JP6535208B2 - Structure identification device for vibration analysis model and identification method thereof - Google Patents

Description

  The present invention relates to a structural identification device of a vibration analysis model that performs structural identification of an analysis model of a vibration system having a plurality of vibration elements, and a method of identifying the same.

  In the model identification device of Patent Document 1 below, mathematical model identification is applied to a rolling load prediction model in a steel rolling process. In the identification of the mathematical expression model, among the classification patterns for classifying the actual data into two groups, the classification pattern in which the difference in statistical information of the prediction error of the mathematical expression model between the two groups is large is determined. The process of identifying a mathematical expression model for each group is repeated while further dividing the group sequentially until the model prediction accuracy satisfies the setting condition, and the parameter when the setting condition is satisfied is set as the parameter of the mathematical expression model in that group . In the determination of the classification pattern, first, the data item to be used for the group classification is selected based on the average value or the standard deviation, and the classification pattern to be adopted from them is determined based on the information amount standard.

  Moreover, in the vibration analysis device of the following patent document 2, multi-dimensional vector data which is a function of time and frequency is generated by acquiring vibration data at a plurality of locations of a vibration system and performing wavelet transform on each vibration data. By performing factor analysis (PARAFAC analysis) using vector data, multiple independent vectors are generated. The plurality of independent vectors have vibrational frequency characteristics, temporal characteristics and spatial characteristics.

JP 2005-157788 A JP, 2010-48684, A Unexamined-Japanese-Patent No. 2004-280450

  In Patent Document 1, the identified mathematical expression model has no physical meaning because it is an identification method that statistically regresses the mathematical expression model. Further, in Patent Document 2, a plurality of independent vectors having frequency characteristics of vibration, temporal characteristics and spatial characteristics are generated by factor analysis, but structural identification of the vibration analysis model is not shown. Also, no consideration is given to the reduction of the vibration analysis model.

  An object of the present invention is to perform structural identification of an analytical model of a vibration system which has physical meaning and is reduced in dimension.

  The structure identification device for a vibration analysis model and the identification method according to the present invention employ the following means in order to achieve the above-mentioned object.

  The structural identification device for a vibration analysis model according to the present invention is a vibration analysis model identification device for structural identification of an analysis model of a vibration system having a plurality of vibration elements, and the vibration element and vibration to be vibration measurement targets in the vibration system. A time-series vibration data acquisition unit that acquires time-series vibration data under a plurality of measurement conditions in which one or more of the measurement directions are different from one another, and each time-series vibration data acquired by the time-series vibration data acquisition unit The factor of each measurement condition corresponding to the factor by performing factor analysis on each piece of vibration data converted by the vibration data converter that converts vibration data with respect to frequency and frequency, and each vibration data converted by the vibration data converter In the vibration data under a plurality of measurement conditions converted by the factor analysis unit that calculates the vibration data and the vibration data conversion unit, the vibration data to be evaluated for vibration and the other vibration data A weight coefficient calculation unit that calculates a weight coefficient corresponding to each measurement condition other than the vibration evaluation target based on the relationship between the contribution degree calculated by the factor analysis unit, and a plurality of measurements converted by the vibration data conversion unit In the vibration data under the conditions, the sum of the vibration data weighted by the weighting factor for all measurement conditions excluding the vibration evaluation target and the vibration data weighted with the weighting factor for some measurement conditions excluding the vibration evaluation target are added The present invention is characterized in that the measurement condition is limited based on an index representing an error with the total sum, and an analysis model identification unit that performs structure identification of the analysis model based on the limited measurement condition is provided.

  In one aspect of the present invention, when the analysis model identification unit calculates the index, it is preferable that the partial measurement condition be selected based on the degree of contribution calculated by the factor analysis unit.

  In one aspect of the present invention, the factor analysis unit performs factor analysis on each vibration data converted by the vibration data conversion unit with a plurality of factor numbers to calculate vibration data for time and frequency for each factor. The factor is selected based on the calculated vibration data for each factor, and the analysis model identification unit calculates the index based on the degree of contribution corresponding to the factor selected by the factor analysis unit. It is preferable to select some of the measurement conditions.

  In one aspect of the present invention, it is preferable that the factor analysis unit select a factor based on the degree of fluctuation of the amplitude in the vibration data for each factor.

  The structure identification method of a vibration analysis model according to the present invention is a vibration analysis model identification method for performing structure identification of an analysis model of a vibration system having a plurality of vibration elements, the vibration element to be a vibration measurement target in the vibration system. Time series vibration data acquisition processing for acquiring time series vibration data under a plurality of measurement conditions in which one or more of the vibration measurement directions are different from each other, and each time series vibration data acquired in the time series vibration data acquisition processing Vibration data conversion processing for converting to vibration data with respect to time and frequency, and performing factor analysis on each vibration data converted by the vibration data conversion processing with a predetermined number of factors, for each measurement condition corresponding to the factor In the factor analysis processing for calculating the degree of contribution, and in the vibration data under a plurality of measurement conditions converted by the vibration data conversion processing, vibration data to be evaluated for vibration, and Based on the relationship between vibration data other than vibration data and contribution degree calculated by factor analysis processing, weight coefficient calculation processing that calculates the weight coefficient corresponding to each measurement condition other than vibration evaluation target, and conversion by vibration data conversion processing In the vibration data under the plurality of measurement conditions, the sum of the vibration data weighted by the weighting factor for all the measurement conditions except for the vibration evaluation target and the weight coefficient for some measurement conditions except the vibration evaluation target Abstract: A measurement condition is limited based on an index representing an error with a sum of added vibration data, and analysis model identification processing for identifying a structure of an analysis model based on the limited measurement condition is included. .

  According to the present invention, the sum obtained by adding the vibration data weighted by the weighting factor for all measurement conditions excluding the vibration evaluation target and the vibration data weighted by the weighting factor for some measurement conditions except the vibration evaluation target are added The structural identification of an analytical model of a vibration system having a physical meaning and reduced in dimension can be performed by performing structural identification of the analytical model under limited measurement conditions based on an index representing an error with the sum. it can.

BRIEF DESCRIPTION OF THE DRAWINGS It is a figure which shows schematic structure of the vehicle which becomes application object of the structure identification apparatus of the vibration analysis model which concerns on embodiment of this invention. It is a figure which shows schematic structure of the powertrain block of a vehicle. It is a figure which shows schematic structure of the powertrain block of a vehicle. It is a figure showing a schematic structure of a structure identification device of a vibration analysis model concerning an embodiment of the present invention. It is a figure which shows an example of three-dimensional array X_ obtained by wavelet transformation. It is a figure explaining the process to decompose | disassemble 3-dimensional array X_ into matrix A, B, C by PARAFAC analysis. It is a figure which shows an example of the basis wavelet of the factor obtained by PARAFAC analysis. Is a diagram illustrating an example of a contribution c _1f to c _Kf of each measurement conditions corresponding to the selected factor f (k = 1~K). It is a figure which shows an example of the result of having calculated the relationship of the parameter | index E _r with respect to ranking r. It is a figure which shows an example of a vibration analysis model.

