JP2007107908A - Method and device for determining inverse filter coefficient, and abnormal vibration detector - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method for determining an inverse filter coefficient having furthermore improved abnormality detection accuracy of a machine or the like than before. <P>SOLUTION: A display part 113 displays a signal waveform acquired by visualizing a normal sound signal recorded in a buffer part 106, and an input reception part 114 receives a range designation including a characteristic part by a user, and a filter coefficient operation part 109 calculates the inverse filter coefficient by using a partial signal corresponding to the range. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a technique for removing normal vibration from input vibration and detecting abnormal vibration, and more particularly to a technique for determining an inverse filter coefficient for removing normal vibration.

  In order to detect the occurrence of an abnormality in a machine or the like, a normal sound signal that is a sound during normal operation of the machine or the like is measured in advance, and an inverse filter coefficient that is a coefficient of a filter for removing the normal sound signal from the acoustic signal is obtained. After the calculation, after receiving the input sound that is the target of the presence or absence of abnormality, apply an inverse filter to the input sound signal to remove the normal sound signal from the input sound signal and only the abnormal sound signal Is used, and the presence or absence of abnormality is determined using the abnormal sound signal.

Here, since it is necessary to perform a complicated calculation to calculate the inverse filter coefficient, a normal sound signal is recorded for a short time (for example, about 25 milliseconds) and used for calculating the inverse filter coefficient. Prevents prolonged computation time.
Modern Digital Signal Processing Theory and its Applications Translated by Yuki Aoyama Maruzen Co., Ltd. December 30, 1992 16-P. 19

  However, when a normal sound signal for a short time is recorded as described above, a characteristic signal such as a pulse signal is generated after the end of the recording, and the characteristic part is used for calculating the inverse filter coefficient. It is assumed that this is not possible. Under such an assumption, an inverse filter coefficient that mainly removes low frequencies is calculated even though an inverse filter coefficient that mainly removes high frequencies corresponding to the pulse signal should be calculated. If the calculated inverse filter coefficient is used, the low frequency portion of the input sound signal is removed when it is abnormal sound, and such abnormal sound detection may be missed. Further, the high frequency portion remains without being removed, and is detected as an abnormal sound. Even if the characteristic part is other than a pulse wave such as a sawtooth wave, an abnormal sound may be missed or erroneously detected for the same reason. In addition, the above problem does not relate only to sound, but may similarly occur for calculation of an inverse filter coefficient for detecting abnormal vibration including abnormal sound.

  In view of the above problems, the present invention provides an inverse filter coefficient determination method, an inverse filter coefficient determination apparatus, and an abnormal vibration detection apparatus that can prevent detection errors and false detections of abnormal vibrations and can improve the detection accuracy of abnormal vibrations as compared with the prior art. The purpose is to provide.

  In order to solve the above-mentioned problems, the present invention is an inverse filter coefficient determination method for calculating an inverse filter coefficient of an inverse filter that detects normal vibration by removing normal vibration from input vibration, and records a normal vibration signal. A recording step, a waveform display step for generating a signal waveform in which the normal vibration signal is visualized and displaying the signal waveform, and a feature of the signal waveform based on an instruction from the user who viewed the displayed signal waveform The method includes: a specifying step of specifying a range including a portion; and a calculating step of calculating an inverse filter coefficient using a partial signal corresponding to the range of the normal vibration signal.

  Since the inverse filter coefficient determination method of the present invention has the above-described configuration, the inverse filter coefficient is calculated using the partial signal including the characteristic vibration that is the characteristic part of the signal waveform. When the inverse filter using is applied, the characteristic vibration in the input vibration can be removed, and erroneous detection of the characteristic vibration as abnormal vibration can be prevented.

  For example, when a pulse signal is included in the input vibration signal, calculating an inverse filter coefficient using a partial signal including the pulse signal provides an inverse filter coefficient that reduces high-frequency components, but does not include the pulse signal. When the inverse filter coefficient is calculated using the partial signal, the inverse filter coefficient that reduces the low-frequency component intensively is obtained. If an inverse filter using an inverse filter coefficient that reduces the low-frequency component preferentially is used for the input vibration signal, the pulse signal included in the input vibration signal is not reduced, and the pulse signal is detected as abnormal vibration. End up.

Further, the specifying step includes a receiving step for receiving the instruction, a temporary specifying step for selecting a temporary range that satisfies a predetermined condition from the signal waveform, and a display for displaying the temporary range on the display of the signal waveform. An updating step and a main specifying step of specifying the range by increasing or decreasing the temporary range based on the instruction may be included.
According to this configuration, the provisional identification step identifies the provisional range that is considered to include the characteristic part in the signal waveform, and the provisional range is changed by the identification step to change the provisional range. Since it specifies, the said range can be specified more rapidly than before.

For example, the provisional specification step specifies a provisional range including a pulse signal, and in this specification step, a user who has seen the signal waveform changes the provisional range by a user operation and specifies the range. In this case, the number of man-hours for the user operation can be reduced as compared with the case where the user specifies the range without specifying the temporary range in the temporary specifying step.
The inverse filter coefficient determination apparatus of the present invention is an inverse filter coefficient determination apparatus that calculates an inverse filter coefficient of an inverse filter that detects abnormal vibration by removing normal vibration from input vibration, and is a recording unit that records a normal vibration signal. And generating a signal waveform that visualizes the normal vibration signal, displaying the signal waveform, and, based on an instruction from the user who viewed the displayed signal waveform, a characteristic portion of the signal waveform. Specifying means for specifying a range to include, and calculating means for calculating an inverse filter coefficient using a partial signal corresponding to the range of the normal vibration signal.

According to this configuration, since the inverse filter coefficient is calculated using the partial signal including the characteristic vibration that is the characteristic part of the signal waveform, when the inverse filter using the inverse filter coefficient is applied to the input vibration, the input is performed. The characteristic vibration during vibration can be removed, and erroneous detection of the characteristic vibration as abnormal vibration can be prevented.
For example, when a pulse signal is included in the input vibration signal, calculating an inverse filter coefficient using a partial signal including the pulse signal provides an inverse filter coefficient that reduces high-frequency components, but does not include the pulse signal. When the inverse filter coefficient is calculated using the partial signal, the inverse filter coefficient that reduces the low-frequency component intensively is obtained. If an inverse filter using an inverse filter coefficient that reduces the low-frequency component preferentially is used for the input vibration signal, the pulse signal included in the input vibration signal is not reduced, and the pulse signal is detected as abnormal vibration. End up.

The specifying means includes a receiving unit that receives the instruction, a temporary specifying unit that selects a temporary range that satisfies a predetermined condition from the signal waveform, and a display that displays the temporary range on the display of the signal waveform. An update unit and a main specifying unit for specifying the range by increasing or decreasing the temporary range based on the instruction may be included. The temporary specifying unit selects a point having the maximum amplitude in the signal waveform. A first selection unit and a second selection unit that selects a predetermined range including the maximum point in the signal waveform as the temporary range may be included.

Further, the provisional specifying unit may select the provisional range in which an amplitude of a predetermined magnitude or more in the signal waveform continues for a predetermined length or more.
The temporary specifying unit sequentially reads a recorded normal vibration signal, a determination unit that determines whether or not the read signal exceeds a predetermined value each time the signal is read, and a continuous A counting unit that counts the number of times determined to exceed a predetermined value, and outputs the temporary range determined to continuously exceed the predetermined value in the normal vibration signal when the number exceeds the predetermined number And an output unit.

