Abnormal noise detection system and abnormal noise detection program

Abstract

To provide an abnormal sound detection system and an abnormal sound detection program which reduce an influence due to environmental conditions so as to be able to detect an abnormal sound with higher accuracy than before.SOLUTION: An abnormal sound detection system 1 includes: a microphone 80A for detecting sound in an environment in which an object is provided; an environment sensor 80B for detecting environmental information indicating predetermined types of environmental conditions in the environment; an adjustment section 11C for adjusting a threshold used when an abnormal sound from the object is detected by using the detected environmental information; and a detection section 11D for detecting the abnormal sound by using the threshold adjusted by the adjustment section 11C.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present invention relates to an abnormal sound detection system and an abnormal sound detection program. [Background technology]

[0002] Energy equipment such as gas engines and gas turbines must be able to generate energy at all times, so inspections are essential.

[0003] Conventionally, there has been a technology that uses sounds picked up by a microphone to inspect such energy facilities and other objects without visiting the site. This technology allows inspections to be carried out at any time, even if the object is in a remote location, and furthermore, it allows automatic detection of abnormalities by computer, making it extremely convenient.

[0004] As such, the following conventional techniques are applicable to inspecting objects and detecting abnormalities using sound.

[0005] Patent Document 1 discloses an abnormal sound detection device that aims to improve the accuracy of identifying whether or not the collected sound contains an abnormal sound, even when the sound collected from the target device contains not only steady sounds but also disturbance sounds such as sudden sounds, and to appropriately detect whether or not there is an abnormality in the target device.

[0006] This abnormal sound detection device includes an input unit connected to a microphone that picks up sound emitted from the target device; a memory that stores sound data for a predetermined period of time of sound emitted from the target device; and a processor that uses the sound data for the predetermined period of time to determine, for each one-frame period that is shorter than the predetermined period, whether the sound corresponding to the sound data for the one-frame period is a steady sound or a sudden sound. The memory stores at least one steady sound as a normal sound from among the one or more one-frame period sound data that are determined to be the steady sound during a learning mode, and the processor identifies whether the sound emitted from the target device includes an abnormal sound by comparing the one-frame period sound data that is determined to be the steady sound during an operation mode different from the learning mode with the sound data of the normal sound stored in the memory.

[0007] Furthermore, Patent Document 2 discloses an abnormal noise detection device that aims to reduce the operational costs of correction procedures when changing the type of sensor or setting conditions without reducing the ability to detect abnormal noise.

[0008] This abnormal sound detection device detects abnormal sounds from the operating sounds of a detected object, and comprises an unchanging interval determination unit that references status information indicating the operating state of the detected object and determines whether the operation of the detected object is an operation in an unchanging interval, which is a time interval in which there is no difference in operating sounds resulting from the difference between normal operation and abnormal operation of the detected object, and a correction parameter generation unit that, when the unchanging interval determination unit determines that the operation is an operation in an unchanging interval, generates correction parameters for correcting the observation signals of the detected object in a time interval outside the unchanging interval from the observation signals that observe the operating sounds of the detected object in the unchanging interval.

[0009] Furthermore, this abnormal sound detection device includes a feature extraction unit that, when the unchanging interval determination unit determines that the operation occurs in a time interval outside the unchanging interval, extracts feature amounts of the operation sound of the detection object in the time interval outside the unchanging interval based on the observed signal of the detection object in the time interval outside the unchanging interval and the correction parameters generated by the correction parameter generation unit, and an abnormal sound determination unit that determines whether or not an abnormal sound is occurring in the detection object based on the feature amounts extracted by the feature extraction unit.

[0010] Furthermore, Patent Document 3 discloses an abnormality diagnosis device that aims to easily identify abnormalities such as breakdowns by measuring measurement targets such as sounds and vibrations during operation of the device to be diagnosed.

[0011] This abnormality diagnosis device comprises a measurement means for measuring the measurement object over time during the operation of the diagnosed device to be diagnosed for the presence or absence of an abnormality, and a control means for handling the measurement data obtained by the measurement means in a state synchronized in time with the operation of the diagnosed device based on synchronization information indicating a predetermined operating state of the diagnosed device. [Prior art documents] [Patent documents]

[0012] [Patent Document 1] Japanese Patent Publication No. 2020-183904 [Patent Document 2] Patent No. 5925397 [Patent Document 3] Japanese Patent Application Publication No. 2019-110392 Summary of the Invention [Problem to be solved by the invention]

[0013] Generally, regardless of whether the sound is indoors or outdoors, noise depending on environmental conditions such as wind and weather is superimposed on the sound, and in order to detect the occurrence of abnormal sounds with high accuracy, it is necessary to take these environmental conditions into consideration.