  Hereinafter, modes for carrying out the present invention (hereinafter referred to as embodiments) will be described according to the drawings.

  1 to 3 are diagrams showing a schematic configuration of a vehicle 10 to which a structural identification apparatus for a vibration analysis model according to an embodiment of the present invention is applied. FIG. 4 is a structure of a vibration analysis model according to an embodiment of the present invention It is a figure which shows schematic structure of an identification apparatus. FIG. 1 shows a schematic configuration of the entire vehicle 10 in the present embodiment, and FIGS. 2 and 3 show a schematic configuration of a power train block 12. A vehicle 10 as a vibration system includes a vehicle body (sprung element) 11, a power train block 12, and an unsprung element 15 as a plurality of vibration elements. The powertrain block 12 is configured to include the engine 13 and the transmission 14, and the unsprung element 15 is configured to include the drive wheel 18 and the lower arm 19. The engine 13 is coupled to a transmission 14, and the transmission 14 is coupled to a drive wheel 18. During actual operation (actual traveling) of the vehicle 10, the engine 13 generates torque (rotational force) as an excitation force, torque is transmitted between the engine 13 and the transmission 14, and torque between the transmission 14 and the driving wheel 18 Is transmitted. The power train block 12 is supported by the vehicle body 11 via the mount 16, and power is transmitted between the power train block 12 and the vehicle body 11 via the mount 16. The unsprung element 15 is supported by the vehicle body 11 via the suspension bush 17, and the force is transmitted between the unsprung element 15 and the vehicle body 11 via the suspension bush 17.

  Here, an xyz three-dimensional coordinate system orthogonal to each other is defined in the vehicle 10 with the vehicle longitudinal direction (vehicle traveling direction) as the x axis, the vehicle lateral direction as the y axis, and the vehicle vertical direction as the z axis. The power train block 12 is capable of rotational vibration (roll vibration) around the x-direction center of gravity axis, rotational vibration (pitch vibration) around the y-direction center of gravity axis, and rotational vibration (yaw vibration) around the z-direction center of gravity axis It is possible. Furthermore, the powertrain block 12 is capable of translational vibration in the x direction, translational vibration in the y direction, and translational vibration in the z direction. Similarly, the unsprung element 15 is also capable of rotational vibration (roll vibration) around the x-direction center of gravity axis, rotational vibration (pitch vibration) around the y-direction center of gravity axis, and rotational vibration around the z-direction center of gravity axis Yaw vibration) is possible. Furthermore, the unsprung element 15 is also capable of translational vibration in the x direction, translational vibration in the y direction, and translational vibration in the z direction. The vehicle body 11 can translate and vibrate in the x direction. Here, the powertrain block 12 and the unsprung element 15 are considered to be mass points with rigid motion with six degrees of freedom. In the following embodiment, an example will be described in which the structure identification of the vibration analysis model of the vehicle 10 is performed in which the x-direction translational vibration of the floor of the vehicle body 11 is evaluated with respect to the torque vibration of the engine 13 which is a vibration source.

As shown in FIGS. 2 and 3, in order to measure the translational vibration and rotational vibration of the powertrain block 12 which is the vibration measurement target during actual traveling of the vehicle 10, each has x-direction translational acceleration and y-direction translational acceleration and z. A plurality of (three) three-axis acceleration sensors 22-1, 22-2, 22-3 for detecting the direction translational acceleration are attached to the power train block 12. In the example of FIGS. 2 and 3, the three-axis acceleration sensor 22-1 and the three-axis acceleration sensor 22-2 are installed in the power train block 12 with a distance l _p1 + l _p2 in the y direction. The 22-1 and 3-axis acceleration sensors 22-3 are installed in the power train block 12 with a distance l _p1 + l _p3 in the y direction. The x-direction position and the z-direction position of the 3-axis acceleration sensor 22-1 and the 3-axis acceleration sensor 22-2 are equal to each other, and the 3-axis acceleration sensors 22-1 and 22-2 and the 3-axis acceleration sensor 22-3 are x The power train block 12 is installed at a distance d _p1 + d _p3 in the direction and in a direction h _p1 + h _p3 in the z direction.

In the power train block 12, the rotational angular acceleration d ² θ _px / dt ² about the x-direction center of gravity axis is expressed by the following equation (1), and the rotational angular acceleration d ² θ _py / dt ² about the y-direction central axis of gravity The rotational angular acceleration d ² θ _pz / dt ² about the z-direction center of gravity axis is expressed by the following expression (2). The x-direction translational acceleration d ² x _p / dt ² at the center of gravity is expressed by the following equation (4), and the y-direction translational acceleration d ² y _p / dt ² at the center of gravity is the following equation (5) The z-direction translational acceleration d ² x _p / dt ² at the center of gravity is expressed by the following equation (6). In equations (1) to (6), d ² z _{p 1} / dt ² is the z-direction translational acceleration at the location where the 3-axis acceleration sensor 22-1 is attached, and d ² z _{p 3} / dt ² is the 3-axis acceleration sensor 22-3 Z-direction translational acceleration at the installation site, d ² x _{p 1} / dt ² is the x-axis translational acceleration at the installation site of the 3-axis acceleration sensor 22-1, d ² x _{p 2} / dt ² is the 3-axis acceleration sensor 22-2 X-direction translational acceleration at the attachment point of the sensor, d ² x _{p 3} / dt ² is the x-direction translational acceleration at the attachment point of the 3-axis acceleration sensor 22-3, and d ² y _{p 1} / dt ² is the 3-axis acceleration sensor 22-1 Y-direction translational acceleration at the attachment point of Further, h _p1 is the z-direction distance between the center of gravity of the power train block 12 and the attachment point of the three-axis acceleration sensor 22-1, 22-2 and d _p1 is the center of gravity of the power train block 12 and the three-axis acceleration sensor 22- The x-direction distance from the attachment point of 1, 22-2 and l _p1 is the y-direction distance between the center of gravity of the power train block 12 and the attachment point of the three-axis acceleration sensor 22-1.

d ² θ _px / dt ² = (d ² z _{p 3} / dt ² −d ² z _{p 1} / dt ² ) / (l _p 1 + l _{p 3} ) (1)
d ² θ _py / dt ² = (d ² x _{p 1} / dt ² −d ² x _{p 3} / dt ² ) / (h _{p 1} + h _{p 3} ) (2)
d ² θ _pz / dt ² = (d ² x _{p 1} / dt ² −d ² x _{p 2} / dt ² ) / (l _p 1 + l _{p 2} ) (3)
d ² x _p / dt ² = d ² x _{p 1} / dt ²
−h _p1 × d ² θ _py / dt ² −l _{p 1} × d ² θ _pz / dt ² (4)
d ² y _p / dt ² = d ² y _p1 / dt ²
−d _{p 1} × d ² θ _pz / dt ² + h _{p 1} × d ² θ _px / dt ² (5)
d ² z _p / dt ² = d ² z _{p 1} / dt ²
+ D _{p 1} x d ² θ _py / dt ² + l _p 1 x d ² θ _px / dt ² (6)

