Scholarlink

<p>This paper introduces an adaptive controller based on a bilinear quantum recurrent neural network (AC-BLQRNN). The proposed neural structure uses five bilinear quantum neurons in the input layer, one quantum neuron in the hidden layer, and one linear neuron is used for the output layer. Thus, the proposed algorithm consists of a (5–1–1) neural structure with only nine tunable parameters that aims to reduce the computation time. Moreover, it merges the merits of the bilinear neural network that requires a little number of hidden neurons and the powerful processing of the quantum neurons. In addition, a BLQRNN identifier is designed with the same proposed neural network structure, i.e., (5–1–1) to find the sensitivity function. All adjustable parameters in the proposed structure are learned using an updating rule derived using the Lyapunov stability criterion to stabilize the proposed learning algorithm. The proposed AC-BLQRNN is implemented practically and applied to a nonholonomic two-wheel mobile robot (NHWMR) to control the wheels' speeds for tracking a specific trajectory. Moreover, several tasks are performed on NHWMR to investigate the robustness of the proposed AC-BLQRNN controller. The experimental results indicate the powerful computing, fast convergence, and robust performance of the proposed AC-BLQRNN over other existing controllers.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Bilinear quantum recurrent neural network-based real-time adaptive controller

  • 期刊论文

  • 发表日期：

    2025年9月28日

查看全文
  • Youssef F. Hanna,
  • Ahmad M. El-Nagar,
  • Mohammad El-Bardini,
  • A. Aziz Khater

摘要

This paper introduces an adaptive controller based on a bilinear quantum recurrent neural network (AC-BLQRNN). The proposed neural structure uses five bilinear quantum neurons in the input layer, one quantum neuron in the hidden layer, and one linear neuron is used for the output layer. Thus, the proposed algorithm consists of a (5–1–1) neural structure with only nine tunable parameters that aims to reduce the computation time. Moreover, it merges the merits of the bilinear neural network that requires a little number of hidden neurons and the powerful processing of the quantum neurons. In addition, a BLQRNN identifier is designed with the same proposed neural network structure, i.e., (5–1–1) to find the sensitivity function. All adjustable parameters in the proposed structure are learned using an updating rule derived using the Lyapunov stability criterion to stabilize the proposed learning algorithm. The proposed AC-BLQRNN is implemented practically and applied to a nonholonomic two-wheel mobile robot (NHWMR) to control the wheels' speeds for tracking a specific trajectory. Moreover, several tasks are performed on NHWMR to investigate the robustness of the proposed AC-BLQRNN controller. The experimental results indicate the powerful computing, fast convergence, and robust performance of the proposed AC-BLQRNN over other existing controllers.