The characteristics of noise and flow field of equal-diameter three cylinders in different symmetric/asymmetric arrangements were tested in an open-jet wind tunnel using a far-field microphone array and the Particle Image Velocimeter （PIV）， and the flow field and noise were controlled by installing solid or porous splitter plates at the trailing edges of the cylinders. The experimental wind speed U₀=20–50 m/s， and the Reynolds number Re_D =2.7×10⁴–6.8×10⁴ based on the diameter of the cylinders D. Noise tests revealed that the total sound pressure of the three cylinders arranged in symmetric centers showed a variation of wave peaks and valleys with the streamwise position x/H of the center-positioned Cylinder 2 between （-2， +2）. At the same wind speed， the maximum difference between the peak noise at the wave peak （x/H =-1.75） and the valleys （x/H =-1.5 or +0.5） can be up to 20 dB， and the noise level in the wave peak range at the upstream position （x/H <0） was also significantly higher than that at the downstream position （x/H >0）. When Cylinder 2 was arranged on an arc with the center points of Cylinders 1 and 3， the noise of the three cylinders did not change significantly with the flow arrangement angle. The control study found that for the three cylinders arranged in the central-flow direction， installing an appropriate splitter plate at the trailing edge of the upstream cylinder can change the flow field and suppress the noise. At the streamwise position x/H =-2， installing the splitter plate with length L/D=1 or the perforated plate with perforation σ=18.4% and the splitter plate with length L/D =2 at the trailing edge of Cylinder 2 can reduce the noise of the three cylinders by 12 dB， and the reduction can even reach 20 dB if continuing to deflect the splitter pater to β=30°. However， at the same streamwise position， when the same type of splitter plates was installed on Cylinders 1 and 3， the solid plate with L/D =1 can achieve the average noise reduction of 5 dB， which is not significant； for the perforated splitter with L/D =2 and σ=18.2%， the noise reduction was gradually enhanced with the increase of the deflection angle β； when the deflection angle β≥20°， the noise can be reduced by more than 10 dB. PIV test and POD analysis showed that the noise in the experimental three-cylinder arrangement mainly comes from the vortex shedding of the upstream cylinder and the vortex-solid interaction on the downstream surface， and the mechanism of noise suppression of the splitter plate is due to the significant suppression of the vortex shedding of the upstream cylinder， i.e.， the suppression of the Karmen vortex street in the flow field.