Abnormal sound diagnosis system

Abstract

To provide a technology to accurately identify abnormal sound.SOLUTION: A computer included in a vehicle abnormal sound diagnosis system acquires sound spectral data for a vehicle and vehicle data that describes a vehicle driving state. The computer receives user input in an evaluation time range for the sound spectral data. The computer sets a first time range before the evaluation time range and a second time range after the evaluation time range to the sound spectral data. The computer identifies as a candidate frequency bandwidth a frequency bandwidth in which a secular change in a sound pressure level does not correspond to a secular change in the vehicle driving state. The computer identifies as an abnormal sound frequency bandwidth the candidate frequency bandwidth including a frequency in which a projection amount of the sound pressure level in the evaluation time range relative to the first time range and the second time range satisfies a predetermined reference, from among the candidate frequency bandwidths.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The technology disclosed in this specification relates to an abnormal sound diagnosis system.

[0002] Patent Document 1 discloses an abnormal sound diagnosis system for vehicles. This abnormal sound diagnosis system accepts a user's specification of a frequency range and a time range that are considered to correspond to an abnormal sound for sound spectrum data recorded in a vehicle over a predetermined period of time. Then, the type of abnormal sound (for example, the source of the abnormal sound) is identified based on the specified frequency range and time range. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2023-33848 A Summary of the Invention [Problem to be solved by the invention]

[0004] In order to specify a frequency range that is considered to correspond to an abnormal sound, knowledge for linking audible sounds to frequencies may be required. Since not all users have such knowledge, it may be difficult to specify a frequency range. [Means for solving the problem]

[0005] According to the abnormal sound diagnosis system of claim 1, the abnormal sound diagnosis system can identify candidate frequency bands that are highly likely to be abnormal sounds. Then, the candidate frequency bands can be identified based on the correspondence between the change over time of the sound pressure level of the sound spectrum data and the change over time of the vehicle operating state. This makes it possible to identify frequency bands that are highly likely to be abnormal sounds as candidate frequency bands with higher accuracy than when a frequency range with a high sound pressure is simply identified as a candidate frequency band. As a result, even when a user does not have knowledge about frequencies, it is possible to identify an abnormal sound frequency band that corresponds to an abnormal sound with high accuracy. It is possible to realize a highly accurate abnormal sound diagnosis. Note that the candidate frequency bands and abnormal sound frequency bands may be identified using machine learning. [Brief description of the drawings]

[0006] [Figure 1] 2A and 2B are a block diagram of an abnormal sound diagnosis system 2 and a diagram showing an example of sound spectrum data. [Diagram 2] 4 is a flowchart of an abnormal sound identification process. [Diagram 3] 5A and 5B are diagrams for explaining an example of a method for identifying candidate frequency bands. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0007] (Configuration of abnormal sound diagnosis system 2) The abnormal sound diagnosis system 2 shown in Fig. 1(A) comprises a diagnostic device 4 and a recording device 6. The diagnostic device 4 comprises a monitor 10, an operation unit 12, a communication interface 14, and a control unit 20. The communication interface 14 is an interface for executing communication with other devices (e.g., the recording device 6). The control unit 20 is composed of a computer including a CPU, ROM, RAM, etc.

[0008] The recording device 6 is mounted on a vehicle (not shown). The recording device 6 records sound data and vehicle data for a predetermined period while the vehicle is traveling. The sound data is data corresponding to sounds generated from the vehicle. The sound data includes vehicle road noise, engine noise, motor noise, PCU switching noise, etc. The vehicle data is data describing the vehicle driving state over a predetermined period. The vehicle driving state includes, for example, vehicle speed, engine rotation speed, motor rotation speed, etc.

[0009] (abnormal sound identification processing) The abnormal sound identification process executed by the control unit 20 of the diagnostic device 4 will be described with reference to the flowchart of Fig. 2. The control unit 20 starts the process of Fig. 2 when an operation for starting the abnormal sound identification process is accepted from the user.

[0010] In S10, the control unit 20 acquires sound spectrum data. The sound spectrum data is data that describes the frequency spectrum of the sound recorded in the vehicle over a predetermined period of time. A more specific description will be given. The control unit 20 receives sound data of the sound recorded in the vehicle from the recording device 6. The control unit 20 acquires the sound spectrum data by performing a short-time Fourier transform (STFT) on the received sound data.

