Abstract: Artificial neural networks were the important part of artificial intelligence,which had a broad space in improving the chemical process of traditional production techniques diagnosis of lag,difficult to optimize and control,large property estimation error and could not deal with complex nonlinear problems. Artificial neural networks have drawn much attention because of superior robustness,fault tolerance,approximation of complex nonlinear correlations,parallel processing and adaptive learning. Artificial neural networks have been applied to chemical processes in the following areas:fault diagnosis,process control and optimization,quality control,quantitative structure-activity/property correlation analysis,property estimation,expert system and clustering analysis. This paper summarized the theory and development history of artificial neural network,and conducted meta-analysis of the literature on the principles and the development of artificial neural networks in chemical processes. Finally,the paper pointed out that the deep learning algorithm had advantages of high-performance and high speed,and then discussed that the future study of neural networks in chemical process would be the direction and hot topic in the development and application of deep learning algorithm.

Key words: neural networks, model, chemical engineering process, production, principle

摘要： 人工神经网络作为人工智能的重要组成部分，以其超强的鲁棒性、容错性、可充分逼近任何复杂的非线性关系、并行处理、可学习和自适应等优点在改善化工过程传统生产技术诊断滞后、难以优化控制、物性估算误差较大以及不能处理非线性复杂情况等问题上有着广阔的发展空间。本文概述了人工神经网络的原理和发展历程，综述并分析了人工神经网络在故障诊断、过程控制和优化、质量控制、定量结构-活性/性质相关性分析、物性估算、专家系统以及聚类分析等化工过程中的应用原理以及研究进展和现状。最后指出卷积神经网络等深度学习算法的性能高、速度快，在化工过程中发展和应用深度学习算法将成为其发展方向和研究热点。

关键词: 神经网络, 模型, 化工过程, 生产, 原理