the identification use information specified to be used for identification includes time-series vibration position information D2, the input unit 306 inputs the time-series vibration position information D2 to the secondary classifier 335. When the identification use information specified to be used for identification includes frequency characteristic information D3, the input unit 306 inputs the frequency characteristic information D3 to the secondary classifier 335.

[0086] The input unit 306 may use a machine learning model to determine a combination of one or more pieces of identification-use information to input to the event identification unit 303. In this case, the machine learning model may be constructed by performing machine learning to estimate a combination pattern of one or more pieces of identification-use information that will provide the best identification result for each identification item.

[0087] <Modification> Modifications of the optical fiber sensor of this embodiment will be described below. The optical fiber sensor shown as a modification can be disposed in the event recognition environment ENV in place of the optical fiber sensor 20 of FIG.

[0088] 14 shows a configuration example of an optical fiber sensor 30 as a modified example of this embodiment. The figure shows a cross section of the optical fiber sensor 30 cut along a direction perpendicular to the extension direction. The optical fiber sensor 30 in the figure includes one optical fiber 22. The optical fiber 22 is covered with a solid inner coating 31. The inner coating 31 is further covered with a solid outer coating 32. The inner coating 31 and the outer coating 32 may be made of, for example, a resin. In the configuration shown in the figure, the portion corresponding to the core material (resin tube 21) of the optical fiber held sensor 20 in Figure 2 may be an inner coating 31 or an outer coating 32. The hardness of the inner coating 31 or the outer coating 32 as the core material in this embodiment may be equivalent to that of the core material (resin tube 21) of the optical fiber held sensor 20, for example.

[0089] When a structure includes one optical fiber 22, as in the optical fiber-carrying sensor 30 shown in the figure, one optical signal Sopt output from the one optical fiber 22 is input from the receiving unit 301 to the information generating unit 302 in the event identification device 300. The information generating unit 302 uses the one input optical signal Sopt to generate identification utilization information (time-series waveform information D1, time-series vibration position information D2, and frequency characteristic information D3).

[0090] FIG. 15 shows a configuration example of an optical fiber sensor 30A as another modified example of this embodiment. The optical fiber-carrying sensor 30A includes two optical fibers 22-1 and 22-2. For example, the optical fiber 22-1 may be used for detection (measurement), and the optical fiber 22-2 may be used for temperature compensation. The optical fiber 22-1 is covered with a solid inner coating 31-1, and the optical fiber 22-2 is covered with a solid inner coating 31-2. The inner coatings 31-1 and 31-2 are in contact with each other and are covered with a solid outer coating 32.

[0091] In the case of the optical fiber-held sensor 30A of this modification, the receiving unit 301 inputs the optical signals Sopt output from the optical fibers 22-1 and 22-2 to the information generating unit 302. The information generating unit 302 performs temperature correction using the optical signal Sopt from the optical fiber 22-2, and generates identification and utilization information (time-series waveform information D1, time-series vibration position information D2, and frequency characteristic information D3) using the optical signal Sopt from the optical fiber 22-1.

[0092] In this embodiment, the identification use information generated by the information generating unit 302 is not limited to the three, namely, the time-series waveform information D1, the time-series vibration position information D2, and the frequency characteristic information D3, and the information generating unit 302 may be configured to generate more identification use information. As an example, the information generating unit 302 may also generate information indicating a change in vibration intensity over time, instead of the vibration amplitude, based on the time-series waveform information D1.

[0093] Alternatively, a program for implementing the functions of the event identification device 300 may be recorded on a computer-readable recording medium, and the program may be loaded into a computer system and executed to perform the processing of the event identification device 300. Here, "loading a program recorded on a recording medium into a computer system and executing it" includes installing the program on a computer system. The term "computer system" here includes hardware such as an operating system and peripheral devices. The term "computer system" may also include multiple computer devices connected via a network, including the Internet, a wide area network (WAN), a local area network (LAN), or a dedicated line. The term "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs, and CD-ROMs, as well as storage devices such as hard disk drives (HDDs) and solid-state drives (SDDs) built into a computer system. Thus, the recording medium storing the program may be a non-transitory recording medium such as a CD-ROM. The recording medium may also include internal or external recording media accessible from a distribution server for distributing the program. The program code stored on the distribution server's recording medium may be different from the program code executable on a terminal device. In other words, the format in which the program is stored on the distribution server does not matter as long as it can be downloaded from the distribution server and installed in a form that is executable on the terminal device. The program may be divided into multiple parts, each of which may be downloaded at different times and then combined on the terminal device, or each of the divided programs may be distributed by a different distribution server. Furthermore, the term "computer-readable recording medium" also includes a storage medium that stores a program for a certain period of time, such as volatile memory (RAM) within a computer system that serves as a server or client when a program is transmitted over a network. The program may also be a program that realizes part of the above-described functions. Furthermore, the program may be a so-called differential file (differential program) that can realize the above-described functions in combination with a program already stored in the computer system. [Explanation of symbols]

