【作者】 司刚全； 曹晖； 张彦斌； 马西奎；

【Author】 Si Gangquan Cao Hui Zhang Yanbin Ma Xikui (School of Electrical Engineering,Xi’an Jiao Tong University Xi’an 710049,China)

【机构】 西安交通大学电气工程学院；

【摘要】 本文针对火电厂磨机负荷传统单一信源检测方法存在的非线性、二义性、局部有效以及灵敏度低等缺点,提出了一种基于特征提取和自适应神经模糊推理系统(ANFIS)的软测量模型。该模型以磨机噪音和振动为输入信息,对特征频段内的能量进行提取获得表征负荷的特征量;特征量进入 ANFIS 进行融合计算获得负荷信息。ANFIS 采用混合学习算法对隶属度函数和线性输出函数参数进行调整,具有很好的自学习、自适应能力。试验仿真表明该软测量模型具有较高的预测精度, 能够显著提高磨机负荷检测的灵敏度和可信度。更多还原

【Abstract】 Based on the characteristic spectra extraction and the adaptive network-based fuzzy inference system (ANFIS),a soft sensing model is designed for measuring the ball mill load of power plant.This scheme avoids the shortcomings of traditional single source information measurement methods such as nonlinearity,ambiguity,local effectivity and low sensitivity.The characteristic extraction is used to generate the characteristic values of the two input variables from both the noise and the vibration of ball mill,and then ANFIS fuses the characteristic values to predicate the ball mill load.In ANFIS,the hybrid learning algorithm is used to adjust the membership functions and the parameters of linear output functions so as to possess the higher properties of the learning and adaptation. The simulation results obtained demonstrate that the soft sensing model can not only gain a good estimated preci- sion,but also improve the detecting sensitivity and reliability of the mill load measurement.更多还原