Practical Application of ARMA Model Method for Blade Tip Timing

Blade tip timing (BTT) is a non-contact measurement technology for rotating blade vibration. However, BTT often has nonuniform undersampling. Building a response mode is an efficient method for BTT signal analysis. The autoregressive moving-average (ARMA) model is famous for modelling a linear stationary time series in discrete time. We modified the classical Yule-Walker equation-based ARMA solving method to adapt to BTT measurement. The ARMA model method is applied in compressor field tests. The tests are conducted in stable states with different rotational speeds. The ARMA models with different order and probe arrangements are performed on the measured data. The results show that the identified natural frequency is credible and has close results for most ARMA models. The identified natural frequency increased as the rotational speed increased, which confirmed the rationality of the results. The results of different probe arrangements indicated that at least four probes are recommended in our test.