Quantitative Identification Method for Glass Panel Defects Using Microwave Detection Based on the CSAPSO-BP Neural Network

Abstract

To address the problem of the quantitative identification of glass panel surface defects, a new method combining the chaotic simulated annealing particle swarm algorithm (CSAPSO) and the BP neural network is proposed for the quantitative evaluation of microwave detection signals of glass panel defects. First, the parameters of the particle swarm optimization (PSO) algorithm are dynamically assigned using chaos theory to improve the global search capability of the PSO. Then, the CSAPSO-BP neural network model is constructed, and the return loss and phase of the microwave detection echo signal of glass panel defects are extracted as the input feature quantity of the network, from which the intrinsic connection between input and output is found through network training and testing to achieve the prediction of the depth and width of glass panel surface defects. The results show that the CSAPSO-BP network model can more accurately characterize the defect geometry of glass panels than the PSO-BP network model.

CSAPSO-BPNN-Based Modeling of End Airbag Stiffness of Nursing Transfer Robot

“…This paper introduces a novel approach for tuning the BP neural network, which is a multilayer feed-forward neural network that utilizes the gradient descent algorithm with weights adjusted by backpropagation [18]. Specifically, this method employs chaos theory to dynamically adjust the random numbers r i by generating a chaotic sequence based on the Logistic model, as shown in Equation (1).…”