CN110737970A - engine acceleration sound quality evaluation method - Google Patents

Abstract

The invention relates to an engine accelerated sound quality evaluation method which comprises the following steps of (1) establishing an engine sound sample database, (2) carrying out consistency screening, (3) extracting sound quality objective parameters by using LMS and MATLAB software, wherein the objective parameters are represented by two-level evaluation indexes, (4) carrying out accelerated sound quality subjective and objective modeling by using a multiple linear regression method, analyzing the influence weight of different objective parameters on subjective evaluation, and (5) according to the weight result of the objective parameters in the step (4), reserving objective evaluation parameters with larger influence, deleting parameters with small influence, carrying out modeling again, and carrying out model correction.

Description

engine acceleration sound quality evaluation method

Technical Field

The invention relates to an novel engine acceleration sound quality evaluation method.

Background

With state and industry legislationThe requirements for vehicle noise are becoming more stringent and the demands on vehicle noise levels and sound quality are also becoming higher and higher for consumers. Engine noise is the main source of noise of the whole vehicle^[1]It can be said that the noise index, the dynamic performance, the reliability, and the economy of the engine are equally important. The traditional engine noise control method is used for researching how to control the noise of the engine, and in recent years, research on the noise quality of the engine has become a key field in research on NVH performance of the engine. Previous engine sound quality research is mostly evaluated and predicted around steady-state working conditions^[2-3]But in practice the engine is operating for 30-70% of its time in an unsteady state, which includes: starting, accelerating, decelerating and flameout, wherein the accelerating noise ratio is large, the time-frequency characteristics are complex and changeable, and the subjective feeling influence on people is large, so that the more in-depth research on the accelerating sound quality of the engine is the key for improving the sound quality of the engine.

Since objective evaluation parameters of sound quality have been researched and developed for a long time, an objective evaluation system consisting of sound pressure level, loudness, sharpness, roughness, jitter and the like is basically formed, and in the past research, researchers have generally adopted these parameters to objectively evaluate sound samples^[4-5]. The objective parameters are applied to the objective evaluation process of the steady-state noise, so that a sound quality prediction model can be established accurately, and the method plays a key role in sound quality research. However, with the gradual and deep research on the sound quality, the research on the acceleration sound quality of the engine is increasing, and researchers find that the previous objective evaluation parameter system of the sound quality is not enough to accurately and objectively evaluate the acceleration noise sample of the engine, which is a disadvantage.

Under the condition of acceleration, the rotating speed of the engine is in the continuous changing process, so that the objective evaluation parameter value of the engine is continuously changed, the acceleration speed is different, and the speed and the trend of the change of the objective evaluation parameter are different.

Reference to the literature

[1]Liu H,Zhang J,Guo P,et al.Sound quality prediction for engine-radiated noise[J].Mechanical Systems&Signal Processing,2015,56-57:277-287.

[2]Bi F,Li L,Zhang J,et al.Sound qualityprediction for diesel engineradiated noise based on EEMD-HT and LSSVM[J].Journal of Tianjin University(Science and Technology),2017:28-34.

[3]Mao J,Hao Z Y,Jing G X,et al.Sound quality improvement for a four-cylinder diesel engine bythe block structure optimization[J].AppliedAcoustics,2013,74(1):150-159.

[4] Bang, diesel engine radiated noise quality study [ D ]. tianjin university, 2013.

[5] Ni key diesel engine radiated noise analysis and subjective and objective assessment study of sound quality [ D ].2012.

Disclosure of Invention

The invention aims to provide novel engine accelerated sound quality evaluation methods which are more reasonable, more comprehensive and more systematic.

The technical scheme is as follows:

A method for evaluating the quality of accelerated sound of engine includes the following steps:

(1) an engine sound sample database is established, and the method comprises the following steps:

step 1: collecting noise samples of the engine under multiple working conditions and at multiple measuring points;

step 2: shearing and preprocessing the collected sound sample, and establishing a sample database;

(2) based on the sound sample database in the step (1), organizing the tested personnel to carry out subjective listening and reviewing tests by adopting a pairwise comparison method, carrying out consistency screening on listening and reviewing results, eliminating data with large deviation, and expressing consistency indexes by using contact ratio C, wherein the calculation formula is as follows:

wherein, a_iIs a subjective evaluation preference value of the sound sample,

the average value of the preference degrees of the sound samples is obtained, n is the number of subjective evaluation results of each sound sample, P is the average absolute deviation value, Q is the preference degree value range, and C represents the contact ratio;

(3) based on the sound sample database in the step (1), LMS and MATLAB software are used for extracting sound quality objective parameters, and the objective parameters are represented by two-level evaluation indexes, wherein the method comprises the following steps:

step 1: based on LMS software, extracting the sound pressure level, loudness, sharpness, roughness, jitter degree and semantic definition of a sound sample as a level 1 evaluation index;

step 2: based on the objective parameters extracted in the step 1 at different times, MATLAB software is adopted to calculate the sound pressure level change rate, the loudness change rate, the sharpness change rate, the roughness change rate, the jitter change rate and the semantic definition change rate as the 2 nd-level evaluation indexes;

(4) taking the two-stage evaluation parameters of the objective sound quality extracted in the step (3) as input, taking the subjective evaluation result in the step (2) as output, performing subjective and objective modeling of the accelerated sound quality by adopting a multivariate linear regression method, and analyzing the influence weight of different objective parameters on subjective evaluation;

(5) and (4) according to the weight result of the objective parameters in the step (4), reserving objective evaluation parameters with large influence, deleting parameters with small influence, modeling again and performing model correction.

