JP5022774B2 - Inspection method and inspection apparatus - Google Patents

Description

  The present invention relates to, for example, an inspection method and an inspection apparatus that collects measurement data for inspection objects such as vibration and noise and determines pass / fail.

  For example, as one of product inspections for A / T (automatic transmission) for automobiles, there is an inspection for measuring vibration or noise (gear noise) while rotating the A / T. In this inspection, vibration or noise data is collected as measurement data to be inspected, and pass / fail is determined based on whether the raw data or the data after statistical processing exceeds a predetermined threshold. As specific inspection methods, various techniques have been proposed so far (for example, Patent Document 1).

However, high-accuracy inspections depend on, for example, the reliability of collected measurement data itself. That is, for example, when collecting measurement data of sound, if the sensor (microphone) is faulty or malfunctioning, data that is supposed to be abnormal although it should be normal is collected as measurement data. Conversely, data that is supposed to be normal but should be normal may be collected. In this case, the inspection result is naturally low quality.
Such a defect can be avoided to some extent by frequently inspecting the inspection apparatus and confirming the reliability of the inspection apparatus itself by a test inspection using reference simulation data or the like. However, there is a limit to increasing the frequency of inspections and test inspections, and some new measures have been required.

JP 2006-258535 A

  The present invention has been made in view of such conventional problems, and an object of the present invention is to provide an inspection method and an inspection apparatus in which the inspection apparatus itself can determine the reliability of the measurement data itself collected by the inspection apparatus. .

The first invention determines whether or not the state of the inspection object is normal using the measurement data Y _i (i = 1 to n, n is a natural number) of the inspection object sequentially collected corresponding to the change of the reference variable. In the inspection method having a pass / fail judgment step,
The inspection object is vibration or noise generated from the inspection object,
The inspection object is an A / T (automatic transmission), the reference variable is the rotation speed of the output shaft of the A / T, and the measurement data Y _i is set to a value determined in advance. Measure every time
In addition to the above pass / fail determination step, the computer has a reliability determination step for determining the reliability of the pass / fail determination step result by a computer.
The reliability determination step includes:
Using the standard deviation σ _i and the average value μ _i obtained for each of the same reference variables for the measurement data Y _i collected in the past,
Whether N (2 ≦ N ≦ n) of the total n measurement data Y _i collected this time exceeds the corresponding confidence range of μ _i ± mσ _i (1 ≦ m ≦ 3) In the inspection method, the reliability of the pass / fail determination step result is determined based on the determination (claim 1).

In the inspection method of the present invention, the reliability determination step is performed separately from the original pass / fail determination step. In this reliability determination step, as described above, the standard deviation σ _i and the average value μ _i obtained for each of the same reference variables using the measurement data Y _i collected in the past are used.

That is, for example, when there is a test of the performance of M times of passes examples in the past, Y ₁₁ as measurement data of the first _{test, Y 12, ··· Y 1i ···} Y 1n of n data Furthermore, there are n data of Y ₂₁ , Y ₂₂ ... Y _2i ,... Y _2n as measurement data for the second inspection, and Y _M1 , Y as measurement data for the M-th inspection. _M2, ··· Y _Mi, and thus there are n data · · · Y _Mn. In this case, since there are M pieces of past measurement data corresponding to the same reference variable, the standard deviation σ _i and the average value μ _i can be obtained from the M pieces of data.

When all the measurement data are obtained, n standard deviations σ _i and average values μ _i are obtained. All n standard deviations σ _i and average values μ _i may be used, or only some (N) may be used.
As will be described later, these standard deviation σ _i and average value μ _i may be calculated and used as one step of the inspection method of the present invention, but using those previously calculated and stored as fixed values. Also good.

In the reliability determination step, N pieces of the measurement data Y _i collected in the current examination are used for determination. N may be all the collected numbers (n) or may be reduced to an appropriate number. Then, for each measurement data Y _i , a determination is made as to whether or not the confidence range of the corresponding μ _i ± mσ _i (1 ≦ m ≦ 3) is exceeded. Various determination methods can be used as specific determination methods. In the present invention, the determination is performed based on the standard deviation σ _i and the average value μ _i .

As described above, in the inspection method of the present invention, the reliability determination step of determining the reliability of the current measurement data itself using the standard deviation σ _i and the average value μ _i of the past measurement data is positively adopted. is doing. Therefore, it can be determined from the result of the reliability determination step whether or not the original result of the pass / fail determination step is reliable.