[0094] 20, 20A, 30, 30A: Optical fiber-supported sensor; 21: Resin pipe; 22, 22-1, 22-2: Optical fiber; 31, 31-1, 31-2: Inner coating; 32: Outer coating; 300: Event classification device; 301: Receiving unit; 302: Information generation unit; 303: Event classification unit; 304: Memory unit; 305: Classification target setting unit; 306: Input unit; 331: Classifier; 332: Primary classifier; 332-1: Primary classifier; 332-2: Primary classifier; 332-3: Primary classifier; 333: Secondary classifier; 334: Primary classifier; 335: Secondary classifier; 341: Classification item / classification usage information table; CR: Vehicle; D1: Time-series waveform information; D2: Time-series vibration position information; D3: Frequency characteristic information; D10: Final classification result

Claims

1. an information generating unit that can generate, as identification information to be used for identifying an event, at least time-series vibration information indicating a change in vibration over time for each specific position, time-series vibration position information indicating a change in vibration position over time, and frequency characteristic information indicating a frequency characteristic of vibration in a unit time for each specific position, based on an optical signal output from an optical fiber arranged corresponding to the moving plane on which the moving body moves; an event identification unit that identifies an identification target using input identification utilization information; an identification target setting unit that sets an identification item to be identified in relation to an event such as the speed of the moving object moving on the moving surface; an input unit that inputs, to the event identification unit, one or more pieces of identification use information to be used for identifying the identification items set as targets by the identification target setting unit from among the time-series vibration information, the time-series vibration position information, and the frequency characteristic information; An event identification device comprising:

2. The identification target setting unit sets, as the identification target, an identification item designated from a plurality of identification items defined in relation to the event as the speed of the moving body.

2. The event identification device according to claim 1.

3. The input unit specifies identification-use information to be used for identifying the identification items set as targets by the identification target setting unit based on an identification-use information table showing identification-use information determined as suitable for identification for each of a plurality of identification items, and inputs the specified identification-use information to the event identification unit.

3. An event identification device according to claim 1 or 2.

4. The input unit selects one or more pieces of identification use information suitable for identifying the identification items set as targets by the identification target setting unit from among the pieces of identification use information including the time-series vibration information, the time-series vibration position information, and the frequency characteristic information, and inputs the selected identification use information to the event identification unit.

3. An event identification device according to claim 1 or 2.

5. The input unit determines one or more pieces of identification-use information to be input to the event identification unit in response to the identification item set as a target by the identification target setting unit, using a machine learning model obtained by performing machine learning to estimate a combination pattern of one or more pieces of identification-use information that will provide the best identification result for the identification item.

5. An event identification device according to claim 4.

6. The event identification unit includes a plurality of primary classifiers, each of which outputs a primary identification result in response to input of any one of the plurality of pieces of identification-use information, and a secondary classifier, each of which outputs a final identification result in response to input of the primary identification results output by the primary classifiers.

3. An event identification device according to claim 1 or 2.

7. The event identification unit includes a single classifier that outputs a final identification result in response to input of all of the identification-use information by the input unit.

3. An event identification device according to claim 1 or 2.

8. The event identification unit includes a predetermined number of primary classifiers that output a primary identification result in response to input of a portion of a plurality of pieces of identification-use information, and a secondary classifier that outputs a final identification result in response to input of the primary identification result and the identification-use information that is not input to the primary classifiers.

3. An event identification device according to claim 1 or 2.

9. An event identification method in an event identification device, comprising: an information generating step in which the information generating unit is capable of generating, as identification and utilization information used for identifying an event, at least time-series vibration information showing a change in vibration over time for each specific position, time-series vibration position information showing a change in vibration position over time, and frequency characteristic information showing a frequency characteristic of vibration in a unit time for each specific position, based on an optical signal output from an optical fiber arranged corresponding to a moving plane on which the moving body moves; an event identification step in which the event identification unit identifies an identification target by using the input identification use information; an identification target setting step in which an identification target setting unit sets an identification item to be identified in relation to an event such as the speed of the moving body moving on the moving surface; an input step in which an input unit inputs, to the event identification step, one or more pieces of identification use information to be used for identifying the identification item set as a target in the identification target setting step from among the time-series vibration information, the time-series vibration position information, and the frequency characteristic information; An event identification method comprising:

10. The computer as an event discriminator an information generating unit that is capable of generating, as identification information to be used for identifying an event, at least time-series vibration information indicating a change in vibration over time for each specific position, time-series vibration position information indicating a change in vibration position over time, and frequency characteristic information indicating a frequency characteristic of vibration in a unit time for each specific position, based on an optical signal output from an optical fiber arranged corresponding to the moving plane on which the moving body moves; an event identification unit that identifies an identification target using input identification utilization information; an identification target setting unit that sets an identification item to be identified in relation to an event such as the speed of the moving object moving on the moving surface; an input unit that inputs, to the event identification unit, one or more pieces of identification use information to be used for identifying the identification items set as targets by the identification target setting unit from among the time-series vibration information, the time-series vibration position information, and the frequency characteristic information; A program to function as a

Images