Similarly, in order to measure the translational vibration and rotational vibration of the unsprung element 15 (lower arm 19), which is the vibration measurement target during actual traveling of the vehicle 10, each has x-direction translational acceleration, y-direction translational acceleration and z-direction translation A plurality of (three) three-axis acceleration sensors 25-1, 25-2, 25-3 for detecting acceleration are attached to the unsprung element 15. The triaxial acceleration sensor 25-1 and the triaxial acceleration sensor 25-2 are installed on the unsprung element 15 with the distance l _s1 + l _s2 mutually set in the y direction, and the triaxial acceleration sensor 25-1 and the triaxial acceleration sensor 25-3 are installed on the unsprung element 15 with a distance l _s1 + l _s3 to each other in the y direction. The x-direction position and z-direction position of the 3-axis acceleration sensor 25-1 and the 3-axis acceleration sensor 25-2 are equal to each other, and the 3-axis acceleration sensor 25-1, 25-2 and the 3-axis acceleration sensor 25-3 are x It is installed in the unsprung element 15 in a state in which the distance d _s1 + d _s3 is mutually spaced in the direction and the distance h _s1 + h _s3 is mutually spaced in the z direction.

In the unsprung element 15 (lower arm 19), the rotational angular acceleration d ² θ _sx / dt ² around the center of gravity in the x direction is expressed by the following equation (7), and the rotational angular acceleration d ² θ _sy around the center of gravity in the y direction / Dt ² is expressed by the following equation (8), and the rotational angular acceleration d ² θ _sz / dt ² about the z-axis center of gravity is expressed by the following equation (9). The x-direction translational acceleration d ² x _s / dt ² at the center of gravity is expressed by the following equation (10), and the y-direction translational acceleration d ² y _s / dt ² at the center of gravity is the following equation (11) The z-direction translational acceleration d ² x _s / dt ² at the center of gravity is expressed by the following equation (12). (7) - (12) In the equation, d ² z _s1 / dt ² is the z-direction translational acceleration in attached position of three-axis acceleration sensor _{^{25-1, d 2 z s3 / dt}} 2 is the three-axis acceleration sensor 25-3 z-direction translational acceleration in attached position of, d ² x _s1 / dt ² is the x-direction translational acceleration in attached position of three-axis acceleration sensor _{^{25-1, d 2 x s2 / dt}} 2 is the three-axis acceleration sensor 25-2 x-direction translational acceleration, d ² x _s3 / dt ² is the x-direction translational acceleration in attached position of three-axis acceleration sensor _{^{25-3, d 2 y s1 / dt}} 2 is 3-axis acceleration sensor 25-1 at the attached portion Y-direction translational acceleration at the attachment point of And h _s1 is the z-direction distance between the center of gravity of the unsprung element 15 and the attachment point of the three-axis acceleration sensor 25-1, 25-2, d _s1 is the center of gravity of the unsprung element 15 and the three-axis acceleration sensor 25 The x-direction distance from the attachment site of 1, 25-2 and l _s1 is the y-direction distance between the center of gravity of the unsprung element 15 and the attachment site of the three-axis acceleration sensor 25-1.

_{^{d 2 θ sx / dt 2 =}} (d 2 z s3 / dt 2 -d 2 z s1 / dt 2) / (l s1 + l s3) (7)
_{^{d 2 θ sy / dt 2 =}} (d 2 x s1 / dt 2 -d 2 x s3 / dt 2) / (h s1 + h s3) (8)
_{^{d 2 θ sz / dt 2 =}} (d 2 x s1 / dt 2 -d 2 x s2 / dt 2) / (l s1 + l s2) (9)
_{^{d 2 x s / dt 2 =}} d 2 x s1 / dt 2
_{^{-H s1 × d 2 θ sy /}} dt 2 -l s1 × d 2 θ sz / dt 2 (10)
_{^{d 2 y s / dt 2 =}} d 2 y s1 / dt 2
−d _s1 × d ² θ _sz / dt ² + h _s1 × d ² θ _sx / dt ² (11)
_{^{d 2 z s / dt 2 =}} d 2 z s1 / dt 2
_{^{+ D s1 × d 2 θ sy}} / dt 2 + l s1 × d 2 θ sx / dt 2 (12)

An angular velocity sensor 28 for detecting an angular velocity dθ _w / dt of the drive wheel 18 is attached to the drive wheel 18 in order to measure rotational vibration of the drive wheel 18 which is an object of vibration measurement during actual traveling of the vehicle 10. Further, in order to measure the translational vibration of the floor of the vehicle body 11 which is an object of vibration evaluation during actual traveling of the vehicle 10, the acceleration sensor 21 detects x-direction translational acceleration d ² x _v / dt ² of the floor of the vehicle body 11. It is attached to the floor of the vehicle body 11. The time series signals detected by the sensors 21, 22-1, 22-2, 22-3, 25-1, 25-2, 25-3, 28 when the vehicle 10 is actually traveling are input to the vibration analysis device 70. Be done. The vibration analysis device 70 performs structure identification of an analysis model for analyzing the vibration of the vehicle 10.

  An example of a functional block diagram of the vibration analysis device 70 is shown in FIG. The vibration analysis device 70 can be configured as a computer with a CPU as a center, and the computer will be described below. The time-series vibration data acquisition unit 71, the setting constant storage device 72, the vibration data conversion unit 73, the factor analysis unit 74, the weights It functions as a coefficient calculation unit 75 and an analysis model identification unit 76.

In vibration analysis device 70, the set constant storage device 72, the distance of the aforementioned _{l p1 + l p2, l p1} + l p3, h p1 + h p3, l p1, h p1, d p1, l s1 + l s2, l s1 + l s3 , H _s1 + h _s3 , l _s1 , h _s1 , d _s1 are stored. The time-series vibration data acquisition unit 71 receives time-series signals detected by the sensors 21, 22-2, 22-2, 22-3, 25-1, 25-2, 25-3, 28. . The time-series vibration data acquisition unit 71 acquires time-series vibration data under a plurality of measurement conditions in which one or more of the vibration element to be subjected to vibration measurement in the vehicle 10 (vibration system) and the vibration measurement direction are different. Here, the x-direction translational acceleration d ² x _v / dt ² of the floor of the vehicle body 11, the rotational angular acceleration d ² θ _w / dt ^{2 of} the drive wheel 18, and power as time-series vibration data under a plurality of measurement conditions Translational acceleration of the train block 12 d ² x _p / dt ² , d ² y _p / dt ² , d ² z _p / dt ² and rotational angular acceleration d ² θ _px / dt ² , d ² θ _py / dt ² , d and ^{2 θ} _pz / dt ^2, translational acceleration ^{_{d 2 x s / dt 2,}} d 2 y s / dt 2, d 2 z s / dt 2 and the rotational angular acceleration d ² theta unsprung element 15 (lower arm 19) _sx / Dt ² , d ² θ _sy / dt ² , d ² θ _sz / dt ² are acquired, and time-series vibrational accelerations under 14 measurement conditions are acquired.