  That is, regardless of the result of the pass / fail determination step, if the result of the reliability determination step is unreliable or low reliability, the result of the pass / fail determination step remains as it is. It can be determined that it cannot be used. Conversely, if the result of the pass / fail determination step is any result, if the result of the reliability determination step is reliable or highly reliable, the result of the pass / fail determination step is It can be judged that it can be used as it is.

Furthermore, if the result of the reliability determination step is a result of no reliability or low reliability, it is confirmed whether there is a failure or malfunction of the inspection apparatus or due to the characteristics of the inspection object itself. It can be determined that re-inspection or inspection of the device should be performed.
As described above, when the inspection method of the present invention is used, the above-described determination that cannot be made conventionally can be added, so that the reliability of the determination by the pass / fail determination step can be significantly improved as compared with the conventional case.

The second invention determines whether or not the state of the inspection object is normal using the measurement data Y _i (i = 1 to n, n is a natural number) of the inspection object sequentially collected corresponding to the change of the reference variable. In the inspection apparatus having a pass / fail determination means for performing the pass / fail determination step,
The inspection object is vibration or noise generated from the inspection object,
The inspection object is an A / T (automatic transmission), the reference variable is the rotation speed of the output shaft of the A / T, and the measurement data Y _i is set to a value determined in advance. Is measured every time
In addition to the above pass / fail determination means, the computer has reliability determination means for performing a reliability determination step for determining the reliability of the pass / fail determination step result by a computer
The reliability determination means includes
Using the standard deviation σ _i and the average value μ _i obtained for each of the same reference variables from the measurement data Y _i collected in the past,
Whether N (2 ≦ N ≦ n) of the total n measurement data Y _i collected this time exceeds the corresponding confidence range of μ _i ± mσ _i (1 ≦ m ≦ 3) The inspection apparatus is configured to determine the reliability of the pass / fail determination step result based on the determination ( claim 6 ).

The inspection apparatus of the present invention has the reliability determination means in addition to the pass / fail determination means. And in this reliability determination means, the reliability determination step in the inspection method described above can be executed, and the above-described operational effects can be obtained as they are.
Therefore, in the inspection apparatus of the present invention, if the result of executing the reliability determination means results in no reliability or low reliability, there is a failure or malfunction of the inspection apparatus, or an inspection. It can be determined that a re-inspection to confirm whether it is due to the characteristics of the object itself or an inspection of the device should be performed.
Therefore, if the inspection apparatus of the present invention is used, the above-described determination that cannot be made in the past can be added, so that the reliability of the determination result by the pass / fail determination means can be significantly improved as compared with the conventional case, and From the determination result of the reliability determination means, it is possible to recognize the possibility that the function of the sensor unit or the like has deteriorated or a defect has occurred, and it is possible to respond early when the occurrence occurs.

The reliability range used in the reliability determination step in the inspection method of the first invention is a range of μ _i ± mσ _i (1 ≦ m ≦ 3), that is, (μ _i −mσ _i ) to (μ _i + mσ _i ). If the measurement data Y _i falls within the range, it can be determined that the data has a certain degree of reliability. Note that the reliability range becomes narrower as the value of m becomes smaller, and the determination becomes stricter. In the present invention, m is in the range of 1 to 3 as described above, and is not necessarily a natural number. If m is less than 1, the criterion may be too strict and a reasonable decision may be difficult. On the other hand, if m is more than 3, the criterion may be too loose and a reasonable decision may be difficult. is there.

Further, by providing the reliability range setting in a plurality of stages of two or more, it is possible to classify the degree of reliability into a plurality of stages and perform a finer determination.
In the present invention, the reliability of the entire n measurement data Y _i group is determined using a plurality (N) of measurement data Y _i . N is most preferably the same as the total number n of the measurement data Y _i , and can be designed so as to be appropriate for the performance of the computer that is the hardware for performing the determination.

Various specific methods can be used as specific determination methods.
That is, in the reliability determination step, whether or not at least N of the measurement data Y _i collected this time exceeds the corresponding μ _i ± mσ _i (1 ≦ m ≦ 3) reliability range. with the determining determines the number a exceeds the above confidence range of the N, it is determined whether or not the value of a / N is greater than the reference value B (0 <B <1) , the acceptance determination step it is possible to determine the results of reliability (claim 2).