[0014] In contrast to this, the conventional technologies described in Patent Documents 1 to 3 do not take environmental conditions into consideration, and have the problem that they are not always able to detect abnormal noises with high accuracy.

[0015] The present invention has been made in consideration of the above circumstances, and aims to provide an abnormal sound detection system and an abnormal sound detection program that can reduce the effects of environmental conditions and detect abnormal sounds with higher accuracy. [Means for solving the problem]

[0016] The abnormal sound detection system according to claim 1 comprises a first detection unit that detects a sound in an environment in which an object is placed, a second detection unit that detects environmental information indicating a predetermined type of environmental condition in the environment, an adjustment unit that adjusts a threshold value used when detecting an abnormal sound from the object using the environmental information detected by the second detection unit, and a detection unit that detects the abnormal sound using the threshold value adjusted by the adjustment unit. The threshold value is derived using a threshold estimation model that has been machine-learned in advance, with input information being a signal obtained by Fourier transforming a normal sound signal indicating the sound detected by the first detection unit when the object is normal and an abnormal sound signal indicating the sound detected by the first detection unit when the object is abnormal, and with output information being threshold information indicating the threshold value. .

[0017] According to the abnormal sound detection system of claim 1, while detecting sounds in the environment in which the object is located, environmental information indicating predetermined types of environmental conditions in the environment is detected, and the detected environmental information is used to adjust the threshold used when detecting abnormal sounds from the object, and the adjusted threshold is used to detect abnormal sounds, thereby reducing the influence of environmental conditions and enabling more accurate detection of abnormal sounds. Furthermore, according to the abnormal sound detection system of claim 1, the threshold value is derived using a threshold estimation model that has been machine-learned in advance and that uses as input information the signals obtained by Fourier transforming a normal sound signal that indicates a sound detected when the object is normal and an abnormal sound signal that indicates a sound detected when the object is abnormal, and threshold information that indicates the threshold value is used as output information, thereby making it possible to detect abnormal sounds with higher accuracy than when the threshold estimation model is not used.

[0020] The abnormal sound detection system according to claim 2 comprises a first detection unit that detects sound in an environment in which an object is placed, a second detection unit that detects environmental information that indicates a predetermined type of environmental condition in the environment, an adjustment unit that adjusts a threshold value used in detecting abnormal sounds from the object using the environmental information detected by the second detection unit, and a detection unit that detects the abnormal sounds using the threshold value adjusted by the adjustment unit, wherein the environmental information includes at least one of weather information related to weather and environmental value information related to environmental values, the weather information includes weather information that indicates the weather, and the environmental value information includes at least one of wind speed, wind direction, and humidity. .

[0021] According to the abnormal sound detection system of claim 2, while detecting sounds in the environment in which the object is located, environmental information indicating predetermined types of environmental conditions in the environment is detected, and the detected environmental information is used to adjust the threshold used when detecting abnormal sounds from the object, and the adjusted threshold is used to detect abnormal sounds, thereby reducing the influence of environmental conditions and enabling more accurate detection of abnormal sounds. Also, Claim 2According to the abnormal sound detection system described in the above, by including environmental information that includes at least one of weather information related to weather and environmental value information related to environmental values, the threshold value can be adjusted according to the included information. Furthermore, according to the abnormal sound detection system described in claim 2, by including weather information indicating the weather and by including at least one of wind speed, wind direction, and humidity as the environmental value information, it is possible to adjust the threshold value according to the included information.

[0024] Claim 3 is an abnormal sound detection program executed in an abnormal sound detection system including a first detection unit that detects sounds in an environment in which an object is placed, and a second detection unit that detects environmental information indicating a predetermined type of environmental condition in the environment, and the abnormal sound detection program adjusts a threshold value used when detecting an abnormal sound from the object using the environmental information detected by the second detection unit, and detects the abnormal sound using the adjusted threshold value. the threshold is derived using a threshold estimation model that has been machine-learned in advance and that uses as input information signals obtained by Fourier transforming a normal sound signal indicating the sound detected by the first detection unit when the object is normal and an abnormal sound signal indicating the sound detected by the first detection unit when the object is abnormal, and threshold information indicating the threshold as output information. The processing is executed by a computer.