The x-direction translational acceleration d ² x _v / dt ² of the floor of the vehicle body 11 is obtained from the detection signal of the acceleration sensor 21. The rotational angular acceleration d ² θ _w / dt ² of the drive wheel 18 is obtained by differentiating the angular velocity dθ _w / dt detected by the angular velocity sensor 28. In the power train block 12, the rotational angular acceleration d ² θ _px / dt ² about the x-direction center of gravity axis is the difference d ² z _{p 3} / of the z-direction translational acceleration detected by the three-axis acceleration sensor 22-3, 22-1. Based on dt ² −d ² z _{p 1} / dt ² , the rotational angular acceleration d ² θ _py / dt ² about the y-direction center of gravity axis is calculated using the three-axis acceleration sensors 22-1 and 22-3. Based on the difference d ² x _{p 1} / dt ² −d ² x _{p 3} / dt ² of the x-direction translational acceleration detected by the equation (2), and the rotational angular acceleration d ² θ _pz / dt ² is calculated by the equation (3) based on the difference d ² x _{p 1} / dt ² -d ² x _{p 2} / dt ² of the x-direction translational acceleration detected by the three-axis acceleration sensors 22-1 and 22-2 Ru. The x-direction translational acceleration d ² x _p / dt ² at the center of gravity is x detected by the rotational angular acceleration d ² θ _py / dt ² , d ² θ _pz / dt ² and the 3-axis acceleration sensor 22-1 Based on the direction translational acceleration d ² x _{p 1} / dt ² , the y-direction translational acceleration d ² y _p / dt ² at the center of gravity is calculated by the equation (4), and the rotational angular acceleration d ² θ _pz / dt ² , Calculated by equation (5) based on d ² θ _px / dt ² and y-direction translational acceleration d ² y _{p 1} / dt ² detected by the 3-axis acceleration sensor 22-1, z-direction translational acceleration at the center of gravity d ² x _p / dt ² is the rotational angular acceleration d ² θ _py / dt ² , d ² θ _px / dt ² and z-axis translational acceleration detected by the 3-axis acceleration sensor 22-1 d ² z _{p 1} / dt ² And is calculated by equation (6).

Similarly, in the unsprung element 15 (lower arm 19), the rotational angular acceleration d ² θ _sx / dt ² about the x-direction center of gravity axis is the z-direction translational acceleration detected by the three-axis acceleration sensor 25-3, 25-1. It is based of the difference _{^{d 2 z s3 / dt 2 -d}} 2 z s1 / dt 2 is calculated by equation (7), the rotational angular acceleration d ² θ _sy / dt ² about the y-direction center of gravity axis, three-axis acceleration sensor based on the difference _{^{d 2 x s1 / dt 2 -d}} 2 x s3 / dt 2 of the detected x-direction translational acceleration in 25-1,25-3 calculated by equation (8), the rotation around the z-direction central axis angular acceleration d ² θ _sz / dt ^2, based on 3 different x-direction translational acceleration detected by the axis acceleration sensor _{^{25-1,25-2 d 2 x s1 / dt 2}} -d 2 x s2 / dt 2 It is calculated by the equation (9). The x-direction translational acceleration d ² x _s / dt ² at the center of gravity is x detected by the rotational angular acceleration d ² θ _sy / dt ² , d ² θ _sz / dt ² and the 3-axis acceleration sensor 25-1 is calculated by on the basis of the direction translational acceleration _{^{d 2 x s1 / dt 2 (}} 10) equation, y-direction translational acceleration d ² y _s / dt ² in the center of gravity, the angular acceleration d ² θ _sz / dt ^2, based on the d ² θ _sx / dt ² and 3-axis acceleration sensor 25 - 1 y-direction translational acceleration detected by d ² y _s1 / dt ² is calculated by the equation (11), z-direction translational acceleration of the center of gravity point d ² x _s / dt ^2, the rotation angular acceleration _{^{d 2 θ sy / dt 2,}} d 2 θ sx / dt 2 and 3 axis z-direction translational acceleration detected by the acceleration sensor 25-1 d ² z _s1 / dt ² And calculated based on Eq. (12).

  Note that a three-axis angular velocity sensor that detects the rotational angular velocity around the x-axis, y-axis, and the z-axis is installed in the powertrain block 12 or the unsprung element 15, 3 It is also possible to acquire the rotational angular acceleration around the center of gravity axis of the power train block 12 or the unsprung element 15 by differentiating the detection signal of the axial angular velocity sensor. If a 3-axis acceleration sensor that detects x-direction translational acceleration, y-direction translational acceleration, and z-direction translational acceleration can be installed directly at the center of gravity of powertrain block 12 or unsprung element 15, detection of the 3-axis acceleration sensor It is possible to obtain the translational acceleration at the center of gravity of the powertrain block 12 or the unsprung element 15 from the signal.

The vibration data conversion unit 73 performs time-frequency analysis on each time-series vibration data acquired by the time-series vibration data acquisition unit 71 to convert it into vibration data with respect to time and frequency. Here, wavelet transformation is performed on each of the time-series vibration data acquired by the time-series vibration data acquisition unit 71 to convert each of the time-series vibration data into vibration data with respect to time and frequency. That is, the x-direction translational acceleration d ² x _v / dt ² of the floor of the vehicle body 11, the rotational angular acceleration d ² θ _w / dt ^{2 of} the drive wheel 18, and the translational acceleration d ² x _p / dt ^{2 of the} powertrain block 12 , D ² y _p / dt ² , d ² z _p / dt ² and rotational angular acceleration d ² θ _px / dt ² , d ² θ _py / dt ² , d ² θ _pz / dt ² and the unsprung element 15 ( Translational acceleration d ² x _s / dt ² , d ² y _s / dt ² , d ² z _s / dt ² and rotational angular acceleration d ² θ _sx / dt ² , d ² θ _sy / dt ² , d ² θ _sz / dt ² is converted to vibrational acceleration with respect to time and frequency by wavelet transform. Thus, as shown in FIG. 5, a matrix X _k (k is an integer from 1 to K, K is the number of measurement conditions and 14 in this embodiment) representing the relationship of vibration acceleration to time and frequency is each measurement condition A three-dimensional array X_ is obtained which is obtained for each time and which represents the relationship between vibration acceleration with respect to time, frequency, and measurement conditions (measurement location and direction). However, in the figures and formulas, the three-dimensional array X_ is represented by _ (under bar) below X (the same applies to other three-dimensional arrays). Note that the wavelet transform itself is known and thus the detailed description is omitted.

  The factor analysis unit 74 performs factor analysis (PARAFAC analysis) on each vibration data (three-dimensional array X_) wavelet-transformed by the vibration data conversion unit 73 with a predetermined number of factors. In PARAFAC analysis, as shown in FIG. 6, the three-dimensional array X_ can be expressed by being decomposed into matrices A, B, and C, and can be expressed. The matrix X obtained by matrixing the three-dimensional array X_ into two dimensions is It can be approximated and expressed by equation (13).