In the reliability determination step, it is determined whether or not N of the measurement data Y _i collected this time exceeds the first reliability range of μ _i ± aσ _i (1 ≦ a <3) corresponding to each of the N measurement data Y _i . Along with the determination, the number A1 exceeding the first reliability range among the N is determined, and it is determined whether or not the value of A1 / N exceeds the reference value B1 (0 <B1 <1),
In the case of A1 / N <B1, the pass / fail determination step result is determined to be the most reliable first level,
In the case of the above A1 / N ≧ B1, N of the measurement data Y _i collected this time exceeds the second confidence range of μ _i ± bσ _i (a <b ≦ 3) corresponding thereto. A number A2 that exceeds the second confidence range among the N is determined, and it is determined whether the value of A2 / N exceeds a reference value B2 (0 <B2 <1). ,
In the case of A2 / N <B2, it is determined that the pass / fail determination step result is a second level lower than the first level, and
In the case of A2 / N ≧ B2, it can be determined that the pass / fail determination step result is a third level lower than the second level (claim 3).

  In this case, since a two-stage determination is performed, a fine determination can be made. For example, when the A1 / N is less than B1, it can be determined that the measurement data group has a very high level of reliability. Even in the case of A1 / N ≧ B1, when A2 / N is less than B2, it can be determined that the measurement data group has a certain level of reliability. In the case of A1 / N ≧ B1, and further in the case of A2 / N ≧ B2, it can be determined that the reliability of the measurement data group is the lowest.

The standard deviation σ _i and the average value μ _i can be fixed values stored in advance.
That is, when there is a track record in which the measurement data Y _i of the same inspection target has been collected many times in the past, the standard deviation σ _i and the average value μ _i calculated in advance from those past records are stored, The fixed value can always be used for determination. In this case, in the case of the same inspection object, it is possible to make a sufficiently accurate determination when the normal actual value hardly changes over time.

In addition, the standard deviation σ _i and the average value μ _i can be re-calculated using the stored past measurement data when performing the reliability determination step. In this case, for example, when the normal actual value shifts with a certain degree of trend, for example, when the average value of the normal value itself tends to change in a better direction at all times. For example, it may be preferable to periodically review the standard deviation σ _i and the average value μ _i that serve as a reference.

More specifically, for example, the standard deviation σ _i and the average value μ _i are calculated again every time 100 times of normal new measurement data is accumulated, and then 100 times of new data is accumulated. In addition, there is a method of recalculating and updating the standard deviation σ _i and the average value μ _i from the data for 100 times. There is also a method of deleting the oldest measurement data and recalculating and updating the standard deviation σ _i and the average value μ _i each time new normal performance data is accumulated.

Further, the inspection target, a vibration or noise generated from the object to be inspected. When the object to be inspected is vibration or noise, it is particularly difficult to determine whether the actual data itself has been collected normally and is highly reliable, and a skilled inspector verifies the data. However, it is often difficult to judge. Therefore, it is very effective to incorporate the above reliability determination step using the statistically processed data of the standard deviation σ _i and the average value μ _i .

Further, the object to be inspected, along with an A / T (automatic transmission), the reference variable is the rotational speed of the output shaft of the A / T, the measurement data Y _i is the speed has predetermined Measure every time the value is reached .
In other words, it is important for A / T to suppress gear noise during operation, and the gear noise (vibration or noise at the time of gear meshing) for a specific rotational speed is measured, and the pass / fail is determined from the measured value. Is done. In this case as well, the reliability of the gear noise inspection itself can be improved by incorporating the above-described reliability determination step.

Next, the inspection apparatus of the second invention preferably includes a configuration corresponding to the above-described content in the inspection method of the first invention described above.
That is, the reliability determination means determines whether or not at least N of the measurement data Y _i collected this time exceeds the corresponding reliability range of μ _i ± mσ _i (1 ≦ m ≦ 3). In addition to determining, the number A exceeding the reliability range is obtained out of the N, and it is determined whether or not the value of A / N exceeds the reference value B (0 <B <1). It may be configured to determine the reliability of the result ( claim 7 ).

