Improved Methods for In－situ Measurement Railway Noise Barrier Insertion Loss

0 Introduction

Noise barriers，the most effective means to mitigate the propagation of sound，are widely applied on urban railway transit systems，especially on elevated lines.The noise reduction effect of a railway noise barrier is thought to depend largely on its height and the relative distance between the source，the barrier，and receiver positions［1］.But there is also a close relationship between acoustic performance and environmental factors，such as ground effect，atmospheric turbulence，air absorption，refraction by wind and temperature gradient profiles［2－3］.To achieve noise reduction effect of barriers on site during the operation of a real urban rail transit system，all factors must be taken into account.To this end，many countries have established guidance standards for in－situ experiments［4－6］.ISO 10847－1997［4］ proposes that barrier performance in a field test can be represented by the difference in sound pressure levels at specified receiver positions before and after the installation of a barrier，provided that the relevant parameters remain unchanged.This is referred to as″insertion loss″or″attenuation″.ISO 10847－1997also proposes that naturally occurring railway traffic，principally the passenger train，should be used as the sound source equivalence for the″before″and″after″measurements.

However，there are no global standards for receiver positions，a state of affairs which produces vagueness.ISO 10847－1997proposes that there are only two conditions：hemi free－field conditions，and reflecting surfaces.These conditions as a very general characterization of the open space behind barriers.In China′s standard HJ/T 90－2004［5］，receivers are defined as being located in the area which is the most sensitive to the noise.TB/T 3050－2002［7］defines the area sensitive to noise as residential buildings，schools，hospitals and other areas which require strong protection from noise.However，under different meteorological conditions，the shapes of areas are vulnerable to noise change.The standard is thus useful for getting aproject accepted，but can be useless as a guide for designers who want to find the best barrier for a particular site.In China，in consequence，these standards have to be supplemented with other standards［7－10］for different receiver positions.The receiver position stated in TB/T 3050－2002，which is concerned with railway lines and used for the investigation of railway boundary noise in GB/T 12525－1990［8］，directs a receiver position 30mfrom the nearest track center and 1.2mabove the mean rail head height of the nearest track in the relevant area.GB/T 5111－2011［9］，which is concerned with railway vehicle noise，directs that receivers be located 7.5maway from lines and at heights of 1.5mand 3.5m.HJ 453－2008［10］，which is concerned with testing the noise intensity of railway traffic，directs that the receiver should be placed 7.5maway from the source and at a height of 1.5m.Thus these given positions can be identified as alternative receivers in the case of comparing barrier performance with different shapes.

Seeking to clarify receiver positions in field experiments，the European Committee for Standardization（ECS）recently made recommendations for the measurement of sound attenuation of given noise barriers at given sites in given meteorological conditions.The ECS′s standard，CEN/TS 16272－7：2015［6］，recommends nine locations to place the receivers，forming agrid，in order to measure the attenuation of a given noise barrier at agiven site including given meteorological conditions.They are placed specifically at a distance of 7.5，12.5and 25maway from lines and at a height of 3.5，6.0and 9.0m.This standard is a useful source of comparison of the noise attenuation capacities of different types of barrier at the same site under the same meteorological conditions.However，although there are many researchers at work on the in－situ measurement of insertion loss in railway noise barriers［1，4－11］，very few base themselves on this European standard［12］.

3) 判断是否达到测量点数，如果没有则发出控制脉冲信号，通过步进电机4和拨盘轮机构带动塞规旋转到下一测量位置，记录步进电机的旋转角度，重复2)；

In consequence，there are no universally accepted receiver positions for the evaluation of barrier performance.This circumstance may be leading to uncertainty with respect to the noise reduction capacities of barriers presently available.

