CN112097894A - Method for detecting radiation noise qualification of horizontal driver of automobile seat - Google Patents
Method for detecting radiation noise qualification of horizontal driver of automobile seat Download PDFInfo
- Publication number
- CN112097894A CN112097894A CN202010826451.6A CN202010826451A CN112097894A CN 112097894 A CN112097894 A CN 112097894A CN 202010826451 A CN202010826451 A CN 202010826451A CN 112097894 A CN112097894 A CN 112097894A
- Authority
- CN
- China
- Prior art keywords
- sound quality
- hdm
- model
- regression
- variable
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000012797 qualification Methods 0.000 title claims abstract description 37
- 230000005855 radiation Effects 0.000 title claims abstract description 9
- 238000011156 evaluation Methods 0.000 claims abstract description 80
- 238000012417 linear regression Methods 0.000 claims abstract description 48
- 238000004458 analytical method Methods 0.000 claims abstract description 22
- 238000001514 detection method Methods 0.000 claims abstract description 15
- 230000001419 dependent effect Effects 0.000 claims description 31
- 238000012360 testing method Methods 0.000 claims description 20
- 238000004364 calculation method Methods 0.000 claims description 13
- 238000000611 regression analysis Methods 0.000 claims description 11
- 238000013178 mathematical model Methods 0.000 claims description 10
- 210000005069 ears Anatomy 0.000 claims description 9
- 230000000694 effects Effects 0.000 claims description 8
- 238000002474 experimental method Methods 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000012353 t test Methods 0.000 claims description 5
- 238000013441 quality evaluation Methods 0.000 claims description 3
- 238000010219 correlation analysis Methods 0.000 claims description 2
- 230000008030 elimination Effects 0.000 claims 1
- 238000003379 elimination reaction Methods 0.000 claims 1
- 230000005236 sound signal Effects 0.000 description 12
- 238000010586 diagram Methods 0.000 description 9
- 238000007689 inspection Methods 0.000 description 7
- 238000000692 Student's t-test Methods 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000000540 analysis of variance Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000007619 statistical method Methods 0.000 description 2
- 238000001134 F-test Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
- G01H17/00—Measuring mechanical vibrations or ultrasonic, sonic or infrasonic waves, not provided for in the preceding groups
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M17/00—Testing of vehicles
- G01M17/007—Wheeled or endless-tracked vehicles
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The method for detecting the qualification of the radiation noise of the horizontal driver of the automobile seat comprises the following steps: 1) acquiring HDM near-field acoustic signals and calculating acoustic quality objective parameters; 2) constructing a multivariate linear regression model of the sound quality qualification evaluation index; the evaluation index formula is as follows: q is 31.917-1.311 multiplied by Loudness-5.84 multiplied by Roughess, wherein Loudness represents Loudness objective parameter index, Roughess represents Roughness objective parameter index, and Q is sound quality objective parameter comprehensive evaluation index obtained through cross-correlation and linear regression analysis; 3) predicting the sound quality qualification degree of the HDM sample; in the detection process, the subjective preference value is obtained by substituting loudness and roughness into a regression model; and judging whether the HDM sound quality is qualified or not according to whether the obtained comprehensive index value exceeds the range of qualified products or not.
Description
The technical field is as follows:
the invention relates to the technical fields of noise control, psychoacoustics, objective parameters of sound quality, product quality inspection, calculation methods and the like.
Background art:
as one of the key moving parts of an automobile seat position adjusting system, a Horizontal Driver (HDM) for an automobile seat is provided, and an electric motor causes impact collision at the meshing part of a worm and a worm wheel, and generates uncomfortable noise. The sound quality of the HDM product has important influence on the overall sound environment comfort of the automobile, the first subjective auditory comfort feeling of consumers is influenced to a great extent, the pricing and selling conditions of high-grade automobiles are also influenced, and the noise qualification inspection of the HDM product is an extremely important step in the whole-section delivery quality inspection production line of the HDM. Therefore, the detection and evaluation of the sound quality of the HDM are important steps for evaluating the comprehensive performance of the NVH.
At present, manufacturers mainly check whether the HDM sound quality is qualified or not by means of subjective auditory perception through a checker close to the HDM reduction box part. The method is simple and easy to implement, but has the defects of high labor cost, strong randomness, weak stability and the like. In order to overcome the defect of manual detection of HDM, the patent provides a method for evaluating the comprehensive index of the sound quality objective parameter of the HDM.
The invention patent CN201610938464.6 discloses an optimized HDM detection platform, which adopts a single sound quality objective parameter to evaluate the radiation noise of an HDM product. The method for acquiring the near-field acoustic signal containing the HDM characteristic information and the corresponding acoustic quality objective parameter information by adopting the near-field acoustic signal acquisition experiment, the acoustic signal time-frequency analysis and the acoustic quality objective parameter calculation analysis method of the low-noise product disclosed by the applicant of the invention in the patent comprises the following steps: loudness, sharpness, roughness, etc.
In order to further improve the detection accuracy, the subjective feeling of human ears is quantified by using a subjective evaluation experiment, a mathematical model describing the relation between the HDM sound quality objective quantity and the subjective feeling is established, a method for judging the qualification by using a new comprehensive index for evaluating the HDM is provided, and the qualification evaluation accuracy of the HDM radiation noise is improved.
The invention content is as follows:
the invention aims to describe the process of subjective evaluation of human ears by using the objective parameters of sound quality and apply the process to the sound quality detection and evaluation work of a horizontal driver of an automobile seat.