According to this configuration, the temporary specifying unit specifies a temporary range that is considered to include a characteristic portion in the signal waveform, and the specific specifying unit increases or decreases the temporary range to thereby determine the range including the characteristic portion. Since it specifies, the said range can be specified more rapidly than before.
For example, a temporary range including a pulse signal is specified by a temporary specifying unit, and in this specifying step, a user who has seen the signal waveform increases or decreases the temporary range based on a user operation, and specifies the range. In this case, the number of man-hours for the user operation can be reduced as compared with the case where the user specifies the range without specifying the temporary range by the temporary specifying unit.

  The specifying means includes a first temporary specifying unit that selects a first temporary range that satisfies a predetermined condition from the signal waveform, and a first display that displays the signal waveform on the display indicating the first temporary range. An update unit, a first temporary specifying unit that increases or decreases the first temporary range based on the instruction, and a second temporary condition that satisfies the predetermined condition among the signal waveforms different from the range selected by the first temporary specific unit. A second temporary specifying unit for selecting a range; a second display updating unit for adding a display indicating the second temporary range to the display of the signal waveform; and a second line for increasing or decreasing the second temporary range based on the instruction. An identification unit and an output unit that outputs the temporary range obtained by combining the first temporary range and the second temporary range may be included.

  According to this configuration, when there are a plurality of feature portions in the normal vibration signal, the inverse filter coefficients corresponding to the plurality of features can be calculated. Therefore, the inverse filter using the inverse filter coefficient is applied to the input vibration. In this case, normal vibration including a plurality of characteristic vibrations in the input vibration can be removed, and erroneous detection of each of the plurality of characteristic vibrations as abnormal vibrations can be prevented.

  The specifying means specifies a first range that specifies a first range that satisfies a predetermined condition in the signal waveform, and specifies a second range that satisfies the predetermined condition and is different from the first range in the signal waveform. A second specifying unit; a selection unit that selects one of the first range and the second range; and a display update unit that adds a display indicating the range selected by the selection unit to the display of the signal waveform. And an output unit that increases / decreases the range selected by the selection unit based on the instruction and outputs the range as the temporary range.

According to this configuration, when there are a plurality of characteristic parts in the normal vibration signal, the inverse filter coefficient corresponding to one of the plurality of characteristics can be calculated. Therefore, the inverse filter coefficient is used for the input vibration. When the inverse filter is applied, it is possible to prevent the selected characteristic vibration from being erroneously detected as abnormal vibration.
The recording means may record a normal sound signal as the normal vibration signal.

According to this configuration, since the inverse filter coefficient is calculated using the partial signal including the characteristic sound that is the characteristic part of the signal waveform, when the inverse filter using the inverse filter coefficient is applied to the input sound, The characteristic sound in the sound can be removed, and the characteristic sound can be prevented from being erroneously detected as an abnormal sound.
The present invention is an abnormal vibration detection device that detects abnormal vibration by removing normal vibration from input vibration, and generates a signal waveform that visualizes the normal vibration signal, recording means for recording the normal vibration signal, and Waveform display means for displaying a signal waveform, identification means for specifying a range including a characteristic portion of the signal waveform based on an instruction from a user who has viewed the displayed signal waveform, and the normal vibration signal among the normal vibration signals Using the partial signal corresponding to the range, a calculating means for calculating an inverse filter coefficient, a receiving means for receiving an input vibration signal, and a normal vibration signal from the input vibration signal using an inverse filter based on the inverse filter coefficient. Detecting means for removing and detecting an abnormal vibration signal included in the input vibration signal.

According to this configuration, the inverse filter coefficient is calculated using the partial signal including the characteristic vibration that is the characteristic part of the signal waveform, and the inverse filter using the inverse filter coefficient is applied to the input vibration. Therefore, it is possible to prevent erroneous detection of the characteristic vibration as abnormal vibration.
The specifying means includes a receiving unit that receives the instruction, a temporary specifying unit that selects a temporary range that satisfies a predetermined condition from the signal waveform, and a display that displays the temporary range on the display of the signal waveform. An update unit and a main specifying unit that specifies the range by increasing or decreasing the temporary range based on the instruction may be included.

According to this configuration, the temporary specifying unit specifies a temporary range that is considered to include a characteristic portion in the signal waveform, and the specific specifying unit increases or decreases the temporary range to thereby determine the range including the characteristic portion. Since it specifies, the said range can be specified more rapidly than before.
For example, a temporary range including a pulse signal is specified by a temporary specifying unit, and in this specifying step, a user who has seen the signal waveform increases or decreases the temporary range based on a user operation, and specifies the range. In this case, the number of man-hours for the user operation can be reduced as compared with the case where the user specifies the range without specifying the temporary range by the temporary specifying unit.

As an embodiment of the present invention, an abnormal sound detection apparatus 1 relating to a sound which is a specific example of vibration will be described.
The abnormal sound detection device 1 according to the present invention is a device that detects an abnormality of a machine or the like. Among normal sound signals that are recorded in advance during normal operation of a machine or the like, a signal that is used to calculate an inverse filter coefficient is used by a user The inverse filter coefficient is calculated from the range-designated signal to prevent the detection of abnormal sound and improve the detection accuracy of abnormal sound.

Hereinafter, embodiments of the present invention will be described together with illustrated examples.
As shown in FIG. 1, the abnormal sound detection apparatus 1 of the present invention includes a microphone 101, an AD conversion unit 102, a switch 103, a normal sound removal unit 104, a time measurement unit 105, a buffer unit 106, a recording control unit 107, and a recording unit 108. , A filter coefficient calculation unit 109, an abnormal sound determination unit 110, a display generation unit 111, a display control unit 112, a display unit 113, an input reception unit 114, and a control unit 115.

Specifically, the abnormal sound detection device 1 is a computer system including a CPU, a ROM, a RAM, a hard disk, a video adapter, a liquid crystal display, and the like. A computer program is stored in the RAM. By operating according to the computer program, the abnormal sound detection device 1 achieves its function.
Hereinafter, the respective parts will be described with their actions.
<Input reception unit 114>
The input receiving unit 114 includes a keyboard, acquires an operation instruction input by a user operating the keyboard, and transmits the acquired operation instruction to the control unit 115.

Examples of the operation instructions include a filter coefficient calculation instruction, a selection cursor change instruction, a cursor right movement instruction, a cursor left movement instruction, a selection range determination instruction, an abnormal sound detection start instruction, and an abnormal sound detection end instruction.
<Microphone 101, AD converter 102, switch 103, timer 105>
The microphone 101 converts the collected acoustic signal into an electric signal in the range of −5 volts to +5 volts and outputs the electrical signal to the AD conversion unit 102.

The AD conversion unit 102 is a 16-bit AD conversion element, and AD converts an electric signal input from the microphone 101 into digital data indicating a range of −32768 to 32767 for each preset sampling period.
In the present embodiment, as an example, the sampling frequency of the AD conversion unit 102 is set to 43.478 KHz.

The AD conversion unit 102 includes an AD conversion result register read from another module, and sequentially sets the AD conversion result in the AD conversion result register.
The switch 103 connects either the normal sound removing unit 104 or the recording control unit 107 to the AD conversion unit 102 in accordance with a selection instruction from the control unit 115.
The timer 105 is a clock that measures the date and time by counting up the counter value every second, and specifically measures the number of seconds from January 1, 1970.

