基于迟滞特性图与ResNet的执行器迟滞检测方法研究

Research on Actuator Hysteresis Detection Method Based on Hysteresis Characteristic Map and ResNet

在工业控制系统中,执行器的迟滞非线性特性会导致控制回路性能下降。为实现执行器迟滞故障的准确检测,提出了一种基于迟滞特性图与残差网络(ResNet)的执行器迟滞检测方法。首先,通过执行器迟滞模型生成正常和迟滞状态下的数据,绘制执行器迟滞特性图。以蝶状图形作为执行器存在迟滞的判据,并将图形数据输入ResNet进行训练,得到执行器迟滞检测模型。然后,根据执行器实际运行数据绘制迟滞特性图,将其输入至训练好的网络中,以得到检测结果。最后,通过执行器半物理试验平台和国际阀门迟滞数据库对所提方法进行了试验验证。试验结果表明:相比传统迟滞检测方法,所提方法能够避免外部干扰和控制器参数调节不佳的影响,具有较高的鲁棒性和准确率。所提方法能够有效检测出控制系统的迟滞故障,对于保障控制系统的安全、稳定运行具有重要意义。

In industrial control systems,the hysteresis nonlinear characteristics of actuators can lead to degradation of control loop performance.To achieve accurate detection of actuator hysteresis faults,an actuator hysteresis detection method based on hysteresis characteristic map and residual network(ResNet)is proposed.Firstly,the actuator hysteresis model is used to generate the data in normal and hysteresis states,draw the actuator hysteresis characteristic graph,use the butterfly graph as the criterion for the existence of hysteresis in the actuator,and input the graph data to the ResNet for training to obtain the actuator hysteresis detection model.Then,the hysteresis characteristic graph is drawn according to the actual actuator operation data and input to the trained network to obtain the detection results.Finally,the proposed method is experimentally validated by the actuator semi-physical testbed and the international valve hysteresis database.The experimental results show that the proposed method can avoid the influence of external disturbances and poorly adjusted controller parameters and has higher robustness and accuracy than the traditional hysteresis detection methods.The proposed method can effectively detect hysteresis faults in the control system,which is important to ensure the safe and stable operation of the control system.