The invention calculates the sound quality objective parameters of the HDM sound signals by collecting the HDM sound signals, carries out cross-correlation and linear regression analysis according to the calculation results of a plurality of sound quality objective parameters, and provides a comprehensive index for carrying out sound quality objective parameter calculation and subjective evaluation analysis on the near-field sound signals obtained by detection. The index quantifies the subjective feeling of the human ear by the objective parameters of the sound quality, can better reflect the subjective feeling of the human ear on the HDM, and enables the evaluation result to be more objective and reasonable. And then judging whether the HDM sound quality is qualified or not according to whether the obtained comprehensive index value exceeds the range of qualified products or not.
The technical scheme adopted by the invention for solving the technical problem is as follows:
the method for detecting the qualification of the radiation noise of the horizontal driver of the automobile seat comprises the following steps:
1) acquiring HDM near-field acoustic signals and calculating acoustic quality objective parameters;
the method for acquiring the near-field acoustic signal containing the HDM characteristic information and the corresponding acoustic quality objective parameter information by adopting the near-field acoustic signal acquisition experiment, the acoustic signal time-frequency analysis and the acoustic quality objective parameter calculation analysis method of the low-noise product disclosed in Chinese patent publication CN2016109384646 comprises the following steps: loudness, sharpness, roughness, etc.
2) Constructing a multivariate linear regression model of the sound quality qualification evaluation index; combining the results of the subjective evaluation and the objective evaluation preliminary analysis, using a mathematical model of multiple linear regression, and carrying out correlation and regression analysis on the sound quality objective evaluation results and the subjective evaluation preference values through the tests of significance and fitting degree, and establishing the relationship between the subjective feeling of human ears on HDM and sound quality objective parameters, so that the evaluation method can combine all the sound quality objective parameters with a certain weight and is characterized as HDM sound quality qualification, and the evaluation index formula is as follows: Q-31.917-1.311X Loudness-5.84X Roughress
Loudness represents an objective parameter index of Loudness, Roughress represents an objective parameter index of Roughness, and Q is an objective parameter comprehensive evaluation index of sound quality obtained through cross-correlation and linear regression analysis.
3) The sound quality qualification degree prediction is carried out on the HDM sample. And in the detection process, the subjective preference value is obtained by substituting loudness and roughness into a regression model. And judging whether the HDM sound quality is qualified or not according to whether the obtained comprehensive index value exceeds the range of qualified products or not.
Preferably, step 2) specifically comprises:
2.1 selecting a mathematical model of multiple linear regression;
the multivariate linear regression model with the number of independent variables p can be expressed as:
in the formula,i～N(0，σ2) I.e. they are independent and identically distributed normal random variables, a, b1,b2,…,bpCalled regression coefficients, for solving the values of these regression coefficients, the least squares method is generally used, i.e. a, b, in which the sum of squared errors is minimized1,b2,…,bpAs the best estimate. That is to say that the first and second electrodes,
let us say that a and bjThe partial derivative of (a) is 0, the normal equation is arranged as follows:
if so:
wherein, b0A. Writing (1-4) in matrix form, the linear regression model can be expressed as:
Y＝Xβ+ (6)
the normal equation available from (6) is:
X'Xβ＝X'Y (7)
obtained from (7):
β＝(X'X)-1X'Y (8)
the method comprises the steps of introducing relevant independent variables and removing non-relevant independent variables by using a stepwise regression method, namely firstly performing significance regression test on a variable with the maximum partial correlation coefficient to determine whether the variable enters an equation, then taking each variable in the equation as a variable finally selected into the equation, performing partial F test, and determining whether the variable can be left in a regression model. Thus, both the introduced and rejected variables are present, and the originally rejected variables may also be introduced into the equation.
2.2 checking significance and fitting degree;
the regression model was tested for significance using the F statistic as follows:
wherein,
in the form of a regression sum of squares,
is the sum of the squares of the residuals. At a given significance level α, if F < FαThe estimated linear regression model is considered significant.
In the multiple linear regression model, to analyze whether the independent variable X has significant influence on the dependent variable Y, the empirical regression coefficient b is requirediA check is made to see if it is 0. When the regression coefficient is not 0, the degree of influence of X on the dependent variable Y is significant. After ensuring that the regression model is significant, the regression coefficients should be further checked for significance. The test statistic t for the regression coefficient for the ith variable is:
wherein, C ═ X' X)-1，CkkFor the elements of matrix C, test statistics calculated from the data can further determine how well each regression model fits.
Next, the magnitude of the correlation index of the linear regression model, i.e., R, is determined2What characterizes is the degree to which the regression model can describe the relationship of independent variables to dependent variables:
wherein R is2The larger the model, the better the fitting degree of the regression model, i.e. the better the prediction effect of the model on the relationship between independent variable and dependent variable. R represents X1The magnitude of the linear relationship to Xp.
2.3, combining subjective and objective evaluation, and obtaining a regression model of the evaluation index according to the acoustic quality objective parameters;
in order to establish a relation model between HDM sound quality psychoacoustics objective parameters and subjective evaluation results, loudness, sharpness, roughness and jitter are selected as independent variables, and x is used for the independent variables1、x2、x3、x4When the dependent variable is a subjective preference value obtained by subjective evaluation of sound quality and is represented as P, the sound quality evaluation multiple linear regression model of HDM is:
P＝a+b1x1+b2x2+b3x3+b4x4 (12)
wherein a is a constant term; b1、b2、b3、b4Are the coefficients of regression.
