基于一维全变分的液压系统压力信号快速实时去噪方法

Fast real time denoising method for pressure signal of hydraulic system based on 1D TV

在载重过程中,工程车辆因受到惯性冲击、工作装置共振以及电磁干扰等,容易造成其液压测量信号的振荡,为了对载重估计模型中的液压系统压力信号进行实时去噪,提出了一种基于一维全变分(1D TV)算法的液压系统压力信号实时去噪方法。首先,阐述了1D TV算法的基本原理;然后,建立了一个主要受液压力影响的载重估计模型,使用MATLAB中的fmincon非线性规划求解器,将载重估计模型作为黑箱函数,载重估值与载重真值之间的均方根误差作为目标函数值,1D TV算法中的参数τ作为优化变量,求解出了能够逼近真实液压信号曲线的参数τ;最后,以某企业伸缩臂叉车为例进行了载重试验,分别采用经验模态分解(EMD)类去噪算法、小波阈值类去噪算法,以及1D TV算法处理液压信号,并将处理结果进行了对比分析。研究结果表明:相比于EMD类去噪算法,采用参数τ优化后的1D TV算法得到的信噪比可以提高5.0964 db,可比小波阈值类去噪算法提高0.653 db,其计算效率是EMD类去噪算法的113倍,是小波阈值类去噪算法的11倍;1D TV算法具有快速、非迭代和逐点扫描的特性,因而更适用于对载重估计模型中的液压系统压力信号进行实时去噪。

Aiming at the oscillation problem of hydraulic measurement signal caused by inertial impact,working device resonance and electromagnetic interference when engineering vehicle was loaded,for real-time denoising of hydraulic system pressure signal in load estimation model,a real-time denoising method of hydraulic system pressure signal based on one-dimensional total variation(1D TV)algorithm was proposed.Firstly,the basic principle of 1D TV was described.Then,a load estimation model which was mainly affected by hydraulic pressure was established,using the fmincon nonlinear programming solver embedded in MATLAB,the load estimation model was taken as the black box function,the root mean square error between the load estimate and the true value of the load was taken as the objective function value,and the parameterτin 1D TV algorithm was taken as the optimization variable,thus the parameterτwhich can approximate the true hydraulic signal curve was optimized.Finally,the load test of an enterprise telescopic handler was carried out,and the results of 1D TV algorithm processing hydraulic signal were compared with the results of empirical mode decomposition(EMD)denoising algorithm and wavelet threshold denoising algorithm.The results show that the signal to noise ratio of the one dimensional total variational algorithm optimized by parameterτis 5.0964 db higher than that of EMD type and 0.653 db higher than that of wavelet threshold type.The calculation efficiency is 113 times higher than that of EMD type and 11 times higher than that of wavelet threshold type.1D TV algorithm with fast,noniteration,and point by point scanning is more suitable for realtime denoising of pressure signal of hydraulic system in load estimation model.