Study On Low-frequency Compensation For Magnetoelectric Moving-coil Sensor

Magnetoelectric moving-coil sensors are widely used in vibration signal measurement due to their simple structure,low cost and high signal-to-noise ratio(SNR).However,due to the internal mechanical structure,the inherent frequency of these sensors is high,the amplitude-frequency characteristic curve is severely attenuated below the inherent frequency.And the response curve is not flat due to overshoot in the effective frequency range.The above-mentioned reasons make such sensors exist for the signal pickup below the intrinsic frequency,in the effective frequency range of the detection signal distortion problem.The vibration of the earth itself,as well as the vibration of roads,bridges,buildings and large scientific instruments are usually below 10 Hz,which is lower than the intrinsic frequency of most magneto-electric dynamic sensors.Hence,it is of great importance and practical value to select a suitable and effective solution to the above problems,improve the low frequency response of such sensors,and widen the effective frequency band range of their measurements.This article firstly analyzes the structure and working principle of the magnetoelectric moving coil sensor,and obtains the dynamics model of the sensor.The transfer function of the sensor is established from the input and output responses of the system,and its amplitude-frequency characteristic curve is obtained.The effects of different characteristic parameters on the amplitude-frequency characteristics of the sensor are then explained,and the reasons for the severe attenuation of the response in the low frequency band and the fluctuation of the response curve in the effective frequency band range of such sensors are analyzed.In order to solve the problems of magnetoelectric moving-coil sensors,the principles of various low-frequency compensation methods are described and analyzed in detail.Based on the implementation ideas and methods of the commonly used negative resistance method and zero-pole compensation method,a new method is proposed to effectively improve the low-frequency response of this type of sensor.The method increases the damping ratio of the system by connecting a band-pass filter in series with a subtractive circuit,thus making the original sensor response flat for velocity into a flat response for acceleration,and improving the ability to pick up signals below the intrinsic frequency.Low-frequency expansion was performed using this method and the zero-pole compensation method for a specific JF-20DX model sensor.Since both methods require a more accurate mathematical model of the sensor,the paper also introduces a variety of methods for testing the characteristic parameters of the sensor and selects the fast and effective DC excitation method for testing,which yields a more accurate transfer function.Based on the transfer functions obtained from the tests,the analog compensation networks of the two methods,the new method and the zero-pole compensation method,and the corresponding digital filtering compensation schemes were designed and completed.A vibration test platform was built using the shaker,signal generator and reference sensor,and the improvement effect was tested and compared with the JF20DX model sensor as the test object.After practical tests,both methods achieved low frequency expansion,and the sensitivity of the compensated system did not change more than 10%within 1～100 Hz,and the fluctuation of the sensitivity after compensation by the newly proposed method was lower than that of the zero-pole method.The sensor low-frequency compensation correction circuit introduces electrical noise,and the noise directly affects the signal-to-noise ratio of the system after lowfrequency compensation.It was found that when measuring the vibration with the same energy at 10 Hz frequency,although the signal-to-noise ratio of the new method is better than that of the zero-pole method by 17.52 dB,there is still a large amount of noise.Therefore,there is a need to investigate how to effectively reduce the noise and improve the signal-to-noise ratio.Wavelet analysis is a powerful tool to deal with real nonstationary signals.In this article,based on wavelet threshold denoising,the approximation coefficients of wavelet decomposition of multiple measurement results are weighted and leveled to reduce the low-frequency random noise in the approximation coefficients,forming an improved denoising algorithm.Through the denoising process of the actual vibration signal,it is shown that the algorithm can improve the signal-to-noise ratio by 16.63 dB.