论文部分内容阅读
针对工业生产过程一类时滞多变量系统,采用BP神经网络实现时滞多变量系统的动态解耦。时滞环节增加了解耦器的设计难度,并导致解耦器在物理上难以实现。针对该问题,对时滞多变量离散系统的解耦进行了讨论,为降低神经网络解耦器的规模,采用离散状态方程模型的均衡实现降阶算法对神经网络解耦器进行降维。以典型的火电机组协调系统进行解耦仿真试验,并采用PID控制器实现控制。结果表明,采用的离散化方法对时滞多变量系统具有良好的解耦效果,合理解决了时滞对解耦过程的影响,并通过模型降阶技术降低了神经网络解耦器的规模,便于神经网络解耦实现在线学习。
Aiming at a class of multivariable systems with time delay in industrial process, the dynamic decoupling of multivariable systems with time delay is implemented by BP neural network. The time lag increases the design difficulty of the decoupler and leads to the fact that the decoupler is physically difficult to achieve. To solve this problem, the decoupling of multivariable discrete-time systems with delay is discussed. In order to reduce the size of the decoupler of neural networks, the decentralized algorithm of reduced-order equations is used to reduce the size of neural network decoupler. A typical thermal power unit coordination system decoupling simulation test, and the use of PID controller to achieve control. The results show that the discretization method has a good decoupling effect for time-delay multivariable systems, reasonably solves the influence of delay on the decoupling process, and reduces the size of the neural network decoupler through model reduction techniques, Neural Network Decoupling to Realize Online Learning.