Further, the reliability determination means determines whether or not N of the measurement data Y _i collected this time exceeds a first reliability range of μ _i ± aσ _i (1 ≦ a <3) corresponding to each of the N pieces of measurement data Y _i . Along with the determination, the number A1 exceeding the first reliability range among the N is determined, and it is determined whether or not the value of A1 / N exceeds the reference value B1 (0 <B1 <1),
In the case of A1 / N <B1, the pass / fail determination step result is determined to be the most reliable first level,
In the case of the above A1 / N ≧ B1, N of the measurement data Y _i collected this time exceeds the second confidence range of μ _i ± bσ _i (a <b ≦ 3) corresponding thereto. A number A2 that exceeds the second confidence range among the N is determined, and it is determined whether the value of A2 / N exceeds a reference value B2 (0 <B2 <1). ,
In the case of A2 / N <B2, it is determined that the pass / fail determination step result is a second level lower than the first level, and
In the case of A2 / N ≧ B2, the pass / fail determination step result may be determined to be a third level lower than the second level ( claim 8 ).

The standard deviation σ _i and the average value μ _i can be fixed values stored in advance ( claim 9 ).
Further, the standard deviation σ _i and the average value μ _i can be re-calculated using the stored past measurement data when the reliability determination means is executed ( claims). 10 ).

  Of course, the inspection method and inspection apparatus according to the present invention include, for example, mass, density, position, time, angle, velocity, acceleration, angular velocity, length, thickness in addition to the A / T gear noise inspection method and inspection device. , Area, volume, flow rate, liquid level, weight, force, stress, pressure, torque, momentum, moment of inertia, energy, light intensity, temperature, calorie, charge, voltage, current, magnetic charge, magnetic field, magnetic flux, etc. The present invention can be applied to a method and an inspection apparatus, and in any inspection method and apparatus, it is possible to obtain an operational effect by including the above-described reliability determination step or reliability determination means.

Example 1
An inspection method and an inspection apparatus according to an embodiment of the present invention will be described with reference to FIGS.
The inspection apparatus 1 of this example is an A / T gear noise inspection apparatus.
As shown in FIG. 1, the inspection apparatus 1 rotates the A / T as the inspection object 80 by the rotation driving means 811 to 816 and collects vibration or noise (gear noise) as an inspection object generated from the inspection object 80. Then, the sensor unit 71 that outputs the sampling signal of the vibration waveform or the sound pressure waveform, and the above sampling signal received from the sensor unit 71 is subjected to arithmetic processing to obtain the measurement data Y _i in a desired form as the reference variable X _i . And a reliability determination for determining the reliability of the measurement data Y _i and a pass / fail determination means for executing a pass / fail determination step for determining whether or not the state of the inspection object is normal. This is an A / T gear noise inspection apparatus having a measurement computer (measurement PC) 73 provided with reliability determination means for executing steps.

  More specifically, the inspection apparatus 1 includes a mechanical system 8 that drives an inspected object 80 that is an A / T, a control device 70 that controls the mechanical system 8, the arithmetic unit 72 connected to the control device 70, and the arithmetic unit. 72 has a sensor unit 71 connected to 72 and a measurement PC 73 connected to the calculation unit 72. Further, a post-processing computer (post-processing PC) 74 and a data server 75 are connected to the measurement PC 73. Each computer is connected to a keyboard as an input device, a display device for display as an output device, and a printer (not shown).

  As shown in the figure, the mechanical system 8 includes a drive motor 811, absorption motors 813 and 815, and shaft bodies 812, 814, and 816 that connect these to the inspection object 80. A portion including these drive motors and shaft bodies 811 to 816 functions as a rotation drive device.

  The control device 70 connected to the mechanical system 8 is configured to control the drive motor 811 so as to change the rotational speed of the inspection object 80 in a predetermined pattern. The rotation number data is collected from the tachometer 818 and the tachometer 819 and can be transmitted to the calculation unit 72.

  In the present example, a microphone is used as the sensor unit 71, and gear noise can be collected. Note that a plurality of sensor units 71 can be connected, and the calculation unit 72 is configured to be able to perform calculations for each of a plurality of processing channels. In this example, only one channel will be described as an example. In addition, the sensor unit 71 is originally set at a location that is defined in advance according to the type of the inspection object 80.

Further, as shown in the figure, the calculation unit 72 is composed of an FFT analyzer that performs frequency analysis of the sampling signal received from the sensor unit 71.
The inspection apparatus 1 is configured to operate while changing the number of rotations of the inspection object 80 in a predetermined pattern according to an instruction from the control apparatus 70.