西安交通大学制冷与低温工程系副主任晏刚着眼技术，以“‘鲜’‘智’—冰箱的技术指向”分析了冰箱产品的升级发展方向。他认为，冰箱行业的技术升级方向主要有保鲜和智能两个大方向，企业创新要“鲜”得有理，“智”得有度。保鲜技术的升级主要体现在保鲜技术的多元化以及保鲜效果的健康化，而智能技术的升级，则主要体现在食物识别、语音识别、智慧物联三个维度。

Since the valid pressure signals at each position were measured at least 10times，equivalent levels had to be expressed as an average.The formula for the averaging method is

1 Improved Methods

1.1 Rearrangement of receiver positions based on the diffraction theory

The microphones at each receiver position were omnidirectional and protected by windscreens.Corresponding frequency responses ranged from 20Hz to 20kHz.The sampling frequency of the sound pressure signals was intended to be 51.2kHz，based on the Nyquist Theorem.To avoid message distortion，this was more than twice the maximum frequency component of the audio frequency （20Hz—20kHz）.The experiments on the″before″and″after″sites were conducted on sunny days only a few days apart.Meteorological conditions were not significantly different and thus were not measured.However，an acoustic amplifier，an electrical charge amplifier，sound pressure collecting equipment，an A/D data collection card and a computer running a data collecting program were prepared.These instru－ments met the requirements of EN 61672－1and the microphones complied with IEC 61672class 1.Pressure signals at all receiver positions were recorded simultaneously and，to ensure the statistical representativeness of the sample，train－passing data for each distance was obtained by taking at least 10measurements.

高等职业院校应该对教师加以正确引导和正向鼓励，促使他们通过大数据平台进行教育观念的创新和教学方法的尝试，通过在传统培养机制中强化数字和大数据技术的方式，在高等职业院校教师心中牢固树立正确的素质发展观念和提升理念，通过增加资源和加大投入的方式优化教师发展、成长和执业环境，为教师在教学中实施大数据平台下的改革提供机制和体系的保障，达到对高等职业院校教育改革大目标的有效适应和不断保障。

Table 1 Rearrangement of receiver positions based on diffraction theory

Distance from the nearest track center≤14m ≥14m Bright zone in near field Bright zone in far field Transition zone in near field Transition zone in far field Shadow zone in near field Shadow zone in far field

The prescribed receiver positions are shown in Fig.1，where the height of barrier above the rail head height is 2m.Different shadows based on diffraction theory show that for Chinese standards（indicated by triangles）all receiver positions are located in the shadow zone in near field，with the exception of the receiver in TB/T 3050－2002，which is located in the shadow zone in far field.All nine positions specified by CEN/TS 16272－7：2015 （indicated by circles）cover four of the acoustic areas.M1－1，M2－1and M2－2represent the performance in the shadow zone in near field；M3－1，M3－2and M3－3represent the performance in the shadow zone in far field；M1－2and M2－3 represent the performance in the transition zone in near field，while M1－3represents the performance in the bright zone in near field.In addition，the sound pressure distribution of the whole of the open space behind the barrier is mapped by the nine grid positions，enabling visualization of the noise reduction effect of a barrier.The gridform method is thus instructive for improving the arrangement.

Fig.1 Comparison of receiver positions prescribed according to the standards［4－10］

By considering the actual need of the Engineering application in urban railway transit systems，the receiver positions have to be rearranged in terms of the changeable locations of the acoustic areas.For instance，if a barrier is installed on a bridge，it is difficult to reach the bright zone.Hence there is no necessity to place receivers at M1－3，M2－3and M3－3unless there are tall residential buildings close to the lines.If the sensitive areas are located far from the lines，receivers at M3－1，M3－2and M3－3can be placed around the sensitive area instead.When the height of the tested barrier is very low，the boundary dividing the shadow zone and the transition zone must be lower than that shown in Fig.1.This means that receiver M1－1can be located in the transition zone in near field，resulting in no receiver positions in the shadow zone in near field close to the barrier.If the shape is near to fully－enclosed，the receiver M1－3should also probably be located in the shadow zone in near field，resulting in no receiver positions in the bright zone.In consequence，receiver positions need to be rearranged in all the acoustic areas as possible.