In Formula (13), T represents transposition (the following other formulas are also the same). The matrix H is a matrix obtained by two-dimensionally matrixing a three-dimensional array H_ (see FIG. 6) in which one is diagonally zero and the other is zero. The matrix A is a matrix of I rows and F columns (I and F are integers of 2 or more) related to the frequency and the factor, and the i-th row and the f-th column (i is 1 to I) as shown in the following equation (14) When an element of an integer of f and an integer of 1 to F is a _if , row (i) corresponds to frequency and column (f) corresponds to a factor. The matrix B is a matrix of J rows and F columns (J is an integer of 2 or more) related to time and factors, and the j-th row and f-th column (j is an integer of 1 to J) as shown in the following equation (15) If the element of b) is b _jf , row (j) corresponds to time and column (f) corresponds to a factor. The matrix C is a matrix of K rows and F columns relating to measurement conditions (measurement points and directions) and factors, and _assuming that the element of the k th row and the f th column is c k f as shown in the following equation (16) (K) corresponds to the measurement condition, and column (f) corresponds to the factor. I is the number of frequency sample points, J is the number of time sample points, and F is the number of factors. I 3D array X_, j, elements X_ _ijk of k is expressed by approximation by the following equation (17). The equation (17) is a rewrite of the equation (13).

The factor analysis unit 74 calculates the matrices A, B, and C substantially satisfying the equation (13) (the equation (17)) by performing the PARAFAC analysis on the three-dimensional array X_ with a plurality of factor numbers. Since the process itself which calculates the matrices A, B, and C by PARAFAC analysis is known, detailed description is omitted. Then, the factor analysis unit 74 calculates the relationship of a _if × b _jf to the frequency (i) and the time (j) for each factor (f = 1 to F) to obtain the frequency and time for each factor. Acquire vibration data.

An example of the basis wavelet of the factor obtained by calculating the relationship of a _if × b _jf to the frequency (i) and the time (j) for each factor (f = 1 to F) is shown in FIG. In PARAFAC analysis, the number of factors is F = 5, and FIG. 7A shows the relationship of a _i1 × b _j1 to the frequency and time at a factor of f = 1, and FIG. = indicates the relationship between a _i2 × b _j2 with respect to frequency and time at 2, FIG. 7 (c) shows the relationship between a _i3 × b _j3 respect to frequency and time of a factor f = 3, FIG. 7 (d) _Shows the relationship of a _i4 × b _j4 to the frequency and time at a factor f = 4, and FIG. 7 (e) shows the relationship of a _i5 × b _j5 to the frequency and time at a factor f = 5. As shown in FIG. 7A, when the factor f = 1, the amplitude of the vibration becomes large at a frequency of 4 Hz, and the amplitude of the 4 Hz vibration becomes a transient vibration that largely fluctuates with respect to time. Similarly, as shown in FIG. 7C, even with the factor f = 3, the amplitude of the vibration becomes large at a frequency of 11 Hz, and the amplitude of the vibration of 11 Hz becomes a transient vibration with a large fluctuation with time. As shown in e), even with the factor f = 5, the amplitude of the vibration becomes large at the frequency of 16 Hz, and the amplitude of the 16 Hz vibration becomes a transient vibration that largely fluctuates with respect to time. On the other hand, as shown in FIG. 7B, in the case of the factor f = 2, the amplitude of the vibration becomes large at a frequency of 6 Hz, but the vibration of 6 Hz becomes a steady vibration hardly fluctuating with time. Similarly, as shown in FIG. 7 (d), even with the factor f = 4, the amplitude of the vibration increases at a frequency of 12 Hz, but the amplitude of the 12 Hz vibration becomes a steady vibration that hardly varies with time.

The factor analysis unit 74 selects a factor based on vibration data corresponding to each factor. Here, the factor (f) is selected based on the fluctuation degree of the amplitude of the vibration data (a _if × b _jf ) corresponding to each factor (f = 1 to F), and the fluctuation degree of the amplitude of a _if × b _jf Is selected as a factor (f) which causes a transient vibration larger than a predetermined amount. In the example of FIG. 7, the factor f = 1, 2, 3 which is a transient vibration is selected without selecting the factor f = 2, 4 which is a steady state vibration.

By calculating the matrix C by PARAFAC analysis, the degree of contribution of each measurement condition can be calculated for each factor, and the elements c _{1f to} c _Kf of the f th column of the matrix C correspond to the factor (f) The degree of contribution of each measurement condition (k = 1 to K). The degree of contribution c _{11 to} c _K1 of each measurement condition (k = 1 to K) corresponding to the selected factor f = 1 is shown in FIG. 8A, and each measurement condition corresponding to the selected factor f = 3 The degree of contribution c _{13 to} c _K 3 of (k = 1 to _K ) is shown in FIG. 8B, and the degree of contribution c _{15 of} each measurement condition (k = 1 to K) corresponding to the selected factor f = 5 c _K5 is shown in FIG. 8 (c). In the example of FIG. 8, K = 14, k = 1 corresponds to the x-direction translational acceleration VE (X) of the floor of the vehicle body 11, and k = 2 corresponds to the rotational angular acceleration WH (Rot) of the drive wheel 18 Do. Then, k = 3, 4, 5, 6, 7, 8 indicate the x-direction translational acceleration LA (X), y-direction translational acceleration LA (Y) at the center of gravity of the unsprung element 15 (lower arm 19), z-direction Translational acceleration LA (Z), Rotational angular acceleration LA (Roll) around the x-axis center of gravity, Rotational angular acceleration LA (Pitch) around the y-axis central axis, Rotational acceleration LA (Yaw) around the z-axis central axis It corresponds. Then, k = 9, 10, 11, 12, 13, and 14 represent the x-direction translational acceleration PB (X), the y-direction translational acceleration PB (Y), and the z-direction translational acceleration PB (Y) at the center of gravity in the powertrain block 12. Z), rotational angular acceleration PB (Roll) around the x-direction center of gravity axis, rotational angular acceleration PB (Pitch) around the y-direction central axis, and rotational angular acceleration PB (Yaw) around the z-direction central axis.

The equation (17) is rewritten as the following equation (18). The matrix X _k is a matrix of I rows and J columns representing the relationship of the vibration acceleration to the frequency (i) and time (j) after wavelet transform in the measurement condition (k). The matrix M _f is a matrix of I rows and J columns representing the relationship of a _if × b _jf (the value of the vertical axis in FIG. 7) to the frequency (i) and time (j) in the factor (f). It is expressed by the equation (19). In equation (19), a _f is a vector corresponding to the f-th column of matrix A (see equation (14)) and b _f is a vector corresponding to f-th row of matrix B (see equation (15)) is there.

_If each element of the matrices X _k and M _f is rearranged into a vertical vector and transposed, then the horizontal vectors vec (X _k ) ^T and vec (M _f ) ^{T are obtained} , and equation (18) is given by equation (20) below Become. The following equation (21) is obtained from the equation (20).

