JP7272631B2 - Sound or vibration determination method and information processing system for determination - Google Patents

Description

The present invention relates to diagnostic techniques for acoustic or vibrational machines and the like.

2. Description of the Related Art In recent years, the state of a machine or the like to be diagnosed (for example, whether it is normal or abnormal) is automatically determined by analyzing the sound or vibration emitted by the machine or the like to be diagnosed. However, since the sound and vibration emitted by the machine to be diagnosed differ depending on the machine to be diagnosed, the same diagnostic logic cannot be used.

In addition, in a document (for example, Patent Document 1), a sound pickup unit that collects sounds to be evaluated from a diagnosis target, a waveform extraction unit that calculates and extracts the sound waveform collected by the sound pickup unit, and a pre-registered A spatial distance calculator that calculates the spatial distance of the sound waveform to be evaluated in the cause-specific reference space represented by the plurality of reference waveforms, and the sound to be evaluated based on the distance calculation result output from the spatial distance calculator. is disclosed. Then, the analysis/judgment unit diagnoses the sound to be evaluated based on whether or not the fluctuation range of the plurality of spatial distance data of the sound to be evaluated, which is captured at a predetermined diagnosis timing, is within the allowable range. there is In this document, office equipment such as copiers are targeted for diagnosis, so the spatial distance is calculated in the reference space for each cause, but after that, evaluation is performed based on whether the fluctuation range of the spatial distance is within the allowable range. Only.

In the case of specific equipment such as office equipment such as copiers, it is sufficient to prepare a diagnostic device such as the one described above that can respond to any state of the specific equipment. Those who do not often do not know which diagnostic logic is suitable for which machine or the like. Unless appropriate diagnostic logic is applied to the machine or the like to be diagnosed, it is not possible to automatically make an appropriate determination of its condition.

Japanese Patent Application Laid-Open No. 2004-205215 JP 2004-309449 A JP 2009-186273 A

Therefore, according to one aspect, an object of the present invention is to provide a technique for easily making appropriate determinations for various diagnosis targets.

The determination method according to the present invention includes (A) a step of determining the type of sound or vibration to be determined from data representing the sound or vibration to be determined; (C) a step of selecting a determination unit corresponding to the type of sound or vibration to be determined from among a plurality of determination units that are prepared and determine a state to be determined; and determining the state to be determined.

According to one aspect, it becomes possible to easily and appropriately determine various diagnostic targets.

FIG. 1 is a diagram showing an outline of a system according to an embodiment. FIG. 2 is a diagram illustrating a configuration example of an analysis unit. FIGS. 3A to 3C are diagrams showing examples of volume time-series data. FIG. 4 is a diagram illustrating a configuration example of a determination unit; FIG. 5 is a diagram illustrating a processing flow according to the embodiment;

FIG. 1 shows a system configuration example according to one embodiment of the present invention.

The information processing apparatus 100 according to the present embodiment includes a microphone 1, an amplifier section 3, an AD (Analog-Digital) conversion section 5, an analysis section 7, a sound type discrimination section 9, a discrimination database 11, It has a switching unit 13 , a hammering sound determination unit 15 , a rotation sound determination unit 17 , a sliding sound determination unit 19 , and an output unit 21 . It should be noted that, here, it is assumed that it corresponds to hitting sounds, rotating sounds, and sliding sounds, but if it corresponds to other sounds such as flowing sounds, a determination unit for that purpose will be added. . Further, data for that purpose is added to the determination database 11 as well.

The microphone 1 outputs a collected sound input signal to the amplifier section 3 . The amplifier section 3 amplifies the sound input signal and outputs the sound signal X(t) to the AD conversion section 5 . The AD conversion unit 5 performs A/D conversion on the sound signal X(t) to generate input sound data {x _i } and outputs it to the analysis unit 7 .

The analysis unit 7 has, for example, the configuration shown in FIG. The analysis unit 7 has a frame division unit 71 , a volume calculation unit 73 and a clipping unit 75 .

The frame division unit 71 divides the input sound data {x _i } by a predetermined frame period (also called frame width), and outputs the input sound data for each frame to the volume calculation unit 73 . The volume calculator 73 calculates the volume p _j for each frame j, and outputs it to the clipping unit 75 . The cutout unit 75 cuts out a portion of the time-series data of the volume p _j that is equal to or greater than the threshold value Thp, and outputs the cutout portion to the sound type determination unit 9 .