Then, as shown in FIG. 2, the measurement data Y _i calculated in the FFT analyzer of the calculation unit 72 is transmitted to the storage device in the measurement PC 72 and temporarily stored in the data processing file 731. The data processing file 731 is used when executing the determination means. After the determination, the measurement data Y _i is transferred from the data storage file 732 in the measurement PC 73 to the data storage file 741 of the post-processing PC 74 and further transferred to the data backup file 742 so as to be stored for a long time. It is configured.

  In addition, the inspection apparatus 1 of this example is configured to set measurement conditions according to the type of the inspection object 80 in the measurement PC 73. Specifically, the A / T changes the gear ratio, the order, and the parameters according to the type and performs the above pass / fail judgment step, so the conditions are set in the measurement PC 73 before the measurement is started.

  Specifically, as shown in FIG. 3, after the measurement start switch on the measurement PC 73 is turned on by the inspector in step S101, the measurement PC 73 receives a number indicating the measurement method in step S102. As shown in Table 1, this measurement method includes measurement conditions such as the number, gear ratio, and order, and parameters. At least a plurality of types are registered in advance and stored in a storage device in the measurement PC 73. It is remembered.

In step S103, it is determined whether or not the input number is registered as shown in Table 1. If it is registered, a measurement start state is entered in step S110. In this state, the actual processing starts when the measurement data Y _i is input from the calculation unit 72.

On the other hand, if it is determined in step S103 that the input number is not registered as shown in Table 1, the parameter input / setting is accepted in step S104. After the parameter input / setting is completed in step S105, the measurement condition input / setting is accepted in step S106. Then, after the input and setting of the measurement conditions are completed in step S107, the current measurement method is registered in the storage device in step S108, and then the measurement start state is entered in step S109. In this state, the actual processing starts when the measurement data Y _i is input from the calculation unit 72.

Further, as described above, the inspection apparatus 1 executes a pass / fail determination unit that executes a pass / fail determination step that determines whether or not the state of the inspection target is normal and a reliability determination step that determines the reliability of the measurement data Y _i. Reliability determination means is provided in the measurement PC 73.
The measurement data Y _i is handled as (X _i , Y _i ) corresponding to each reference variable X _i (rotational speed). Therefore, when one inspection is performed, n pieces of data such as (X ₁ , Y ₁ ), (X ₂ , Y ₂ )... (X _i , Y _i ) (X _n , Y _n ) Is obtained.

And the said pass / fail judgment means, for example, sets a threshold value of the value of Y _i corresponding to each of the above X _i in advance, and a method of failing when the number exceeding that exceeds a predetermined number Various can be adopted. Since this acceptance / rejection judging means judges acceptance / rejection by the same method as in the prior art, detailed description in this example is omitted.

On the other hand, the reliability determination means is an unprecedented means, using the standard deviation σ _i and the average value μ _i obtained for each of the same reference variables X _{i based} on the measurement data Y _i collected in the past, and the total collected this time Based on the determination of whether or not N (2 ≦ N ≦ n) of the n pieces of measurement data Y _i exceeds the corresponding confidence range of μ _i ± mσ _i (1 ≦ m ≦ 3). It is configured to determine the presence or absence or degree of reliability. As the standard deviation σ _i and the average value μ _i in this example, a value calculated from a past record is stored as a fixed value.

As shown in FIG. 4, the reliability determination step performed by the reliability determination means first corresponds to each of all N (n in this example) of the measurement data Y _i collected this time in sub-step S201. In addition to determining whether or not the first confidence range of μ _i ± 2σ _i to be exceeded, the number A1 exceeding the first confidence range among the n is obtained, and the value α of A1 / N is calculated. Next, in sub-step S202, it is determined whether or not the value α of A1 / N exceeds the reference value B1 (B1 = 0.3).

If A1 / N <B1, it is further determined in sub-step S203 whether or not the value α of A1 / N exceeds 0.1.
If α ≦ 0.1, it is determined in sub-step S204 that the current measurement data Y _i is “having 90% or more reliability”, and the content is displayed on the display device of the measurement PC 73.
If α> 0.1, in sub-step S205, it is determined that the current measurement data Y _i has “70 to 90% reliability”, and the content is displayed on the display device of the measurement PC 73.