Substituting HDM sound quality psychoacoustic objective parameters into the independent variable x of formula (12)1、x2、x3、x4And (3) substituting the HDM sound quality subjective evaluation result into the position P to establish a relation model between the HDM sound quality subjective evaluation result and the position P. Calculating coefficient b of independent variable by stepwise regression method in linear regression method1、b2、b3、b4And the significance and the regressiveness of the model are checked, the coefficient in the model is determined to accord with the logical relation between the independent variable and the dependent variable, and the determined equation model is put into the sound quality detection of more HDM products to evaluate whether the HDM sound quality is qualified.
And (3) importing the HDM sound quality subjective evaluation result, namely the subjective preference of a subjective evaluator on each sound sample and the sound quality objective parameter data thereof into SPSS software for regression analysis, wherein the regression model adopts a step-by-step method on the basis of the existing loudness parameters, the improper independent variable is removed from the regression analysis, the sound quality objective parameter which can influence the subjective preference is left as the independent variable of the regression model, the subjective preference value is a dependent variable, and the accurate regression model is obtained. The dependent variable is a subjective preference value of each sound sample after subjective evaluation by HDM sound quality.
And selecting a model containing two sound quality objective parameters of loudness and roughness as a regression model of the HDM sound quality qualification evaluation index.
And through analysis of variance of the linear regression model, with the introduction of roughness, the mean square error of the model is continuously reduced, and the regression effect of the representation model is remarkably improved. The significance probability of the model when the F test is performed is 0(sig. ═ 0), and is smaller than α ═ 0.01. It can therefore be concluded that: the original assumption that the overall regression coefficient is 0 can be rejected so the regression model is meaningful and significant.
Loudness and roughness variable coefficients in the model are subjected to t-test, and the significance probability of the loudness and roughness variable coefficients is less than 0.05(Sig. <0.05), so that all the variable coefficients in the model have significance.
Combining the results of the above regression analysis can yield: the influence of roughness and loudness on the subjective preference value of people is obvious, and the regression coefficient value and the constant term value are substituted into a regression model equation to obtain a multiple linear regression model of the HDM sound quality qualification index Q:
Q＝31.917-1.311×x1-5.84×x3 (13)
will be provided withX of1And x3Instead of loudness and roughness values, respectively, the regression model can be written as:
Q＝31.917-1.311×Loud.-5.84×Rough. (14)
wherein, loud.
The invention has the beneficial effects that:
1) according to the method, a prediction mathematical model of HDM sound quality qualification inspection combining an HDM sound quality objective parameter and manual inspection subjective evaluation is established through a multiple linear regression method, an index for evaluating the HDM sound quality qualification is provided, the comprehensive index reflects the subjective feeling of human ears on the HDM sound quality and the main contribution of the objective sound quality parameter, and qualified products and unqualified products can be effectively distinguished.
2) The invention selects two sound quality objective parameters which can most influence the qualification of human ears on the HDM sound quality: loudness and roughness, a design that improves the acoustic quality of HDM.
Description of the drawings:
FIG. 1 is a schematic diagram of the HDM acoustic signal acquisition hardware system connection of the method of the present invention.
FIG. 2 is a diagram illustrating the comparison between the predicted value and the subjective evaluation value of the regression model of the present invention.
Fig. 3 is a flow chart of subjective and objective combination evaluation of sound quality according to the present invention.
FIGS. 4a to 4b are the sound quality calculation results of 0 to 3.0s for the qualified and unqualified products of the present invention, wherein FIG. 4a is a schematic diagram of the sound quality calculation results of 0 to 3.0s for the unqualified product, and FIG. 4b is a schematic diagram of the sound quality calculation results of 0 to 3.0s for the qualified product.
Fig. 5a to 5d are schematic diagrams illustrating the correlation between loudness, sharpness, roughness and jitter and subjective evaluation value of the present invention, wherein fig. 5a is a schematic diagram illustrating the correlation between loudness and subjective preference, fig. 5b is a schematic diagram illustrating the correlation between roughness and subjective preference, fig. 5c is a schematic diagram illustrating the correlation between sharpness and subjective preference, and fig. 5d is a schematic diagram illustrating the correlation between jitter and subjective preference.
Detailed Description
The invention is further described below with reference to the accompanying drawings.
The invention discloses a method for detecting the qualification of radiation noise of a horizontal driver of an automobile seat, which comprises the following steps:
1) acquiring HDM near-field acoustic signals and calculating acoustic quality objective parameters;
the method for acquiring near-field sound signals containing HDM characteristic information and corresponding sound quality objective parameter information by adopting a near-field sound signal acquisition experiment, sound signal time-frequency analysis and sound quality objective parameter calculation analysis method of a low-noise product disclosed in CN2016109384646 by the applicant of the invention comprises the following steps: loudness, sharpness, roughness, etc.
2) Constructing a multivariate linear regression model of the sound quality qualification evaluation index;
2.1 mathematical model of multiple linear regression
A research algorithm and an evaluation index for evaluating the HDM sound quality are established through a mathematical model of multiple linear regression, whether the HDM sound quality is qualified or not is automatically evaluated through the evaluation algorithm in the whole method, the method replaces a traditional manual testing method for the HDM sound quality, and the method is high in efficiency, low in cost and relatively reliable.