[0025] Claim 3 According to the abnormal sound detection program described in the above, while detecting sounds in an environment in which an object is placed, environmental information indicating predetermined types of environmental conditions in the environment is detected, and the detected environmental information is used to adjust a threshold value used when detecting abnormal sounds from the object, and the adjusted threshold value is used to detect abnormal sounds, thereby reducing the influence of environmental conditions and enabling more accurate detection of abnormal sounds. Furthermore, according to the abnormal sound detection program of claim 3, the threshold value is derived using a threshold estimation model that has been machine-learned in advance and that uses as input information signals obtained by Fourier transforming a normal sound signal that indicates a sound detected when the object is normal and an abnormal sound signal that indicates a sound detected when the object is abnormal, and threshold information that indicates the threshold value is used as output information, thereby making it possible to detect abnormal sounds with higher accuracy than when the threshold estimation model is not used. The abnormal sound detection program described in claim 4 is an abnormal sound detection program executed in an abnormal sound detection system including a first detection unit that detects sounds in an environment in which an object is located, and a second detection unit that detects environmental information that indicates a predetermined type of environmental condition in the environment, the abnormal sound detection program causing a computer to execute processing that adjusts a threshold value used when detecting abnormal sounds from the object using the environmental information detected by the second detection unit, and detects the abnormal sound using the adjusted threshold value, the environmental information including at least one of weather information related to weather and environmental value information related to environmental values, the weather information including weather information that indicates the weather, and the environmental value information including at least one of wind speed, wind direction, and humidity. According to the abnormal sound detection program of claim 4, while detecting sounds in the environment in which the object is located, environmental information indicating predetermined types of environmental conditions in the environment is detected, and the detected environmental information is used to adjust the threshold used when detecting abnormal sounds from the object, and the adjusted threshold is used to detect abnormal sounds, thereby reducing the influence of environmental conditions and enabling more accurate detection of abnormal sounds. Furthermore, according to the abnormal sound detection program described in claim 4, by including at least one of weather information related to weather and environmental value information related to environmental values, the threshold can be adjusted according to the included information. Furthermore, according to the abnormal sound detection program of claim 4, by including weather information indicating the weather and including at least one of wind speed, wind direction, and humidity as the weather information, the threshold value can be adjusted according to the included information. [Effects of the Invention]

[0026] As described above, according to the present invention, it is possible to reduce the influence of environmental conditions and detect abnormal noise with higher accuracy. [Brief explanation of the drawings]

[0027] [Figure 1]1 is a block diagram showing an example of the overall configuration of an abnormal sound detection system according to an embodiment. FIG. [Figure 2] FIG. 1 is a block diagram illustrating an example of a hardware configuration of an abnormal sound detection device according to an embodiment. [Figure 3] FIG. 2 is a block diagram showing an example of the functional configuration of the abnormal sound detection device according to the embodiment when learning a threshold estimation model. [Figure 4] FIG. 2 is a block diagram showing an example of the functional configuration of the abnormal sound detection device according to the embodiment when a threshold estimation model is in operation. [Figure 5] FIG. 2 is a schematic diagram illustrating an example of a configuration of a learning signal database according to the embodiment. [Figure 6] 10 is a flowchart illustrating an example of the flow of a learning process according to the embodiment. [Figure 7] 10 is a flowchart illustrating an example of the flow of an abnormal sound detection process according to the embodiment. [Figure 8] FIG. 10 is a front view showing an example of the configuration of an abnormality presentation screen according to the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0028] Hereinafter, an example of an embodiment of the present invention will be described in detail with reference to the drawings. Note that in this embodiment, the object of the present invention will be described as an energy facility such as a gas engine or a gas turbine owned by an energy supply company, but the present invention is not limited thereto. For example, any object that is the object of the present invention, such as other facilities such as air conditioning facilities or water supply facilities, or information processing devices such as supercomputers, can be used as the object of the present invention as long as it is an object to be detected for the occurrence of an abnormality.

[0029] First, the configuration of an abnormal sound detection system 1 according to this embodiment will be described with reference to Fig. 1. Fig. 1 is a block diagram showing an example of the overall configuration of the abnormal sound detection system 1 according to this embodiment.

[0030] 1, the abnormal sound detection system 1 according to this embodiment includes an abnormal sound detection device 10 that plays a central role in the system. The abnormal sound detection system 1 according to this embodiment also includes an information storage device 30 that stores various databases, and a weather information server 90 that stores various weather information.

[0031] Furthermore, the abnormal sound detection system 1 of this embodiment includes a microphone (hereinafter referred to as a "mic") 80A and an environmental sensor 80B, each of which is provided in correspondence with an energy facility 80 owned by each of the multiple energy supply companies 70 that are subject to management by the abnormal sound detection system 1.

