复杂过程对象混合建模策略的研究

提出一种面向复杂过程的混合建模策略并应用于精馏塔的建模。混合模型包含基于基本原理的简化机理模型和基于神经网络的稳态、动态补偿机构。通过工况的识别以及模型参数的离线辨识和在线优化，混合模型可以准确地模拟复杂过程在大范围内的动态特性。仿真结果证明了混合建模策略的有效性。

Study of Hybrid Modeling Strategy for Complex Processes

A hybrid modeling strategy for complex process is presented and applied to simulate a distillation. It comprises a reduced first principles model and a steady neural network model with a dynamic regulation structure. By recognition of work condition, off-line identification and on-line optimization of parameters, hybrid model can be used to simulate dynamics of complex process correctly in a large scale. Simulation results show that hybrid modeling strategy is valid in general purpose.