论文部分内容阅读
针对目前采用经验模态分解(empirical model decomposition,EMD)得到的系列子信号构建的磨机负荷参数软测量模型泛化性能差、难以进行清晰物理解释,以及EMD算法存在的模态混叠等问题,本文提出了基于选择性融合多尺度筒体振动频谱的建模方法.首先采用EMD、集合EMD(ensemble EMD,EEMD)、希尔伯特振动分解(Hilbert vibration decomposition,HVD)共3种多组分信号自适应分解算法获得磨机筒体振动多尺度子信号的集合,接着通过相关性分析剔除虚假无关部分,然后再将与原始信号相关性强的那部分多尺度子信号变换至频域,进而更有利于构建这些多尺度频谱与磨机负荷参数间的映射模型,最后通过改进分支定界选择性集成(improved branch and bound based selective ensemble,IBBSEN)算法建立软测量模型,实现对多源多尺度筒体振动频谱的最优选择性信息融合.基于实验球磨机运行数据的仿真实验表明所提方法在模型可解释性和泛化性能上均优于之前研究所提出方法.
In view of the poor generalization performance of the mill load parameter soft-sensing model constructed by the series of sub-signals obtained by the empirical model decomposition (EMD), it is difficult to explain clearly the physical problems and the modal aliasing existing in the EMD algorithm , A modeling method based on selective fusion of multi-scale cylinder vibrational spectrum is proposed in this paper.First, EMD, ensemble EMD (EEMD) and Hilbert vibration decomposition (HVD) Then the correlation coefficient is used to remove the false unrelated part and then convert the part of multiscale subsignals with strong relativity to the original signal into the frequency domain. And then it is more conducive to construct the mapping model between these multi-scale spectrum and mill load parameters. Finally, an improved branch and bound based selective ensemble (IBBSEN) algorithm is used to establish the soft-sensing model, Optimal Selective Information Fusion of Vibrational Spectrum of Scale Cylinder Based on the experimental data of the experimental ball mill, The proposed method is superior to the one proposed in previous studies in terms of model interpretability and generalization performance.