Visualizing Noise Sources of KTX High-Speed ​​Trains Using LabVIEW

　　The Korea Railroad Research Institute and SM Instruments Co., Ltd. (a member of the National Instruments IAlliance Partner Network, specializing in sound and vibration related test applications) developed a mobile sound source beamforming system using a phased microphone array in the LabVIEW environment and used the system to visualize the noise sources on the entire train in normal operation. The test was mainly to compare the noise of two different models of trains: one is the KTX-1, which evolved from the TGV Réseau type and was put into use in 2004; the other is the new KTX-Sancheon (KTX-II), which is a commercial high-speed train developed by South Korea.

　　Beamforming is a method of mapping noise sources using an acoustic array. It detects the time delay of the sound reaching the microphone array to identify the direction from which the sound is coming. If the noise target is moving, the test becomes more complicated because the object will move past the microphone array (such as in a pass-through test), and the Doppler effect will cause the frequency of the sound to be distorted, which is a key disadvantage of traditional real-time beamforming methods. To compensate for this, we continuously adjust the time delay in the software to match the moving sound source. This method automatically eliminates the Doppler effect. Although it requires more processing time, we can average the power of the moving beam. We use triggerable sensors to determine the exact location of the moving noise source at each point. In our software, we assume that the sound source has a fixed speed.

　　The hardware configuration is basically the same as in the standard beamforming application. An additional feature is that the beamforming of moving sound sources requires trigger sensors. We used two photoelectric sensors for position triggering and calculating the train speed.

　　For the high-speed train test, we designed a 144-channel microphone array to improve the resolution of the sound image. We used the NIPXI-4496 dynamic signal acquisition module to collect the measurement signal and used a special photoelectric sensor to assist the ICP/IEPE microphone to trigger the train position. We conducted early tests on KTX in 2006, shortly after Korail launched its high-speed train service, using a 48-channel microphone array and successfully captured the noise source on the high-speed KTX train at 297 km/h.

　　The performance of a microphone array is affected by two parameters: (1) the width of the main lobe of beam power, which determines the resolution of the image; and (2) the maximum side lobe, which determines the ghosting rate of the image. Different array patterns have different performance indicators. We compared four different patterns, and the spiral pattern showed a very balanced result.

　　For the 144-channel microphone array, we fuse three different modes together to improve performance. Each mode has the same shape but a different diameter. The smaller diameter mode mainly measures the low maximum sidelobe level of high frequency components, while the larger diameter mode mainly measures the low frequency components with high resolution. To reduce the noise caused by wind, we added a windshield to the microphone. [page]

　　At high frequencies, the noise source from the wheels is quite obvious. Here it is shown that each wheel has a different noise amplitude. Therefore, this technology has great potential to be used to monitor the working condition of the wheels for maintenance.

　　Future trains will need to increase their speed, which will generate more severe noise, especially the noise generated by airflow. To develop quieter trains, we first need to have a deep understanding of the noise sources of trains. Because LabVIEW can visualize the location and size of noise sources, we can use it to complete noise reduction tests.