基于频域特征提取与RBF相结合的磨矿浓度软测量

Soft sensing for grinding concentration based on frequency-domain-feature extraction and RBF network

为解决磨矿浓度难以直接检测的问题,提出一种通过磨机振动、磨音信号频域特征提取利用特征频谱与径向基函数(RBF)神经网络相结合的非线性建模方法。采用快速傅里叶变换(FFT)将时域振动及磨音信号转换为频谱变量,对频谱变量通过主元分析(PcA)进行谱特征提取,采用径向基函数(RBF)变换实现谱特征的非线性映射。实验表明,该方法可以实现对磨矿浓度的准确软测量,提高测量精度1%,方法有效。

Because it is very difficult to detect the grinding concentration,in order to solve this problem, the nonlinear modeling method of radial basis network（RBF） combining with feature-extraction in frequency domain of the vibration and mill sound signals is proposed.The time domain vibration and mill sound signals are converted to frequency spectrum by using fast Fourier transform（FFT）. Then,the spectral features of them are extracted by using the principal component analysis（PCA） algorithm, and the power spectrum non-linear mapping is achieved by using the radial basis function （RBF） network.The experiments show that the modeling method can precisely achieve a soft sensing to the grinding concentration,and the experimental results verify the effectiveness of the method.