A Gas Leakage Source Positioning and Estimation Algorithm Based on Deep Neural Network (DNN)

Gas leakage source positioning and estimation is a crucial approach to ensuring urban safety. This paper presents a gas leakage source estimation method based on deep neural network. The gas leakage diffusion model and machine learning model are applied to the estimation method of leakage gas source. Taking into account the characteristics of the leakage gas, we proposed the usage of both the Gaussian plume model and the DNN model. Then corresponding observation models are compiled based on these different models. The Particle Swarm Optimization algorithm is applied to analyze and predict the observation data. Simulation results show that compared with traditional modeling methods, the improved DNN algorithm has better performance, further reducing the source estimation and positioning error.