Article
Mathematics, Applied
Francesc Arandiga, Rosa Donat, Daniela Schenone
Summary: This paper presents and analyzes several prediction strategies in the 2D setting based on multi-quadric radial basis function interpolation with either linear or Weighted Essentially Non Oscillatory (WENO) shape parameter approximation. These prediction operators give rise to sparse multi-scale representations of 2D signals, and their compression capabilities are demonstrated through numerical experiments. The local adaptive estimates of the shape parameters lead to non-separable, fully 2D reconstruction strategies, which in turn result in efficient compression algorithms.
APPLIED MATHEMATICS AND COMPUTATION
(2023)
Article
Computer Science, Interdisciplinary Applications
Yang Bai, Maojin Tan
Summary: The study proposed a dynamic committee machine with fuzzy-c-means clustering (DCMF) to predict the total organic carbon (TOC) content in shale reservoirs. DCMF utilizes multiple experts and subtasks decomposition to improve prediction accuracy and model performance.
COMPUTERS & GEOSCIENCES
(2021)
Article
Computer Science, Information Systems
Kun Cao, Cong Zhang, Liangliang Li, Shuaifeng Li
Summary: This study proposes a dynamic neural network optimization model to improve the accuracy of soil heavy metal content prediction. Experimental results demonstrate that the proposed model outperforms other commonly used models in terms of accuracy and prediction.
Article
Computer Science, Artificial Intelligence
Yukun Zheng, Yixiang Liu, Rui Song, Xin Ma, Yibin Li
Summary: This study proposes an adaptive robust control strategy based on RBFNN and a state observer for a mobile manipulator with uncertain dynamics and external disturbances. The proposed method achieves precise position tracking in the task space by implementing virtual speed tracking control and control torque conversion. Theoretical analysis demonstrates the global asymptotic stability of the system under the control of the proposed method.
Article
Materials Science, Multidisciplinary
Jianfei Huang, Kai Guo, Xiaotao Liu, Zhen Zhang
Summary: This paper proposes an eigenstrain reconstruction method based on radial basis function (RBF) for predicting residual stress. By using the least squares method, the full-field residual stress can be reconstructed by solving an inverse eigenstrain problem. The novel elliptical radial basis function is used to accurately predict complex residual stresses.
MECHANICS OF MATERIALS
(2023)
Article
Multidisciplinary Sciences
Jingwei Liu, Peixuan Li, Xuehan Tang, Jiaxin Li, Jiaming Chen
Summary: By introducing Wavelet Neural Network (WNN) and Wavelet-based Convolutional Neural Network (WCNN), the problems of BPNN and RBFNN can be solved and the performance of neural networks can be improved. The proposed Convolutional Wavelet Neural Network (CWNN) based on WNN effectively reduces the mean square error and error rate of CNN, achieving better maximum precision than CNN.
SCIENTIFIC REPORTS
(2021)
Article
Engineering, Aerospace
Yong Zhao, Ning Wang, Quan Chen, Sunquan Yu, Xiaoqian Chen
Summary: This paper proposes a new method using subregion radial basis function to predict satellite coverage traffic volume (SCTV) and guide the design of adaptive ADS-B antenna strategy. Simulation results show that the proposed method achieves a better balance between predictive accuracy and computational cost, while maintaining high detection probability and power savings.
Article
Computer Science, Artificial Intelligence
Xi Meng, Yin Zhang, Junfei Qiao
Summary: An adaptive task-oriented radial basis function (ATO-RBF) network was developed to design prediction models for accurate timely acquirements of effluent BOD and TN in wastewater treatment plants. The network combined error correction-based growing strategy and second-order learning algorithm to enhance learning ability and generalization performance of prediction models. The ATO-RBF network analysis based on the Lyapunov criterion showed superior prediction accuracy compared with conventional methods.
NEURAL COMPUTING & APPLICATIONS
(2021)
Article
Thermodynamics
Ali Sohani, Siamak Hoseinzadeh, Saman Samiezadeh, Ivan Verhaert
Summary: An enhanced design for a solar still desalination system was employed to develop artificial neural network (ANN) models, with FF and RBF types identified as the best structures for predicting distillate production and water temperature. Error analysis on data not used for ANN model development showed varying errors in different months.
JOURNAL OF THERMAL ANALYSIS AND CALORIMETRY
(2022)
Article
Engineering, Multidisciplinary
Linxiong Hong, Huacong Li, Kai Peng
Summary: This paper introduces an efficient sequential sampling method combined with radial basis function based on the Kriging method for reliability analysis, aiming to reduce modeling complexity and uncertainties. A novel active learning function is developed to effectively search for sequential samples, while a convergence criterion based on failure probability from cross-validation is constructed to terminate the sampling process. The method demonstrates high precision, efficiency, and applicability in structural reliability analysis through numerical examples.
APPLIED MATHEMATICAL MODELLING
(2021)
Article
Mathematics
Lin-Tian Luh
Summary: In this paper, we propose a method of directly choosing the shape parameter c in the multiquadrics using MN-curve theory for the RBF collocation method. Experiments show that the quality of the obtained c value is very close to the best approximation error among all possible choices.
Article
Computer Science, Artificial Intelligence
Zhiyong Zhou, Dongbing Tong, Qiaoyu Chen, Wuneng Zhou, Yuhua Xu
Summary: This paper discusses the use of radial basis function-neural networks for approximation in nonlinear systems and dynamic surface control method to address complexity issues, ensuring global asymptotic stability through Lyapunov stability theory. The effectiveness of the proposed control technique is validated through simulation examples.
Article
Mathematics, Applied
Xin Xu, Xiaopeng Luo
Summary: This paper proposes two adaptive approximations for multivariate scattered data, namely sparse residual tree (SRT) and sparse residual forest (SRF). SRT provides sparse and stable approximations in areas with sufficient or redundant data, and identifies possible local regions with insufficient data; while SRF combines SRT predictors to enhance approximation accuracy and stability. The hierarchical parallel SRT algorithm is based on tree decomposition and adaptive radial basis function explorations, achieving convergence results with time complexity O(N log(2) N) and worst case storage requirement O(N log(2) N). Numerical experiments validate the effectiveness of the proposed methods.
JOURNAL OF SCIENTIFIC COMPUTING
(2022)
Article
Physics, Multidisciplinary
Menghan Li, Shaobo Li, Junxing Zhang, Fengbin Wu, Tao Zhang
Summary: This paper proposes an adaptive funnel dynamic surface control method for the permanent magnet synchronous motor with time delays. The method integrates an improved prescribed performance function with a modified funnel variable to achieve unconstrained output. It utilizes a disturbance observer and radial basis function neural networks to estimate disturbances and unknown nonlinearities. By constructing Lyapunov-Krasovskii functionals, it compensates for time delays and enhances control performance. Through detailed stability analysis, the boundedness and binding ranges of all signals are ensured.
Article
Multidisciplinary Sciences
Osama Siddig, Hany Gamal, Pantelis Soupios, Salaheldin Elkatatny
Summary: This paper presents the application of two AI approaches in predicting total organic carbon content in Devonian Duvernay shale. The results show that the ANFIS method yields the best predictions. The study found that gamma ray has the most significant impact on TOC prediction.
SN APPLIED SCIENCES
(2022)