Based on early empirical prediction models and the test data of turbine noise

the key parameters of turbine which may influence the turbine noise were analyzed. And an improved empirical prediction model for multi-stage turbine noise was developed. This improved model was composed of three components: computation for single stage turbine noise

computation for the noise attenuation effect of each downstream turbine stage and superposition for the noise of each stage turbine. The calculation method for single stage turbine noise had been validated by use of turbine noise test data. Result shows that

the calculation deviation is less than 1.5 dB. An empirical prediction method fitting by experimental data for the propagation attenuation caused by downstream blade rows is also used in the improved multi-stage turbine noise prediction model. With comparison to early empirical prediction models

the calculation method of the improved model is more reasonable and the noise prediction result is more reliable.