The multivariate linear regression analysis method is a mathematical statistical method for processing the interdependence relation among a plurality of variables, not only can analyze the influence of independent variables on dependent variables, but also can predict and control the dependent variables through a regression model. By means of a multiple linear regression method, the evaluation value obtained by subjecting each HDM sound sample to sound quality subjective evaluation, namely the subjective preference value of human ears is used as a dependent variable, the sound quality objective parameter value is used as an independent variable, the influence of the sound quality objective parameter on the HDM sound quality subjective evaluation value is analyzed, a linear regression model is formed, the sound quality qualification of the HDM is predicted according to the model, and the aim of completing HDM sound quality inspection work is achieved.
The multivariate linear regression model with the number of independent variables p can be expressed as:
in the formula,i～N(0，σ2) I.e. they are independent and identically distributed normal random variables, a, b1,b2,…,bpCalled regression coefficients, for solving the values of these regression coefficients, the least squares method is generally used, i.e. a, b, in which the sum of squared errors is minimized1,b2,…,bpAs the best estimate. That is to say that the first and second electrodes,
let us say that a and bjThe partial derivative of (a) is 0, the normal equation is arranged as follows:
if so:
wherein, b0A. Writing (1-4) in matrix form, the linear regression model can be expressed as:
Y＝Xβ+ (6)
the normal equation available from (6) is:
X'Xβ＝X'Y (7)
obtained from (7):
β＝(X'X)-1X'Y (8)
in a regression model, not all the independent variables introduced will have a large impact on the dependent variables. In the method, a stepwise regression method is adopted to introduce relevant independent variables and remove non-relevant independent variables, namely, significance regression test is firstly carried out on the variable with the maximum partial correlation coefficient to determine whether the variable enters an equation, then each variable in the equation is used as the variable finally selected into the equation to carry out partial F test, and whether the variable can be left in a regression model is determined. Thus, both the introduced and rejected variables are present, and the originally rejected variables may also be introduced into the equation.
2.2 testing of significance and Fit
In the multiple linear regression model study for practical problems, it is impossible to determine in advance whether or not there is a linear relationship between the dependent variable Y and the plurality of independent variables X. Therefore, after the multiple linear regression model is obtained, F-test is performed on the significance of the equation to determine whether the relationship between the independent variable and the dependent variable is significant. The regression model may be tested for significance using the F statistic as follows:
wherein,
in the form of a regression sum of squares,
is the sum of the squares of the residuals. At a given significance level α, if F < FαThe estimated linear regression model is considered significant.
In the multiple linear regression model, to analyze whether the independent variable X has significant influence on the dependent variable Y, the empirical regression coefficient b is requirediA check is made to see if it is 0. When the regression coefficient is not 0, the degree of influence of X on the dependent variable Y is significant. After ensuring that the regression model is significant, the regression coefficients should be further checked for significance. The test statistic t for the regression coefficient for the ith variable is:
wherein, C ═ X' X)-1，CkkIs an element of the matrix C, formed by dataThe calculated test statistic can further judge the fit of each regression model.
Next, the magnitude of the correlation index of the linear regression model, i.e., R, is determined2What characterizes is the degree to which the regression model can describe the relationship of independent variables to dependent variables:
wherein R is2The larger the model, the better the fitting degree of the regression model, i.e. the better the prediction effect of the model on the relationship between independent variable and dependent variable. R represents X1The magnitude of the linear relationship to Xp.
2.3, combining subjective and objective evaluation, and obtaining a regression model of the evaluation index according to the acoustic quality objective parameters;
the invention aims to complete the detection and evaluation of the HDM sound quality, and an evaluation algorithm of the method calculates an evaluation index value according to a sound signal to judge whether the HDM sound quality is qualified.
In order to establish a relation model between HDM sound quality psychoacoustics objective parameters and subjective evaluation results, loudness, sharpness, roughness and jitter are selected as independent variables, and x is used for the independent variables1、x2、x3、x4When the dependent variable is a subjective preference value obtained by subjective evaluation of sound quality and is represented as P, the sound quality evaluation multiple linear regression model of HDM is:
P＝a+b1x1+b2x2+b3x3+b4x4 (12)
wherein a is a constant term; b1、b2、b3、b4Are the coefficients of regression.
The research idea of the evaluation algorithm is to carry out HDM sound quality subjective and objective evaluation and substitute HDM sound quality psychoacoustic objective parameters into independent variables x in formula (12)1、x2、x3、x4And (3) substituting the HDM sound quality subjective evaluation result into the position P to establish a relation model between the HDM sound quality subjective evaluation result and the position P. Tong (Chinese character of 'tong')Calculating to obtain coefficient b of independent variable by stepwise regression method in linear regression method1、b2、b3、b4And the significance and the regressiveness of the model are checked, the coefficient in the model is determined to accord with the logical relation between the independent variable and the dependent variable, and the determined equation model is put into the sound quality detection of more HDM products, so that the purpose of evaluating whether the sound quality of the HDM is qualified is achieved.
And (3) importing the HDM sound quality subjective evaluation result, namely the subjective preference of a subjective evaluator on each sound sample and the sound quality objective parameter data thereof into SPSS software for regression analysis, wherein the regression model adopts a step-by-step method on the basis of the existing loudness parameters, the improper independent variable is removed from the regression analysis, the sound quality objective parameter which can influence the subjective preference is left as the independent variable of the regression model, the subjective preference value is a dependent variable, and the accurate regression model is obtained. The dependent variable is a subjective preference value of each sound sample after subjective evaluation by HDM sound quality.
