基于GRU-NN预测模型的压电作动器MPC-KAN控制方法  

MPC-KAN Control Method for Piezoelectric Actuators Based on GRU-NN Prediction Model

作　　者：郭辰星 李自成[1] 徐瑞瑞 GUO Chenxing;LI Zicheng;XU Ruirui(School of Electrical and Information Engineering,Wuhan Institute of Technology,Wuhan 430205,China)

机构地区：[1]武汉工程大学电气信息学院,湖北武汉430205

出　　处：《压电与声光》2025年第1期157-162,171,共7页Piezoelectrics & Acoustooptics

基　　金：湖北省高等学校中青年科技创新团队项目(T2022012)。

摘　　要：为了提高压电作动器(PEAs)的轨迹跟踪性能,提出了一种基于门控递归单元(GRU)神经网络(NN)预测模型的Kolmogorov-Arnold网络前馈模型预测控制(MPC-KAN)。与神经网络逆模型控制不同,该方法使用GRU-NN正向建模,并根据模型预测结果调整模型预测控制(MPC)的输出。首先,根据线性化模型选择GRU-NN的训练输入特征,并训练该网络。然后,为了提高优化效果和缩短优化时间,将麻雀搜索算法(SSA)用作MPC优化器,并建立Kolmogorov-Arnold网络(KAN)以替代SSA优化。该方法的有效性在PEAs平台上得到验证,与传统方法相比,控制精度提高了约30%。To improve the trajectory tracking performance of piezoelectric actuators(PEAs),this study proposes a Kolmogorov-Arnold network feedforward model predictive control(MPC-KAN)based on a gated recurrent unit(GRU)neural network(NN)prediction model.Unlike neural network inverse model control,this method uses GRU-NN forward modeling and adjusts the model predictive control(MPC)output based on the model prediction results.First,this study selects the training input features of GRU-NN based on a linearized model and trains the network.Then,to improve optimization performance and shorten optimization time,the sparrow search algorithm(SSA)was used as the MPC optimizer,and a Kolmogorov-Arnold network(KAN)was established to replace the SSA optimization.The effectiveness of this method has been verified on the PEA platform.Compared with traditional methods,the control accuracy has been improved by approximately 30%.

关 键 词：压电陶瓷作动器 高精度跟踪 模型预测控制 GRU网络 KAN网络 

分 类 号：TN384[电子电信—物理电子学]