[0032] The microphone 80A according to this embodiment detects sounds in an environment in which an object (in this embodiment, the energy equipment 80) is installed, and is used to detect abnormal sounds from the object. Therefore, the microphone 80A according to this embodiment has sensitivity that allows it to detect abnormal sounds that are expected to be generated from the object.

[0033] In this embodiment, a single microphone 80A is used to collect sound emitted from the corresponding energy equipment 80 from one direction, but this is not limited to this. For example, a configuration may be adopted in which a plurality of microphones 80A are provided and the plurality of microphones 80A collect sound emitted from the corresponding energy equipment 80 thoroughly. Note that in this embodiment, a directional microphone is used as the microphone 80A, but this is not limited to this, and an omnidirectional microphone may be used as the microphone 80A.

[0034] Furthermore, the environmental sensor 80B according to this embodiment is for detecting environmental value information relating to environmental values, and is installed near the corresponding energy facility 80 within the premises of the corresponding energy supply company 70. The environmental sensor 80B according to this embodiment measures wind speed as the environmental value information, but is not limited to this. For example, the environmental sensor 80B may be configured to measure at least one of wind direction and humidity as the environmental value information in addition to wind speed.

[0035] On the other hand, the weather information server 90 according to this embodiment is a server that sequentially accumulates weather information related to the weather (weather information indicating the weather in this embodiment) that cannot be measured by the environmental sensor 80B, among the environmental information that is the detection target of the abnormal sound detection system 1. The weather information stored by the weather information server 90 can be obtained from information made public by the Japan Meteorological Agency, Weathernews, etc.

[0036] The information storage device 30 according to this embodiment also includes a non-volatile storage unit 32. The storage unit 32 is realized by a hard disk drive (HDD), a solid state drive (SSD), a flash memory, or the like. A training signal database 32A is stored in the storage unit 32 as a storage medium. The training signal database 32A will be described in detail later.

[0037] The abnormal sound detection device 10, the information storage device 30, the microphone 80A and the environmental sensor 80B provided in each energy supply company 70, and the weather information server 90 are connected via a network N. The abnormal sound detection device 10 is capable of communicating with the information storage device 30, the microphone 80A, the environmental sensor 80B, and the weather information server 90 via the network N. In this embodiment, the network N is a public communication line such as the Internet or a telephone line, but is not limited to this form. The network N may also be an in-company communication line such as a LAN (Local Area Network) or a WAN (Wide Area Network), or a combination of these in-company communication lines and public communication lines. In this embodiment, the network N is a wired communication line, but is not limited to this form. A wireless communication line or a combination of wired and wireless communication lines may also be used.

[0038] Note that examples of the abnormal sound detection device 10, the information storage device 30, and the weather information server 90 according to this embodiment include various information processing devices such as personal computers and server computers.

[0039] Next, the hardware configuration of the abnormal sound detection device 10 according to this embodiment will be described with reference to Fig. 2. Fig. 2 is a block diagram showing an example of the hardware configuration of the abnormal sound detection device 10 according to this embodiment.

[0040] The abnormal sound detection device 10 according to this embodiment includes a CPU (Central Processing Unit) 11, a memory 12 as a temporary storage area, a non-volatile storage unit 13, an input unit 14 such as a keyboard and mouse, a display unit 15 such as a liquid crystal display, a medium read / write device (R / W) 16, and a communication interface (I / F) unit 18. The CPU 11, memory 12, storage unit 13, input unit 14, display unit 15, medium read / write device 16, and communication I / F unit 18 are connected to one another via a bus B. The medium read / write device 16 reads information written in a recording medium 17 and writes information to the recording medium 17.

[0041] The storage unit 13 is realized by an HDD, SSD, flash memory, etc. A learning program 13A and an abnormal sound detection program 13B are stored in the storage unit 13 as a storage medium. Each of these programs is stored (installed) in the storage unit 13 by setting a recording medium 17, on which the respective programs have been written, in the medium reading and writing device 16, and having the medium reading and writing device 16 read each of the above programs from the recording medium 17. The CPU 11 sequentially reads each of the learning program 13A and the abnormal sound detection program 13B from the storage unit 13, expands them in the memory 12, and sequentially executes the processes of each of the programs.

[0042] The storage unit 13 also stores a threshold estimation model 13C that estimates a threshold used when detecting abnormal noise from an object.

[0043] The threshold estimation model 13C according to this embodiment receives as input information a signal obtained by Fourier transforming a normal sound signal indicating a sound detected by the microphone 80A when the object is normal under predetermined reference environmental conditions (hereinafter referred to as "reference environmental conditions"). The threshold estimation model 13C according to this embodiment also receives as input information a signal obtained by Fourier transforming an abnormal sound signal indicating a sound detected by the microphone 80A when the object is abnormal under the reference environmental conditions. The threshold estimation model 13C according to this embodiment receives as output information threshold information indicating the threshold. The threshold estimation model 13C according to this embodiment has been machine-learned in advance using this input information and output information.

