过程控制的对象有些是复杂非线性系统，其控制依赖于人的智能经验，它们是仅靠常规控制策略难以完成的。本文采用神经学习机制和模糊推理集成的思想来实现这类控制器，而我们提出的一种隐节点校正学习算法保证了这一思想的实现，以此思想为基础的优化控制系统已试运行在广东茂名石化公司二重整装置上，初步的记录数据表明这一方案是可行的 Some of the objects of process control are complex non-linear systems whose control depends on human’s intelligent experience, which is difficult to accomplish with conventional control strategies. In this paper, neural learning mechanism and fuzzy reasoning to achieve the integration of these ideas to achieve such controllers, and we propose a hidden node correction learning algorithm to ensure the realization of this idea, based on the idea that the optimal control system has been commissioned in Guangdong Maoming Petrochemical Company double reformer, the preliminary records show that this program is feasible