借助专用差分信号采集探头，将示波器接入线路，通过协议分析仪配合端口触发进行波形抓取。发现设备D3的端口0x350在进入RPT之前回复从帧正常，但是经过RPT之后，从帧消失。为了排除RPT自身原因，更换RPT后再进行试验，故障依旧存在。

1.2 One－third－octave－band analysis

中国联塑建有国际领先、国内一流的研究院，拥有各类科研人员1000多名，设有博士后科研工作站、CNAS国家认可实验室。目前，中国联塑拥有和正在申请的专利有近1000项。科研成果先后入选国家火炬计划项目、国家重点新产品、全国建设行业科技成果推广项目和政府绿色采购清单；先后被国家有关部门授予国家高新技术企业、国家认定企业技术中心、中国建设科技自主创新优势企业、知识产权优势企业、建设部产业化示范基地、广东省政府质量奖等荣誉称号和奖项。

However，in the interests of reliability and applicability，real site testing is necessary.In our view，it is advisable to design an in－situ insertion loss experiment for railway noise barriers based on standards and our findings.This way，it is possible to compare the results of different analysis methods and to offer practicable suggestions.

Table 2 Dominant frequency range of noise sources（other than rolling stock noise）in urban rail transit systems

Special situation Curve squeal Brake screech Bridge Frequency range Pure tone，high frequency（up to 10kHz）Pure tone，high frequency（during braking）Low frequency

2 Experiment Design

Based on the improved methods，an in－situ experiment was conducted on Jiading Campus，Tongji University.A straight barrier，placed at ground level，with a thickness of 0.1m，was tested.The length of the barrier was 10mand its height above the track was 1.5m，which is relatively lower than other railway noise barriers.Since the barrier could be removed during the period of experiment，utilizing the direct measurement method［4］，sound pressures at receiver positions were tested by microphones before and after barrier installation.The noise source in the experiment was naturally occurring railway traffic：two－carriage passenger trains，each 22mlong.Since the experiment sites were located in the middle of the lines，the noise induced by trains in brake mode could not be considered.The trains traveled at 40km/h as they passed the test field.

The circles in Fig.2indicate the directed receiver grid formation.Applying the improved methodology，it was evident that M1－2and M1－3 could not be located in the transition zone，on account of the low profile of the barrier.The attenuation property of receiver M1－1，located far from the barrier and close to the transition zone in near field，might underestimate the performance of the barrier in the shadow zone in near field.Moreover，since the barrier was installed at the ground line，it proved possible to choose the receiver above ground at 1.2mto simulate pedestrian hearing.Thus receiver positions were better reset close to the barrier and to the ground.In consideration of the low profile，it was possible，by applying the grid－formation criteria of the standards，to determine receiver positions.These are indicated by crosses in Fig.2.They were at a distance from the nearest track center of 3.3，9and 18mand at a height above the mean rail head of the nearest track of 1.2，1.8and 2.5m，respectively.As per the discussion above，the receiver positions in our experiment were set in four areas of the open space behind the barrier.P1－1，P2－1，P2－2and P2－3are represented in the shadow zone in near field.P3is represented in the shadow zone in far field.P1－2is represented in the transition zone in near field.P1－3is represented in the bright zone in near field.

[18]Wenzhou,which is to say almost everyone in Wenzhou,has a horror story about the financial crisis that struck in mid-2011.（2016-06-11）