The weight coefficient calculation unit 75 calculates vibration data to be evaluated for vibration (in this embodiment, x of the floor of the vehicle body 11 in the vibration data X _{1 to} X _K under a plurality of measurement conditions wavelet-transformed by the vibration data conversion unit 73). Based on the relationship between the direction translational acceleration) X ₁ and the other vibration data X _{2 to} X _K and the contribution degree C calculated by the factor analysis unit 74, each measurement condition other than the vibration evaluation target (k = 2 Weight coefficients d _{2 to} d _K corresponding to to K) are calculated. If Formula (21) is divided into vibration data X ₁ to be subjected to vibration evaluation and other vibration data X _{2 to} X _K , the following Formulas (22) and (23) are obtained. If vec (M ₁ ) ^T to vec (M _F ) ^T are eliminated from the equations (22) and (23), the following equation (24) is obtained. In equation (24), the lateral vector D is expressed by the following equation (25) based on the lateral vector C ₁ and the matrix C ₂ . That is, (24) represents the vibration data X ₁ to the vibration evaluated, otherwise the vibration data X ₂ to X _K of the relationship between the contribution C of each measurement conditions corresponding to each factor. The weighting factors d _{2 to} d _K are elements of the horizontal vector D, and are calculated by equation (25) based on the horizontal vector C ₁ and the matrix C ₂ (degree of contribution C).

As shown in the following equation (26), the right side of the equation (24) is defined as y. y for all measurement conditions except for vibration evaluated (k = 2~K), the sum obtained by adding the d _k × vec (X _k) ^T, i.e. the sum of the vibration data X _k weighted by weighting coefficients d _k Represents the sum. Further, as shown in the following equation (27), a sum obtained by adding vibration data X _k (d _k × ve c (X _k ) ^T ) weighted by the weighting coefficient d _k for a part of measurement conditions excluding the vibration evaluation target Define y ^ _r . However, in the formula, y ^ _r is represented by adding ^ (hat) on y _r .

The analysis model identification unit 76 weights all of the measurement conditions (k = 2 to K) except the vibration evaluation target in the vibration data X _{1 to} X _K under a plurality of measurement conditions wavelet-transformed by the vibration data conversion unit 73. the sum y obtained by adding the vibration data X _k weighted by coefficients d _k, the error between the sum y ^ _r obtained by adding the vibration data X _k weighted by weighting coefficients d _k for some measurement conditions, except for vibration evaluation The measurement conditions to be considered in the vibration analysis model of the vehicle 10 are limited based on the index E _r to be represented. The total sum y is calculated by the equation (26), and the total sum ^ _r is calculated by the equation (27). For example, the following equation (28) can be used as the index E _r representing the error between y and y _r . However, the x-direction translational acceleration VE (X) (k = 1) of the floor of the vehicle body 11 that is the vibration evaluation target, and the rotational angular acceleration WH (Rot) of the drive wheel 18 driven by the torque of the engine 13 that is the vibration source. ) (K = 2) is included in the measurement conditions considered in the vibration analysis model.

A part of measurement conditions used to calculate the total sum ^ _{r in} the analysis model identification unit 76 is selected based on the degree of contribution c _kf of each measurement condition calculated in the factor analysis unit 74. More specifically, a part of the measurement conditions used to calculate the total sum ^ _r is the degree of contribution c corresponding to the factor f = 1, 3, 5 selected by the factor analysis unit 74 as shown in FIG. It is selected based on _kf , and for example, for each factor (f = 1, 3, 5), the measurement condition (k) is selected by ranking in descending order of the absolute value of the degree of contribution c _kf . However, the rotational angular acceleration WH (Rot) (k = 2) of the drive wheel 18 included in the measurement conditions considered in the vibration analysis model is included in some of the measurement conditions used for calculating the total sum y _r . Therefore, among the measurement conditions of k = 3 to K, the measurement condition (k) is selected for each factor (f = 1, 3, 5) in the descending order of the absolute value of the degree of contribution c _kf .

In the example of FIG. 8, the absolute value of the degree of contribution c _k1 of the measurement condition (k = 3 to K) corresponding to the factor f = 1, and the contribution of the measurement condition (k = 3 to K) corresponding to the factor f = 3 As for the absolute value of the degree c _k3 , the measurement condition (k = 3 to K) corresponding to the factor f = 5 is the largest in rotational angular acceleration PB (Pitch) (k = 13) about the y-direction center axis of the powertrain block 12 The absolute value of the degree of contribution c _k5 ) is the largest in the x-direction translational acceleration LA (X) (k = 3) at the center of gravity of the unsprung element 15 (lower arm 19). Therefore, in the ranking r = 1, WH (Rot) (k = 2), LA (X) (k = 3), and PB (Pitch) (k = 13) are selected as a part of measurement conditions, and the ranking is The total sum ^ _r at r = 1 is calculated by the following equation (29). Then, at ranking r = 1, an index E _r representing an error between y and ^ _r is calculated by equation (28).

y ^ _r = d ₂ x vec (X ₂ ) ^T + d ₃ x vec (X ₃ ) ^T + d ₁₃ x vec (X ₁₃ ) ^T (29)

The absolute value of the degree of contribution c _k1 of the measurement condition (k = 3 to K) corresponding to the factor f = 1 is 2 in the x direction translational acceleration PB (X) (k = 9) at the center of gravity of the powertrain block 12 The absolute value of the contribution degree c _k3 of the measurement condition (k = 3 to K) corresponding to the factor f = 3 is the z direction translational acceleration LA (Z (Z)) at the center of gravity of the unsprung element 15 (lower arm 19) The absolute value of the degree of contribution c _k5 of the measurement condition (k = 3 to K) corresponding to the factor f = 5, where (k = 5) is the second largest, is the rotation of the powertrain block 12 about the center of gravity in the z direction. The angular acceleration PB (Yaw) (k = 14) is the second largest. Therefore, in the ranking r = 2, WH (Rot) (k = 2), LA (X) (k = 3), LA (Z) (k = 5), PB (X) (K = 5) as a part of measurement conditions. k = 9), PB (Pitch) (k = 13), and PB (Yaw) (k = 14) are selected, and the total sum ^ _r at ranking r = 2 is calculated by the following equation (30). Then, at ranking r = 2, an index E _r representing an error between y and ^ _r is calculated by equation (28).

y ^ _r = d ₂ x vec (X ₂ ) ^T + d ₃ x vec (X ₃ ) ^T + d ₅ x vec (X ₅ ) ^T
+ D ₉ × vec (X ₉ ) ^T + d ₁₃ × vec (X ₁₃ ) ^T + d ₁₄ × vec (X ₁₄ ) ^T (30)

The absolute value of the degree of contribution c _k1 of the measurement condition (k = 3 to K) corresponding to the factor f = 1 is 3 in the z direction translational acceleration PB (Z) (k = 11) at the center of gravity of the powertrain block 12 The absolute value of the degree of contribution c _k3 of the measurement condition (k = 3 to K) corresponding to the factor f = 3 is the rotational angular acceleration PB (Yaw) (k of the powertrain block 12 about the z axis) = 14) is the third largest, and the absolute value of the degree of contribution c _k5 of the measurement condition (k = 3 to K) corresponding to the factor f = 5 is the z axis of gravity of the unsprung element 15 (lower arm 19) The rotational angular acceleration LA (Yaw) (k = 8) is the third largest. Therefore, in the ranking r = 3, WH (Rot) (k = 2), LA (X) (k = 3), LA (Z) (k = 5), LA (Yaw) (LAw) as a part of measurement conditions. Select k = 8), PB (X) (k = 9), PB (Z) (k = 11), PB (Pitch) (k = 13), and PB (Yaw) (k = 14), and rank them The total sum ^ _r at r = 3 is calculated by the following equation (31). Then, at ranking r = 3, an index E _r representing an error between y and y _r is calculated by equation (28). Furthermore, by repeating the same process, in ranking r 表 す 4, an index E _r representing an error between y and y _r is calculated by equation (28).

