基于RNN补偿器的欠驱动过程控制方法应用

Application of Underactuated Process Control Method Based on RNN Compensator

化工过程中存在一类操作变量维度小于目标被控变量维度的过程。针对该类过程,现行控制方法的控制效果仍有待提高。为实现对上述化工过程的良好控制,类比欠驱动机械系统提出了基于循环神经网络(recurrent neural network,RNN)补偿器的控制方法。该方法基于模糊逻辑与信息融合设计主控器,并通过额外串联模糊控制器实现了传统模糊控制器的量化因子与比例因子的自适应整定,又运用RNN设计了补偿控制器,通过以上改进解决了模型与实际过程之间的未建模动态性能与参数不确定性偏差导致的动态性能较差和存在稳态误差的问题。最后,以二级二路压缩过程为例进行了仿真,仿真结果证明了所提控制方法的有效性。

In chemical process,there is a kind of process in which the dimension of the operating variable is smaller than the dimension of the target controlled variable.The control effect of the current control methods for this kind of process still needs to be improved.In order to achieve a good control of the above chemical process,a control method based on recurrent neural network(RNN)compensator is proposed by analogy to the underactuated mechanical system.The method is based on fuzzy logic and information fusion to design the main controller,and the adaptive tuning of the quantization factor and scale factor of the traditional fuzzy controller is realized through an additional series fuzzy controller,and a compensation controller is designed by using RNN.The above improvements solve the problems of poor dynamic performance and steady-state errors due to unmodeled dynamic performance and parameter uncertainty deviation between the model and the actual process.Finally,the simulation is carried out by taking the two-stage two-way compression process as an example,and the simulation results prove the effectiveness of the proposed control method.