Evaluating parameters of passive SAW torque sensing signal using Genetic algorithms

When detecting the torque with passive wireless SAW (surface acoustic wave) resonator sensor, the response signal is of narrow band, high frequency, low SNR and transient attenuation. The response signal is produced only in the case that the interrogation covers the operational frequency band of the SAW resonator. Burst of sinusoidal is used in the experiment to excite the resonator, and analysis of the sensing signal reveals that the response signal is an exponential decay signal of single frequency, and changes of strain lead to a shift of the resonance frequency. Torque applied to the shaft can be acquired from changes of the center frequency of the resonator. The frequency resolution of traditional FFT spectrum analysis method is limited by sampling length, which can't meet the accuracy requirement of SAW torque measurement. Parameter estimation method, such as MLE (Maximum likelihood estimate) or LSE (Least Square estimate) can be used, but it is time-consuming. In this paper, GA (Genetic algorithm) is employed to estimate parameters of the sensing signal, in particular, the center frequency. Before the introduction of genetic algorithms, response signal should be converted to sinusoid with Hilbert envelope-demodulation. This can simplify the waveform greatly. Hence, the work is turned into extracting sinusoidal signal parameters from the limited sampling, including frequency, amplitude, phase and DC offset. For the demodulated single frequency signal, the resonance frequency can be got directly in time domain by genetic algorithm. The results show that this method can estimate the frequency more accurately and faster.