Research On Signal De-noising Method Of Displacement Sensor In Wide Frequency Inertial Stabilization Platform

The broadband inertial stabilization platform is an attitude measurement and control system that uses inertial sensors as feedback.It can isolate the carrier vibration in a wide frequency band to ensure the platform load accurately tracking the target.As the angular position sensor of the inertial stabilization platform,the measurement accuracy of the displacement sensor directly affect the accuracy of the system.In order to ensure the tracking accuracy of the system,this paper intends to study the signal denoising of the displacement sensor.(1)The co-circular position distribution of the displacement sensor in the inertial stabilization platform is proposed.This method adopts the measurement method of averaging and re-differential,which can increase the measurement accuracy by 2 times and reduce the installation volume.In addition,the paper analyzed the characteristics of the sensor noise and the impact of noise on the system error,then the following two methods of noise reduction are introduced.(2)The threshold de-noising method based on wavelet transform and the wavelet correlation de-noising method are analyzed.The simulation results show that the denoising effect of wavelet threshold method is affected by the threshold criterion,and the overall de-noising effect is not as good as wavelet correlation de-noising.However,the wavelet correlation method also has the problem of poor adaptability.The denoising effect is affected by the wavelet basis and the number of decomposition layers.(3)To solve the problem of poor self-adaptability of wavelet de-noising,the empirical mode decomposition method based on local features of the signal is studied.To solve the problem of the empirical mode decomposition method that the noise dominant mode and the signal dominant mode are difficult to determine and the useful components may contained in the noise dominant component,the empirical mode decomposition-de-trend analysis-non-local mean de-noising method is proposed.(4)An experimental test system is built to acquire and de-noise the displacement signal.The experimental results show that the peak-to-peak value of the sensor static noise can be reduced from 11 m V to 1 m V after wavelet correlation de-noising and EMD-DFA-NLM de-noising.For sensor dynamic signals,the EMD-DFA-NLM method has better de-noising effect under different sinusoidal excitation signals,and the signal-to-noise ratio is more than 10 d B higher than the wavelet correlation denoising method.However,the wavelet correlation de-noising method has better realtime performance,and processes 15,000 points online,which only takes hundreds of milliseconds.Therefore,when the system requires high measurement accuracy and no response time is required,EMD-DFA-NLM de-noising method can be adopted;when fast response is required,wavelet correlation de-noising method can be adopted.