The timekeeping unit 105 includes a time reading register that is read from the recording control unit 107 of the abnormal sound detection device 1, and converts the count value into a format of “year / month / day hour / minute / second” at each count-up. Then, the time information is set in the time reading register.
For example, time information “2005/6/14 10:31:40” indicating 10:31:40 on June 14, 2005 is set in the time reading register, and the recording control unit 107 stores the time information. read out.
<Buffer unit 106, recording unit 108>
The buffer unit 106 is composed of a buffer memory, and is a digital data such as a signal data file written when calculating the inverse filter coefficient for normal sound removal by the recording control unit 107 and abnormal sound detection target data written when detecting abnormal sound. Retain data.

The signal data file includes a signal data file name, a time stamp, and a signal data array.
The signal data file name is a name for identifying the signal data file, the time stamp is time information read from the time measuring unit 105 by the recording control unit 107, and the signal data array is 16 bits which is an AD conversion result. An array that stores data.

As an example, the buffer unit 106 records a signal data file 201 as shown in FIG.
The signal data file 201 includes a signal data file name “data file 1” 202, a time stamp “2005/6/14 10:31:40” (203), and a signal data array 204. , (N _max ) 16-bit data Sig [0] to Sig [N _max −1].

In this embodiment, N _max is 43478.
The abnormal sound detection target data is a data array having 16-bit data as elements, and the normal sound removal unit 104 reads the 16-bit data from the AD conversion unit 102 and transfers the data to the recording control unit 107. Is stored.
The abnormal sound detection target data is held in a ring buffer so that the storage area of the data does not increase infinitely.

The recording unit 108 is a recording device including a hard disk, and is used for abnormal data files created from abnormal sound detection target data determined to be abnormal, written by the recording control unit 107, and for inverse filter coefficient and inverse filter coefficient calculation. Hold the signal data file.
<Recording control unit 107>
When receiving a recording start instruction from the control unit 115, the recording control unit 107 generates a signal data file and stores the signal data file in the buffer unit 106.

The recording control unit 107 reads the AD conversion result from the AD conversion result register of the AD conversion unit 102 at every predetermined sampling time, and writes it into the signal data array of the signal data file.
When the recording control unit 107 receives an abnormal sound detection instruction from the control unit 115, the recording control unit 107 creates an abnormal sound detection target file in the buffer unit 106, and detects 16-bit data sent from the normal sound removal unit 104 as abnormal sound detection. The target data is sequentially stored in a ring buffer as a data array. Further, when an abnormal signal storage instruction is received from the control unit 115, after data for a certain period has been recorded, the abnormal data file created by rearranging this data in time series is stored in the recording unit 108.
<Display generator 111>
The display generation unit 111 receives an instruction from the control unit 115, performs processing according to the instruction, generates a display image such as the display image 230 illustrated in FIG. 3 as an example, and image data indicating the display image Is written in the image memory included in the display control unit 112. The display image 230 includes a cursor position display image 231, a cursor width display image 232, and a graph image 233. The graph image 233 includes a graph 234, a start cursor 235, and an end cursor 236.

The cursor position determination process performed by the display generation unit 111 will be described with reference to FIGS.
The display generation unit 111 acquires a waveform display instruction including the signal data file name from the control unit 115 (step S11).
The display generation unit 111 reads out the signal data file identified by the signal data file name from the buffer unit 106 (step S12).

The display generation unit 111 uses the signal data array Sig [n] in the read signal data file, as shown in FIG. 3 as an example, on the x-axis and y-axis orthogonal coordinates with the origin 237 as the origin, A graph 234 in which (n, Sig [n]) is plotted as (x coordinate, y coordinate) is generated (step S13). Here, n is a natural number and takes a value of 0 to N _max −1.
The display generation unit 111 manages three pieces of information regarding the start cursor 235 and the end cursor 236: SigSelect indicating the selection cursor, SigStart indicating the x coordinate of the start cursor, and SigEnd indicating the x coordinate of the end cursor.

SigSelect is a variable for identifying the cursor to be moved, and indicates either a value “0” indicating the end cursor or a value “1” indicating the start cursor.
The display generation unit 111 initializes SigStart with 0, initializes SigEnd with N _max , and initializes SigSelect with 0 (step S14).
For example, as illustrated in FIG. 3, the display generation unit 111 superimposes a start cursor 235 indicated by a broken line parallel to the y axis on a position where the x coordinate is SigStart, and the x coordinate is SigEnd. A graph image 233 that is an image in which an end cursor 236 indicated by a broken line parallel to the y-axis is superimposed at a certain position is generated (step S15). Here, it is assumed that a display image is generated so that a cursor indicated by SigSelect is displayed in red so that it can be recognized as an operation target, and the other cursor is displayed in black.

  As an example, as illustrated in FIG. 7, the display generation unit 111 generates a cursor position display image 231 that displays the value of SigStart, generates a cursor width display image 232 that displays the value of SigEnd-SigStart, and graph image 233. And the display image 230 is generated, and the display image 230 is written in the image memory included in the display control unit 112 (step S16).

For example, in FIG. 7, the start cursor 235 has the x coordinate displayed at the position of the SigStart value “11060”, and the SigStart value “11060” is displayed in the cursor position display image 231.
The end cursor 236 is displayed at the position where the x coordinate is the value “12084” of the SigEnd, and the value “1024” which is the calculation result of 12084-11060 is displayed as the cursor width display image 232.

The display control unit 112 generates a video signal for displaying the display image 230 written in the image memory and outputs the video signal to the display unit 113. The display unit 113 receives the video signal and displays the display image. 230 is displayed (step S17).
The display control unit 112 receives an instruction from the control unit 115 (step S18).
When the instruction is a cursor right movement instruction (step S31: YES) and SigSelect is “0” (step S32: YES), if SigEnd is less than N _max (step S33: NO), SigEnd is set to 1. If it is incremented (step S34) and is equal to or greater than N _max (step S33: YES), the process proceeds to step S18.

If SigSelect is not “0” (step S32: NO), it is determined whether SigStart is greater than or equal to N _max or whether SigStart is greater than or equal to SigEnd (step S35).
If it is determined negative in step S35 (step S35: NO), SigStart is incremented by 1 (step S36), and if it is determined positive (step S35: YES), the process proceeds to step S18.

  If the instruction is a cursor left movement instruction (step S41: YES) and SigSelect is “0” (step S42: YES), it is determined whether SigEnd is greater than 0 and SigEnd> SigStart (step S42: YES). Step S43). If it is determined to be affirmative in step S35 (step S43: YES), SigEnd is decremented by 1 (step S44), and if it is determined to be negative (step S43: NO), the process proceeds to step S18.

If SigSelect is not “0” (step S42: NO), if SigStart is greater than 0 (step S45: YES), SigStart is decremented by 1 (step S46), and if it is 0 or less (step S33: NO), step The process proceeds to S18.
If the instruction is a selection cursor change instruction (step S61: YES) and SigSelect is “0” (step S62: YES), SigSelect is rewritten with a value “1”, and SigSelect is not “0” (step S62). : NO), SigSelect is rewritten with the value “0”, and the process proceeds to step S15.

When the instruction is not a selection range determination instruction (step S65: NO), the process proceeds to step S18. When the instruction is a selection range determination instruction (step S65: YES), the display generation unit 111 instructs the control unit 115 to , SigStart, SigEnd, and the signal data file name are transmitted, and the process ends.
<Display control unit 112, display unit 113>
The display control unit 112 is a graphic adapter including the image memory, generates a video signal representing a display image based on the image data written in the image memory by the display generation unit 111, and sends the video signal to the display unit 113. Output.

