基于DCS的神经网络控制的工程实现

介绍了LN2000分散控制系统的软件设计方案。在系统功能模块的设计过程中,充分利用先进控制策略在热工过程控制中的应用研究成果,设计并开发了先进控制算法模块。详细介绍了该模块的输入输出变量和参数设置,论述了模块内部进行的手自动无扰切换的处理方法。针对火电厂主汽温被控对象,采用基于单个神经元的自适应控制器组成串级控制系统,利用LN2000分散控制系统平台,进行实时仿真研究。结果表明,采用基于单个神经元的自适应控制器的主汽温串级控制系统具有良好的跟踪性能和较强的鲁棒性。

Engineering Implement of Neural Networks Control Based on DCS

The software scheme of LN2000 system is introduced. In the processing of design for modules of system, modern control algorithm modules are designed and developed, making use of application results of modern control strategy in thermal processing control sufficiently. For neural networks control especially, the module with adaptive control based on a neuron is developed successfully. Input variables, output variables and parameters of the module is introduced particularly, and the method of non-turbulent switch from manual control to automatic control is discussed. According to main steam temperature object in the thermal power plant, a cascade control system is established with the adaptive controller based on a neuron, and real time simulation is carried out with LN2000 DCS platform. Tracking performance and strong robust of the system are proved by real time simulation results.