Research on pulse laser detection anti-interference method in smoke environment

Abstract

To address the problem of smoke medium interference on pulse laser detection signals, this study employs Monte Carlo method to precisely establish the backscattered signals model of pulsed lasers in smoke conditions, and proposes an anti-interference strategy combining particle swarm optimization (PSO) and convolutional neural network (CNN) by echo signals collected experimentally. Specifically, this study uses the Monte Carlo method and scattering phase function to simulate particle collisions and constructs a sampling model of reflected photons based on the target surface’s reflective properties. Semi-analytical reception technology is used to extract the received backscattered photon signals. Finally, a CNN is built taking all the backscattered data points as the input. The parameter of the network is optimized using PSO algorithm. To verify the accuracy of the simulated backscattered echo signals and the effectiveness of the anti-interference algorithm, a laser detection experiment in a smoke environment was conducted. The experimental results show that the simulated detection backscattered signals and the experimental backscattered signals have a high degree of consistency, with a maximum root mean square error of only 0.043. The recognition accuracy of the CNN optimized by the PSO algorithm is 96.154 %, with a computational parameter size of 193.008 KB. This study provides solid theoretical support and technical assurance for the pulse laser detection technology in the anti-interference field.