The Electro-hydraulic Servo System Based On Chirp Signal Frequency Characteristic Test Method Research

In the process of performance evaluation of electro-hydraulic servo system and its controller design, the frequency characteristic is one of the most important system characteristics. Due to the existence of clearance, dead zone nonlinearity, sampling quantization error, and external electromagnetic interference and other issues in electro-hydraulic servo system, when using Chirp (chirp sweep) as the excitation source signal to system frequency characteristic test, if noise ratio of the test data signal is less than10dB, the directly use of FFT (Fast Fourier Transform) to deal with the data often makes a serious distortion for the frequency characteristic curve and cannot obtain satisfactory results. To solve the problems, based on the analysis of frequency response characteristics of the electro-hydraulic servo system under these factors, this paper viewed the impact of these factors on the response signals as random interference whose spectral density constants are different, and used of white noise model simulating these factors for testing influence. This paper proposed to use of the band-pass filtering and wavelet packet transform to get rid of the noise for the preprocessing of the system response signal adding white noise, and using the least squares method to segment fitting for high frequency response. Then this paper analyzed and processed by the FFT to obtain satisfactory the system frequency characteristic curve. Used this method to electrical analog experiments of electric hydraulic servo system adding white noise, and actual frequency characteristics experiments of the inverted pendulum load electro-hydraulic servo system, the simulation and the results show that:for the band below the cutoff frequency, when noise ratio of the signal is less than10dB, even-20dB, amplitude-frequency characteristic curve that the test error of not is less than2dB and the error of not less than6°of phase frequency response curve can be get, which can be better to meet the system performance evaluation and controller design application requirements.