The determination database 11 stores volume time series data for each type of sound to be determined. For example, for each type of sound to be determined, a standard sound is input from the microphone 1 in advance, the AD conversion unit 5 generates input sound data, and the analysis unit 7 generates and stores volume time series data. Keep

FIG. 3 shows an example of volume time-series data stored in the determination database 11 . FIG. 3(a) shows an example of volume change over time for a tapping sound, FIG. 3(b) shows an example of volume change over time for a rotating sound, and FIG. 3(c) shows a sliding sound. shows an example of time change in the case of In the case of hitting sound, the volume is concentrated at the impact portion. In the case of the rotating sound, it includes an acceleration section where the volume gradually increases, a steady rotation section where the steady volume continues, and a deceleration section where the volume gradually decreases. In the case of a sliding sound, a sliding portion with a large volume is sandwiched between stop portions with a low volume. In this way, for each sound type, a characteristic volume change over time occurs for that sound type. Spectral information related to timbre other than volume is not used to discriminate a sound type having such characteristics. This is because the timbre is different for each device model, and it is not realistic to prepare data for discrimination for each model. The change in volume over time is information representing the correspondence relationship with the structure or operation to be determined, regardless of the model.

The sound type discriminating unit 9 compares the time-series data of the sound volume for each sound type stored in the discriminating database 11 and the time-series data of the sound volume p _j to be determined output from the analysis unit 7 to determine distance or similarity. A degree calculation is performed to determine, for example, the sound type with the shortest distance or the highest degree of similarity. Then, the sound type determination unit 9 outputs to the switching unit 13 so that the output of the AD conversion unit 5 is processed by the determination unit corresponding to the determined sound type.

The switching unit 13 inputs the output of the AD conversion unit 5 to one of the hitting sound determination unit 15, the rotation sound determination unit 17, and the sliding sound determination unit 19 according to the output from the sound type determination unit 9. do.

The hammering sound determining unit 15, the rotating sound determining unit 17, and the sliding sound determining unit 19 each have a logic for determining whether the hammering sound, the rotating sound, and the sliding sound are in a normal state or an abnormal state, respectively. is built in.

FIG. 4 shows a basic configuration example of the determination unit.

The determination unit (15-19) described above has a second analysis unit 31, a collation unit 33, a state determination unit 35, and a database 37 for determination.

The second analysis unit 31 cuts out a characteristic portion related to a specific sound type from the input sound data, executes Fast Fourier Transform (FFT) or the like to generate spectrum data, and performs Executes a predetermined analysis process. If the sound type is a hammering sound, the impact part of the hammering sound is cut out and analyzed. If the sound type is a rotating sound, an acceleration portion, a deceleration portion, and a steady rotation portion are cut out, and analysis processing is performed for each of them. If the sound type is sliding sound, the sliding portion is cut out and analysis processing is executed.

The determination database 37 stores analysis results of normal sounds of specific sound types. That is, if the sound type is hammering sound, the analysis results for the spectrum data of the impact portion of the normal hammering sound are stored. If the sound type is rotating sound, the analysis results are stored for each spectrum data of the acceleration portion, deceleration portion, and steady rotation portion of the normal rotating sound. If the sound type is sliding sound, the analysis result of the spectral data of the sliding portion of the normal sliding sound is stored.

The matching unit 33 compares the analysis result of the second analysis unit 31 with the analysis result of the normal state stored in the determination database 37, calculates a predefined distance or similarity, and performs state determination. Output to unit 35 . The state determination unit 35 compares the distance or the degree of similarity with a predetermined threshold to determine, for example, whether it is normal. That is, if the distance exceeds the threshold or if the degree of similarity is less than the threshold, it is determined to be abnormal. Then, the state determination section 35 outputs to the output section 21 . The output unit 21 outputs the determination result to, for example, a display device.

Such a configuration of the determination unit is an example, and a configuration that can appropriately determine the state of the determination target by various logics can be adopted. For example, normal sounds and abnormal sounds may be learned in advance by machine learning, and whether or not the sound is normal may be determined using this learned model.

Next, operation contents of the information processing apparatus 100 will be described with reference to FIG.

First, a sound input signal from the microphone 1 is amplified by the amplifier section 3, and the sound signal X(t) is digitized by the AD conversion section 5 to generate input sound data {x _i } (step S1).

In the AD converter 5, the sampling frequency (sampling frequency) is set to twice or more the frequency of the highest frequency component contained in the input sound based on the sampling theorem. For example, if the input sound contains components up to 5 KHz, the sampling frequency is set to 10 KHz or higher, which is twice the frequency. If the sampling frequency is 10 KHz, the sampling interval (the time interval between x _i and x i+1 ) will be 0.0001 seconds.