[0044] In this embodiment, the conditions applied as the reference environmental conditions are fine weather and a wind speed of 5 (m / s), but it goes without saying that the conditions are not limited to this. Also, in this embodiment, one threshold estimation model 13C is constructed corresponding to each energy facility 80, but the present invention is not limited to this. For example, a threshold estimation model 13C may be constructed corresponding to each energy facility 80 and for each type of abnormal noise to be detected.

[0045] The threshold estimation model 13C according to this embodiment is an AI (Artificial Intelligence) model using an RNN (Recurrent Neural Network), but is not limited thereto. For example, other machine learning models such as AI other than an RNN, such as an AI model using a multilayer perceptron, may be applied as the threshold estimation model 13C.

[0046] Next, the functional configuration of the abnormal sound detection device 10 according to this embodiment when learning the threshold estimation model 13C will be described with reference to Fig. 3. Fig. 3 is a block diagram showing an example of the functional configuration of the abnormal sound detection device 10 according to this embodiment when learning the threshold estimation model 13C.

[0047] 3, when learning the threshold estimation model 13C, the abnormal sound detection device 10 includes a signal processing unit 11A and a learning unit 11B. The CPU 11 of the abnormal sound detection device 10 executes the learning program 13A, thereby functioning as the signal processing unit 11A and the learning unit 11B.

[0048] The signal processing unit 11A according to this embodiment performs signal processing (hereinafter simply referred to as "signal processing") that performs a Fourier transform on the above-described normal sound signal and abnormal sound signal (hereinafter collectively referred to as "learning signals"). In this embodiment, the learning signals are acquired by reading normal sound signals and abnormal sound signals that have been obtained in advance by a microphone 80A and registered from a learning signal database 32A (see also FIG. 5), which will be described later. However, this is not a limitation, and normal sound signals and abnormal sound signals that are publicly available on the Internet or the like may also be acquired via a network N or the like.

[0049] The learning unit 11B according to this embodiment uses the learning signal processed by the signal processing unit 11A as input information and the threshold information described above as output information to perform machine learning of the threshold estimation model 13C. Note that in this embodiment, the threshold information is also acquired by reading it from a learning signal database 32A (described later), but it goes without saying that this is not limitative.

[0050] Next, the functional configuration of the abnormal sound detection device 10 according to this embodiment when the threshold estimation model 13C is in operation will be described with reference to Fig. 4. Fig. 4 is a block diagram showing an example of the functional configuration of the abnormal sound detection device 10 according to this embodiment when the threshold estimation model 13C is in operation.

[0051] 4, the abnormal sound detection device 10 when the threshold estimation model 13C is in operation includes an adjustment unit 11C and a detection unit 11D in addition to the signal processing unit 11A that is the same as that applied during learning. The CPU 11 of the abnormal sound detection device 10 executes the abnormal sound detection program 13B, thereby functioning as the signal processing unit 11A, adjustment unit 11C, and detection unit 11D.

[0052] The signal processing unit 11A according to this embodiment performs the above-described signal processing on the audio signal detected by the microphone 80A. The audio signal after the signal processing by the signal processing unit 11A is input to the machine-learned threshold estimation model 13C. In response to this, the threshold estimation model 13C outputs the above-described threshold information.

[0053] Therefore, the adjustment unit 11C according to this embodiment adjusts the threshold indicated by the threshold information output from the threshold estimation model 13C using the environmental information acquired from the environmental sensor 80B and the weather information server 90.

[0054] In the adjustment unit 11C according to this embodiment, when the wind speed is higher than the wind speed applied as the reference environmental condition (5 (m / s) in this embodiment), the threshold value is adjusted to be larger as the wind speed increases. In contrast, in the adjustment unit 11C according to this embodiment, when the wind speed is lower than the wind speed applied as the reference environmental condition, the threshold value is adjusted to be smaller as the wind speed decreases.

[0055] In the adjustment unit 11C of this embodiment, if the weather is accompanied by the sound of rain, sleet, hail, sleet, thunder, etc., a value predetermined depending on the type of weather is further added to the adjusted threshold value based on the wind speed.

