null

论文期刊

论文标题    SVM-based Multisensor Data Fusion for Phase Concentration Measurement in Biomass-coal co-combustion
作者    XiaoxinWang, Hongli Hu, Huiqin Jia and Kaihao Tang
发表/完成日期    2018-05-08
期刊名称    Rev. Sci. Instrum.
期卷    89（5）
相关文章   
论文简介    Abstract: In this paper, the electrical method combines electrostatic sensor and capacitance sensor to measure the phase concentration of pulverized coal/biomass/air three-phase flow through data fusion technology. In order to eliminate the effects of flow regimes and improve the accuracy of phase concentration measurement, the Mel Frequency Cepstrum Coefficient (MFCC) features extracted from electrostatic signal are used to train the Continuous Gaussian Mixture Hidden Markov Model (CGHMM) for flow regime identification. Support Vector Machine (SVM) is introduced to establish the concentration information fusion model under identified flow regimes. The CGHMM models and SVM models are transplanted on DSP to realize on-line accurate measurement. DSP flow regime identification time is 1.4ms, and the concentration predict time is 164s, which can fully meet the real-time requirement. The average absolute value of relative error of the pulverized coal is about 1.5% and that of the biomass is about 2.2%.
返回