JP2020094805A - Factor estimation system for vibration noise problem - Google Patents

Abstract

To solve the problem of suppressing vibration and noise generated by a product that it is necessary to grasp a sound source, a vibration source, and a transmission path because data having large variations are often obtained in the vibration/noise phenomenon, and it is desired to probabilistically treat estimation results for grasping, and to determine an examination of structural changes made according to the results based on the probability and the degree of influence, and a probabilistic rationale for cost-effectiveness of the structural changes is required as a criterion for the structural changes toward the suppression of vibration and noise.SOLUTION: This can be solved by expressing the relationship between measured signals with a Bayesian network.SELECTED DRAWING: Figure 4

Description

The present invention is based on a sound source or a vibration source that causes a vibration/noise problem, an estimation method of a vibration source/sound source and a transmission path for transmission of vibration/noise originating from them, and an estimation result. Regarding the effect judgment of the structural change.

Vibrations and noises generated from products are widely recognized as factors that deteriorate comfort when using the products. Therefore, it is important to understand and suppress vibration and noise. Toward this end, it is necessary to understand the source of vibration and noise and the transmission path in advance at the design stage, and to study the structural change based on the result. It is important to measure the transfer characteristics in order to understand the source of vibration and noise and the transfer path to be taken. In Patent Document 1, in order to obtain the vibration transfer characteristics between the input and output, both the input and output are accelerated in the actual operating state. Is measured and the transfer characteristics between the input and output (output acceleration/input acceleration) are calculated.

On the other hand, from the idea that the input signal should originally measure the excitation force or the size of the sound source, not the acceleration, and therefore, in Patent Document 2, there are multiple excitation forces that are difficult to measure directly in actual operation. It shows a method of deterministically identifying from the observable signal of, and a method of calculating the transfer characteristic using the estimation results of the excitation force and the size of the sound source obtained on it.

JP, 2007-57460, A JP, 2016-017808, A

In order to suppress the vibration and noise generated by the product, it is first necessary to understand the sound source, the vibration source, and the transmission path. To realize this, generally, the sound source and the transfer characteristic are identified using the measured data. Furthermore, the dominant transmission path and the contribution rate indicating the magnitude of the contribution from each path are determined. However, in the measurement of vibration and noise phenomena, data with large variations are often obtained. Therefore, it is desirable that the identification result also be treated probabilistically, including the uncertainty due to the variation in the data, and the examination of the structural change that is made out of the result is also considered as the probability distribution of the vibration noise estimation value. It is desirable that the judgment be made based on the degree of influence.

On the other hand, in the transfer characteristic acquisition method described in Patent Document 1 or Patent Document 2, a deterministic transfer characteristic is calculated, and the result does not take into account data variations and uncertainties. For example, as shown in FIG. 1, when identifying the transfer characteristics between a plurality of signals, the proportional constant between the signals is generally determined by a method called a least squares method, a multiple regression analysis, or an inverse matrix method. However, in this case, since the variation contained in the actual measurement data is not taken into consideration, the difference between the left diagram and the center diagram of FIG. 1 can be expressed only by a slight inclination of the approximation curve. In addition, as shown in the figure on the right, a problem may occur in which the proportional constant indicating the transfer characteristic is unintentionally negative. Therefore, it is not possible to handle vibration and noise using such transfer characteristics expressed as a uniform numerical value, of course, because it is not possible to handle the variation of data. I can't make a logical judgment.

In addition, regarding the method of suppressing vibration noise, general products still mainly use vibration noise reduction measures by changing the structure, and even when making a decision on this structure change, there are variations in data and inconsistencies regarding the cost-effectiveness. Probabilistic discussion considering certainty is required.

The above problem is solved by expressing the relationship between the excitation source, the sound source, and the signal based on the measured signal with a Bayesian network. As a result, the relationship between signals is expressed by the likelihood, which is a conditional probability density distribution, instead of the sound source/transfer characteristics that were conventionally represented by uniform numerical values, and the degree of influence of structural change is calculated by the probability theory. Expressive is achieved. In order to realize this, claim 1 of the present invention provides a system configuration in which necessary data collection, storage, and analysis methods are combined, and claim 2 presents a method of displaying contributions and the effect of the element on the entire system at the time of element selection and , When a display system characterized by having a function of displaying elements in consideration of vibration and noise suppression in order of priority of countermeasures is selected, an element for which structural modification is considered in claim 3 is selected. Shows a structure characterized by having a function of displaying the cost of changing the structure and the effect of suppressing vibration and noise.

