本文以一个三元精馏塔为研究对象，详细介绍了精馏过程的神经网络控制系统（ＮＮＣＳ）的设计过程。本ＮＮＣＳ主要包括ＮＮ预估器和ＮＮ控制器，前者被训练后用来精确表达精馏塔的输入到输出的非线性映射，后者被训练后用来对该ＮＮ预估器实现解耦控制。所采用的神经网络模型为ＢＰ网络，是在一个以ＰＣ机为宿主机带９片ＴＲＡＮＳＰＵＴＥＲ的并行处理系统上实现的 In this paper, a ternary distillation column is taken as the research object, and the design process of the neural network control system (NNCS) of the distillation process is introduced in detail. The NNCS mainly includes an NN predictor and an NN controller. The former is trained to accurately represent the input-output nonlinear mapping of the rectification column, and the latter is trained to decouple the NN predictor . The neural network model used is BP network, which is implemented on a parallel processing system with 9 TRANSPUTERs as PC host