The display unit 113 is a liquid crystal display, receives the video signal output from the display control unit 112, and displays the display image on the liquid crystal screen according to the received video signal.
<Filter coefficient calculation unit 109>
The filter coefficient calculation unit 109 performs filter coefficient calculation processing for normal sound removal by autocorrelation function calculation processing and calculation by the Levinson method.

  The filter coefficient calculation unit 109 acquires a signal data file name, SigStart, and SigEnd as an instruction to start autocorrelation function calculation processing from the control unit 115. SigStart indicates the start position of the portion used for calculation in the signal data array and is the same value as the value of the start cursor position information. SigEnd indicates the end position of the portion used for calculation in the signal data array. It is the same value as the cursor position information.

The filter coefficient calculation unit 109 sets a value calculated by (SigEnd−SigStart) as an initial value in the constant L.
The filter coefficient calculation unit 109 obtains the linear prediction coefficient σ, a [1] to a [m] from the signal values X [0] to X [L−1] by the Levinson method, and further, the inverse filter coefficient b [0 ] = A / σ, b [i] = (A / σ) a [i] (where i is 1 ≦ i ≦ m).

The filter coefficient calculation unit 109 transmits b [0] to b [m] to the normal sound removal unit 104 and stores them in the recording unit 108 as an inverse filter coefficient file.
<Normal sound removal unit 104, abnormal sound determination unit 110>
Normal sound removal section 104 calculates a filter coefficient calculator 109 b receives the [0] ~b [m], the received b [0] ~b [m] to the normal sound canceller filter output average amplitude _{P N} based _PN is transmitted to the abnormal sound determination unit 110.

The abnormal sound determination unit 110 acquires _PN from the normal sound removal unit 104, determines whether or not the acquired _PN is _{equal to} or greater than the abnormality detection threshold P _TH , and determines P _N that is _{equal to} or greater than P _TH as T If it is detected not less than _min times and not more than T _max times, it is determined that an abnormal sound has occurred.
The autocorrelation function calculation process used for calculating the inverse filter coefficient and the calculation based on the Levinson method are described in detail in Non-Patent Document 1 described above, and a description thereof will be omitted.
<Control unit 115>
The control unit 115 controls the overall operation of the abnormal sound detection device 1 based on the operation instruction received from the input reception unit 114. The processing of the control unit 115 for each operation instruction is as follows.

When the control unit 115 receives a filter coefficient calculation instruction from the input reception unit 114, the control unit 115 instructs the switch 103 to connect the AD conversion unit 102 and the recording control unit 107, and then instructs the recording control unit 107. In response, a recording start instruction is transmitted.
The control unit 115 waits for the completion of signal data file creation by the recording control unit 107, acquires the signal data file name from the recording control unit 107, and transmits a waveform display instruction including the signal data file name to the display generation unit 111. To do.

The control unit 115 transmits a selection cursor change instruction to the display generation unit 111 when a selection cursor change instruction is received from the input reception unit 114, and displays a cursor right movement instruction when a cursor right movement instruction is received. When it is transmitted to the generation unit 111 and a cursor left movement instruction is received, the cursor left movement instruction is transmitted to the display generation unit 111.
When receiving the cursor position determination instruction from the input receiving unit 114, the control unit 115 acquires the signal data file name, start cursor position information as SigStart, and end cursor position information as SigEnd from the display generation unit 111. Then, the signal data file name, the SigStart, and the SigEnd are transmitted to the filter coefficient calculation unit 109.

When the control unit 115 receives an abnormal sound detection start instruction from the input reception unit 114, the control unit 115 instructs the switch 103 to connect the AD conversion unit 102 and the normal sound removal unit 104, and removes the normal sound. The abnormal sound detection start instruction is transmitted to the unit 104 and the recording control unit 107.
When receiving the abnormal sound detection end instruction from the input receiving unit 114, the control unit 115 transmits an abnormal sound detection end instruction to the normal sound removing unit 104 and the recording control unit 107.
<Operation>
Hereinafter, with respect to the operation of the abnormal sound detection apparatus 1, (1) a filter coefficient calculation process for calculating a coefficient of an inverse filter for removing normal sound from an acoustic signal, and (2) an inverse using the inverse filter coefficient Description will be made in the order of abnormal sound detection processing for detecting abnormal sound included in a signal to be inspected using a filter.
(1) Filter coefficient calculation process The operation of the filter coefficient calculation process will be described with reference to FIGS.

The user instructs the input reception unit 114 to calculate a filter coefficient using the keyboard.
The input receiving unit 114 acquires a filter coefficient calculation instruction, and transmits a filter coefficient calculation instruction to the control unit 115.
The control unit 115 acquires the filter coefficient calculation instruction (step S501).

The control unit 115 instructs the switch 103 to connect the AD conversion unit 102 and the recording control unit 107, and the switch 103 receives the instruction and connects the AD conversion unit 102 and the recording control unit 107. Connect (step S502).
The control unit 115 transmits a signal recording start instruction to the recording control unit 107 (step S503).

The recording control unit 107 receives the signal recording start instruction and generates a signal data file in the buffer unit 106 (step S505).
The recording control unit 107 initializes the number of samplings held as a variable with the value “0” (step S506).
The recording control unit 107 determines whether or not the number of samplings is equal to or less than N _max (step S508). If it is equal to or less than N _max (step S508: YES), the AD conversion result register of the AD conversion unit 102 The AD conversion result is read (step S510), and the AD conversion result is recorded in an unused area of the signal data array of the signal data file (step S511).

The recording control unit 107 increments the sampling count by “1” (step S512), and returns to step S508.
If the number of samplings is greater than N _max (step S508: NO), the recording control unit 107 ends the signal recording and notifies the control unit 115 of the end (step S521).
As a result of the loop from step S507 to step S512, the signal data array in the signal data file is Sig [0] = 8, Sig [1] = 84, Sig [2] = 84... Sig [43477] = 192 Is recorded.

The control unit 115 transmits a waveform display instruction including the signal data file name to the display generation unit 111 (step S523).
The display generation unit 111 receives the waveform display instruction, is identified by the signal data file name included in the waveform display instruction, and is based on the signal data arrangement in the signal data file recorded in the buffer unit 106 as described above. A display image as shown in FIG. 6 is generated, written in the image memory, and instructed to be displayed (step S524).

Next, cursor position determination processing is performed (step S531).
Details of step 531 are the cursor position determination processing already described with reference to FIGS.
Next, the filter coefficient calculation unit 109 obtains a signal data file name, SigStart, and SigEnd from the control unit 115 as an instruction to start autocorrelation function calculation processing, and is recorded in the buffer unit 106, and the signal data file name The filter coefficient calculation process is executed using Sig [SigStart] to Sig [SigEnd-1] in the signal data file identified by (Step S532).

Here, Sig [SigStart] to Sig [SigEnd-1] correspond to the aforementioned X [0] to X [L-1].
In the filter coefficient calculation process, the filter coefficient calculation unit 109 obtains linear prediction coefficients σ, a [1] to a [m] from the signal values X [0] to X [L−1] by the Levinson method, and Inverse filter coefficients b [0] = A / σ, b [i] = (A / σ) a [i] (where i is 1 ≦ i ≦ m) are calculated.