In order to compare the insertion loss values of different types of barriers at the same site under given meteorological conditions，it is quite important to offer an approach to determine the receiver positions in the full－scaled experiment of barrier performance.In the case of noise barriers，receiver positions are located on the opposite side of the sound source.In accordance with diffraction theory［21］，the open area behind the barrier can be divided into three zones：A bright zone where all frequencies transmit directly，a transition zone where low and middle frequencies bend around the barrier during the direct transition of high frequencies，and a shadow zone where，as a result of the vibration and the diffraction，only low－frequency sounds are transmitted.Since the noise reduction effects of barriers vary substantially by zone，and variations are a function of frequency［22］，in－situ measurements of insertion loss for all frequencies in each zone are necessary.Depending on distance of receiver from sound source，the acoustic energy produced by the source will behave quite differently.In far field，the spherical shape of the sound waves can be reasonably approximated as a plane－wave，with no curvature［23］.It is important to understand this difference，and place the receiver positions in near field and far field separately when taking measurements.Generally，a far field acoustic begins two wavelengths from the sound source，and extends outward to infinity.In the case of traffic barriers，the start of the far field is at least around 14m［23］.Receiver positions should therefore better be placed less and greater than 14m，respectively.It is considered that receiver positions represent barrier performance at all the acoustic areas given above.A conservative estimate is that six positions meet the requirements（Table 1）.

3 Results

3.1 Signal processing procedure

According to the relevant standards［4－6］，the equivalent continuous A－weighted sound pressure level can be represented as follow

where Tpis the train pass－by time interval，pAis the A－weighted instantaneous sound pressure，and p0is the reference sound pressure（20μPa）.During the post－processing procedure，the sound pressure signals were first filtered by the bandpass of audio frequency range and A－weighting filter，and then，by utilizing the time interval of the train′s passing，the equivalent continuous A－weighted sound level LpAeq，Tpasswas obtained.

The present study aims to shed light on the measurement of insertion loss in railway noise barriers.In order to specify a set of reasonable receiver positions for comparing different types of barriers and to introduce a descriptor for the prediction of the acoustic performance of railway noise barriers，based on relevant standards and sound diffraction theory，an improved arrangement of receiver positions is put forward and a current indicator is taken into account as a supplement to the A－weighting method.Utilizing these，we designed and carried out an in－situ experiment to investigate insertion loss of railway noise barriers.For the experimental analysis，differences between recommended receiver positions and the reset receivers，sound pressure level（SPL）and attenuation in A－weighting method and the improved method were researched.In the following，we first describe the determination of receiver positions in the field experiment.We then describe sound pressure level acquisitions by different descriptors on the″before″and″after″ measurements.Finally，we give the attenuation results for all receiver positions.

where LAeqis the sound level used to calculate the noise attenuation of the barrier，and LAE，iis the i th pass－by level computed by Eq.（1）.Hence，it was easy to obtain the attenuation single－number rating for barrier performance.

硫磺尾气改造前后吸收塔和再生塔主要操作条件有所不同，吸收塔主要操作条件如表3所示，再生塔主要操作条件见表4。

The C－weighted level and the 1/3octave band level was acquired in the same way，producing an effective supplement to the A－weighting method.However，for the sake of simplification，the attenuation of each 1/3octave band was obtained by calculating the ratio of sound energy in the field with and without the barrier.This is given by

where p（f）is the sound pressure with respect to a certain frequency，calculated by applying the Fast Fourier Transform formula.foctis the central frequency of the 1/3octave band.

3.2 Sound pressure level

To reach an assessment of barrier performance，the experimental results of the sound pressure level at all receiver positions on the″before″and″after″measurement will be illustrated first.This is in order to comprehend the characteristics of railway noise at a speed of 40km/h.In Fig.2，the bold numbers denote the continuous equivalent A－weighted sound pressure level（LAeq）at all receiver positions before the installation of the barrier.The italicized numbers denote LAeqon the″after″site.It can be seen that before the installation of the barrier，LAeqat P1－1was marginally higher than that at P1－2and higher than that at P1－3.Since the P1－1and P1－2positions were much closer to the track，the results confirmed that rolling stock noise produced by wheel－rail contact vibrations，could be a predominant component of railway traffic noise.However，P1－1 level was significantly lower than P1－2and P1－3 around 5dB（A）after the barrier was installed，indicating that the barrier was able to suppress rolling stock noise effectively.Performance of the barrier was particularly good in the shadow zone in near field.At the receiver－source distance of 9 m，the levels of each of the three receiver positions were almost the same on the″before″site，whereas the P2－3level was much higher than P2－1 and P2－2on the″after″site.With increase of receiver－source distance，LAeqshowed a tendency to decrease at the same height，regardless of the installation of the barrier：P1－1＞P2－1＞P3（LAeq），P1－2＞P2－2（LAeq）and P1－3＞P2－3（LAeq）.Interestingly，in contrast to the″before″site，the downward trend of the A－weighted level near the ground became slower on the″after″site.

