Guizhong Xie
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Guizhong Xie has a total of 43 co-authors. Here are the top co-authors by number of shared publications:
Hao Li
5 shared publications- Hongrui Geng
College of mathematics and information science, Zhengzhou University of Light Industry, Zhengzhou 450002, China
4 shared publications Liangwen Wang
3 shared publications- Yudong Zhong
Henan Provincial Key Laboratory of Intelligent Manufacturing of High-End Equipment, Zhengzhou University of Light Industry, Zhengzhou 450002, China
3 shared publications - Hao Li
Henan Provincial Key Laboratory of Intelligent Manufacturing of High-End Equipment, Zhengzhou University of Light Industry, Zhengzhou 450002, Henan, P. R. China
2 shared publications Jie Zhou
2 shared publications
Latest Publications
- Open Access
Efficient Thermal Analysis of On-Orbit Satellites Using a Nondimensionalized Physics-Informed Neural Network
This study proposes a nondimensional physics-informed neural network (Nd-PINN) framework for predicting the infrared radiation characteristics of low-Earth-orbit satellites under complex space thermal environments. The Laplace heat conduction equation and nonlinear boundary conditions involving solar irradiation, Earth albedo, and Earth infrared radiation are embedded into the loss function, while nondimensional preprocessing improves training stability by balancing residual magnitudes. Four numerical cases of increasing complexity were used for validation. For cases with analytical solutions, Nd-PINN achieved absolute temperature errors below 0.2 K, relative infrared radiation errors below 0.3%, and energy conservation errors below 0.001%. Under unidirectional solar irradiation, the model shows good consistency with the azimuthal temperature gradient predicted by Lambert’s cosine law, with an energy conservation error below 0.58%. In the multi-source coupled radiation case, it maintained strong physical consistency, with an energy conservation error below 0.54%. These results demonstrate that Nd-PINN can effectively solve both linear and strongly nonlinear thermal problems without complex meshing, while providing continuous and smooth global temperature fields. The proposed framework offers an accurate, mesh-free alternative for satellite thermal analysis and infrared feature identification.
Open AccessEfficient Thermal Analysis of On-Orbit Satellites Using a Nondimensionalized Physics-Informed Neural Network
This study proposes a nondimensional physics-informed neural network (Nd-PINN) framework for predicting the infrared radiation characteristics of low-Earth-orbit satellites under complex space thermal environments. The Laplace heat conduction equation and nonlinear boundary conditions involving solar irradiation, Earth albedo, and Earth infrared radiation are embedded into the loss function, while nondimensional preprocessing improves training stability by balancing residual magnitudes. Four numerical cases of increasing complexity were used for validation. For cases with analytical solutions, Nd-PINN achieved absolute temperature errors below 0.2 K, relative infrared radiation errors below 0.3%, and energy conservation errors below 0.001%. Under unidirectional solar irradiation, the model shows good consistency with the azimuthal temperature gradient predicted by Lambert’s cosine law, with an energy conservation error below 0.58%. In the multi-source coupled radiation case, it maintained strong physical consistency, with an energy conservation error below 0.54%. These results demonstrate that Nd-PINN can effectively solve both linear and strongly nonlinear thermal problems without complex meshing, while providing continuous and smooth global temperature fields. The proposed framework offers an accurate, mesh-free alternative for satellite thermal analysis and infrared feature identification.
- 1 March 2026
- Published by IOP Publishing in Engineering Research Express
- Vol. 8 (6) , 065517
A stress prediction approach employing finite element neural network integrated with Runge-Kutta
Stress concentration in hazardous zones of mechanical structural components during service can lead to fatigue cracks and structural failure. Accurate assessment of their stress fields is crucial for structural health monitoring. This paper proposes an evaluation method that combines finite element analysis with neural networks to establish a highly accurate mapping relationship between the boundary data and the maximum stress in the defect area. Latin hypercube sampling is employed to determine representative boundary measurement locations, and finite element simulations are used to construct the training dataset. To address the issues of neural networks getting trapped in local minima and slow convergence speed, the Runge–Kutta method is introduced to optimize the update process of the network’s weights and biases, aiming to enhance the global search ability of the neural network, making it more likely to break free from the constraints of local minima and find the global optimal solution. Several structural components are provided to validate the effectiveness and generalization capability of the proposed network model. The results demonstrate that, the 95th-percentile error range obtained by proposed method remains within 3.1%, significantly lower than that of neural networks (17.5%) and Elman networks (15.9%). Segmented error analysis across low-, medium-, and high-stress regimes further confirms that the proposed method suppresses error amplification under high-load conditions, maintaining stable accuracy across the entire response domain. These show that the proposed approach exhibits superior performance in terms of prediction accuracy, convergence speed, and robustness.