As can be seen from fig. 5c and 5d, the degree of partial correlation between the sharpness and the jitter and the subjective preference of the human ear is low, and the degree of visible correlation is small; in addition, after the t-test of each objective parameter of acoustic quality, only the test result of roughness was significant (sig ≦ α ≦ 0.05), representing that the t-test at a level of 0.05, and the result was significant. Therefore, the roughness enters the regression model, and the sharpness and the jitter are eliminated.
Introducing roughness, correlation coefficient (R) and determination coefficient (R) of regression model2) And determining the adjustment value of the coefficient (adjusting R)2) Are all improved, R20.865 indicates that there is a 86.5% probability that the relationship between the independent variable and the dependent variable may fall into the prediction model, indicating that the regression model has a better fit and better represents the relationship between the subjective preference P of the human ear for HDM and the objective parameter of sound quality. The introduction of roughness has a significant effect on the integrity of the regression model, and the standard deviation of the estimated value is continuously reduced. The variation of regression model parameters shows that the introduction of roughness greatly contributes to the regression degree of the model, so that the method selects the sound quality including loudness and roughnessThe model of the objective parameters is used as a regression model of the HDM sound quality qualification evaluation index.
And through analysis of variance of the linear regression model, with the introduction of roughness, the mean square error of the model is continuously reduced, and the regression effect of the representation model is remarkably improved. The significance probability of the model when the F test is performed is 0(sig. ═ 0), and is smaller than α ═ 0.01. It can therefore be concluded that: the original assumption that the overall regression coefficient is 0 can be rejected so the regression model is meaningful and significant.
Loudness and roughness variable coefficients in the model are subjected to t-test, and the significance probability of the loudness and roughness variable coefficients is less than 0.05(Sig. <0.05), so that all the variable coefficients in the model have significance.
Combining the results of the above regression analysis can yield: the influence of roughness and loudness on the subjective preference value of people is obvious, and the regression coefficient value and the constant term value are substituted into a regression model equation to obtain a multiple linear regression model of the HDM sound quality qualification index Q:
Q＝31.917-1.311×x1-5.84×x3 (13)
x therein is1And x3Instead of loudness and roughness values, respectively, the regression model can be written as:
Q＝31.917-1.311×Loud.-5.84×Rough. (14)
wherein, loud.
Therefore, the qualification of the HDM sound quality is mainly affected by the loudness and roughness of the objective sound quality parameter. The regression model obtained by data analysis shows the relationship between the subjective feeling (namely subjective preference value) of human ears on the HDM sound quality and the sound quality objective parameter, and the coefficients in front of loudness and roughness in the regression model represent the influence of the coefficients on the subjective preference value.
3) Judging whether the sound quality is qualified or not according to the HDM sound quality qualification index;
by collecting HDM sound signals, calculating loudness and roughness value in the sound quality objective parameters, substituting the loudness and roughness value into the regression model, predicting the sound quality qualification degree of the HDM sample, and judging whether the HDM sound quality is qualified according to whether the obtained comprehensive index value exceeds the range of qualified products.
In this embodiment, (1) the hardware measurement system setup of this experiment was composed of the following components:
(a) an HDM clamping platform;
(b) two motors (the rotating speed is 3000r/s, and one of the motors is standby);
(c) a direct current power supply of the motor;
(d) a B & K1/4 inch microphone (model 4958);
(e) an NI PIXe industrial personal computer (with an built-in NI PXIe-8135 embedded controller);
(f) NI 4496-1 data acquisition card;
(g) a display;
(h) a B & K1/4 inch microphone holder structure.
The specific layout is shown in figure 1.
(2) Substituting the objective parameters of the sound quality into a regression model (14) obtained by statistical analysis, calculating the sound quality qualification degree, and comparing the qualification degree with an actual subjective preference value, wherein the comparison chart is shown in figure 2: in fig. 2, the abscissa is the sample number of 18 HDM, the ordinate is the subjective preference value, wherein the blue line portion is the actual subjective preference evaluation value of 18 samples, the orange line portion is the subjective preference value predicted by the regression model obtained in the foregoing, and the green line is the difference between the regression model prediction value and the subjective evaluation value, that is, the absolute error. The actual evaluation curve in the graph is compared with the prediction curve of the regression model, the two numerical values are very similar, the fitting effect of the regression model is better, the regression phenomenon is obvious, and the model can represent the relationship between the subjective preference value of human ears on the HDM sound quality and the sound quality objective parameter, namely the loudness and the roughness. In the figure, the red dotted line represents the subjective preference qualified threshold, the upper part of the red line is a qualified product area, and the lower part of the red line is an unqualified product area.
Therefore, the qualified product threshold value of the HDM subjective preference value is set to 8.8, and in the detection process, the qualified product is judged to be a qualified product if the subjective preference value calculated by substituting the loudness and the roughness into the regression model is lower than 8.8.
(3) And subsequently, performing fitting analysis on the regression model, collecting more sound signals of the HDM sample, calculating objective parameters of sound quality of the HDM sample, substituting the objective parameters into the regression model, and using a subjective preference value obtained by calculation of the regression model to check whether the HDM is qualified.
The objective parameters of sound quality such as loudness, sharpness, roughness and jitter are compared with the subjective evaluation value, the subjective preference value is multiplied by-1 for comparison because the objective parameters of sound quality and the subjective preference are in a negative correlation relationship, and the jitter value is over-small and mostly about 0.02, so that 100 times of the jitter value is compared with the HDM subjective preference value. The results are shown in fig. 5a to 5d, so that the degree of correlation between the known loudness value and the known roughness value and the subjective evaluation preference value is high, the degree of correlation between the sharpness and the jitter and the subjective preference value is low, and the effectiveness and the accuracy of the regression model are verified.
