Scholarlink

<p>Deterministic learning theory combines the concepts of adaptive control and dynamics, based on radial basis function (RBF) neural network (NN), to obtain knowledge and rules from data. This paper utilizes the pressure of the Rijke tube control system platform to establish a dynamic model of thermoacoustic unstable system which depends on the deterministic learning theory. Furthermore a deterministic learning control algorithm based on the dynamic model is proposed to stabilize the system. Firstly, when the thermoacoustic instability occurs inside the Rijke tube, the pressure data at multiple different sampling instants are taken as the inputs of the RBF NN to train the constant RBF NN model. Secondly, the constant RBF NN is used to construct a dynamic predictor to predict the future states of the thermoacoustic unstable system. Finally, based on the constant RBF NN model, an RBF NN closed-loop control system is constructed, the deterministic learning control algorithm is proposed to effectively suppress the pressure and velocity oscillation caused by the thermoacoustic instability.</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

Deterministic Learning-based Modeling and Control for Thermoacoustic Systems in a Rijke Tube

  • 期刊论文

  • 发表日期：

    2025年6月11日

查看全文
  • Biao Wang,
  • Dan Ma,
  • Yuchen Quan

摘要

Deterministic learning theory combines the concepts of adaptive control and dynamics, based on radial basis function (RBF) neural network (NN), to obtain knowledge and rules from data. This paper utilizes the pressure of the Rijke tube control system platform to establish a dynamic model of thermoacoustic unstable system which depends on the deterministic learning theory. Furthermore a deterministic learning control algorithm based on the dynamic model is proposed to stabilize the system. Firstly, when the thermoacoustic instability occurs inside the Rijke tube, the pressure data at multiple different sampling instants are taken as the inputs of the RBF NN to train the constant RBF NN model. Secondly, the constant RBF NN is used to construct a dynamic predictor to predict the future states of the thermoacoustic unstable system. Finally, based on the constant RBF NN model, an RBF NN closed-loop control system is constructed, the deterministic learning control algorithm is proposed to effectively suppress the pressure and velocity oscillation caused by the thermoacoustic instability.