基于优化BP神经网络的非稳态精馏过程建模

Modeling of unsteady distillation process based on optimized BP neural network

间歇精馏已广泛应用于小批量、高附加值的精细化工和制药行业。然而在非稳态精馏过程中,一些重要的质量参数如产品组成等难以被直接测量,导致常规控制方法效果不佳。为解决这一问题,提出基于BP神经网络的非稳态精馏过程软测量模型,网络结构为3-12-1,再利用遗传算法、麻雀搜索算法等智能优化算法对网络进行优化,使用优化后的BP神经网络对非稳态精馏过程的产品质量进行预测。结果表明:经麻雀搜索算法优化后,BP神经网络的均方误差为4.32×10^(-5),与标准BP神经网络相比降低约59%,与遗传算法优化后的BP神经网络相比降低约26%。因此采用麻雀搜索算法优化后的BP神经网络建立非稳态精馏过程软测量模型,能够实现更高的预测精度。

Batch distillation has been widely used in low-volume,high-value added fine chemical and pharmaceutical industries.However,some important quality parameters,such as product composition,are difficult to measure.Conventional control methods are not effective.In order to solve this problem,a soft sensing model of unsteady distillation process based on BP neural network was proposed,and the network structure was 3-12-1.Then intelligent optimizatiion algorithms such as Genetic Algorithm and Sparrow Search Algorithm were used to optimize BP neural network.The results show that the mean square error of BP neural network is 4.32×10^(-5) after optimized by Sparrow Search Algorithm.Compared with BP neurral network,the mean square error is reduced by about 59%.Compared with BP neural network optimized by Genetic Algorithm,the mean square error is reduced by about 26%.Therefore,using BP neural network optimized by Sparrow Search Algorithm to establish the soft sensing model of unsteady distillation process can achieve higher prediction accuracy.