  As an example, as a result of the autocorrelation function calculation process in the filter coefficient calculation process, the filter coefficient calculation unit 109 performs r [0] = 1.21429e + 10, r [1] = 9.45524e + 9,... R [15] = −9.06491e + 08 is obtained (where m = 15). Further, the filter coefficient calculation unit 109 obtains, as an example, linear prediction coefficients σ = 48720.7, a [1] = − 0.985659... A [15] = − 0.0251498 by the Levinson method.

Next, the filter coefficient calculation unit 109 performs an inverse filter coefficient determination process. As a result, the filter coefficient calculation unit 109, as an example, uses the inverse filter coefficient b [0] = 1.345575 and b [1] = − 1. 32442,..., B [15] = − 0.0326441 (however, gain A = 2.0).
(2) Abnormal Sound Detection Process The abnormal sound detection process performed using the inverse filter coefficient calculated in the filter coefficient calculation process will be described with reference to FIG.

The normal sound removal unit 104 includes a memory area for storing the inverse filter coefficient, receives b [0] to b [m] from the filter coefficient calculation unit 109, and stores the received b [0] to b [m] in the memory area.
In addition, an abnormal sound detection target file including abnormal sound detection target data for recording an AD conversion result acquired from the AD conversion unit 102 is generated in the buffer unit 106.
Here, it is assumed that the abnormal sound detection target file constitutes a ring buffer having a certain length (for example, N _max ).

When the normal sound removal unit 104 receives an abnormal sound detection start instruction from the control unit 115, the normal sound removal unit 104 starts the abnormal sound detection described below.
Normal sound removal section 104, sequence w [i] (where, 0 ≦ i ≦ m), and secure a temporary storage area for storing variables _{P N, w [0] =} w [1] = · ... = W [m] = 0, P _N = 0 is executed (step S401), and iw = 0, n = 0, q = 0, Tp = 0 is executed (step S402).

Further, the normal sound removal unit 104 reads the AD conversion result by the AD conversion unit 102 from the AD conversion result register and sends it to the recording control unit 107. The recording control unit 107 transmits this to the abnormal sound of the abnormal sound detection target file. The data is stored in the detection target data array S [n] (step S403).
Next, the normal sound removal unit 104 executes the result of w [iw] = AD conversion (step S404).
next,

を計算し（ステップＳ４０５）、次に、式
ｉｗ＝Ｍｏｄ（ｉｗ＋ｍ，ｍ＋１）
を計算し（ステップＳ４０６）、式
Ｐ_Ｎ＝λＰ_Ｎ＋（１−λ）｜Ｙ_Ｎ｜
を計算し（ステップＳ４０７）、Ｐ_Ｎを、異常音判定部１１０に送信する。

(Step S405), and then the equation iw = Mod (iw + m, m + 1)
Is calculated (step S406), and the expression P _N = λP _N + (1−λ) | Y _N |
The calculated (step S407), the _{P N,} and transmits the abnormal sound determination unit 110.

Here, the description “Mod (A, B)” indicates the result of the remainder operation of A by B.
Abnormal sound determination unit 110 acquires the _{P N} from the normal sound removal section 104, the obtained _{P N,} it is determined whether or _{P TH} (step _S408), when it is determined that the _{P TH} or more (step S408 : YES), incremented by 1 Tp (step _S409), when it is determined that _{P TH} is smaller than (step S408: NO), the Tp is judged whether _{T max} below (step S410).

When it is determined that Tp is not equal to or less than _Tmax (step S410: NO), the process proceeds to step S414, which will be described later. When it is determined that Tp is equal to or less than _Tmax (step S410: YES), further, Tp is equal to T _It is determined whether or not it is not less than _min .
If Tp is determined to not more than _{T min} (step S411: NO), the process proceeds to step S412 to be described later, if Tp is determined to be equal to or greater than _{T min} (step S410: YES), the abnormality in the control unit 115 A sound generation signal is transmitted (step S412).

As a result, the control unit 115 transmits an abnormal signal storage instruction to the recording control unit 107. In response to this, the recording control unit 107 captures a certain number of data, and records the abnormal data file created from the data stored in the ring buffer in the abnormal sound detection target file stored in the buffer unit 106 at that time. The data is stored in the unit 108 (step S413).
In the present embodiment, as shown in step S411 from step _S407, the data is not less than _{P TH,} if it detects less than _{T max} times _{T min} times, it is determined as abnormal sound has occurred . When _PTH times or more are detected, it is determined that there is some abnormality in the abnormal sound detection process.

Next, Tp is initialized to 0 (step S414), and the normal sound removing unit 104 determines whether or not to continue the process, and when continuing (step S415: YES), n = Mod (n + 1, N _max ) Is calculated (step S416), the process proceeds to step S403, and if not continued (step S415: NO), the process ends.
In step S415, for example, whether or not to continue the process is determined to be the process end when an abnormal sound detection end instruction is received from the control unit 115 as an example.
<Supplement>
On the other hand, unlike the present application, the user does not select data to be used for filter coefficient calculation, and 1024 pieces of data from the beginning of the signal data array determined in advance, that is, Sig [0] = 8, Sig [1] = 84. , Sig [2] = 84... Sig [1023] = 80, the following results are obtained as an example.

r [0] = 1.66365e + 08, r [1] = 1.64333e + 08,..., r [15] = 1.40538e + 08.
σ = 1844.92, a [1] = 1.111129,..., a [15] = 0.0733347.
b [0] = 1.78564, b [1] =-1.96717,..., b [15] = 0.155876 (where gain A = 0.1).

FIG. 10 shows (a) a filter characteristic diagram of the inverse filter coefficient obtained by the method of the present application, and (b) a filter characteristic diagram of the inverse filter coefficient obtained by using 1024 pieces of data from the head of the signal data array. The horizontal axis of the filter characteristic diagram indicates the frequency, and the vertical axis indicates the input / output ratio (decibel) of the input signal to the output signal.
In FIG. 10B, the characteristic curve 302 does not sufficiently reflect the characteristics of the normal sound because the calculation is performed using a portion that does not include a characteristic signal such as a pulse signal in a normal sound. The attenuation factor is large in the low frequency area, and there is a danger of overlooking the abnormality of the low frequency component.

On the other hand, in FIG. 10A, since the calculation is performed using a signal including a characteristic part such as a pulse signal in a normal sound in the calculation, the characteristic curve 301 sufficiently reflects the characteristic of the normal sound. It shows the characteristics and can detect low-frequency abnormalities.
<Modification>
Although the present invention has been described based on the above embodiment, it is needless to say that the present invention is not limited to the above embodiment. The following cases are also included in the present invention.
(1) In the above embodiment, an apparatus for detecting abnormal sound has been described. However, the present invention is not limited to sound but can be applied to vibrations other than sound.

  For example, the microphone 101 in the above embodiment may be changed to a piezoelectric element that converts a force applied to the piezoelectric body into a voltage. When a force is applied to the piezoelectric element due to vibration generated by the operation of the machine, the piezoelectric element converts the force into an electric signal, and outputs the electric signal to the AD conversion unit 102. The configuration other than the piezoelectric element is the same as that of the above embodiment.

  Using the above-described configuration, a vibration signal indicating vibration measured by the piezoelectric element is recorded as a normal vibration signal while the machine that is the target of detection of abnormal vibration is operating normally. The inverse filter coefficient is calculated using the normal vibration signal in the same procedure as. After the calculation, when the presence / absence of abnormal vibration is determined, the piezoelectric element measures the vibration that is the object of the determination, and thereafter, the abnormality indicating the measured vibration in the same procedure as described in the embodiment. The presence or absence of abnormal vibration is determined using a signal obtained by removing the normal vibration signal from the vibration detection target signal.