- 1 March 2026
- Published by IOP Publishing in Engineering Research Express
- Vol. 8 (6) , 065517
A stress prediction approach employing finite element neural network integrated with Runge-Kutta
Stress concentration in hazardous zones of mechanical structural components during service can lead to fatigue cracks and structural failure. Accurate assessment of their stress fields is crucial for structural health monitoring. This paper proposes an evaluation method that combines finite element analysis with neural networks to establish a highly accurate mapping relationship between the boundary data and the maximum stress in the defect area. Latin hypercube sampling is employed to determine representative boundary measurement locations, and finite element simulations are used to construct the training dataset. To address the issues of neural networks getting trapped in local minima and slow convergence speed, the Runge–Kutta method is introduced to optimize the update process of the network’s weights and biases, aiming to enhance the global search ability of the neural network, making it more likely to break free from the constraints of local minima and find the global optimal solution. Several structural components are provided to validate the effectiveness and generalization capability of the proposed network model. The results demonstrate that, the 95th-percentile error range obtained by proposed method remains within 3.1%, significantly lower than that of neural networks (17.5%) and Elman networks (15.9%). Segmented error analysis across low-, medium-, and high-stress regimes further confirms that the proposed method suppresses error amplification under high-load conditions, maintaining stable accuracy across the entire response domain. These show that the proposed approach exhibits superior performance in terms of prediction accuracy, convergence speed, and robustness.
- 20 January 2026
- Published by World Scientific Pub Co Pte Ltd in International Journal of Structural Stability and Dynamics
High Efficiency Estimation of Failure Probability: A Fusion Method of Cross-Entropy Optimization and Back Propagation Neural Network
The estimation of failure probability in complex engineering structures suffers from low computational efficiency and challenges in handling with high-dimensional nonlinearity. Traditional methods are inefficient and require large sample sizes, making them difficult to meet the demands of practical engineering applications. This paper proposes a failure probability estimation framework that integrates cross-entropy (CE) optimization with a back propagation neural network (BPNN). The CE method is employed to adaptively optimize the sampling distribution, thereby enhancing the sampling efficiency for rare failure domains. By leveraging BPNN’s powerful nonlinear approximation capability, a surrogate model for the limit state function (LSF) is constructed, significantly reducing the number of LSF evaluations. Validation through numerical engineering examples demonstrates that, compared to cross-entropy-based importance sampling (CE-IS) and cross-entropy-based Gaussian mixture sampling (CE-GM) methods, the proposed CE-BPNN approach stably approximates the reference failure probability across varying sample sizes, with lower coefficient of variation (CoV) and mean absolute percentage error (MAPE). The predictive model achieves [ ]-square ([ ]) values consistently exceeding 0.97. Under identical sample sizes, CE-BPNN exhibits significant accuracy improvement and high stability. The results indicate that the CE-BPNN method offers superior accuracy and efficiency for structural reliability analysis involving high-dimensional nonlinearity and small failure probabilities, providing a promising new approach for reliability assessment of complex engineering structures.
- 20 January 2026
- Published by World Scientific Pub Co Pte Ltd in International Journal of Structural Stability and Dynamics
High Efficiency Estimation of Failure Probability: A Fusion Method of Cross-Entropy Optimization and Back Propagation Neural Network
The estimation of failure probability in complex engineering structures suffers from low computational efficiency and challenges in handling with high-dimensional nonlinearity. Traditional methods are inefficient and require large sample sizes, making them difficult to meet the demands of practical engineering applications. This paper proposes a failure probability estimation framework that integrates cross-entropy (CE) optimization with a back propagation neural network (BPNN). The CE method is employed to adaptively optimize the sampling distribution, thereby enhancing the sampling efficiency for rare failure domains. By leveraging BPNN’s powerful nonlinear approximation capability, a surrogate model for the limit state function (LSF) is constructed, significantly reducing the number of LSF evaluations. Validation through numerical engineering examples demonstrates that, compared to cross-entropy-based importance sampling (CE-IS) and cross-entropy-based Gaussian mixture sampling (CE-GM) methods, the proposed CE-BPNN approach stably approximates the reference failure probability across varying sample sizes, with lower coefficient of variation (CoV) and mean absolute percentage error (MAPE). The predictive model achieves [ ]-square ([ ]) values consistently exceeding 0.97. Under identical sample sizes, CE-BPNN exhibits significant accuracy improvement and high stability. The results indicate that the CE-BPNN method offers superior accuracy and efficiency for structural reliability analysis involving high-dimensional nonlinearity and small failure probabilities, providing a promising new approach for reliability assessment of complex engineering structures.