Drawings

FIG. 1 novel engine acceleration sound quality evaluation flow

FIG. 2 Rate of change of loudness with time

Detailed Description

The invention relates to a method for evaluating the accelerated sound quality of an engine, which comprises the steps of testing noise data of the engine under various accelerated working conditions, taking loudness, sharpness, roughness, jitter degree, semantic definition and the like as primary objective evaluation indexes, taking the change rate of the primary indexes along with time as a secondary evaluation index, constructing a two-stage sound quality evaluation index system for sound quality objective evaluation, carrying out subjective evaluation on accelerated sound samples by a comparison method in pairs, establishing an engine accelerated sound quality subjective and objective evaluation prediction model by a multiple linear regression method, and analyzing the influence weight of each objective evaluation index on subjective preference, wherein novel engine accelerated sound quality evaluation processes are shown in figure 1, and the method comprises the following specific steps:

(1) establishing an engine sound sample database

The engine noise sampling test is carried out in a precise and semi-free sound field noise laboratory, the noise of air inlet and exhaust of the engine is led out of a room according to international standard ISO6798:1995, a fan is disassembled, the measuring points of a sound sensor are arranged according to GB/T1859-2000, and the multi-working-condition and multi-measuring-point noise samples of the engine are collected.

The collected sound samples are cut and preprocessed, a sample database is established, the requirements of the sound samples are 1) the cut is not suitable for being too long or too short, the length is 3-8s like , and 2) the sound samples are guaranteed to be WAV waveform files and cannot be stored into files such as MP3, and the sound distortion caused by signal conversion is avoided.

(2) Subjective evaluation of acceleration noise

Based on the sound sample database in the step (1), organizing the tested personnel to carry out subjective listening and examining tests by adopting a pair-wise comparison method, carrying out consistency screening on the listening and examining results, and eliminating data with larger deviation. consistency indexes are represented by contact ratio C, and the calculation formula is as follows:

wherein, a_iIs a subjective evaluation preference value of the sound sample,the average value of the preference degrees of the sound samples is shown, n is the number of subjective evaluation results of each sound sample, P is the average absolute deviation value, Q is the preference degree value range, and C represents the contact ratio.

(3) Objective evaluation parameter of accelerated noise

And (2) based on the sound sample database in the step (1), performing sound quality objective parameter extraction by using LMS (least mean square) and MATLAB (matrix laboratory) software, wherein the objective parameters are represented by two-level evaluation indexes. Feature extraction is performed based on LMS software, and Sound Pressure Level (SPL), Loudness (Loudness), Sharpness (Sharpness), Roughness (Roughness), jitter (fluctuationStrength) and semantic clarity (AI) are used as evaluation indexes of level 1. Based on the objective parameters of the 1 st level evaluation, MATLAB software is adopted to calculate the sound pressure level change rate, the loudness change rate, the sharpness change rate, the roughness change rate, the jitter change rate and the semantic definition change rate as the 2 nd level evaluation indexes, as shown in FIG. 2.

(4) Establishment of prediction model

The method comprises the steps of (1) taking two-stage evaluation parameters of the objective sound quality extracted in the step (3) as input, taking the subjective evaluation result in the step (2) as output, and establishing an accelerated sound quality subjective and objective prediction model, wherein the construction principle of the multiple linear regression model is simple, too many parameters do not need to be set, and meanwhile, coefficients of every objective evaluation indexes can be definitely obtained, so that the influence weight of the objective evaluation indexes can be analyzed.

(5) Optimized adjustment of models

And (4) according to the weight result of the objective parameters in the step (4), reserving objective evaluation parameters with large influence, eliminating parameters with small influence, and obtaining the most accurate prediction result by repeatedly adjusting the type of the objective parameters input by the model, thereby completing the optimization adjustment of the prediction model. In tuning the model, a balance between the accuracy and generalization ability of the predictive model is noted.

Claims (1)

1, A method for evaluating the quality of engine accelerating sound, comprising the following steps:

(1) an engine sound sample database is established, and the method comprises the following steps:

step 1: collecting noise samples of the engine under multiple working conditions and at multiple measuring points;

step 2: and shearing and preprocessing the collected sound sample, and establishing a sample database.

(2) Based on the sound sample database in the step (1), organizing the tested personnel to carry out subjective listening and reviewing tests by adopting a pairwise comparison method, carrying out consistency screening on listening and reviewing results, eliminating data with large deviation, and expressing consistency indexes by using contact ratio C, wherein the calculation formula is as follows:

wherein, a_iIs a subjective evaluation preference value of the sound sample,the average value of the preference degrees of the sound samples is obtained, n is the number of subjective evaluation results of each sound sample, P is the average absolute deviation value, Q is the preference degree value range, and C represents the contact ratio;

(3) based on the sound sample database in the step (1), LMS and MATLAB software are used for extracting sound quality objective parameters, and the objective parameters are represented by two-level evaluation indexes, wherein the method comprises the following steps:

step 1: based on LMS software, extracting the sound pressure level, loudness, sharpness, roughness, jitter degree and semantic definition of a sound sample as a level 1 evaluation index;

step 2: based on the objective parameters extracted in the step 1 at different times, MATLAB software is adopted to calculate the sound pressure level change rate, the loudness change rate, the sharpness change rate, the roughness change rate, the jitter change rate and the semantic definition change rate as the 2 nd-level evaluation indexes;

(4) taking the two-stage evaluation parameters of the objective sound quality extracted in the step (3) as input, taking the subjective evaluation result in the step (2) as output, performing subjective and objective modeling of the accelerated sound quality by adopting a multivariate linear regression method, and analyzing the influence weight of different objective parameters on subjective evaluation;

(5) and (4) according to the weight result of the objective parameters in the step (4), reserving objective evaluation parameters with large influence, deleting parameters with small influence, modeling again and performing model correction.

Images