Further, instead of the piezoelectric element, a device capable of measuring vibration such as an accelerometer or a gyro may be used.
(2) In the above-described embodiment, all the processes are realized in the abnormal sound detection apparatus 1. However, the present invention is not limited to this, and the processes may be shared by a plurality of apparatuses.
For example, as shown in FIG. 11, the abnormal sound detection apparatus 1 includes an abnormal sound detection unit 401 that is in charge of the inverse filter coefficient calculation process and the abnormal sound detection process of the filter, and obtains user operation instructions and displays an image. The personal computer 402 may be configured separately. The abnormal sound detection unit 401 and the personal computer 402 are connected via a network 403 such as IEEE1394.

As shown in FIG. 12, the abnormal sound detection unit 401 includes a microphone 101, an AD conversion unit 102, a switch 103, a normal sound removal unit 104, a time measurement unit 105, a buffer unit 106, a recording control unit 107, a recording unit 108, and a filter coefficient. The calculation unit 109, the abnormal sound determination unit 110, the control unit 115, and the communication unit 411 are configured.
Microphone 101, AD conversion unit 102, switch 103, normal sound removal unit 104, timing unit 105, buffer unit 106, recording control unit 107, recording unit 108, filter coefficient calculation unit 109, abnormal sound determination unit 110, and control unit 115 Shall perform the same operation as that in the abnormal sound detection device 1 of the above embodiment.

The communication unit 411 transmits the signal data file stored in the buffer unit 106 to the personal computer 402 via the network 403.
As shown in FIG. 13, the personal computer 402 includes a display generation unit 111, a display control unit 112, a display unit 113, an input reception unit 114, a communication unit 421, and a control unit 422.

The display generation unit 111, the display control unit 112, the display unit 113, and the input reception unit 114 perform the same operations as those in the abnormal sound detection device 1 of the above embodiment.
The communication unit 421 performs data communication with the abnormal sound detection unit 401 via the network 403.
In the embodiment, the processing performed by the control unit 115 is performed by the control unit 115 of the abnormal sound detection unit 401 and the control unit 422 of the personal computer 402.

  In the embodiment, the control unit 115 directly receives an operation instruction by a user's keyboard operation from the input reception unit 114. However, in the present modification, the input reception unit 114 is acquired by the user operating the keyboard. The control unit 422 that has transmitted the operation instruction to the control unit 422 and has received the operation instruction transmits the operation instruction to the control unit 115 via the communication unit 421 and the communication unit 411.

In the embodiment, the display generation unit 111 receives a waveform display instruction including a signal data file name from the control unit 115 and is identified by the signal data file name, and is recorded in the buffer unit 106. The display image is generated based on the content of the image and the image data indicating the display image is written in the image memory included in the display control unit 112. However, in the present modification, the display generation unit 111 is controlled by the control unit 115. A waveform display instruction including the signal data file name is received via the communication unit 411, the network 403, and the communication unit 421, and a signal data file read request including the signal data file name is transmitted via the communication unit 421 and the communication unit 411. The control unit 115 transmits the signal data file included in the signal data file read request to the communication unit 411. An instruction to transmit the signal data file indicated by the name in the personal computer 402. The communication unit 411 reads out the signal data file from the buffer unit 106 and transmits the signal data file to the display generation unit 111 via the network 403 and the communication unit 421. The display generation unit 111 acquires the signal data file.
(3) After determining the inverse filter coefficient, the display generation unit 111 in the abnormal sound detection apparatus 1 displays the display image 230 shown in FIG. 7 as shown in FIG. 14 and the determined inverse filter coefficient shown in FIG. A filter characteristic diagram representing the above characteristic may be displayed in an overlapping manner.

This allows the user to know at a glance the characteristics of the inverse filter coefficient calculated using the partial signal between the selected start cursor position and end cursor position by operating the keyboard. It is easy to determine whether or not to calculate the inverse filter coefficient by reselecting the start cursor position and the end cursor position because the characteristics of the are significantly different from the expected ones. become.
(4) The display generation unit 111 displays one set of cursors of the start cursor and end cursor in the display image 230. However, the display generation unit 111 is not limited to this, and displays two or more cursor sets. Also good.

As an example, an example in which two cursor sets are shown in the display image 230 will be described with reference to FIGS.
The display generation unit 111 acquires a waveform display instruction including the signal data file name from the control unit 115 (step S11).
The display generation unit 111 reads out the signal data file identified by the signal data file name from the buffer unit 106 (step S12).

The display generation unit 111 uses the signal data array Sig [n] in the read signal data file, as shown in FIG. 19 as an example, on the x-axis and y-axis orthogonal coordinates with the origin 237 as the origin, A graph 234 in which (n, Sig [n]) is plotted as (x coordinate, y coordinate) is generated (step S13). Here, n is a natural number and takes a value of 0 to N _max −1.

  The display generation unit 111 includes SigStart [0] indicating the x coordinate of the start cursor 235, SigEnd [0] indicating the x coordinate of the end cursor 236, SigSelect [0] indicating either the start cursor 235 or the end cursor 236, and start. SigStart [1] indicating the x coordinate of the cursor 238, SigEnd [1] indicating the x coordinate of the end cursor 239, and SigSelect [1] indicating either the start cursor 238 or the end cursor 239 are managed. In addition, the display generation unit 111 includes a first cursor set, which is a set of the start cursor 235 and an end cursor 236, and a second set, which is a set of the start cursor 238 and the end cursor 239. SigGroup indicating which of the cursor sets is managed.

Here, when SigGroup is a value “0”, the first cursor set is indicated, and when the value is “1”, the second cursor set is indicated.
The display generation unit 111 initializes SigStart [0] and SigStart [1] with 0, initializes SigEnd [0] and SigEnd [1] with N _max , and initializes SigSelect [0] and SigSelect [1] with 0. And SigGroup is initialized with “0” (step S1014).

  As shown in FIG. 19 as an example, the display generation unit 111 superimposes a start cursor 235 indicated by a broken line parallel to the y axis on the position where the x coordinate is SigStart [0], as shown in FIG. An end cursor 236 indicated by a broken line parallel to the y axis is superimposed on a position where the coordinates are SigEnd [0], and a start cursor indicated by a broken line parallel to the y axis is positioned at a position where the x coordinates are SigStart [1]. 238 is superimposed, and a graph image 233 is generated, which is an image in which an end cursor 239 indicated by a broken line parallel to the y axis is superimposed at a position where the x coordinate is SigEnd [1] (step S1015).

Here, it is assumed that a display image is generated so that a cursor indicated by SigSelect [SigGroup] is displayed in red so that it can be recognized that it is an operation target, and other cursors are displayed in black.
As an example, as illustrated in FIG. 15, the display generation unit 111 generates a cursor position display image 231 that displays a value of SigStart [SigGroup], and displays a cursor width display that displays a value of SigEnd [SigGroup] -SigStart [SigGroup]. An image 232 is generated, a cursor set number display image 240 for displaying SigSelect [SigGroup] is generated, a display image 230 combined with the graph image 233 is generated, and the display image 230 is stored in the image memory included in the display control unit 112. Writing is performed (step S1016).

