论文部分内容阅读
基于神经网络方法研究了一类随机动态系统的建模和最优化问题。首先将采样样本和相关的时间区域段上的平均值样本,用来训练一个多层反传网络,以得到该随机过程的神经网络模型;然后,基于一给定的随机目标函数,该网络模型用来寻优以获得系统操作变量的最优均值设定值。这一网络建模和优化方法用于解决一尿素合成塔的优化问题。仿真结果表明了这一方法的有效性。
Based on the neural network method, a kind of stochastic dynamic system modeling and optimization problems are studied. Firstly, a sample of the samples and the mean sample in the relevant time zone segment are used to train a multi-layer back propagation network to obtain the neural network model of the stochastic process. Then, based on a given stochastic objective function, the network model Used to find the optimal mean-value setting for system manipulated variables. This network modeling and optimization method is used to solve the optimization problem of a urea synthesis tower. Simulation results show the effectiveness of this method.