Fig.2 Configurations of the in－situ experiments with the straight barrier on the ground line（frequency range：20Hz—20kHz）

By one－third octave analysis，SPLs at all receiver positions on the″before″site in Fig.3（a）show that the dominant frequency range of railway noise was quite wide：five 1/3octave bands of 400，500，630，800and 1 000Hz，which are given relatively small weights by A－weighting［27］.Moreover，the maximum values at all receiver positions were，coincidently，all at 800Hz，and the differences between the LAeqs and the maximum values at all the positions were no more than 1dB（Table 3）.Therefore，without one－third－octaveband analysis，the continuous equivalent A－weighted sound pressure level could present almost the same level as that of the predominant component.This indicated that A－weighting was suitable to describe the annoyance induced by railway noise.On the″after″site（Fig.3（b）），sound levels in the dominant range of railway noise were roughly consistent with the levels at low frequencies，which were reduced considerably by the barrier.In order to understand the importance of low frequency noise，the difference between C－and A－weightings has been considered as a predictor since it indicates the amount of low frequency energy in the noise［14］.If the difference is greater than 15dB，there is a potential for low frequencies.In Table 4，we see that the differences between A－weighted and C－weighted levels at all positions were as large as 6dB.Although the differences were not too large，it is noteworthy that low frequency noise played the same significant role as the middle and high frequencies on the″after″site.This should not be neglected in the future attenuation research.

Fig.3 Sound pressure levels in the one－third－octave band at all receiver positions

Table 3 Comparison between LAeqand maximum value of 1/3octave band on the″before″site

Receiver position P1－1 P1－2 P1－3 P2－1 P2－2 P2－3 P3 LAeq（20－20kHz）/（dB（A）） 95.95 95.55 93.05 84.70 84.33 85.64 78.69 800Hz of 1/3octave band/dB 96.76 96.24 93.75 84.88 84.69 85.64 79.02 Absolute difference 0.81 0.69 0.70 0.18 0.36 0.00 0.33

Table 4 Differences between A－weighted and C－weighted SPL at all positions on the″after″site dB

P1－1 P1－2 P1－3 P2－1 P2－2 P2－3 P3 4.73 3.06 2.88 4.38 4.09 3.73 6.15

3.3 Insertion loss

Insertion loss，also called attenuation，is defined［4］as the difference in sound levels at a specified receiver position before and after the installation of a barrier.Using the results of SPLs at all the receiver positions analyzed above，barrier attenuation was obtained and listed in Table 5.It appears that the barrier varied in effectiveness depending on where the receiver position was located.Of the seven positions，the attenuations in the shadow zone in near field （P1－1，P2－1，P2－2 and P2－3）were the highest，being at least 10dB（A）.The next were in the transition zone in near field（P1－2）and in the shadow zone in far field（P3）.The lowest was in the bright zone in near field（P1－3）.Hence the area in the shadow zone in near field appears to be the major area of competence for the noise reduction effect of the barrier.Since the difference between the maximum and the minimum values of the attenuations（P1－1and P1－3）was around 7dB（A），it is evident that a single measurement point cannot provide a comprehensive presentation of barrier performance.In other words，the significant difference between these attenuations is attributable to variation among receiver positions.These must be taken into account when evaluating the acoustic performance of barriers.