The method comprises the following specific implementation steps:
(1) connecting a display and an NI PXIe industrial personal computer, starting a power supply of the display and the NI industrial personal computer, supporting hot plugging of the NI industrial personal computer, connecting a data connecting line and a B & K4958 microphone with a PXI1 slot5/ai1 channel of the industrial personal computer after starting, sending a single-frequency sound signal, starting to collect data, observing signal characteristics collected by the microphone, checking whether the microphone and a data collection program have problems, and if the microphone and the data collection program have the problems, carrying out the next step;
(2) arranging a microphone: the distance is 2.5-3 mm (red arrow in the figure), the 4958 microphone is clamped on a support, the support is fixed on the left side of an HDM clamping platform, a mark is made, the position of the whole support is ensured to be unchanged, and the distance between the microphone and the HDM reduction gearbox is further ensured to be unchanged;
(3) turning on the direct current-alternating current conversion power supply to confirm that the motor can work normally;
(4) after the position of the microphone is arranged, the motor and the HDM are installed on the clamping platform, before HDM sound pressure data are formally collected, the motor is started, the HDM is operated in a trial mode, the situation that the part of the HDM reduction gearbox is clamped without a loose part is ensured, the tail end of the screw rod is fixed with the rubber ring, and unnecessary structural noise is avoided;
(5) and after the steps are ensured to have no problem, carrying out sound pressure data acquisition work of the HDM. The acquisition time is 15s, the sampling frequency is 12800Hz, a data storage path is set, and the acquisition of the sound pressure signal is started.
(6) The method for acquiring the near-field acoustic signal containing the HDM characteristic information and the corresponding acoustic quality objective parameter information by adopting the near-field acoustic signal acquisition experiment, the acoustic signal time-frequency analysis and the acoustic quality objective parameter calculation analysis method of the low-noise product disclosed by the applicant of the invention in the CN2016109384646 patent comprises the following steps: loudness, sharpness, roughness, etc.
(7) And (4) carrying out subjective and objective joint evaluation on the HDM sound quality. The flow chart of subjective and objective combination evaluation is shown in fig. 3. The objective evaluation comprises HDM sound signal acquisition experiments, sound signal time-frequency analysis and sound quality objective parameter calculation and analysis. The subjective evaluation comprises evaluation scheme design, evaluation result summarization and data reliability and validity check.
(8) And establishing a relation between the HDM sound quality subjective evaluation result and the sound quality objective parameter through a multiple linear regression mathematical model, and establishing a sound quality qualification judgment index. And determining a threshold value of the qualification range of the HDM sound quality qualification index Q.
(9) And (4) utilizing an HDM sound quality qualification evaluation mathematical model to carry out the inspection of the HDM sample. In the detection process, judging that the subjective preference value obtained by substituting loudness and roughness into the regression model is lower than a threshold value as an unqualified product, and otherwise, judging that the subjective preference value is qualified. And judging whether the HDM sound quality is qualified or not according to whether the obtained comprehensive index value exceeds the range of qualified products or not.
Claims (2)
1. The method for detecting the qualification of the radiation noise of the horizontal driver of the automobile seat comprises the following steps:
1) acquiring HDM near-field acoustic signals and calculating acoustic quality objective parameters;
the method for acquiring the near-field acoustic signal containing the HDM characteristic information and the corresponding acoustic quality objective parameter information by adopting the near-field acoustic signal acquisition experiment, the acoustic signal time-frequency analysis and the acoustic quality objective parameter calculation analysis method of the low-noise product disclosed in Chinese patent publication CN2016109384646 comprises the following steps: loudness, sharpness, roughness;
2) constructing a multivariate linear regression model of the sound quality qualification evaluation index; combining the results of the subjective evaluation and the objective evaluation preliminary analysis, using a mathematical model of multiple linear regression, and carrying out correlation and regression analysis on the sound quality objective evaluation results and the subjective evaluation preference values through the tests of significance and fitting degree, and establishing the relationship between the subjective feeling of human ears on HDM and sound quality objective parameters, so that the evaluation method can combine all the sound quality objective parameters with a certain weight and is characterized as HDM sound quality qualification, and the evaluation index formula is as follows: Q-31.917-1.311X Loudness-5.84X Roughress
Loudness represents an objective parameter index of Loudness, Roughress represents an objective parameter index of Roughness, and Q is an objective parameter comprehensive evaluation index of sound quality obtained through cross-correlation and linear regression analysis;
3) predicting the sound quality qualification degree of the HDM sample; in the detection process, the subjective preference value is obtained by substituting loudness and roughness into a regression model; and judging whether the HDM sound quality is qualified or not according to whether the obtained comprehensive index value exceeds the range of qualified products or not.