Next, the frame division unit 71 of the analysis unit 7 divides the input sound data {x _i } into frames, and the volume calculation unit 73 performs volume calculation for each frame to obtain time-series data of the volume p _j . Generate (step S3).

The frame period is set to the reciprocal of a frequency that is at least twice the maximum frequency of changes in volume. For example, if the maximum frequency of volume change is 5 Hz, the frame frequency is 10 Hz or higher, and the frame period is 0.1 seconds or less (when the sampling frequency is 10 kHz, the input sound data is 1000 or less).

The volume calculator 73 calculates the volume p _j (dB value) according to the following formula for each frame j.
p _j =10×log ₁₀ (Σ _{k∈frame j} × _k ² )

As a result, the volume for each frame is calculated, and the change in volume over time is obtained.

Further, the cutout unit 75 cuts out the volume time-series data at a predetermined threshold Thp (step S5). That is, a portion of the time-series data of the volume p _j that is equal to or greater than the threshold value Thp is cut out and output to the sound type discrimination section 9 .

Then, the sound type discrimination unit 9 calculates the distance or similarity between the volume time series data of each sound type registered in the discrimination database 11 and the volume time series data output from the analysis unit 7. (Step S7).

In general, the volume time-series data registered in the determination database 11 and the volume time-series data from the analysis unit 7 have different lengths (number of frames). Calculate similarity. Alternatively, the distance or similarity may be calculated after performing optimal nonlinear expansion and contraction using dynamic programming.

Sound type discrimination 9 discriminates the sound type based on the calculated distance or similarity (step S9). For example, the sound type with the shortest distance or the highest degree of similarity is specified. However, when the distance of the identified sound type is longer than the discrimination threshold, or when the similarity is smaller than the discrimination threshold, there is a possibility that appropriate discrimination cannot be made.

Accordingly, the sound type discriminating section 9 determines whether or not the sound type has been discriminated in step S9 (step S11). For example, when the distance of the specified sound type is equal to or less than the discrimination threshold value, it is assumed that the similarity is equal to or greater than the discrimination threshold value, the discrimination is possible, and otherwise the discrimination is not possible.

If the sound type cannot be determined, the sound type determining unit 9 indicates to the user that the sound type could not be determined, and prompts the user to input the sound type. For example, the sound type with the shortest distance or the highest degree of similarity may be presented. The user designates an appropriate sound type for this. The sound type discrimination section 9 accepts this specification (step S13).

When the determination is made or after step S<b>13 , the sound type determining unit 9 outputs the determined or designated sound type to the switching unit 13 .

The switching unit 13 outputs the input sound data to the determination unit of the determined or designated sound type (step S15). The determination unit to which the input sound data is input executes state determination processing based on the input sound data (step S17). Any one of the hitting sound determination unit 15, the rotation sound determination unit 17, and the sliding sound determination unit 19 executes determination processing. The output unit 21 outputs the processing result to a display device or the like.

By using such an information processing apparatus 100, even a worker who is unfamiliar with the sound or vibration of a machine can appropriately determine the state of the machine.

Although the embodiment of the present invention has been described above, the present invention is not limited to this. For example, the functional block configurations of FIGS. 1, 2, and 4 are examples, and may not match the program module configuration. The processing flow shown in FIG. 5 may also include steps that may be executed in parallel or whose order may be changed as long as the processing results are the same.

In the above, the microphone 1 is provided for sound, but when vibration is the object, a vibration sensor may be used instead. In addition, time-series data for that purpose is accumulated in the discrimination database 11 . As for the determination unit (15-19), a determination unit for that purpose is also prepared.

In addition, although one type of hammering sound, rotating sound, and sliding sound is shown, one or more types for each (for example, rotating sound 1 (for example, the rotating sound of a Ferris wheel) and rotating sound 2 (for example, a small size for a model) (motor), etc.) may be adapted. As described above, sound types may be added or changed.

Further, the information processing apparatus 100 may be implemented as a single apparatus or may be implemented as a plurality of apparatuses. For example, the microphone 1 and the AD converter 5 may be placed near the device to be determined, and the other components may be placed at other locations. In this way, the information processing apparatus 100 may be constructed as an information processing system realized by one or more apparatuses, and may be called an information processing system including the case of one apparatus.

