Method for identifying the impact load condition of thin-walled structure damage based on PSO-BP neural network

Area of Science:

• Structural mechanics
• Computational intelligence
• Materials science

Background:

• Thin-walled structures (TWS) are prevalent in engineering applications and susceptible to impact-induced damage.
• Determining the precise impact load conditions (position and velocity) responsible for TWS damage is a significant challenge.

Purpose of the Study:

• To develop and validate a novel method for identifying impact load conditions in damaged thin-walled structures.
• To leverage a particle swarm optimization-backpropagation (PSO-BP) neural network for accurate impact parameter prediction.

Main Methods:

• A finite element model (FEM) of TWS was used to simulate permanent plastic deformation under known impact conditions.
• A multivariate polynomial function fitted the characteristic deformation shapes, creating a basic dataset with impact parameters as input and function coefficients as output.
• An extended PSO-BP neural network was trained on this dataset, and the sample set was expanded to enhance model robustness.
• A predictive PSO-BP neural network model was established using functional coefficients of deformed surfaces to predict impact position and velocity.

Main Results:

• The PSO-BP neural network model successfully predicted impact position and velocity with high accuracy.
• Comparison with conventional BP neural network predictions demonstrated the superior performance of the PSO-BP algorithm.
• FEM analysis validated the accuracy of the PSO-BP method in reconstructing impact events.

Conclusions:

• The developed PSO-BP neural network method offers a highly accurate and reliable approach for identifying impact load conditions in damaged thin-walled structures.
• This technique addresses a critical gap in understanding impact events and their consequences on structural integrity.
• The findings have significant implications for damage assessment, structural health monitoring, and safety analysis in engineering applications involving TWS.