Long Short Term Memory Based Self Tuning Regulator Design for Nonlinear Systems

Sanatel C., Öke Günel G.

NEURAL PROCESSING LETTERS, 2022 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Publication Date: 2022
  • Doi Number: 10.1007/s11063-022-10997-1
  • Journal Name: NEURAL PROCESSING LETTERS
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, PASCAL, Applied Science & Technology Source, Compendex, Computer & Applied Sciences, Information Science and Technology Abstracts, INSPEC, zbMATH, DIALNET
  • Keywords: Adaptive PID Controller, Long short term memory, System identification, Self tuning regulator
  • Istanbul Technical University Affiliated: Yes

Abstract

In this paper, a Long Short Term Memory (LSTM) based Self Tuning Regulator (STR) for trajectory tracking problem of nonlinear systems is proposed. In the STR, a Proportional Integral Derivative (PID) controller is used as an adaptive parametric controller. The system model is estimated at every time step since it is utilized in computing the system Jacobian, hence controller design involves an inherent system identification problem. In the proposed architecture, LSTM is employed for both system model estimation and for updating the parameters of the PID controller. Namely, the K-P, K-I and K-D gains are computed at every time step by LSTM, so that a cost function which is obtained from tracking error is minimized. The performance of the proposed method has been evaluated on two different nonlinear systems by extensive simulations. Simulation results justify the success of the introduced control architecture.