Location Method of Smoke Pollution Source based on LMBP Neural Network

Abstract

In complex outdoor scenes, most applicable neural networks can only detect and identify smoke, but cannot accurately locate the source of its pollution. In response to this problem, this paper proposes a smoke pollution source location method based on LMBP neural network to improve the prediction and location results of outdoor smoke pollution sources. This paper first analyzes the related knowledge of artificial neural network (ANN) and Levenberg-Marquardt algorithm (LM algorithm). Then it studies the ANN-BP model based on gradient descent method and the ANN-LMBP model based on the LM algorithm. Finally, experimental simulations verify the feasibility of the ANN-LMBP model in the problem of smoke pollution source location and its strong generalization ability. The error between the latitude and longitude of the ANN-LMBP model proposed in this paper and the actual latitude and longitude in the actual scene are both within 200 meters, which is of great significance for studying the location of smoke pollution sources in complex scenes.