INDUSTRIAL APPLICABILITY The present invention enables probabilistic inference regarding the sound source, the excitation source, and the transmission path, the degree of influence of arbitrary elements on the whole, and the cost-effectiveness of structural change, which is a subject, and suppresses vibration and noise. It is possible to provide a judgment material for making a rational judgment.

It is a figure which shows the subject which cannot consider the variation in the conventional vibration noise transmission identification. It is a figure which shows the concept of the variation consideration in vibration noise transmission identification proposed in the embodiment of the present invention. It is a figure explaining the concept of the sound source and transfer characteristic which considered the variation proposed in the embodiment of the present invention. It is a figure explaining the structure of the vibration noise problem factor estimation system of embodiment of this invention. It is a figure explaining another structure of the vibration noise problem factor estimation system of embodiment of this invention. It is a figure explaining another structure of the vibration noise problem factor estimation system of embodiment of this invention. It is a figure explaining the example of a display in the contribution rate display system of an embodiment of the present invention. It is a figure explaining another example of a display in the contribution rate display system of an embodiment of the present invention.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

FIG. 2 is a diagram showing a concept of considering variation in vibration noise characteristic identification proposed in the embodiment of the present invention. As described above, when the proportional constant between signals is approximately identified as a uniform numerical value by using the least squares method, etc., the influence of signal variation cannot be considered, and the subsequent processing using this data Uncertainty due to data dispersion cannot be included as a criterion. As a method to solve the above problem, Fig. 2 shows the concept of maintaining the relationship between data in a form that includes variations. By this, the difference between the left figure and the central figure is expressed from the viewpoint of data variation. Not only can it be done, but even if the least squares approximation causes the proportionality constant to become a negative value as shown in the figure on the right, the data can be managed and retained without causing any doubt about the interpretation of that matter.

FIG. 3 is a diagram showing the concept of the sound source/transfer characteristics in consideration of the variations calculated based on these results. Considering the uncertainties to be solved by the way of managing/holding such characteristic data. The above probabilistic judgment can be realized, and in the case of the structural change, the probabilistic discussion in consideration of the variation in data and uncertainty regarding the cost-effectiveness can also be realized.

FIG. 4 is a diagram showing a system configuration for implementing the embodiment of the present invention. Various data of vibration and noise in the actual operation state of the measurement target product 10 are extracted by the on-site data extraction system 11, and the on-site data extraction system 11 stores the extracted actual operation data of the on-site in the on-site database storage 12. The user 20 estimates the transmission path of vibration and noise of the product to be measured, prepares a transmission path model 16a that represents this in a graphical model, and stores it in the transmission path model data storage 13 by the laboratory data storage system 17. The sound source/transmission characteristic estimation engine 14 transmits based on the plurality of actual operation site data stored in the site database storage 12 and the transmission route model data storage 13 and the transmission route model 16a selected by the user. The sound source/transfer characteristic probability distribution corresponding to each transfer path of the path model 16 a is estimated, and the estimated sound source/transfer characteristic probability distribution is stored in the sound source/transfer characteristic probability distribution database 15. The contribution rate estimation engine 18 uses the transfer path model 16a stored in the transfer path model data storage 13, the sound source/transfer characteristic probability distribution data 16b stored in the sound source/transfer characteristic probability distribution database 15, and the field database storage 12 as well. The vibration and noise contribution rate is calculated based on the plurality of actual operation site data stored in, and the calculated vibration and noise contribution rate is passed to the contribution rate display system 19. The user 20 operates the contribution rate display system 19 to display the contribution of the measurement target product 10 to a predetermined partial structural element, and based on this, the degree of influence on the measurement target product as a whole, and when planning a structural change, Probabilistic inference results regarding cost effectiveness can be obtained. It should be noted that the sound source/transfer characteristic probability distribution data 16b acquired in the pre-shipment test or the periodic inspection test of the measurement target product 10 is stored in the sound source/transfer characteristic probability distribution database 15 by the user 20 by the laboratory data storage system 17 described above. It is also possible to store and update to.

As described above, the transmission path model 16a and the sound source/transfer characteristic probability distribution data 16b that the user 20 can store in the system by the laboratory data storage system 17 are examined or calculated based on the measurement result and the analysis result in the laboratory. This makes it possible to update and improve the vibration noise model of the product under consideration. Therefore, here, the transfer path model 16a and the sound source/transfer characteristic probability distribution data 16b are collectively referred to as laboratory data 16.