For example, in FIG. 15, it is assumed that SigGroup is “0”, and the start cursor 235 is displayed at the position where the x coordinate is the value “11060” of SigStart [SigGroup], and the value “11060” of SigStart [SigGroup] is The cursor position display image 231 is displayed.
The end cursor 236 is displayed at the position of the value “12084” of the SigEnd [SigGroup] x coordinate, and the value “1024” that is the calculation result of 12084-11060 is displayed as the cursor width display image 232. .

The display control unit 112 generates a video signal for displaying the display image 230 written in the image memory and outputs the video signal to the display unit 113. The display unit 113 receives the video signal and displays the display image. 230 is displayed (step S17).
The display control unit 112 receives an instruction from the control unit 115 (step S18).
If the instruction is an instruction to move the cursor to the right (step S31: YES), and SigSelect [SigGroup] is “0” (step S1032: YES), if SigEnd [SigGroup] is less than N _max (step S1033: NO), SigEnd [SigGroup] is incremented by 1 (step S1034), and if it is N _max or more (step S1033: YES), the process proceeds to step S18.

  If SigSelect [SigGroup] is not “0” (step S32: NO), it is determined whether SigStart [SigGroup] ≧ Nmax or SigStart [SigGroup] ≧ SigEnd [SigGroup] (step S1035). If it is determined to be negative in step S1035 (step S1035: NO), SigStart [SigGroup] is incremented by 1 (step S1036), and if it is determined to be positive (step S1035: YES), step The process proceeds to S18.

  When the instruction is a cursor left movement instruction (step S41: YES) and SigSelect [SigGroup] is “0” (step S42: YES), SigEnd [SigGroup]> 0 and SigEnd [SigGroup]> SigStart [SigGroup] ] (Step S1043). In step S1043, when it is determined to be affirmative (step S1043: YES), SigEnd [SigGroup] is decremented by 1 (step S1044), and when it is determined to be negative (step S1043: NO), step The process proceeds to S18.

  When SigSelect [SigGroup] is not “0” (step S1042: NO), if SigStart [SigGroup] is larger than 0 (step S1045: YES), SigStart [SigGroup] is decremented by 1 (step S1046), and is 0 or less. If (step S1045: NO), the process proceeds to step S18.

  If the instruction is a selection cursor change instruction (step S1061: YES) and the selection cursor information SigSelect [SigGroup] is “0” (step S1062: YES), the SigSelect [SigGroup] is rewritten with the value “1”, and SigSelect If [SigGroup] is not “0” (step S1062: NO), SigSelect [SigGroup] is rewritten with the value “0”, and the process proceeds to step S1015.

If the instruction is a selection range determination instruction (step S1065: YES), the process ends. If the instruction is not a selection range determination instruction (step S1065: NO), the instruction is a selection cursor set change instruction (step S1065: NO) In step S1067: YES), the display generation unit 111 replaces SigGroup with Mod (SigGroup + 1, the number of sets) (step S1068), and proceeds to step S1015. Here, the “number of sets” is the number of cursor sets to be displayed. In this modification, the number of sets = 2.
(5) Although the initial positions of the start cursors 235 and 238 are set to “0” and the initial positions of the end cursors 236 and 239 are set to N _max in step S1014 of the modification (4) described above, this is not restrictive. The initial position of the cursor may be moved to a position that is assumed to be easy for the user to operate, as shown in FIG. 14 as an example.

For example, among Sig [0] to Sig [N _max −1], a predetermined number (for example, 1024) of array elements including the array element holding the maximum value (for example, Sig [m]) is set as the range. Alternatively, a pair of SigStart [0] and SigEnd [0] or a pair of SigStart [1] and SigEnd [1] may be determined. If the array element having the maximum value is Sig [m], SigStart [0] = (m−511), SigEnd [0] = (m + 512). However, SigStart [0] and SigEnd [0] are determined to be 0 or more and N _max or less. When m = 50, m−511 is smaller than “0”, so SigStart [0] = 0, SigEnd [0] = 1024, and the initial value is determined so that 1024 data are used. Alternatively, SigStart [0] may be set to 0 and SigEnd [0] = m + 512 = 562.

In addition, among Sig [0] to Sig [N _max −1], a predetermined number (for example, 1024) of array elements including an array element (for example, Sig [m]) that holds the minimum value instead of the maximum value. May be determined as a set of SigStart [0] and SigEnd [0], or a set of SigStart [1] and SigEnd [1].
(6) The initial position of the cursor may be as shown below in addition to the above-described modification (5).

For example, among Sig [0] to Sig [N _max −1], a set of SigStart [0] and SigEnd [0] or SigStart [0] so as to indicate a range in which a value equal to or greater than a predetermined threshold value continues for a predetermined number or more. A set of [1] and SigEnd [1] may be determined.
For example, when the values of Sig [2300] to Sig [2700] exceed a predetermined threshold (for example, 4000), initial values are set as SigStart [0] = 2300 and SigEnd [0] = 2700. .

Further, SigStart [0] and SigEnd [0] are used so that a predetermined number (for example, 500) of signal data is used before and after the middle of a predetermined range (in this modification, 2500 which is the middle between 2300 and 2700). It may be set. In this case, SigStart [0] = 2000 and SigEnd [0] = 3000.
In addition, as described above, SigStart [[Smax [Smax [ _Nmax- 1]] is not a value equal to or greater than a predetermined threshold value, but indicates a range in which a value equal to or greater than the predetermined threshold value is continuous. 0] and SigEnd [0], or SigStart [1] and SigEnd [1].
(7) Each of the above devices is specifically a computer system including a microprocessor, ROM, RAM, a hard disk unit, a display unit, a keyboard, a mouse, and the like. A computer program is stored in the RAM or the hard disk unit. Each device achieves its functions by the microprocessor operating according to the computer program. Here, the computer program is configured by combining a plurality of instruction codes indicating instructions for the computer in order to achieve a predetermined function.
(8) A part or all of the constituent elements constituting each of the above devices may be configured by one system LSI (Large Scale Integration). The system LSI is an ultra-multifunctional LSI manufactured by integrating a plurality of components on a single chip, and specifically, a computer system including a microprocessor, ROM, RAM, and the like. . A computer program is stored in the RAM. The system LSI achieves its functions by the microprocessor operating according to the computer program.
(9) A part or all of the constituent elements constituting each of the above devices may be constituted by an IC card or a single module that can be attached to and detached from each device. The IC card or the module is a computer system including a microprocessor, a ROM, a RAM, and the like. The IC card or the module may include the super multifunctional LSI described above. The IC card or the module achieves its function by the microprocessor operating according to the computer program. This IC card or this module may have tamper resistance.
(10) The present invention may be the method described above. Further, the present invention may be a computer program that realizes these methods by a computer, or may be a digital signal composed of the computer program.

  The present invention also provides a computer-readable recording medium such as a flexible disk, hard disk, CD-ROM, MO, DVD, DVD-ROM, DVD-RAM, BD (Blu-ray Disc). ), Recorded in a semiconductor memory or the like. Further, the present invention may be the computer program or the digital signal recorded on these recording media.

Further, the present invention may transmit the computer program or the digital signal via an electric communication line, a wireless or wired communication line, a network represented by the Internet, a data broadcast, or the like.
The present invention may be a computer system including a microprocessor and a memory, wherein the memory stores the computer program, and the microprocessor operates according to the computer program.

In addition, the program or the digital signal is recorded on the recording medium and transferred, or the program or the digital signal is transferred via the network or the like, and is executed by another independent computer system. It is good.
(11) The above embodiment and the above modifications may be combined.