On the other hand, if the determination in sub-step S202 is A1 / N ≧ B1, the second confidence of μ _i ± 3σ _i corresponding to each of all the N pieces of the measurement data Y _i collected this time in sub-step S206. A determination is made as to whether or not the range is exceeded, and the number A2 that exceeds the second confidence range among the N is determined, and the value β of A2 / N is calculated. Subsequently, in sub-step S207, it is determined whether or not the value β of A2 / N exceeds the reference value B2 (B2 = 0.5).

When β ≦ 0.5, in sub-step S208, it is determined that the current measurement data Y _i is “50 to 70% reliable”, and the content is displayed on the display device of the measurement PC 73.
If β> 0.5, it is determined in sub-step S209 that the current measurement data Y _i is “reliability of 50% or less and re-examination is appropriate”, and the contents are stored in the measurement PC 73. Display on the display device.

As described above, in the inspection method using the inspection apparatus 1 of the present example, the reliability determination step is performed separately from the original pass / fail determination step. Then, by performing this reliability determination step, as described above, the current measurement data Y _i is “having a reliability of 90% or more”, “having a reliability of 70 to 90%”, “50 to It is possible to make a four-step determination of “having 70% reliability” or “having only 50% or less reliability and appropriate re-examination”.
Therefore, it can be determined from the result of the reliability determination step whether or not the original result of the pass / fail determination step is reliable.

  That is, regardless of the result of the pass / fail determination step, the result of the reliability determination step has only the lowest level of reliability that “there is only 50% or less reliability and re-examination is appropriate”. In the case of the determination result, it can be determined that the result of the pass / fail determination step cannot be used as it is.

On the other hand, when the highest level of reliability is obtained that the result of the reliability determination step is “having 90% or more reliability”, the result of the pass / fail determination step is very reliable and can be used as it is. It can be judged.
In addition, when the second highest level of reliability that the result of the reliability determination step has “70 to 90% reliability” is obtained, the result of the pass / fail determination step can be trusted. It can be judged that it can be used.
In addition, when the second lowest level of reliability that the result of the reliability determination step has “50 to 70% reliability” is obtained, the result of the pass / fail determination step is changed to the above two results. Although it is less reliable, it can be determined that it can be used as it is.

  Thus, in this example, since the determination which cannot be performed conventionally can be added, the reliability of the determination by the pass / fail determination step can be significantly improved as compared with the conventional case.

Here, FIG. 5 shows an example of measurement data measured in this example. In the figure, A / T rotation speed (rpm) is taken on the horizontal axis, and gear noise (dB) is taken on the vertical axis, and past measurement data is plotted. The average value μ _i and the value of standard deviation σ _i × 3 (3σ _i ) for all the data are plotted together. Although −3σ _i is not plotted, the above-described second confidence range is within a region sandwiched between 3σ _i and -3σ _{i (} not shown).

FIG. 3 is an explanatory diagram illustrating a configuration of an inspection apparatus according to the first embodiment. FIG. 3 is an explanatory diagram illustrating a file configuration for storing measurement data in the first embodiment. FIG. 3 is an explanatory diagram illustrating steps up to the start of measurement with a measurement PC in the first embodiment. FIG. 3 is an explanatory diagram illustrating a flow of a reliability determination step in the first embodiment. FIG. 3 is an explanatory diagram showing an example of measured measurement data in Example 1.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Inspection apparatus 70 Control apparatus 71 Sensor part 72 Calculation part (FFT analyzer)
73 PC for measurement
8 Mechanical system 80 Inspected object (A / T)

Claims (10)