Table 5 Attenuations in LAeqat all receiver positions（frequency range：20Hz—20kHz） （dB（A））

P1－1 P1－2 P1－3 P2－1 P2－2 P2－3 P3 14.72 8.44 6.97 11.50 11.46 10.60 7.08

For comprehension of the frequency characteristics of the attenuation，the attenuations in the one third octave band from 50to 2 000Hz are computed and shown in Fig.4.The barrier performed well in the dominant frequency range of these five bands，especially in the shadow zone in near field（P1－1，P2－1，P2－2and P2－3）.Attenuations at all receiver positions were as high at the frequencies above the dominant range.However，the attenuations in the range of low frequencies were ultra－low，and，below the band of 100Hz，even negative.Among all the receiver positions，the maximum of excess attenuation was 6.3dB in the band of 50Hz at receiver P3.Ground effect and diffraction of low frequencies at the top of the barrier might be the cause of negative values of attenuation located primarily in the shadow zone in far field（P3），the transition zone in near field（P1－2）and the bright zone in near field（P1－3）.In summary，the barrier performed quite well in the range of mid and high frequencies，but relatively badly at low frequencies.

The ISO standard［4］ recommends octaveband or one－third－octave－band sound pressure levels as indicators when it is necessary to obtain frequency characteristics of barrier insertion loss.Since the dominant frequency components are easily recognizable from one－third－octave－band analysis，this method has been adopted by low frequency noise standards of Polish，Swedish and German in general environment［13］.This has implications for the placement of barriers in areas proximate to urban main road traffic.It appears that such noise makes a smaller contribution to reported annoyance than might be inferred from the objective or physical dominance of the noise［24］.In such a case，it is unnecessary to analyze in－situ experimental results by employing one－third－octave－band analysis.The same applies with railway noise barriers，since it is well known that rolling stock noise is the predominant component of urban railway noise and that the latter is normally within a rather broad frequency range of 800Hz to 2 500Hz［25］.There are many other sources of noise，such as curve squeal，brake screech and bridges.The acoustic characteristics［26］of these are shown in Table 2.When a train crosses a viaduct，the low－frequency rumble noise induced is a significant annoyance for those in the station and residents in the vicinity，even at considerable distances.This is because lower frequency noise travels farther than higher frequency noise［23］.When sound barriers are installed on a viaduct，the additional low－frequency noise which radiates from the viaduct，the barrier and their related connectors cannot be neglected.One－thirdoctave－band analysis must be deployed here，since it produces data helpful to the attenuation of such low－frequency noise.

As the most common descriptor for assessing the barrier performance，the equivalent continuous A－weighted sound pressure level［4－6］is introduced to calculate the attenuation of a barrier.The ISO standard［4］ minimally requires field measurements of equivalent A－weighted sound levels，with and without barrier，for all receiver positions，producing a single－number attenuation rating.Chinese［5］and European［6］standards also adopt the latter as evaluation indicator.However，it is impossible to assess the performance of barriers at different sound frequencies using this single－number rating.In addition， A－weighting tends to devalue the effects of low frequency noise，making its suitability for the evaluation of noise barrier performance dubious.In recent years，many researchers have concluded［13－17］that A－weighting underestimates the annoyance produced by low－frequency and predominantly lowfrequency noise，even at low volume levels.Despite the masking effects of higher level components in complex sound environments，the weakness of A－weighting have been identified［18－19］as well.Since barriers are mostly erected on the elevated section of lines，while relevant sound emissions are mainly concentrated at low frequencies［20］，A－weighting is not a useful guide.

Fig.4 Attenuations in the one－third－octave band at all receiver positions（20—5 000Hz）

4 Discussion

By utilizing the grid－form method，attenuation at the recommended positions could be estimated by known results.It appears that the maxi－mum value of attenuation at all recommended positions can be located at receiver M1－1and M1－2.As receiver M1－1is close to the boundary between the transition zone and the shadow zone，based on diffraction theory，the attenuation at receiver M1－1must be a little lower than that at receiver P1－2，which is located in the transition zone in near field.In the same way，attenuation at receiver M2－1must be a little lower than that at receiver P2－3.It follows that maximum attenuation in the shadow zone in near field is no more than 10dB（A）.Compared with the experimental results in section 3，the attenuations at the recommended positions definitely underestimate the performance of the barrier.As such，the recommended receiver positions seem to be higher than the evaluation requirement，and hence unsuited to lowheight barriers.With the rearrangement of receiver positions much closer to the barrier and to the ground，the attenuations in LAeqin the bright zone，the transition zone and in the shadow zone in near and far field can be demonstrated more completely and distinguished more clearly.In consequence，it would be better to rearrange receiver positions to suit the actual need of the barrier.