2. The method of claim 1 for detecting the acceptability of radiation noise for a horizontal driver of a vehicle seat, wherein: the step 2) specifically comprises the following steps:
2.1 selecting a mathematical model of multiple linear regression;
the multivariate linear regression model with the number of independent variables p can be expressed as:
in the formula,i～N(0,σ2) I.e. they are independent and identically distributed normal random variables, a, b1,b2,…,bpCalled regression coefficients, for solving the values of these regression coefficients, the least squares method is generally used, i.e. a, b, in which the sum of squared errors is minimized1,b2,…,bpAs the best estimate; that is to say that the first and second electrodes,
let us say that a and bjThe partial derivative of (a) is 0, the normal equation is arranged as follows:
if so:
wherein, b0A; writing (1-4) in matrix form, the linear regression model can be expressed as:
Y＝Xβ+ (6)
the normal equation available from (6) is:
X'Xβ＝X'Y (7)
obtained from (7):
β＝(X'X)-1X'Y (8)
adopting a stepwise regression method to carry out relevant independent variable introduction and non-relevant independent variable elimination, namely firstly carrying out significance regression test on the variable with the maximum partial correlation coefficient to determine whether the variable enters an equation, then taking each variable in the equation as the variable finally selected into the equation and carrying out partial F test to determine whether the variable can be left in a regression model; thus, the introduced variable and the removed variable exist, and the originally removed variable can be introduced into the equation;
2.2 checking significance and fitting degree;
the regression model was tested for significance using the F statistic as follows:
wherein,
in the form of a regression sum of squares,
is the sum of the squares of the residuals; at a given significance level α, if F < FαThen the estimated linear regression model is considered significant;
in the multiple linear regression model, to analyze whether the independent variable X has significant influence on the dependent variable Y, the empirical regression coefficient b is requirediChecking whether the value is 0; when the regression coefficient is not 0, the influence degree of X on the dependent variable Y is obvious; after ensuring the regression model is significant, the significance of the regression coefficients should be further checked; the test statistic t for the regression coefficient for the ith variable is:
wherein, C ═ X' X)-1，CkkThe element of the matrix C is used, and the fitting quality of each regression model can be further judged according to the test statistic calculated by data;
next, the magnitude of the correlation index of the linear regression model, i.e., R, is determined2Characterized by the regression model describing the relationship between independent variables and dependent variablesDegree:
wherein R is2The larger the model is, the better the fitting degree of the regression model is, namely the better the prediction effect of the model on the relationship between the independent variable and the dependent variable is; r represents X1The magnitude of the linear relationship to Xp;
2.3, combining subjective and objective evaluation, and obtaining a regression model of the evaluation index according to the acoustic quality objective parameters;
in order to establish a relation model between HDM sound quality psychoacoustics objective parameters and subjective evaluation results, loudness, sharpness, roughness and jitter are selected as independent variables, and x is used for the independent variables1、x2、x3、x4When the dependent variable is a subjective preference value obtained by subjective evaluation of sound quality and is represented as P, the sound quality evaluation multiple linear regression model of HDM is:
P＝a+b1x1+b2x2+b3x3+b4x4 (12)
wherein a is a constant term; b1、b2、b3、b4Is a regression coefficient;
substituting HDM sound quality psychoacoustic objective parameters into the independent variable x of formula (12)1、x2、x3、x4Substituting the HDM sound quality subjective evaluation result into the position P, and establishing a relation model between the HDM sound quality subjective evaluation result and the position P; calculating coefficient b of independent variable by stepwise regression method in linear regression method1、b2、b3、b4Checking the significance and the regressiveness of the model, determining that the coefficient in the model accords with the logical relationship between the independent variable and the dependent variable, and putting the determined equation model into sound quality detection of more HDM products to evaluate whether the HDM sound quality is qualified;
the method comprises the steps that the subjective evaluation result of the HDM sound quality, namely the subjective preference of a subjective evaluator on each sound sample and sound quality objective parameter data of the subjective evaluator are led into SPSS software for regression analysis, the regression model is based on the existing loudness parameters, the current regression analysis adopts a step-by-step method to remove improper independent variables, sound quality objective parameters which can influence the subjective preference are left as the independent variables of the regression model, the subjective preference value is a dependent variable, and an accurate regression model is obtained; the dependent variable is a subjective preference value of each sound sample after the HDM sound quality is subjectively evaluated;
selecting a model containing two sound quality objective parameters of loudness and roughness as a regression model of the HDM sound quality qualification evaluation index;
the variance analysis of the linear regression model, along with the introduction of roughness, the mean square error of the model is continuously reduced, and the significance of the regression effect of the representation model is improved; in addition, the significance probability of the model when the F test is performed is 0(sig. ═ 0), and is smaller than α ═ 0.01; it can therefore be concluded that: the original assumption that the overall regression coefficient is 0 can be rejected, so the regression model is meaningful and significant;
loudness and roughness variable coefficients in the model are subjected to t test, and the significance probability of the loudness and roughness variable coefficients is less than 0.05(Sig. <0.05), so that all the variable coefficients in the model have significance;