y ^ _r = d ₂ x vec (X ₂ ) ^T + d ₃ x vec (X ₃ ) ^T + d ₅ x vec (X ₅ ) ^T
+ D ₈ × vec (X ₈ ) ^T + d ₉ × vec (X ₉ ) ^T + d ₁₁ × vec (X ₁₁ ) ^T
+ D ₁₃ × vec (X ₁₃ ) ^T + d ₁₄ × vec (X ₁₄ ) ^T (31)

An example of the result of calculating the relationship between the index E _{r and} the ranking r is shown in FIG. However, the index E _{r on} the vertical axis in FIG. 9 is normalized by the following equation (32). FIG. 9 also shows the total number of measurement conditions (including the x-direction translational acceleration VE (X) of the floor of the vehicle body 11 and the rotational angular acceleration WH (Rot) of the drive wheel 18) considered in the vibration analysis model.

In the example of FIG. 9, the index E _r at the ranking r = 2 is significantly reduced compared to the index E _r at the ranking r = 1. On the other hand, the index E _r of ranking r ≧ 3 has little difference indicative E _r of the ranking r = 2. Therefore, the analysis model identification unit 76 determines the measurement condition to be considered in the vibration analysis model of the vehicle 10, the rotational vibration of the drive wheel 18, which is the measurement condition selected by the ranking r = 2 immediately after the index _Er rapidly decreases. , X-direction translational vibration and z-direction translational vibration at the center of gravity of the unsprung element 15 (lower arm 19), x-direction translational vibration at the center of gravity of the powertrain block 12 and rotational vibration around the center of gravity y-axis It is limited to rotational vibration around the center of gravity axis and x-direction translational vibration of the floor of the vehicle body 11 that is the object of vibration evaluation. And the analysis model identification part 76 performs structure identification of the vibration analysis model of the vehicle 10 based on this limited measurement condition. The analysis model identification unit 76 determines the measurement conditions considered in the vibration analysis model of the vehicle 10, the x-direction translational vibration of the floor of the vehicle body 11 that is the vibration evaluation target, and the minimum ranking where the index E _r is less than the set value. It is also possible to limit to the measurement conditions selected by r.

An example of the vibration analysis model of the vehicle 10 structurally identified by the analysis model identification unit 76 is shown in FIG. 10 and the following equations (33) to (41). The equations (33) to (41) can be rewritten as the following equation (42). However, (33) in the - (42) below, represents the time derivative in. (Dot) represents the angular acceleration of the engine 13 are given the ... on the theta _e, the angular acceleration of the transmission 14 theta _m represents subjected to ... over the angular acceleration of the driving wheel 18 stands subjected to ... on the theta _w, a x-direction translational acceleration of the vehicle body 11 denoted by the · on the x _v Representing the x-direction translational acceleration of the unsprung element 15 and the z-direction translational acceleration by appending x _LA , z _LA on the x-direction translational acceleration of the powertrain block 12, an angle around the y-direction center of gravity axis The acceleration and the angular acceleration around the center of gravity axis in the z direction are indicated by adding ··· onto x _PB , β _PB , γ _PB . Further, (33) to (42) below, J _e is the moment of inertia of the engine 13, J _m moment of inertia of the transmission 14, J _w is the moment of inertia of the drive wheels 18, m _v is the mass of the vehicle body 11, m _LA is the mass of unsprung elements 15, m _PB is the mass of the powertrain block 12, Jβ _PB is the moment of inertia about the y-direction central axis of the power train block 12, Jγ _PB is around the z-direction central axis of the power train block 12 It is a moment of inertia. τ _e is the torque of the engine 13 (excitation force of the vibration system), τ _m is the torque transmitted between the engine 13 and the transmission 14, d _m is the transmission gear ratio of the transmission 14, and τ _ds is the drive with the transmission 14 The torque transmitted between the wheels 18, F _d is an x-direction translational force acting between the drive wheels 18 and the road surface, r _w is the radius of the drive wheels 18, F ^x _SS is transmitted between the unsprung element 15 and the vehicle body 11 that the x-direction translational force, F ^x _PB is x-direction translation force transmitted between the powertrain block 12 and the vehicle body 11, F ^z _SS is the z-direction translational force acting on the unsprung element 15, τβ _PB powertrain block 12 Is a torque about the center of gravity y axis in the y direction, .tau..gamma. _PB is a torque about the center of gravity z axis in the z direction acting on the power train block 12. As shown in FIG.

  The vibration analysis model according to the equations (33) to (41) (equation (42)) is a principle model representing a connection relationship between a plurality of vibration elements as shown in FIG. (33) represents the rotational motion equation of the engine 13, (34) represents the rotational motion equation of the transmission 14, (35) represents the rotational motion equation of the drive wheel 18, and (36) represents the vehicle body Equations (37) and (38) represent the x-direction translational motion equation and the z-direction translational motion equation of the unsprung element 15, and the equations (39), (40) and (41) These represent the x-direction translational motion equation of the powertrain block 12, the rotational motion equation about the y-axis, and the rotational motion equation about the z-axis. However, in the vibration analysis model (principle model), the y-direction translational vibration at the center of gravity of the unsprung element 15, the rotational vibration around the x-axis, and the rotational vibration around the y-axis, and the rotation around the z-axis The vibration and the y-direction translational vibration at the center of gravity of the powertrain block 12, the z-direction translational vibration and the rotational vibration around the x-axis center of gravity are not taken into consideration. That is, the x-direction translational vibration of the vehicle body 11, the rotational vibrations of the engine 13, the transmission 14, and the drive wheel 18, the x-direction translational vibration and the z-direction translational vibration at the center of gravity of the unsprung element 15, and the center of gravity of the powertrain block 12 It is limited to a vibration analysis model that takes into account only the x-direction translational vibration at the point, the rotational vibration around the y-direction center of gravity axis, and the rotational vibration around the z-direction center of gravity axis.

In the vibration analysis device 70, for example, the angular acceleration of the engine 13, the angular acceleration of the transmission 14, the angular acceleration of the drive wheel 18, the x direction translational acceleration of the vehicle body 11, the x direction translational acceleration of the unsprung element 15 and the z direction translational acceleration, And the angular acceleration around the x-direction translational acceleration of the powertrain block 12 and the y-direction center of gravity axis and the angular acceleration around the z-direction center of gravity axis as observation variables, the equations (33) to (41) (42) Transfer force τ _m , τ _ds , F _d , F ^x _SS , F ^z _SS , F that propagates each vibration element by the vibration force τ _e of the vibration system, which is unknown variable, and the vibration force τ _e . It is possible to estimate ^x _PB , τβ _PB , τγ _PB . Then, it is possible to calculate a transfer function representing the relationship between input and output when the torque of the engine 13 (the excitation force of the vibration system) is input and the x-direction translational acceleration of the vehicle body 11 is output. It is also possible to estimate the elastic coefficient and damping coefficient between the vibrating elements based on the transfer force transmitted through the vibrating elements.