[0011] The control unit 20 displays the acquired sound spectrum data on the monitor 10. An example of sound spectrum data is shown in Fig. 1(B). The vertical axis indicates the frequency of the sound, and the horizontal axis indicates time. The shading of each pixel in the sound spectrum data indicates the sound pressure level (dB). In other words, the sound spectrum data is data that indicates the change over time in the frequency spectrum of the sound.

[0012] In S20, the control unit 20 acquires vehicle data. Specifically, the control unit 20 receives from the recording device 6 the vehicle data recorded in the vehicle.

[0013] In S30, the control unit 20 accepts an input of an evaluation time range TE by the user. The evaluation time range TE is a time range for evaluating the presence or absence of abnormal sounds in the sound spectrum data (i.e., a time range in which the user feels that an abnormal sound is occurring). Specifically, the user operates the operation unit 12 while looking at the monitor 10 of the diagnostic device 4. This allows the user to select the evaluation time range TE on the sound spectrum data displayed on the monitor 10. For example, the evaluation time range TE is selected as shown in FIG. 1(B).

[0014] In S40, the control unit 20 sets a first time range T1 and a second time range T2 for the sound spectrum data. The first time range T1 is a time range before the evaluation time range TE. The second time range T2 is a time range after the evaluation time range TE. There may be various methods for setting the ranges of the first time range T1 and the second time range T2. For example, as shown in the example of FIG. 1B, the entire period from the start time P1 of the recording of the sound spectrum data to the start time P2 of the evaluation time range TE may be set as the first time range T1. Also, the entire period from the end time P3 of the evaluation time range to the end time P4 of the recording of the sound spectrum data may be set as the second time range T2. Also, for example, a part of the entire period may be appropriately extracted and set as the first time range T1 or the second time range T2.

[0015] In S50, the control unit 20 acquires a change over time in the vehicle driving state for a period from the first time range T1 through the evaluation time range TE to the second time range T2. Various parameters can be used for the vehicle driving state. For example, the vehicle speed, the engine speed, the motor speed, etc. can be used. In addition, the vehicle driving state can be used by combining a plurality of parameters. For example, the vehicle speed, the engine speed, and the motor speed may be combined. In addition, for example, when the vehicle speed is used as the vehicle driving state, if the change over time in the vehicle speed is small, the change over time in the engine speed or the change over time in the motor speed can be used. In addition, various methods can be used for calculating the change over time.

[0016] In this embodiment, a case where the vehicle speed is used as the vehicle driving state will be described. In addition, as a method of calculating the change over time, a case where the average value of the vehicle speed is calculated in each of the first time range T1, the evaluation time range TE, and the second time range T2, and the average values ​​are compared will be described. In this embodiment, a case where the change over time of the vehicle driving state is divided into three types of patterns, "constant vehicle speed", "deceleration", and "acceleration". A specific description will be given. In this embodiment, the average vehicle speed S1 in the first time range T1, the average vehicle speed SE in the evaluation time range TE, and the average vehicle speed S2 in the second time range T2 are each calculated. Then, when the maximum value of the difference between the average vehicle speeds S1, SE, and S2 is within a predetermined range (e.g., ±3 km / h), the vehicle driving state is determined to be "constant vehicle speed". In addition, when the average vehicle speed S1 is the highest and the vehicle speeds decrease in the order of the average vehicle speeds S1, SE, and S2, the vehicle driving state is determined to be "deceleration". Also, when the average vehicle speed S1 is the lowest and the average vehicle speeds increase in the order of S1, SE, and S2, it is determined that the vehicle is in an "accelerating" driving state.

[0017] If it is determined in S50 that the vehicle is in a "constant vehicle speed" driving state (S50: constant vehicle speed), the process proceeds to S60. In S60, the control unit 20 identifies candidate frequency bands from the sound spectrum data. The candidate frequency bands are frequency bands that may correspond to abnormal sounds. In other words, the candidate frequency bands are frequency bands in which the change over time in the sound pressure level of the sound spectrum data does not correspond to the change over time in the vehicle driving state. In other words, the candidate frequency bands are frequency bands in which an unnatural change over time in sound pressure level occurs in view of the change over time in the vehicle driving state, and are frequency bands in which a change over time that is not normally expected occurs.