[0056] The various values ​​that are adjusted by adjusting these thresholds are assumed to be adjusted according to the environmental conditions to which the thresholds correspond by applying these values, and values ​​obtained in advance by computer simulations using actual equipment or the design specifications of the equipment are applied. However, this is not a limited form, and for example, the various values ​​may be input by the user according to the accuracy of abnormality detection required of the abnormal sound detection device 10. Furthermore, the method of adjusting the thresholds is not limited to the above method, and for example, adjustment may be made by taking into account only the influence of wind speed, without taking into account the influence of weather.

[0057] The detection unit 11D according to this embodiment detects abnormal noise using the threshold value adjusted by the adjustment unit 11C. Note that in this embodiment, the detection result by the detection unit 11D is presented by display on the display unit 15, but this is not limiting. For example, the detection result by the detection unit 11D may be presented by printing using an image forming device or by audio using an audio playback device.

[0058] Next, the learning signal database 32A according to this embodiment will be described with reference to Fig. 5. Fig. 5 is a schematic diagram showing an example of the configuration of the learning signal database 32A according to this embodiment.

[0059] As shown in FIG. 5, the learning signal database 32A according to this embodiment stores information such as equipment ID (Identification), normal sound signal, abnormal sound signal, and threshold value in association with each other.

[0060] The facility ID is information that is assigned in advance and is different for each energy facility 80 in order to identify each energy facility 80 that is subject to management by the abnormal sound detection system 1. Furthermore, the normal sound signal, the abnormal sound signal, and the threshold value are information that respectively indicate the normal sound signal, the abnormal sound signal, and the threshold value themselves.

[0061] In this embodiment, a value designated by a person is applied as the threshold information as a value capable of distinguishing between a sound indicated by a corresponding normal sound signal and a sound indicated by a corresponding abnormal sound signal, but this is not limited to this. For example, a median value between the average level of the corresponding normal sound signal per predetermined period and the average level of the corresponding abnormal sound signal per predetermined period may be automatically applied as the threshold information.

[0062] Next, the operation of the abnormal sound detection device 10 according to this embodiment will be described with reference to Figs. 6 to 8. First, the operation of the abnormal sound detection device 10 when executing a learning process will be described with reference to Fig. 6. Fig. 6 is a flowchart showing an example of the flow of the learning process according to this embodiment. Note that, in order to avoid confusion, the case will be described where a learning signal database 32A has been constructed in advance, in which the amount of information required for the learning process has been registered.

[0063] When a user of the abnormal sound detection device 10 or the like inputs an instruction to execute a learning process via the input unit 14, the CPU 11 of the abnormal sound detection device 10 executes the learning program 13A, thereby executing the learning process shown in FIG. 6.

[0064] 6, the CPU 11 reads out from the learning signal database 32A one set of information on a normal sound signal, an abnormal sound signal, and a threshold value (hereinafter referred to as "processing target information") for any energy facility 80 (hereinafter referred to as "processing target facility") that is subject to management by the abnormal sound detection system 1. In step 102, the CPU 11 performs the above-described signal processing on the read normal sound signal and abnormal sound signal.

[0065] In step 104, the CPU 11 uses the processed normal sound signal and abnormal sound signal as input information, and the information indicating the read threshold as output information (correct answer information), to machine-learn the threshold estimation model 13C.

[0066] In step 106, the CPU 11 determines whether or not the machine learning in step 104 has been completed for all information stored in the learning signal database 32A, and if the determination is negative, the process returns to step 100, whereas if the determination is positive, the process proceeds to step 108. When repeatedly executing the processes of steps 100 to 106, the CPU 11 sets information that has not been subject to processing up to that point as information to be processed.

[0067] In step 108, the CPU 11 determines whether or not the processing of steps 100 to 106 has been completed for all energy facilities 80. If the determination is negative, the process returns to step 100, whereas if the determination is positive, the learning process ends. When repeatedly executing the processing of steps 100 to 108, the CPU 11 designates an energy facility 80 that has not been the processing target up to that point as the processing target facility.

[0068] Through the above learning process, the threshold estimation model 13C is learned.

[0069] Next, the operation of the abnormal sound detection device 10 when performing the abnormal sound detection process will be described with reference to Figures 7 and 8. Figure 7 is a flowchart showing an example of the flow of the abnormal sound detection process according to this embodiment.

[0070] The CPU 11 of the abnormal sound detection device 10 executes the abnormal sound detection program 13B at predetermined times (every hour of each day in this embodiment), thereby executing the abnormal sound detection process shown in Fig. 7. Note that the case where learning of the threshold estimation model 13C has been completed will be described here.

[0071] In step 200 of FIG. 7, the CPU 11 starts inputting, into the corresponding threshold estimation model 13C, an audio signal that has undergone the above-described signal processing (hereinafter referred to as the "audio signal to be processed"), which is acquired in real time from the microphone 80A for any of the energy equipment 80 (equipment to be processed) that is subject to management by the abnormal sound detection system 1.