Inspection having a pass / fail determination step of determining whether or not the state of the inspection object is normal using measurement data Y _i (i = 1 to n, n is a natural number) of the inspection object sequentially collected corresponding to changes in the reference variable In the method
The inspection object is vibration or noise generated from the inspection object,
The inspection object is an A / T (automatic transmission), the reference variable is the rotation speed of the output shaft of the A / T, and the measurement data Y _i is set to a value determined in advance. Measure every time
In addition to the above pass / fail determination step, the computer has a reliability determination step for determining the reliability of the pass / fail determination step result by a computer.
The reliability determination step includes:
Using the standard deviation σ _i and the average value μ _i obtained for each of the same reference variables for the measurement data Y _i collected in the past,
Whether N (2 ≦ N ≦ n) of the total n measurement data Y _i collected this time exceeds the corresponding confidence range of μ _i ± mσ _i (1 ≦ m ≦ 3) An inspection method characterized by determining the reliability of the pass / fail determination step result based on the determination. In claim 1, in the reliability determination step, at least whether or not N of the measurement data Y _i collected this time exceeds a corresponding reliability range of μ _i ± mσ _i (1 ≦ m ≦ 3). In addition to determining whether or not the number A exceeds the reliability range, the A / N value exceeds the reference value B (0 <B <1). An inspection method characterized by determining the reliability of the pass / fail determination step result. In claim 1, in the reliability determination step, whether or not N of the measurement data Y _i collected this time exceeds a first reliability range of μ _i ± aσ _i (1 ≦ a <3) corresponding thereto. And determining the number A1 that exceeds the first confidence range among the N, and determining whether the value of A1 / N exceeds the reference value B1 (0 <B1 <1),
In the case of A1 / N <B1, the pass / fail determination step result is determined to be the most reliable first level,
In the case of the above A1 / N ≧ B1, N of the measurement data Y _i collected this time exceeds the second confidence range of μ _i ± bσ _i (a <b ≦ 3) corresponding thereto. A number A2 that exceeds the second confidence range among the N is determined, and it is determined whether the value of A2 / N exceeds a reference value B2 (0 <B2 <1). ,
In the case of A2 / N <B2, it is determined that the pass / fail determination step result is a second level lower than the first level, and
When A2 / N ≧ B2, it is determined that the pass / fail determination step result is a third level lower than the second level. 4. The inspection method according to claim 1, wherein a fixed value stored in advance is used as the standard deviation σ _i and the average value μ _i . 5. The standard deviation σ _i and the average value μ _i in any one of claims 1 to 4 are newly calculated using the stored past measurement data when performing the reliability determination step. A characteristic inspection method.   Measurement data Y to be inspected sequentially corresponding to changes in the reference variable _i In the inspection apparatus having a pass / fail determination means for performing a pass / fail determination step of determining whether or not the state of the inspection target is normal using (i = 1 to n, n is a natural number),
  The inspection object is vibration or noise generated from the inspection object,
  The inspection object is an A / T (automatic transmission), and the reference variable is the rotational speed of the output shaft of the A / T, and the measurement data Y _i Is measured every time the rotational speed reaches a predetermined value,
  In addition to the above pass / fail determination means, the computer has reliability determination means for performing a reliability determination step for determining the reliability of the pass / fail determination step result by a computer.
  The reliability determination means includes
  Measurement data Y collected in the past _i Standard deviation σ obtained for each same reference variable by _i And mean value μ _i And
  A total of n measurement data Y collected this time _i Μ of N (2 ≦ N ≦ n) _i ± mσ _i An inspection apparatus configured to determine the reliability of a pass / fail determination step result on the basis of determination of whether or not a reliability range of (1 ≦ m ≦ 3) is exceeded.   7. The reliability determination means according to claim 6, wherein at least the measurement data Y collected this time _i For N of each _i ± mσ _i It is determined whether or not the reliability range exceeds (1 ≦ m ≦ 3), and the number A exceeding the reliability range among the N is obtained, and the value of A / N is the reference value B (0 <B < 1) The inspection apparatus is configured to determine whether or not it exceeds 1) and to determine the reliability of the pass / fail determination step result.   In Claim 6, the said reliability determination means is said measurement data Y collected this time. _i Of N of the _i ± aσ _i It is determined whether or not the first reliability range of (1 ≦ a <3) is exceeded, and the number A1 that exceeds the first reliability range among the N is obtained, and the value of A1 / N is the reference value B1 ( Determine whether 0 <B1 <1)
  In the case of A1 / N <B1, the pass / fail determination step result is determined to be the most reliable first level,
  In the case of A1 / N ≧ B1, the measurement data Y collected this time _i Of N of the _i ± bσ _i It is determined whether or not the second confidence range of (a <b ≦ 3) is exceeded, and the number A2 that exceeds the second confidence range among the N is obtained, and the value of A2 / N is the reference value B2 ( Whether 0 <B2 <1) is exceeded,
  In the case of A2 / N <B2, it is determined that the pass / fail determination step result is a second level lower than the first level, and
  An inspection apparatus configured to determine that the pass / fail determination step result is a third level lower than the second level when A2 / N ≧ B2.   The standard deviation σ according to any one of claims 6 to 8. _i And the average value μ _i Uses a fixed value stored in advance.   The standard deviation σ according to any one of claims 6 to 9. _i And the average value μ _i Is configured to re-calculate using the stored past measurement data when executing the reliability determination means.

Images