[0018] Using the example of FIG. 3, an example of a method for identifying candidate frequency bands in a vehicle driving state at a constant vehicle speed will be described. The vertical axis of FIG. 3(A) is sound pressure level [dBA], and the horizontal axis is frequency [Hz]. Graphs G1, GE, G2, and Genv are displayed in FIG. 3(A). Each of the graphs G1, GE, and G2 shows the analysis results of FFT analysis (one-third octave band analysis) of sound spectrum data in the first time range T1, the evaluation time range TE, and the second time range T2, respectively. The graph Genv corresponds to the upper envelope of the graphs G1 and G2. That is, the graph Genv is a graph that includes the higher sound pressure level of the graphs G1 and G2 at each frequency. The control unit 20 then identifies a frequency band in which the sound pressure of the graph GE is higher than that of the graph Genv. In the example of FIG. 3(A), five candidate frequency bands CF1-CF5 are identified.

[0019] In S70, the control unit 20 calculates the level difference for each of the candidate frequency bands. Various methods can be used to calculate the level difference. In this embodiment, the energy is added for each of the candidate frequency bands. As a result, as shown in the example of FIG. 3(B), the difference DV [dB] is calculated for each of the candidate frequency bands CF1-CF5.

[0020] In S200, the control unit 20 identifies an abnormal sound frequency band from one or more identified candidate frequency bands. The abnormal sound frequency band is a frequency band corresponding to an abnormal sound. By identifying the abnormal sound frequency band, the type of the abnormal sound (for example, the source of the abnormal sound) can be identified. Various methods can be used to identify the abnormal sound frequency band. In this embodiment, a candidate frequency band including a specific frequency is identified as the abnormal sound frequency band. The specific frequency is a frequency at which the prominence of the sound pressure level in the evaluation time range TE relative to the sound pressure level in the first time range T1 and the second time range T2 satisfies a predetermined criterion. Various criteria can be used as the predetermined criterion. In this embodiment, the "maximum value" is used as the predetermined criterion. That is, the "specific frequency" in this embodiment is a frequency at which the difference value of the sound pressure of the graph GE with respect to the graph Genv is positive and maximum.

[0021] 3A, the specific frequency SF is 1600 Hz. Therefore, the control unit 20 identifies the candidate frequency band CF4 that includes the specific frequency SF as the abnormal sound frequency band NF (see arrow A1).

[0022] In S210, the control unit 20 converts the ⅓ octave band frequencies of the identified abnormal sound frequency band NF into narrow band frequencies. The narrow band frequencies are the frequencies of abnormal sounds. In the example of Fig. 3, the ⅓ octave band frequencies (1250 to 3150 Hz) of the abnormal sound frequency band NF (see region R1) are converted into narrow band frequencies (1113.6 to 3535.9 Hz). Then, the flow ends.

[0023] On the other hand, if it is determined in S50 that the vehicle driving state is "deceleration" (S50: deceleration), the process proceeds to S80. In S80, the control unit 20 identifies candidate frequency bands from the sound spectrum data. An example of a method for identifying candidate frequency bands in a deceleration vehicle driving state will be described. As in the example of FIG. 3(A), a graph G1 of a first time range T1, a graph GE of an evaluation time range TE, and a graph G2 of a second time range T2 are obtained by 1 / 3 octave band analysis. Then, the control unit 20 identifies, as candidate frequency bands, frequency bands in which the sound pressure levels are in the order of "graph GE>graph G1>graph G2". The reason will be described. Since the vehicle driving state is deceleration, the vehicle speeds are in the order of "graph G1>graph GE>graph G2". Therefore, it is natural that the sound pressure levels of the sound spectrum data are also in the order of "graph G1>graph GE>graph G2". On the other hand, the frequency band in which the sound pressure levels are in the order of "Graph GE>Graph G1>Graph G2" can be identified as a frequency band in which an unnatural change in sound pressure level occurs over time in terms of the change over time of the vehicle driving state. Therefore, the frequency band in which this unnatural change in sound pressure level occurs over time can be identified as a candidate frequency band.

[0024] In S90, the control unit 20 calculates the level difference for each of the candidate frequency bands. The specific contents have been explained in S70, so detailed explanation will be omitted. Then, the process proceeds to S200.

