larger at high frequencies. As shown in Fig. 24, the results at 100and 6300 Hz were raised by 7.65 and 16.05 dB when filling centrifugal glass cotton with a thickness of 136 mm.After the installation of the noise barrier, the SPLs between itselfand the train body tend to be larger. Taking the cavity next to thepoint V0 as an example, the increment of the SPL was up to 2.5 dB(A), which was reduced by 0.35 dB(A) after filling internal soundabsorption material with a thickness of 136 mm (Fig. 25(a)). Thenegative effect of the barrier can be controlled effectively in thefrequency range of 100–1000 Hz and the maximum reductionwas approximately 1.2 dB(A). Nevertheless, this effect is muchweaker in the higher frequencies.Taking the point P2 as an example, the effect of sound absorption material on the insertion losses was studied (Fig. 25(b)). Theoverall insertion loss increased 3.27 dB(A) when the filling thickness was 136 mm. The noise reduction effect at the frequencyrange of 100–1000 Hz can be enhanced effectively, which has acertain relationship with the control of the noise between the barrier and the train body. As the thickness of the material increases,its improvement effect is weakened.In general, the role of centrifugal glass cotton is mainly concentrated in the low- and mid- frequency bands and its absorption ofhigh-frequency noise is small. Better improvement can be obtainedby cooperating with other optimization measures.4.4. Effect of sound leakage