combining the results of the above regression analysis can yield: the influence of roughness and loudness on the subjective preference value of people is obvious, and the regression coefficient value and the constant term value are substituted into a regression model equation to obtain a multiple linear regression model of the HDM sound quality qualification index Q:
Q＝31.917-1.311×x1-5.84×x3 (13)
x therein is1And x3Instead of loudness and roughness values, respectively, the regression model can be written as:
Q＝31.917-1.311×Loud.-5.84×Rough. (14)
wherein, loud.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010826451.6A CN112097894A (en) | 2020-08-17 | 2020-08-17 | Method for detecting radiation noise qualification of horizontal driver of automobile seat |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010826451.6A CN112097894A (en) | 2020-08-17 | 2020-08-17 | Method for detecting radiation noise qualification of horizontal driver of automobile seat |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112097894A true CN112097894A (en) | 2020-12-18 |
Family
ID=73754243
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010826451.6A Pending CN112097894A (en) | 2020-08-17 | 2020-08-17 | Method for detecting radiation noise qualification of horizontal driver of automobile seat |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112097894A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113361874A (en) * | 2021-05-24 | 2021-09-07 | 纳铁福传动***(平湖)有限公司 | Automobile part EOL test evaluation method |
| CN114383855A (en) * | 2021-11-29 | 2022-04-22 | 江铃汽车股份有限公司 | Method and device for detecting sound quality of electric seat, storage medium and electronic equipment |
| CN117951545A (en) * | 2024-03-25 | 2024-04-30 | 山东科技大学 | Method for quickly and accurately identifying correlation of red noise signals |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101672690A (en) * | 2009-09-27 | 2010-03-17 | 吉林大学 | Method for objectively and quantifiably evaluating noise fret degree in vehicle based on auditory model |
| JP2012047483A (en) * | 2010-08-24 | 2012-03-08 | Railway Technical Research Institute | Evaluation method for noise in railway vehicle |
| CN103471709A (en) * | 2013-09-17 | 2013-12-25 | 吉林大学 | Method for predicting noise quality of noise inside passenger vehicle |
| CN105021275A (en) * | 2015-07-15 | 2015-11-04 | 吉林大学 | Time domain dynamic characteristic analysis-based intra-vehicle sound quality prediction method |
| CN106568501A (en) * | 2016-10-25 | 2017-04-19 | 浙江工业大学 | Low-noise product sound quality objective parameter near-field detection method |
| CN110737970A (en) * | 2019-09-24 | 2020-01-31 | 天津大学 | engine acceleration sound quality evaluation method |
-
2020
- 2020-08-17 CN CN202010826451.6A patent/CN112097894A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101672690A (en) * | 2009-09-27 | 2010-03-17 | 吉林大学 | Method for objectively and quantifiably evaluating noise fret degree in vehicle based on auditory model |
| JP2012047483A (en) * | 2010-08-24 | 2012-03-08 | Railway Technical Research Institute | Evaluation method for noise in railway vehicle |
| CN103471709A (en) * | 2013-09-17 | 2013-12-25 | 吉林大学 | Method for predicting noise quality of noise inside passenger vehicle |
| CN105021275A (en) * | 2015-07-15 | 2015-11-04 | 吉林大学 | Time domain dynamic characteristic analysis-based intra-vehicle sound quality prediction method |
| CN106568501A (en) * | 2016-10-25 | 2017-04-19 | 浙江工业大学 | Low-noise product sound quality objective parameter near-field detection method |
| CN110737970A (en) * | 2019-09-24 | 2020-01-31 | 天津大学 | engine acceleration sound quality evaluation method |
Non-Patent Citations (1)
| Title |
|---|
| 郭栋 等: "变速器敲击噪声心理学评价研究", 《噪声与振动控制》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113361874A (en) * | 2021-05-24 | 2021-09-07 | 纳铁福传动***(平湖)有限公司 | Automobile part EOL test evaluation method |
| CN114383855A (en) * | 2021-11-29 | 2022-04-22 | 江铃汽车股份有限公司 | Method and device for detecting sound quality of electric seat, storage medium and electronic equipment |
| CN117951545A (en) * | 2024-03-25 | 2024-04-30 | 山东科技大学 | Method for quickly and accurately identifying correlation of red noise signals |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112097894A (en) | Method for detecting radiation noise qualification of horizontal driver of automobile seat | |
| CN111275307B (en) | Quality control method for high-frequency continuous observation data of automatic online station for water quality | |
| CN108663115A (en) | A kind of car inside idle noise objective quantification evaluation method | |
| CN106547265B (en) | A kind of live reliability estimation method and system of rail traffic electronic control unit | |
| CN111983020B (en) | System and method for detecting and identifying internal defects of metal component through knocking | |
| CN106568501B (en) | Near-field detection method for sound quality objective parameters of low-noise product | |
| JP2000171529A5 (en) | ||
| CN111177655B (en) | Data processing method and device and electronic equipment | |
| CN115577325A (en) | Method for evaluating sound quality and determining limit value of electric drive assembly | |
| Blommer et al. | Sound quality metric development for wind buffeting and gusting noise | |
| CN117233347B (en) | Carbon steel spheroidization grade measuring method, system and equipment | |
| TW200512467A (en) | System and method for classifying defects in and identifying process problems for an electrical circuit | |
| CN111798109B (en) | Material friction abnormal sound matching method | |
| CN111898924B (en) | Comprehensive evaluation method and system for equipment testability use | |
| CN116046415B (en) | Whole vehicle functional part reliability bench verification method, system and whole vehicle verification bench | |
| CN109029711B (en) | Dynamic bridge structure multi-order frequency identification method | |
| Cho et al. | Quality evaluation of car window motors using sound quality metrics | |
| CN115307721A (en) | Method, device and equipment for evaluating quality of automobile acceleration sound and storage medium | |
| CN115421045A (en) | Method and device for analyzing influence factors of battery health degree | |
| CN113567146A (en) | Method for evaluating road noise based on masking effect | |
| CN114299907B (en) | Abnormal sound detection method for shock absorber assembly | |
| CN104794290B (en) | A kind of modal parameter automatic identifying method for mechanized equipment structure | |
| CN114544194B (en) | Vehicle road noise evaluation method based on spectrum analysis | |
| CN115855658B (en) | Performance evaluation system applied to hydraulic brake hose product | |
| CN117705941A (en) | Stone relic damage detection and prediction method, device and storable medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201218 |