Based on diffraction theory，we have proposed here an improved method for the arrangement of receiver positions for the in－situ measurement railway noise barrier insertion loss.The method is capable of optimizing the performance of railway noise barriers in all the areas behind barriers.Our in－situ investigation of insertion loss with low－height barriers validates the claim that this method is more effective than CEN/TS 16272－7.The A－weighted SPL led to the overrating of the railway noise barrier performance with respect to the SPL.We conclude that one－thirdoctave band analysis provides superior frequency domain results，and is a good supplement to the A－weighting method.The one－third－octave－band values seem to provide a better general description of barrier performance than do A－weighted results.

5 Conclusions

Using one－third－octave－band analysis，the predominant frequency range of railway noise can be identified.Moreover，the frequency characteristics of the attenuation can also be recognized：The barrier performed well at the predominant frequency range of railway noise but relatively poorly at low frequencies and at frequencies below 100Hz in particular.It is of interest to note that there is considerable variation in attenuation even in the same frequency band.Although A－weighting is inapplicable here，the single－number rating can be still utilized as a railway noise indicator.O－verall，the A－weighting method is inadequate to the analysis of the performance of railway noise barriers.However，a combination of A－weighting method and one－third－octave analysis can rectify the problem.

This study may have limitations.Our experiment was conducted on barriers situated at ground level and as such our directions might not suffice to indicate low frequencies after barrier installation.Future research will seek to rectify this through experiments on elevated line sections.Nevertheless，to the extent that our study indicates how to achieve railway noise barrier performance from the measurement of sound pressure levels by the improved arrangement of receiver positions and one－third－octave analysis，it is a step toward better understanding.

语音是英语学习的重点和难点。然而，目前对语音教学的研究却少之又少。英汉语音的比较研究、对比教学实属必要。

Acknowledgements

This work was supported in part by the National Natural Science Foundations of China （Nos.51708422，51678446，51408434）.The authors would like to acknowledge many people who were involved in this work，extra thanks go to LIU Yan，LV Guosheng，BU Zheng and ZHANG Tianqi for their valuable additional help.The authors also wish to thank China Scholarship Council and cole Nationale des Ponts et Chaussées，for providing necessary financial assistance to LI Qiutong to pursue her Ph.D.in France.

References：

［1］ JOLIBOIS A，DEFRANCE J，KORENEFF H，et al.In situ measurement of the acoustic performance of a full scale tramway low height noise barrier prototype［J］.Applied Acoustics，2015，94：57－68.

［2］ B RENGIER M C， GAUVREAU B，BLANCBENON P，et al.Outdoor sound propagation：A short review on analytical and numerical approaches［J］.Acta Acustica united with Acustica，2003，89（6）：980－991.

［3］ MAKAREWICZ R.A simple model of outdoor noise propagation［J］.Applied Acoustics，1998，54（2）：131－140.

［4］ Acoustics—In－situ determination of insertion loss of outdoor noise barriers of all types：ISO 10847：1997［S/OL］.［1997－07］.https：//www.iso.org/standard/1314.html.

［5］ Norm on Acoustical Design and Measurement of Noise Barriers：HJ/T 90－2004［S/OL］.［2004－10－01］.http：//kjs.mep.gov.cn/hjbhbz/bzwb/jcffbz/200410/t20041001＿63070.shtml.

［6］ Railway applications－Track－Noise barriers and related devices acting on airborne sound propagation：CEN/TS 16272－7：2015［S/OL］.［2015－09－30］.http：//store.uni.com/catalogo/index.php/cen－ts－16272－7－2015.html？josso＿back＿to＝http：//store.uni.com/josso－security－check.php＆josso ＿cmd ＝ login ＿optional＆josso＿partnerapp＿host＝store.uni.com.