[0025] On the other hand, if it is determined in S50 that the vehicle driving state is "accelerating" (S50:accelerating), the process proceeds to S100. In S100, the control unit 20 identifies candidate frequency bands from the sound spectrum data. An example of a method for identifying candidate frequency bands in the vehicle driving state of accelerating will be described. As in the example of FIG. 3(A), a graph G1 of the first time range T1, a graph GE of the evaluation time range TE, and a graph G2 of the second time range T2 are obtained by 1 / 3 octave band analysis. Then, the control unit 20 identifies, as candidate frequency bands, frequency bands in which the sound pressure levels are in the order of "graph GE>graph G2>graph G1". The reason will be described. Since the vehicle driving state is accelerating, the vehicle speeds are in the order of "graph G1<graph GE<graph G2". Therefore, it is natural that the sound pressure levels of the sound spectrum data are also in the order of "graph G1<graph GE<graph G2". On the other hand, the frequency band in which the sound pressure levels follow the order of "Graph G1 < Graph G2 < Graph GE" can be identified as a frequency band in which an unnatural change in sound pressure level occurs over time in terms of the change over time of the vehicle driving state. Therefore, the frequency band in which this unnatural change in sound pressure level occurs over time can be identified as a candidate frequency band.

[0026] In S110, the control unit 20 calculates the level difference for each of the candidate frequency bands. The specific contents have been explained in S70, so detailed explanation will be omitted. Then, the process proceeds to S200.

[0027] (effect) According to the abnormal sound diagnosis system 2 of the present specification, the candidate frequency bands with a high probability of abnormal sound can be identified automatically by the abnormal sound diagnosis system 2, not manually by the user. Then, the candidate frequency bands can be identified as frequency bands in which the change over time in the sound pressure level of the sound spectrum data does not correspond to the change over time in the vehicle operating state (S60, S80, S100). Furthermore, the abnormal sound diagnosis system 2 can automatically identify an abnormal sound frequency band from among one or more identified candidate frequency bands (S200). This makes it possible to identify an abnormal sound frequency band with a high probability of abnormal sound with higher accuracy than when a frequency range with a high sound pressure is simply identified as the abnormal sound frequency band. It is possible to realize a highly accurate abnormal sound diagnosis even when the user does not have knowledge about frequencies.

[0028] Although the embodiments have been described in detail above, these are merely examples and do not limit the scope of the claims. The technology described in the claims includes various modifications and changes to the specific examples exemplified above. The technical elements described in this specification or drawings exhibit technical utility alone or in various combinations, and are not limited to the combinations described in the claims at the time of filing. In addition, the technology exemplified in this specification or drawings achieves multiple objectives simultaneously, and achieving one of the objectives itself has technical utility.

[0029] (Modification) The method of specifying the evaluation time range TE in S30 is not limited to the method of visually checking the sound spectrum data by STFT, and various methods can be used. For example, it may be a method of specifying while listening to the sound, or a method of specifying while visually checking the time axis waveform. These methods may also be combined.

[0030] There may be various methods for identifying the abnormal sound frequency band in S200. For example, the candidate frequency band in which the difference DV is the largest may be identified as the abnormal sound frequency band. Alternatively, for example, the candidate frequency band in which the difference DV is larger than a predetermined threshold may be identified as the abnormal sound frequency band.

[0031] Identification of candidate frequency bands by the abnormal sound diagnosis system 2 (S60, S80, S100), Identification of the abnormal sound frequency band (S200) may be performed using machine learning. [Explanation of symbols]

[0032] 2: abnormal sound diagnosis system, 4: diagnosis device, 6: recording device, 10: monitor, 12: operation unit, 14: communication interface, 20: control unit

Claims

[Claim 1] An abnormal sound diagnosis system for a vehicle, comprising: At least one computer, the computer comprising: obtaining sound spectrum data describing a frequency spectrum of sounds recorded in the vehicle over a predetermined period of time; acquiring vehicle data describing vehicle operating conditions of the vehicle over the predetermined period of time; A process of receiving a user input of an evaluation time range, which is a time range for evaluating the presence or absence of an abnormal sound with respect to the sound spectrum data; A process of setting a first time range, which is a time range before the evaluation time range, and a second time range, which is a time range after the evaluation time range, for the sound spectrum data; a process of identifying, as a candidate frequency band, from the sound spectrum data, a frequency band in which a change over time in sound pressure level described in the sound spectrum data does not correspond to a change over time in the vehicle driving state described in the vehicle data, for a period from the first time range through the evaluation time range to the second time range; a process of identifying, from the identified one or more candidate frequency bands, a candidate frequency band including a frequency whose prominence amount of sound pressure level in the evaluation time range satisfies a predetermined criterion with respect to the first time range and the second time range, as an abnormal sound frequency band; This abnormal noise diagnosis system performs the following:

Images