Authors:
Kazuhiko Takahashi
;
Yuka Shiotani
and
Masafumi Hashimoto
Affiliation:
Doshisha University, Japan
Keyword(s):
Quantum Neural Network, Qubit neuron, Self-tuning Controller, PID Controller, Fuzzy Logic Controller.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Artificial Intelligence and Decision Support Systems
;
Computational Intelligence
;
Engineering Applications
;
Enterprise Information Systems
;
Informatics in Control, Automation and Robotics
;
Intelligent Control Systems and Optimization
;
Machine Learning in Control Applications
;
Neural Networks Based Control Systems
;
Robotics and Automation
;
Signal Processing, Sensors, Systems Modeling and Control
;
Soft Computing
Abstract:
This paper presents a self-tuning controller based on a quantum neural network and investigates the controller’s characteristics for control systems. A multi-layer quantum neural network which uses qubit neurons as an information processing unit is utilized to design an adaptive-type self-tuning controller which conducts
the training of the quantum neural network as an online process. As an example of designing the self-tuning controller, either a proportional integral derivative controller or a fuzzy logic controller is utilized as a conventional controller for which parameters are tuned by the quantum neural network. To evaluate the learning
performance and capability of the adaptive-type quantum neural self-tuning controller, we conduct computational experiments to control the single-input single-output non-linear discrete time plant. The results of the computational experiments confirm both feasibility and effectiveness of the proposed self-tuning controller.