[0072] In step 202, the CPU 11 acquires threshold information output from the threshold estimation model 13C in response to the input of the audio signal to be processed, and in step 204, the CPU 11 acquires environmental information at this time from the corresponding environmental sensor 80B and the weather information server 90.

[0073] In step 206, the CPU 11 uses the acquired environmental information to adjust the threshold indicated by the threshold information acquired from the threshold estimation model 13C as described above.

[0074] In step 208, the CPU 11 compares the audio signal to be processed with the adjusted threshold value to determine whether or not an abnormal sound is occurring. If the determination is negative, the CPU 11 proceeds to step 214, whereas if the determination is positive, the CPU 11 proceeds to step 210.

[0075] In step 210, the CPU 11 controls the display unit 15 to display an abnormality display screen having a predetermined configuration, and in step 212, the CPU 11 waits until predetermined information is input.

[0076] An example of an abnormality notification screen according to this embodiment is shown in Fig. 8. As shown in Fig. 8, the abnormality notification screen according to this embodiment displays a message indicating that an abnormality may have occurred in the target equipment, as well as information indicating the target equipment and a message indicating that an abnormal noise is occurring. Therefore, the user can grasp this information by referring to the abnormality notification screen.

[0077] 8 is displayed on the display unit 15, the user checks the displayed content and then selects the end button 15A via the input unit 14. In response to this, the determination in step 212 is affirmative, and the process proceeds to step 214.

[0078] In step 214, the CPU 11 determines whether a predetermined end timing has arrived, and if the determination is negative, the process returns to step 202, whereas if the determination is positive, the process proceeds to step 216. Note that in this embodiment, the end timing is set to every minute, but it goes without saying that the present invention is not limited to this.

[0079] In step 216, the CPU 11 ends the input of the processing target audio signal to the corresponding threshold estimation model 13C, which was started in the processing of step 200. In step 218, the CPU 11 determines whether or not the processing of steps 200 to 216 has been completed for all energy facilities 80, and if the determination is negative, the process returns to step 200, whereas if the determination is positive, the abnormal sound detection process ends. Note that when repeatedly executing the processing of steps 200 to 218, the CPU 11 designates an energy facility 80 that has not been designated as a processing target up to that point as a processing target facility.

[0080] As described above, according to this embodiment, while detecting sounds in an environment in which an object is located, environmental information indicating predetermined types of environmental conditions in the environment is detected, the detected environmental information is used to adjust the threshold used when detecting abnormal sounds from the object, and the adjusted threshold is used to detect abnormal sounds. Therefore, the influence of environmental conditions is reduced, and abnormal sounds can be detected with higher accuracy.

[0081] Furthermore, according to this embodiment, the threshold value is derived using a threshold estimation model that has been machine-learned in advance, with input information being signals obtained by Fourier transforming a normal sound signal indicating a sound detected when the object is normal and an abnormal sound signal indicating a sound detected when the object is abnormal, and with threshold information indicating the threshold value being output information. Therefore, abnormal sounds can be detected with higher accuracy than when a threshold estimation model is not used.

[0082] Furthermore, according to this embodiment, the environmental information includes at least one of weather information related to the weather and environmental value information related to the environmental values, so that the threshold value can be adjusted depending on the included information.

[0083] Furthermore, according to this embodiment, the weather information includes weather information indicating the weather, and the environmental value information includes at least one of wind speed, wind direction, and humidity. Therefore, the threshold value can be adjusted according to the included information.

[0084] In the above embodiment, the weather information and environmental value information in the environmental information are acquired from the weather information server 90 and the environmental sensor 80B, but the present invention is not limited to this. For example, the weather information and environmental value information may be acquired from information published on the Internet or the like.

[0085] In the above embodiment, the case where the threshold value before adjustment by the adjustment unit 11C is derived using AI has been described, but the present invention is not limited to this. For example, the threshold value may be derived using a statistical method such as regression analysis, or the threshold value under the reference environmental conditions may be set by the user for each of the corresponding energy facilities 80.

[0086] In the above embodiment, the case where the threshold value before adjustment by the adjustment unit 11C is derived using the threshold value estimation model 13C has been described, but the present invention is not limited to this. For example, the threshold value after adjustment may be derived using the threshold value estimation model 13C. In this case, a threshold value estimation model 13C may be constructed for each conceivable environmental condition, and the adjusted threshold value may be derived using the threshold value estimation model 13C that matches or corresponds to the closest environmental condition to the actual environmental condition.

