Precise Positioning Model of Back Propagation Neural Network Based on Genetic Algorithm Optimization

Abstract

Damage identification and location is a critical problem in structural health monitoring. The fundamental idea is to employ features like amplitude thresholds and signal timing differences to identify and localize structural damage by acquiring anomalous signals brought on by damage. The method proposed in this paper addresses the shortcomings of existing localization methods, such as slow localization efficiency, low localization accuracy, and poor model generalization. Firstly, by employing data cleaning principles to clean invalid data, then the decision tree classification model is used to distinguish the presence of interference signals, and finally, the BP neural network localization model based on genetic algorithm optimization is established to identify and localize the damage. Both signal interference and no signal interference were used in the studies with pulsed radio transmission. By changing the position of the anchor point of the excitation signal and the target point of the acquisition signal, the distance data from the anchor point to the target point at different locations was collected using the time of arrival (TOF) based ranging principle, and the validity of the positioning model was finally verified. Without taking into account the location of the target and anchor, the model can precisely identify and localize damage. It can be used as a reference for further structural health monitoring studies, with good prospects for application.