【摘要】 提出了一种在线测量混炼胶黏度的新方法。该方法以神经网络建模技术为基础,综合考虑混炼过程各因素对胶料黏度的影响,建立起胶料黏度的在线测量模型。在建模过程中,采用RBF神经网络描述胶料黏度变化的非线性过程,研究了一种自适应训练算法自动确定合适的隐层单元数目,从而获得了结构紧凑、精度较高的测量模型。根据此方法,研制了以DSP为核心的测量仪,并进行了现场试验。试验结果表明该仪器能够取得良好的测量效果。更多还原

【Abstract】 A novel approach for on-line measurement of viscosity of rubber mixture was proposed.In the approach,neural network modeling technology was applied to construct the viscosity measurement model.During modeling,RBF neural network was employed to approximate the non-linearity of viscosity of rubber mixture,a self-adaptation learning algorithm was proposed to determine the optimum number of hidden units in network.It make measurement model compact and accurate.Based on it,an apparatus was developed and test on an F370 mixer was conducted.The test result shows the apparatus can obtain almost the same measurement precitsion as that obtained by lab experiment.更多还原