  The inverse filter coefficient determination method of the present invention is used for detecting an abnormality of a machine or the like in a manufacturing industry that manufactures a product using a machine or the like.

It is a block diagram which shows the structure of the abnormal sound detection apparatus which is one Embodiment of this invention. It is a figure which shows the structure of the signal data file recorded on a buffer part. It is a figure which shows an example of the display image displayed on a display part. It is a flowchart which shows the operation | movement of a cursor position determination process. It is a flowchart which shows the operation | movement of a cursor position determination process. It is a flowchart which shows the operation | movement of a cursor position determination process. It is a figure which shows an example of the display image displayed on a display part. It is a flowchart which shows a filter coefficient calculation process. It is a flowchart which shows an abnormal sound detection process. It is a figure which shows an example of the frequency characteristic of the inverse filter using the filter coefficient computed by this invention, and the frequency characteristic of the conventional inverse filter. It is the schematic of the abnormal sound detection system comprised from an abnormal sound detection unit and a personal computer. It is a block diagram which shows the structure of an abnormal sound detection unit. It is a block diagram which shows the structure of a personal computer. It is a figure which shows the display image which displays a filter characteristic figure and a signal waveform collectively. It is a figure which shows the display image which displays multiple sets of a start and an end cursor. It is a figure which shows operation | movement of the cursor position determination process in the case of displaying multiple sets of cursors. It is a figure which shows operation | movement of the cursor position determination process in the case of displaying multiple sets of cursors. It is a figure which shows operation | movement of the cursor position determination process in the case of displaying multiple sets of cursors.

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 Microphone 102 Conversion part 103 Switch 104 Normal sound removal part 105 Control part 105 Timekeeping part 106 Buffer part 107 Recording control part 108 Recording part 109 Filter coefficient calculating part 110 Abnormal sound determination part 111 Display generation part 112 Display control part 113 Display part 114 Input reception unit 115 Control unit 201 Signal data file 204 Signal data array 230 Display image 231 Cursor position display image 232 Cursor width display image 233 Graph image 234 Graph 235 Start cursor 236 End cursor 237 Origin 301 Characteristic curve 302 Characteristic curve

Claims (12)

An inverse filter coefficient determination method for calculating an inverse filter coefficient of an inverse filter that detects abnormal vibration by removing normal vibration from input vibration,
A recording step for recording a normal vibration signal;
A waveform display step of generating a signal waveform visualizing the normal vibration signal and displaying the signal waveform;
Based on an instruction from a user who viewed the displayed signal waveform, a specifying step for specifying a range including a characteristic portion of the signal waveform;
And a calculation step of calculating an inverse filter coefficient using a partial signal corresponding to the range of the normal vibration signal. The specific step includes
A reception step for receiving the instruction;
A provisional specifying step of selecting a provisional range satisfying a predetermined condition among the signal waveforms;
A display update step of attaching a display indicating the temporary range to the display of the signal waveform;
The inverse filter coefficient determination method according to claim 1, further comprising: a main specifying step of specifying the range by increasing or decreasing the temporary range based on the instruction. An inverse filter coefficient determination device that calculates an inverse filter coefficient of an inverse filter that detects abnormal vibration by removing normal vibration from input vibration,
A recording means for recording a normal vibration signal;
Waveform display means for generating a signal waveform visualizing the normal vibration signal and displaying the signal waveform;
Based on an instruction from a user who viewed the displayed signal waveform, a specifying means for specifying a range including a characteristic portion of the signal waveform;
An inverse filter coefficient determination device, comprising: a calculation unit that calculates an inverse filter coefficient using a partial signal corresponding to the range of the normal vibration signal. The specifying means is:
A reception unit for receiving the instruction;
A provisional specifying unit for selecting a provisional range that satisfies a predetermined condition among the signal waveforms;
A display update unit for attaching a display indicating the temporary range to the display of the signal waveform;
The inverse filter coefficient determination apparatus according to claim 3, further comprising: a main specifying unit that specifies the range by increasing or decreasing the temporary range based on the instruction. The temporary specific part is:
A first selection unit for selecting a point having the maximum amplitude in the signal waveform;
The inverse filter coefficient determination device according to claim 4, further comprising: a second selection unit that selects a predetermined range including the maximum point in the signal waveform as the temporary range. The temporary specific part is:
The inverse filter coefficient determination device according to claim 4, wherein the temporary range in which an amplitude of a predetermined magnitude or more in the signal waveform is continuous for a predetermined length or more is selected. The provisional specific part is:
A reading unit for sequentially reading the recorded normal vibration signals;
A determination unit that determines whether or not the read signal exceeds a predetermined value each time the signal is read;
A counting unit for counting the number of times determined to continuously exceed a predetermined value;
The reverse unit according to claim 6, further comprising: an output unit that outputs the temporary range determined to continuously exceed a predetermined value in the normal vibration signal when the number of times exceeds a predetermined number. Filter coefficient determination device. The specifying means is:
A first temporary specifying unit for selecting a first temporary range that satisfies a predetermined condition among the signal waveforms;
A first display update unit for attaching a display indicating the first temporary range to the display of the signal waveform;
A first book specifying unit for increasing or decreasing the first temporary range based on the instruction;
A second temporary specifying unit that selects a second temporary range that satisfies the predetermined condition from the signal waveform, different from the range selected by the first temporary specifying unit;
A second display update unit for attaching a display indicating the second temporary range to the display of the signal waveform;
A second main identifying unit that increases or decreases the second temporary range based on the instruction;
The inverse filter coefficient determination apparatus according to claim 3, further comprising: an output unit that outputs the temporary range obtained by combining the first temporary range and the second temporary range. The specifying means is:
A first specifying unit for specifying a first range satisfying a predetermined condition in the signal waveform;
A second specifying unit that satisfies a predetermined condition of the signal waveform and specifies a second range different from the first range;
A selection unit for selecting one of the first range and the second range;
A display update unit for attaching a display indicating the range selected by the selection unit to the display of the signal waveform;
The inverse filter coefficient determination apparatus according to claim 3, further comprising: an output unit that increases or decreases a range selected by the selection unit based on the instruction and outputs the range as the temporary range. The inverse filter coefficient determination device according to claim 3, wherein the recording unit records a normal sound signal as the normal vibration signal. An abnormal vibration detection device that detects normal vibration by removing normal vibration from input vibration,
A recording means for recording a normal vibration signal;
Waveform display means for generating a signal waveform visualizing the normal vibration signal and displaying the signal waveform;
Based on an instruction from a user who viewed the displayed signal waveform, a specifying means for specifying a range including a characteristic portion of the signal waveform;
Calculating means for calculating an inverse filter coefficient using a partial signal corresponding to the range of the normal vibration signal;
Receiving means for receiving an input vibration signal;
An abnormal vibration detection device comprising: a detecting unit that removes a normal vibration signal from the input vibration signal using an inverse filter based on the inverse filter coefficient and detects an abnormal vibration signal included in the input vibration signal. . The specifying means is:
A reception unit for receiving the instruction;
A temporary identification unit that selects a temporary range that satisfies a predetermined condition among the signal waveforms;
A display update unit for attaching a display indicating the temporary range to the display of the signal waveform;
The abnormal vibration detecting apparatus according to claim 11, further comprising: a main specifying unit that specifies the range by increasing or decreasing the temporary range based on the instruction.

Images