利用B样条神经网络控制聚合物相对分子质量分布新方法

Control of molecular weight distribution of polymerization via B-spline neural networks

文中利用B样条神经网络可以逼近连续函数的良好性质,构造合适的网络模型从而得到相对分子质量分布模型的线性表示方法,根据最优化方法中的梯度法,求得最优控制序列。应用在苯乙烯聚合反应的仿真实验中,收到了很好的控制效果,证明了方法的可行性。

The characteristics of B-spline functions and the structure of B-spline neural networks was introduced and a new way was presented to build the neural networks and to get a linear express of molecular weight distribution. An optimization algorithm based on the linear express was used to calculate the optimal control sequences. A new control procedure was utilized on a simulated object and a well control led result was got that demonstrated the feasibility of this new way.