摘要
研究了油水两相流温度波动信号的特点,提出了一种基于Hilbert-Huang translation（HHT）的含油率的PSO-LSSVM测量模型。将HHT作为自适应滤波滤掉温度信号中的噪声成分,对分解出的各个高频分量进行了Fourier能量谱分析,发现最大的能量值与含油率密切相关。同时对分解出的各个趋势分量的瞬时频率进行概率密度统计,当某一频率成分的能量占总能量的20%以上时,即可以停止EMD筛选,从剩余分量中提取了温度信号的幅值波动最大值,发现与含油率有较好的对应关系。因此将Fourier能量谱最大值和幅值波动最大值与油水总流量一起作为LSSVM模型输入量,含油率作为输出量进行训练学习。为了提高含油率模型的测量精度和运算速度,采用改进的粒子群算法（PSO）优化了含油率的LSSVM模型参数。实验结果表明,改进的PSO-LSSVM模型在0%~70%含油率范围内,含油率的测量精度较高。
A modified PSO-LSSVM model based on Hilbert-Huang translation（HHT） was proposed,which is capable to be used for the oil holdup measurement of oil-water two phase flow by researching on the characters of temperature fluctuated signal.Noises are removed from temperature signal by the empirical mode decomposition（EMD） with its ideal performance on local adaptability.The high frequency components were analyzed by the Fourier energy spectrum,and it was found that the maximum energy value is related to oil holdup.Then the residue components were analyzed by the Hilbert transform.When the probability density of a certain frequency component exceeds 20%,the sifting process will be terminated.The instantaneous fluctuated amplitude that relates to oil holdup was extracted from the last residue component.Subsequently,the maximum Fourier energy value,the instantaneous fluctuated amplitude and the total flux of oil-water two phase flow were employed as the inputs of LSSVM model,and the oil holdup was used as the output.In order to improve the measurement accuracy and the calculation velocity of the oil holdup,an improved PSO was adopted to determine the optimal LSSVM model parameters automatically.The experimental results indicate that,when the oil holdup is in the range of 0% to 70%,the proposed PSO-LSSVM model has good precision for the oil holdup measurement of oil-water two phase flow.
出处
《高校化学工程学报》
EI
CAS
CSCD
北大核心
2010年第5期758-763,共6页
Journal of Chemical Engineering of Chinese Universities
基金
国家高技术研究发展计划项目(2007AA04Z180
08JCYBJC11800)