As a probabilistic estimation algorithm shown above, a method using a method called Bayesian network is appropriate. Bayesian network assumes a causal relationship between the event H and the data D obtained under the event H, and then calculates the conditional probability density distribution in which the data D occurs under the occurrence of the event H as the likelihood P(D|H). , And when the data D ₁ is obtained at the actual site, the probability of occurrence of the event H for which the data D ₁ can be obtained is the likelihood P(D ₁ |H).

For example, when the signals of the excitation force F and the sound source N and the vibration/noise S _i at the evaluation point can be directly measured in the actual operation test, first, in the learning phase, the transfer characteristic H ₁ =S _i /F, And H ₂ =S _i /N are calculated backward, and the conditional probability density distribution of the vibration/noise S _{i at} the evaluation points with respect to H ₁ and H ₂ is stored in the database as likelihood P(S _i |H ₁ ∩H ₂ ). It is converted and stored in the sound source/transfer characteristic probability distribution database 15.

Next, in the estimation phase, after obtaining the vibration/noise S _n at the evaluation point during actual operation, the likelihood P(S _n |H ₁ ∩H ₂ regarding the transfer characteristics H ₁ and H ₂ that can be observed by this data is obtained. ) Is sequentially multiplied by the number of observed signals, the transfer characteristics H ₁ and H ₂ in actual operation can be identified stochastically.

It should be noted that the sound source and the transfer characteristic are commutative in that they become the vibration and noise at the evaluation point by being multiplied. Therefore, if the above method is taken in reverse, it can be used for identifying the excitation source and sound source that are difficult to measure directly in actual operation. For example, the likelihood P(S _i |F∩N) between an arbitrary excitation force F or sound source P and the vibration/noise S _{i at} the evaluation point is acquired in a pre-shipment test, and this is stored in a database to generate a sound source. Stored in the transfer characteristic probability distribution database 15. Next, from the vibration/noise S _n at the evaluation points acquired during actual operation, the probability density P(S _n |F∩N) of the excitation force F or the sound source N that can be observed by this data is calculated from the observed signal. By sequentially multiplying by the number, it is possible to probabilistically identify the excitation force and the size of the sound source during actual operation.

In this way, by estimating the excitation force/sound source characteristics and transfer characteristics of the current model, the contribution of the evaluation point vibration/noise in the current model is calculated stochastically, and based on this, the specifications of the next model and Discuss design guidelines.

After that, at the time of the completion of the next model, in the pre-shipment test, the likelihood data of the excitation force of the next model and the transmission characteristics of the evaluation point vibration and noise with respect to the sound source were acquired. By multiplying the probabilistic identification results of sound sources, it is possible to predict the vibration and noise at the evaluation point of the next model during operation probabilistically before operation.

In FIG. 5, the sound source/transfer characteristic estimation engine 14 and the contribution rate estimation engine 18, which are the two calculation engines shown in FIG. 4, are integrated, and further, the laboratory data storage system 17 and the contribution rate display system 19 which are user interfaces are integrated. Also shows an integrated configuration. In this way, the functions such as the calculation engine and the user interface may be integrated. Further, as shown in FIG. 6, a transmission path model data storage 13 which is a data storage and a sound source/transmission characteristic probability distribution database 15 may be integrated.

FIG. 7 shows an example of a contribution ratio display method in the contribution ratio display system 19 shown in the first embodiment. As shown in the figure, the contribution rate display system 19 displays a transmission path graphical model display area 19a for displaying a transmission path model represented by a graphical model and a contribution rate two-dimensional graph display for displaying a two-dimensional graph showing frequency characteristics of the contribution rate. An area 19b and an influence degree/probability judgment element display area 19c for displaying design judgment materials are provided. When an arbitrary element is selected on the graphical model that represents the transfer path, the extent to which variations in the sound source and transfer characteristics related to that element affect the overall estimation result is displayed on a two-dimensional graph. Specifically, the upper and lower limits or the confidence intervals of the contribution ratio of vibration and noise passing through the element are displayed from the variation of the sound source and transfer characteristics related to the selected element. Based on this figure, the vibration of the selected element is displayed. It is possible to determine the effect of variations in noise characteristics on the vibration noise characteristics of the entire system, specifically, the degree of upward and downward deflection and its probability. For example, in FIG. 7, the frequency characteristic of the evaluation point noise is shown by a thick black line in the contribution rate two-dimensional graph display area 19b, the median value of the selected element contributions among these is shown by a gray line, and the variation is shown by a confidence interval. The fluctuation range of the evaluation point noise due to the uncertainty of this contribution is shown by the gray dashed line. In addition, as a result, the possibility that the evaluation point noise will move upward by 3 dB with a probability of 35% and downward by 2 dB with a probability of 25% is shown in the influence degree/probability judgment element display area 19c.