The information processing apparatus 100 described above is a computer apparatus, and includes a memory, a CPU (Central Processing Unit), a hard disk drive (HDD), a display control unit connected to a display device, and a removable display device. A disk drive device, an input device, and a communication control unit for connecting to a network are connected by a bus. An operating system (OS) and an application program for executing the processing in this embodiment are stored in the HDD, and are read from the HDD into the memory when executed by the CPU. The CPU controls the display control unit, the communication control unit, and the drive device according to the processing content of the application program, and causes them to perform predetermined operations. In addition, the data being processed is mainly stored in the memory, but may be stored in the HDD. In the embodiment of the present invention, an application program for carrying out the processing described above is stored and distributed in a computer-readable removable disk, and installed from the drive device to the HDD. It may be installed in the HDD via a network such as the Internet and a communication control section. Such a computer device implements the various functions described above through organic cooperation between hardware such as the CPU and memory described above and programs such as the OS and application programs.

The present embodiment described above can be summarized as follows.

The information processing system according to the present embodiment includes (A) a plurality of determination units that are prepared in advance corresponding to the type of sound or vibration and determines the state of the determination target, and (B) the sound or vibration that is the determination target. From the data represented, the data is sent to a determination unit that determines the type of sound or vibration to be determined, and (C) a determination unit that corresponds to the type of sound or vibration that is determined to be determined among a plurality of determination units. and a switching unit for processing. By doing so, an appropriate determination unit corresponding to the current determination target is automatically selected from among a plurality of assumed determination targets, and appropriate determination is performed. As described above, the types of sound or vibration are not limited to types such as hammering sound, rotating sound, and sliding sound, and may also be configured to identify a plurality of subordinate types included in at least one of them. . Also, it may be determined that the type is one of a plurality of subordinate types.

Note that the determination unit described above (b1) calculates the time change in the amount of sound or vibration to be determined from the above data, and (b2) the time change in the amount of sound or vibration to be determined, The type of sound or vibration to be determined may be determined based on the distance or similarity to the temporal change in the amount of sound or vibration for a plurality of types stored in the device. This is because the temporal change in volume or vibration amount is an important parameter for determining the type.

It should be noted that a program for executing the above processing can be created, and the program can be read by a computer, such as a flexible disk, an optical disk (CD-ROM, DVD-ROM, etc.), a magneto-optical disk, a semiconductor memory, a hard disk, etc. stored in a suitable storage medium or storage device. Note that intermediate processing results are temporarily stored in a storage device such as a main memory.

100 information processing device 1 microphone 3 amplifier unit 5 AD conversion unit 7 analysis unit 9 sound type determination unit 11 determination database 13 switching unit 15 hammering sound determination unit 17 rotation sound determination unit 19 sliding sound determination unit 21 output unit

Claims (4)

a determination step of determining the type of sound or vibration to be determined from input data representing the sound or vibration to be determined;
After the determination step, out of a plurality of determination units that are prepared in advance corresponding to the type of sound or vibration and that determine a state to be determined by extracting a characteristic portion related to the type of sound or vibration from the input data , a step of selecting a determination unit corresponding to the type of sound or vibration to be determined determined in the determining step ;
causing only the determination unit selected from the plurality of determination units to process the input data;
A program that causes a computer to run In the determination step,
calculating a time change in the amount of sound or vibration to be determined from the input data;
Based on the distance or similarity between the time change in the amount of sound or vibration to be determined and the time change in the amount of sound or vibration for a plurality of types stored in advance in a storage device, the sound or vibration to be determined The program according to claim 1, wherein the type of vibration is discriminated. a determination step of determining the type of sound or vibration to be determined from input data representing the sound or vibration to be determined;
After the determination step, out of a plurality of determination units that are prepared in advance corresponding to the type of sound or vibration and that determine a state to be determined by extracting a characteristic portion related to the type of sound or vibration from input data , a step of selecting a determination unit corresponding to the type of sound or vibration to be determined determined in the determination step ;
a step of processing the data only by the determination unit selected from the plurality of determination units and determining the state of the determination target;
A computer-implemented determination method. a plurality of determination units that are prepared in advance corresponding to the type of sound or vibration and that determine a state to be determined by extracting a characteristic portion related to the type of sound or vibration from input data;
a determination unit that determines the type of sound or vibration to be determined from input data representing the sound or vibration to be determined;
a switching unit that, after processing by the determining unit, causes only the determining unit corresponding to the type of the sound or vibration to be determined determined by the determining unit to process the data, among the plurality of determining units;
An information processing system having

Images