- 29 August 2025
- Published by World Scientific Pub Co Pte Ltd in International Journal of Structural Stability and Dynamics
Mixed-Mode Fatigue Crack Growth Reliability Analysis: An Extended Finite Element Method-Based Framework
Structural fatigue cracking is an important cause of the failure of engineering structures. Accurately predicting the crack growth behavior of structures and evaluating the structural reliability are crucial for ensuring structural safety and extending the service life. To address this challenge, this paper proposes a reliability analysis method for fatigue crack growth based on the extended finite element method (XFEM). Due to the nonlinear relationship between the uncertain parameters of the cracked structure and the crack length, as well as the stress intensity factor (SIF), the limit state function (LSF) is highly complex. Consequently, the surrogate model is used to establish this relationship. The uncertain structural parameters are sampled using Latin hypercube sampling (LHS), and then XFEM is employed to simulate the crack propagation to obtain a sample dataset. Based on the sample data set, the response surface method (RSM) is applied to establish a surrogate model of the LSF between the uncertain parameters of the cracked structure and the crack growth length, as well as the SIF. The reliability index and failure probability are obtained through the first-order reliability method (FORM), the second-order reliability method (SORM) and the subset simulation (SS) method. By comparing with the Monte Carlo (MC) method, it is found that the SS method has a high accuracy and can provide a new idea for the reliability analysis of structural crack growth.
- 29 August 2025
- Published by World Scientific Pub Co Pte Ltd in International Journal of Structural Stability and Dynamics
Mixed-Mode Fatigue Crack Growth Reliability Analysis: An Extended Finite Element Method-Based Framework
Structural fatigue cracking is an important cause of the failure of engineering structures. Accurately predicting the crack growth behavior of structures and evaluating the structural reliability are crucial for ensuring structural safety and extending the service life. To address this challenge, this paper proposes a reliability analysis method for fatigue crack growth based on the extended finite element method (XFEM). Due to the nonlinear relationship between the uncertain parameters of the cracked structure and the crack length, as well as the stress intensity factor (SIF), the limit state function (LSF) is highly complex. Consequently, the surrogate model is used to establish this relationship. The uncertain structural parameters are sampled using Latin hypercube sampling (LHS), and then XFEM is employed to simulate the crack propagation to obtain a sample dataset. Based on the sample data set, the response surface method (RSM) is applied to establish a surrogate model of the LSF between the uncertain parameters of the cracked structure and the crack growth length, as well as the SIF. The reliability index and failure probability are obtained through the first-order reliability method (FORM), the second-order reliability method (SORM) and the subset simulation (SS) method. By comparing with the Monte Carlo (MC) method, it is found that the SS method has a high accuracy and can provide a new idea for the reliability analysis of structural crack growth.
- Retracted
- 1 July 2025
- Published by Elsevier in Measurement
- Vol. 256
RETRACTED: A quantification method for model parameter uncertainty based on an improved sequential Monte Carlo filter
- 1 July 2025
- Published by Elsevier in Measurement
- Vol. 256
RetractedRETRACTED: A quantification method for model parameter uncertainty based on an improved sequential Monte Carlo filter
- 10 June 2025
- Published by IOP Publishing in Physica Scripta
- Vol. 100 (7) , 075214
Crack identification approach integrating partition of unity XFEM and network-optimized fusion model
In engineering components, crack identification plays an important role in ensuring structural safety, prolonging service life, reducing maintenance costs and improving detection efficiency. This paper proposes a crack identification approach integrating the partition of unity extended finite element method (PU-XFEM) and network-optimized fusion prediction model. Firstly, the PU-XFEM is employed to analyze the cracks of the structural component and obtain the displacement field variables of the structural surface. The sampling method of statistical analysis is used to arrange measuring points on the surface of the component. Subsequently, the displacement changes at the selected measuring points are employed as sample data and input into the constructed neural network model. To solve the limitation that the network model is prone to over-fitting and falling into local optimization, four optimization algorithms (Marine Predator Algorithm, Honey Badger Algorithm, Moth Firefly Optimization Algorithm, and Aquila Optimizer) are employed to optimize and enhance the model parameters. Then, a network-optimized fusion model of crack identification is developed by integrating the optimization algorithms with the neural network. To assess the ability of the optimization algorithms, nine test functions with diverse characteristics are selected. The efficacy of these algorithms in optimizing neural network parameters is evaluated by examining their convergence precision and rate. Ultimately, the proposed approach is validated by a series of examples. The outcomes of the network-optimized fusion model are promising, particularly in terms of data fitting and crack identification.