Further, by further processing such an impact degree calculation result, it is possible to provide a function for displaying structural elements that need to be examined in order to suppress vibration and noise at the evaluation point in order of priority of measures. ..

FIG. 8 shows another example of the contribution rate display method in the contribution rate display system 19 shown in the first embodiment. Similarly to the above, the contribution rate display system 19 is composed of a transfer path model represented by a graphical model and a two-dimensional graph showing the frequency characteristics of the contribution rate. Select and set this element change policy in a separate dialog that opens. Then, in the contribution ratio two-dimensional graph display area 19b, the contributions of the predetermined element before and after the structural change are displayed together with the variations, and the increase or decrease of the evaluation point noise due to this change can be discussed stochastically. For example, in FIG. 8, the frequency characteristics of the evaluation point noise before the structure change is shown by a thick gray line in the contribution rate two-dimensional graph display area 19b, and the contribution of the selected elements among these is shown by the gray thin line and other than the selected elements. The contribution is shown by the black dashed line. Although the contribution that was initially a gray thin line decreases as the structure changes, the effect of reducing that contribution is expressed with a confidence interval with the median value as the black thin line. As a result, it is expected that the evaluation point noise will be a thick black line as the sum of the contributions of the change elements indicated by the black thin line and the contributions other than the change elements indicated by the black dashed-dotted line. The subsequent evaluation point noise also has a probability distribution or confidence interval. Summarizing these results, when an element for which a structural change is planned is selected, the cost and vibration noise suppression effect of the structural change are displayed in the influence degree/probability judgment element display area 19c. Compared with the specification value of 65 dB, the cost of the structure change is 3 M¥ and the noise specification achievability is 75%.

In this way, if the effects associated with structural changes can be expressed stochastically, by adding this and the cost of structural changes, the cost-effectiveness of vibration noise reduction and policy application, which are often trade-offs, can be reduced. It enables rational discussion and formalization of design guidelines.

10: Product to be measured 11: Field data extraction system 12: Field database storage 13: Transmission path model data storage
14: Sound source/transfer characteristic estimation engine
15: Sound source/transfer characteristic probability distribution database
16: Laboratory data
16a: transfer path model 16b: sound source/transfer characteristic probability distribution data
17: Laboratory data storage system
18: Contribution rate estimation engine
19: Contribution rate display system
19a: Transmission route graphical model display area
19b: Contribution rate two-dimensional graph display area
19c: influence degree/probability judgment element display area 20: user

Claims (3)

A site data extraction system that measures site data from the products to be evaluated,
A site database storage for storing the site data,
A transmission path model data storage for storing a transmission path model,
Estimate the probability distribution data of the sound source/transfer characteristics corresponding to the elements and paths of the transfer path model in the product to be evaluated from the field data stored in the field database storage and the transfer path model stored in the transfer path model data storage Sound source/transfer characteristic estimation engine
A sound source/transfer characteristic probability distribution database storing sound source/transfer characteristic probability distribution data calculated by the sound source/transfer characteristic estimation engine;
The assumed transfer path model is stored in the transfer path model data storage, or the sound source/transfer characteristic probability distribution data obtained by measurement or numerical analysis in a laboratory is stored in the sound source/transfer characteristic probability distribution database. A laboratory data storage system, a transfer path model stored in the transfer path model data storage, a sound source/transfer characteristic probability distribution stored in the sound source/transfer characteristic probability distribution database, and a field data storage stored in the field database storage. A contribution rate estimation engine that estimates the contribution rate of vibration and noise from field data,
A contribution rate display system for displaying the contribution rate calculated by the contribution rate estimation engine;
A noise and noise problem factor estimation system comprising: Regarding the vibration noise problem factor estimation system according to claim 1,
Vibration and noise transfer path and its contribution are displayed in a graphical model.A transfer path graphical model display area is displayed, and a contribution rate two-dimensional graph display area is displayed to display a two-dimensional graph representing the frequency characteristics of the contribution rate. Then, when any element on the graphical model that represents the transmission path is selected, the degree to which the transmissibility related to that element affects the whole is displayed on the two-dimensional graph and in order to suppress vibration and noise. A vibration and noise problem factor estimation system, which has a function of displaying structural elements that need to be examined in order of priority of countermeasures. The vibration noise problem factor estimation system according to claim 1 or 2,
A vibration and noise problem factor estimation system having a function of displaying the cost and vibration and noise suppression effect of the structure change when an element for which the structure change is planned is selected.

Images