［7］ Specifications for the Measuring Technique of Environmental Noise along Railway Lines.，China Railway Publishing House：TB/T 3050－2002［S/OL］.［2002－05－17］.http：//csres.com/detail/160748.html.

［8］ Emission standards and measurement methods of railway noise on the boundary alongside railway line：GB 12525－90［S/OL］.［1990－11－09］.http：//csres.com/detail/1957.html.

［9］ Acoustics－Measurement of noise emitted by railbound vehicle：GB/T 5111－2011［S/OL］.［2011－12－30］.http：//csres.com/detail/221289.html.

［10］Technical guidelines for environment impact－Assessment of urban rail transit：HJ 453－2008［S/OL］.［2009－04－01］.http：//kjs.mep.gov.cn/hjbhbz/bzwb/other/pjjsdz/200812/t20081229＿133078.htm.

［11］BUSCH T A，NUGENT R E.A reduced－scale railway noise barrier＇s insertion loss and absorption coefficients：Comparison of field measurements and predictions［J］.Journal of Sound and Vibration，2003，267（3）：749－759.

［12］CLAIRBOIS J P，GARAI M.The European standards for roads and railways noise barriers：State of the art 2015［C］//Proc Euronoise.2015.

［13］PAWLACZYK－USZCZYN＇SKA M，DUDAREWICZ A，WASZKOWSKA M，et al.Assessment of annoyance from low frequency and broadband noise［J］.Int J Occup Environ Med，2003，16：337－43.

［14］LEVENTHALL H G.Low frequency noise and annoyance［J］.Noise and Health，2004，6（23）：59.

［15］BRONER N，LEVENTHALL H G.A criterion for predicting the annoyance due to higher level，low frequency noise［J］.Journal of Sound and Vibration，1982，84（3）：443－448.

［16］BRONER N，LEVENTHALL H G.Low frequency noise annoyance assessment by low frequency noise rating（LFNR）curves［J］.Journal of Low Frequency Noise，Vibration and Active Control，1983，2（1）：20－28.

［17］ISHAC N.Low frequency noise and environmental assessment［J］.

［18］PERSSON K，BJ RKMAN M.Annoyance due to low frequency noise and the use of the dB（A）scale［J］.Journal of Sound and Vibration，1988，127（3）：491－497.

［19］KJELLBERG A，GOLDSTEIN M，GAMBERALE F.An assesment of dB（A）for predicting loudness and annoyance of noise containing low frequency components［J］.Journal of Low Frequency Noise，Vibration and Active Control，1984，3（3）：10－16.

［20］LI X，LIU Q，PEI S，et al.Structure－borne noise of railway composite bridge：Numerical simulation and experimental validation［J］.Journal of Sound and Vibration，2015，353：378－394.

［21］PIERCE A D.Diffraction of sound around corners and over wide barriers［J］.The Journal of the Acoustical Society of America，1974，55（5）：941－955.

［22］KURZE U J，ANDERSON G S.Sound attenuation by barriers［J］.Applied Acoustics，1971，4（1）：35－53.

［23］BERANEK L L，MELLOW T J.Acoustics：sound fields and transducers［M］.Academic Press，2012.

［24］TORIJA A J，FLINDELL I H.The subjective effect of low frequency content in road traffic noise［J］.The Journal of the Acoustical Society of America，2015，137（1）：189－198.

［25］THOMPSON D J.Wheel－rail noise generation，part I：introduction and interaction model［J］.Journal of sound and vibration，1993，161（3）：387－400.

［26］Licitra，Gaetano，ed.Noise mapping in the EU：models and procedures［M］.Baca Raton：CRC Press，2012.

［27］HELLMAN R，ZWICKER E.Why can a decrease in dB（A）produce an increase in loudness？［J］.The Journal of the Acoustical Society of America，1987，82（5）：1700－1705.