摘要
针对静电传感器无法给出颗粒质量流量绝对值以及多相流流动形态和结构变化影响传感器输出等问题,提出了一种基于分解合成的多模型加权平均的固相质量流量非线性软测量模型。在高压密相气力输送系统上,通过静电传感器获得大量试验数据,提取信号特征,利用模糊聚类算法将输入数据进行空间分区,每一区间上用径向基函数(RBF)神经网络辨识出一个子模型,再利用模糊推理将各子模型输出加权求和得到颗粒质量流量的估计值。该模型减小了流型对测量结果的影响,提高了测量精度。
Aiming to resolve the problems of the electrodynamic sensor's deficiency in absolute mass flow rate measurement and the effect of flow regime on the output of the sensor, a multi-modeling based nonlinear soft sensor for particle mass flow rate is introduced. In the dense phase pneumatic conveying system under high pressure, abundant experimental data could be obtained by using the electrodynamic sensor and the signal characteristics of the experimental results could be extracted. The characteristic data space is a non-linear sub-model is etwork. Finally the whole tsed sub-models. The soft es an effective solution to partitioned into some local regions by the fuzzy clustering algorithm firstly, then established for each local region by using the radial basis function (RBF) neural n soft measurement model could be accurately described by a set of fuzzy rules ba model reduces the influence of flow regime on the measurement results and provid on-line mass flow rate measurement of pneumatically conveyed solid particles.
出处
《化工学报》
EI
CAS
CSCD
北大核心
2007年第9期2225-2231,共7页
CIESC Journal
基金
国家重点基础研究发展计划项目(2004CB17702-04)
东南大学博士基金项目(9203002354)~~
关键词
颗粒流量
静电传感器
RBF神经网络
模糊聚类
气固两相流
particle flow rate
electrodynamic sensor
radial basis function neural network
fuzzyclustering
gas-solid two phase flow