- 10 June 2025
- Published by IOP Publishing in Physica Scripta
- Vol. 100 (7) , 075214
Crack identification approach integrating partition of unity XFEM and network-optimized fusion model
In engineering components, crack identification plays an important role in ensuring structural safety, prolonging service life, reducing maintenance costs and improving detection efficiency. This paper proposes a crack identification approach integrating the partition of unity extended finite element method (PU-XFEM) and network-optimized fusion prediction model. Firstly, the PU-XFEM is employed to analyze the cracks of the structural component and obtain the displacement field variables of the structural surface. The sampling method of statistical analysis is used to arrange measuring points on the surface of the component. Subsequently, the displacement changes at the selected measuring points are employed as sample data and input into the constructed neural network model. To solve the limitation that the network model is prone to over-fitting and falling into local optimization, four optimization algorithms (Marine Predator Algorithm, Honey Badger Algorithm, Moth Firefly Optimization Algorithm, and Aquila Optimizer) are employed to optimize and enhance the model parameters. Then, a network-optimized fusion model of crack identification is developed by integrating the optimization algorithms with the neural network. To assess the ability of the optimization algorithms, nine test functions with diverse characteristics are selected. The efficacy of these algorithms in optimizing neural network parameters is evaluated by examining their convergence precision and rate. Ultimately, the proposed approach is validated by a series of examples. The outcomes of the network-optimized fusion model are promising, particularly in terms of data fitting and crack identification.
- 21 October 2024
- Published by Elsevier in Engineering Analysis with Boundary Elements
- Vol. 169
Crack growth evaluation based on the extended finite element and particle filter combined method
- 21 October 2024
- Published by Elsevier in Engineering Analysis with Boundary Elements
- Vol. 169
Crack growth evaluation based on the extended finite element and particle filter combined method
- Top Cited Papers
- 3 June 2024
- Published by Elsevier in Engineering Analysis with Boundary Elements
- Vol. 166
A gradient-enhanced physics-informed neural networks method for the wave equation
- 3 June 2024
- Published by Elsevier in Engineering Analysis with Boundary Elements
- Vol. 166
Top Cited PapersA gradient-enhanced physics-informed neural networks method for the wave equation
- 12 October 2023
- Published by IOP Publishing in Physica Scripta
- Vol. 98 (11) , 115109
A combination of extended finite element method and the Kriging model based crack identification method
In this paper, we proposed a crack identification method in which the extended finite element method (XFEM) and a surrogate model are employed. The XFEM is used for accurate modeling of fractures, while the employment of Latin hypercube sampling (LHS) ensures a representative sample space for the input parameters. Then, we use a Kriging surrogate model to establish the response surface between the input and output data and to verify the accuracy of the model predictions. The Kriging model is based on a Gaussian process that models the correlation between the sample points, and it provides an efficient way to interpolate between known data points. To find the optimal solution, we combine the Kriging surrogate model with the particle swarm optimization (PSO) algorithm. From the numerical examples, it can be found that the optimal solutions are in good agreement with the exact solutions.
- 12 October 2023
- Published by IOP Publishing in Physica Scripta
- Vol. 98 (11) , 115109
A combination of extended finite element method and the Kriging model based crack identification method
In this paper, we proposed a crack identification method in which the extended finite element method (XFEM) and a surrogate model are employed. The XFEM is used for accurate modeling of fractures, while the employment of Latin hypercube sampling (LHS) ensures a representative sample space for the input parameters. Then, we use a Kriging surrogate model to establish the response surface between the input and output data and to verify the accuracy of the model predictions. The Kriging model is based on a Gaussian process that models the correlation between the sample points, and it provides an efficient way to interpolate between known data points. To find the optimal solution, we combine the Kriging surrogate model with the particle swarm optimization (PSO) algorithm. From the numerical examples, it can be found that the optimal solutions are in good agreement with the exact solutions.