In the embodiment described above, processing for acquiring time-series vibration data under a plurality of measurement conditions in which any one or more of the vibration element to be subjected to vibration measurement and the vibration measurement direction are different from each other A process of converting the vibration data X _{1 to} X _K with respect to time and frequency by wavelet transformation; a process of calculating a contribution degree C by performing PARAFAC analysis on each of the vibration data X _{1 to} X _K ; Based on Eq. (24) representing the relationship between the vibration data X ₁ and the other vibration data X _{2 to} X _K and the degree of contribution C. The weighting coefficients d _{2 to} d _K corresponding to the measurement conditions other than the vibration evaluation target And a process of identifying the structure of the vibration analysis model under a measurement condition limited based on the index E _r representing the error between y and ^ _r . In the process of structural identification of the vibration analysis model, the measurement conditions to be considered in the vibration analysis model are limited to the minimum necessary measurement conditions in which the contribution degrees c _{1f to} c _Kf corresponding to the factor (f) are in the upper rankings The vibration analysis model can be reduced in dimension. Furthermore, the structure identification method of the vibration analysis model in the present embodiment is not an identification method that statistically regresses the vibration analysis model, but is not an identification method that derives a simplified model from a complex physical model by a balanced termination method or the like. The structurally identified vibration analysis model (principle model) has physical meaning. Therefore, the accuracy can be improved while reducing the dimension of the vibration analysis model. In addition, in the case of structure identification of a vibration analysis model, it can carry out only from the vibration data at the time of real operation (actual run) of vehicle 10, and the simple measurement which dropped parts from vehicle 10 is unnecessary. Moreover, comparison with other vehicles 10 is also easy, and it can be easily grasped what kind of idea the other vehicles 10 are designed.

Furthermore, in this embodiment, in the process of performing PARAFAC analysis, evaluation is performed by selecting the factor (f) based on the vibration data (a _if x b _jf ) calculated for each factor (f = 1 to F). A vibration mode corresponding to a desired factor (f) can be extracted, and structural identification of a vibration analysis model for evaluating the extracted vibration mode can be performed. Furthermore, by selecting the factor f = 1, 3, 5 in which the degree of fluctuation of the amplitude of a _if × b _jf is large, the transient vibration corresponding to the factor f = 1, 3, 5 to be evaluated is extracted It is possible to perform structural identification of a vibration analysis model for evaluating transient vibration.

  Although the above embodiment demonstrated the example whose vibration system was the vehicle 10, this embodiment is applicable also to vibration systems other than a vehicle.

  As mentioned above, although the form for implementing this invention was demonstrated, this invention is not limited at all by such embodiment, It can be implemented with various forms in the range which does not deviate from the summary of this invention. Of course.

  Reference Signs List 10 vehicle, 11 vehicle body, 12 power train block, 13 engine, 14 transmission, 15 unsprung element, 16 mount, 17 suspension bush, 18 drive wheel, 19 lower arm, 21 acceleration sensor, 22-1, 22-2, 22 -3, 25-1, 25-2, 25-3 3-axis acceleration sensor, 28 angular velocity sensor, 70 vibration analysis device, 71 time series vibration data acquisition unit, 72 setting constant storage device, 73 vibration data conversion unit, 74 factor Analysis unit, 75 weight coefficient calculation unit, 76 analysis model identification unit.

Claims (2)

A structural identification device of a vibration analysis model for identifying the structure of an analysis model of a vibration system having a plurality of vibration elements,
A time-series vibration data acquisition unit that acquires time-series vibration data under a plurality of measurement conditions in which one or more of a vibration element to be subjected to vibration measurement and a vibration measurement direction in a vibration system are different from each other;
A vibration data conversion unit that converts each time-series vibration data acquired by the time-series vibration data acquisition unit into vibration data with respect to time and frequency;
There rows by changing the number of factors Factor analysis on the vibration data converted by the vibration data conversion unit selects a factor depending on the variation degree of the vibration data in the result of the factor analysis for each number of factors, the factors A factor analysis unit that calculates the degree of contribution of each measurement condition corresponding to
In the vibration data under a plurality of measurement conditions converted by the vibration data conversion unit, the vibration data measured under the measurement conditions targeted for vibration evaluation, the other vibration data, and the contribution degree calculated by the factor analysis unit A weighting factor calculation unit that calculates weighting factors corresponding to measurement conditions other than the vibration evaluation target based on the relationship of
In the vibration data under a plurality of measurement conditions converted by the vibration data conversion unit, the sum of the sum of the vibration data weighted by the weighting factor for all measurement conditions excluding the vibration evaluation target and the factor analysis unit excluding the vibration evaluation target The measurement conditions are limited based on an index that represents an error from the sum of the sum of the vibration data weighted by the weighting factor for a part of the measurement conditions selected based on the calculated degree of contribution , and the limited measurement conditions and the vibration An analysis model identification unit that identifies the structure of the analysis model based on the measurement conditions to be evaluated ;
A structural identification device of a vibration analysis model, comprising: A structure identification method of a vibration analysis model for identifying a structure of an analysis model of a vibration system having a plurality of vibration elements,
Time-series vibration data acquisition processing for acquiring time-series vibration data under a plurality of measurement conditions in which any one or more of a vibration element to be subjected to vibration measurement and a vibration measurement direction in a vibration system are different from each other;
Vibration data conversion processing for converting each time-series vibration data acquired in the time-series vibration data acquisition processing into vibration data with respect to time and frequency;
There rows by changing the number of factors Factor analysis on the vibration data converted by the vibration data conversion process, and select the factor in accordance with the fluctuation degree of the vibration data in the result of the factor analysis for each number of factors, the factors Factor analysis processing to calculate the degree of contribution of each measurement condition corresponding to
In the vibration data under a plurality of measurement conditions converted by the vibration data conversion process, the vibration data measured under the measurement conditions to be evaluated for vibration, the other vibration data, and the contribution degree calculated by the factor analysis process Weight coefficient calculation processing for calculating a weight coefficient corresponding to each measurement condition other than the vibration evaluation target based on the relationship of
In the vibration data under multiple measurement conditions converted by the vibration data conversion process, the sum of the sum of the vibration data weighted by the weighting factor for all measurement conditions excluding the vibration evaluation target, and the factor analysis unit excluding the vibration evaluation target The measurement conditions are limited based on an index that represents an error from the sum of the sum of the vibration data weighted by the weighting factor for a part of the measurement conditions selected based on the calculated degree of contribution , and the limited measurement conditions and the vibration Analysis model identification processing for identifying the structure of the analysis model based on the measurement condition to be evaluated ;
How to identify the structure of vibration analysis model, including