[0087] Furthermore, in the above embodiment, for example, the following various processors can be used as the hardware structure of the processing units that execute the processes of the signal processing unit 11A, the learning unit 11B, the adjustment unit 11C, and the detection unit 11D. As described above, the various processors include a CPU, which is a general-purpose processor that executes software (programs) and functions as a processing unit, as well as dedicated electrical circuits that are processors having a circuit configuration specifically designed to execute specific processes, such as a programmable logic device (PLD), which is a processor whose circuit configuration can be changed after manufacture, such as an FPGA (Field-Programmable Gate Array), and an ASIC (Application Specific Integrated Circuit).

[0088] The processing unit may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA).The processing unit may also be configured with a single processor.

[0089] Examples of configuring a processing unit with a single processor include, first, a form in which one processor is configured with a combination of one or more CPUs and software, and this processor functions as the processing unit, as typified by computers such as client and server. Second, a form in which a processor is used to realize the functions of the entire system, including the processing unit, on a single IC (Integrated Circuit) chip, as typified by systems on chips (SoCs). In this way, the processing unit is configured using one or more of the above-mentioned various processors as a hardware structure.

[0090] Furthermore, more specifically, the hardware structure of these various processors can be an electric circuit that combines circuit elements such as semiconductor elements. [Explanation of symbols]

[0091] 1. Abnormal noise detection system 10 Abnormal noise detection device 11 CPU 11A Signal Processing Unit 11B Learning Department 11C Adjustment section 11D Detection unit 12 Memory 13 Storage section 13A Study Program 13B Abnormal noise detection program 13C threshold estimation model 14 Input section 15 Display section 16 Media reading and writing device 17 Recording Media 18 Communication I / F section 30 Information storage device 32 Storage section 32A Training Signal Database 70 Energy supply companies 80 Energy Facilities 80A Microphone 80B Environmental Sensor 90 Weather Information Server

Claims

1. a first detection unit that detects a sound in an environment where an object is located; a second detection unit that detects environmental information indicating a predetermined type of environmental condition in the environment; an adjustment unit that adjusts a threshold value used when detecting abnormal noise from the object, using the environmental information detected by the second detection unit; a detection unit that detects the abnormal noise using the threshold value adjusted by the adjustment unit; and Equipped with The threshold value is derived using a threshold estimation model that has been machine-learned in advance, with input information being a signal obtained by Fourier transforming a normal sound signal indicating the sound detected by the first detection unit when the object is normal and an abnormal sound signal indicating the sound detected by the first detection unit when the object is abnormal, and with output information being threshold information indicating the threshold value. Abnormal noise detection system.

2. a first detection unit that detects a sound in an environment where an object is located; a second detection unit that detects environmental information indicating a predetermined type of environmental condition in the environment; an adjustment unit that adjusts a threshold value used when detecting abnormal noise from the object, using the environmental information detected by the second detection unit; a detection unit that detects the abnormal noise using the threshold value adjusted by the adjustment unit; and Equipped with the environmental information includes at least one of weather information related to weather and environmental value information related to environmental values, The weather information includes weather information indicating weather, and the environmental value information includes at least one of wind speed, wind direction, and humidity. Abnormal noise detection system.

3. An abnormal sound detection program executed in an abnormal sound detection system including a first detection unit that detects a sound in an environment in which an object is placed, and a second detection unit that detects environmental information that indicates a predetermined type of environmental condition in the environment, adjusting a threshold value used when detecting an abnormal sound from the object using the environmental information detected by the second detection unit; detecting the abnormal sound using the adjusted threshold value; The threshold value is derived using a threshold estimation model that has been machine-learned in advance, with input information being a signal obtained by Fourier transforming a normal sound signal indicating the sound detected by the first detection unit when the object is normal and an abnormal sound signal indicating the sound detected by the first detection unit when the object is abnormal, and with output information being threshold information indicating the threshold value. An abnormal noise detection program that causes a computer to execute a process.

4. An abnormal sound detection program executed in an abnormal sound detection system including a first detection unit that detects a sound in an environment in which an object is placed, and a second detection unit that detects environmental information that indicates a predetermined type of environmental condition in the environment, adjusting a threshold value used when detecting an abnormal sound from the object using the environmental information detected by the second detection unit; detecting the abnormal sound using the adjusted threshold value; the environmental information includes at least one of weather information related to weather and environmental value information related to environmental values, The weather information includes weather information indicating weather, and the environmental value information includes at least one of wind speed, wind direction, and humidity. An abnormal noise detection program that causes a computer to execute a process.

Images