Research on Machine Vision Detection Method of Ship Sulfur Emission Based on Convolutional Neural Network

Abstract

Listen

Translate article

The traditional method of ship exhaust gas detection has the shortcomings of poor timeliness, blindness and hysteresis, which cannot meet the needs of marine supervision departments for comprehensive, efficient and real-time supervision of ship sulfur emissions. To solve this problem, a convolution neural network(CNN) based machine vision detection method for ship sulfur emission is introduced, and the plume characteristics of heavy sulfur black smoke ship exhaust are adaptively extracted by convolution operation. An improved LeNet-5 network model structure is proposed.By changing the activation function and increasing the depth of the model, the batch normalization operation is introduced to improve the accuracy and generalization ability of the model.The experimental results show that the detection accuracy of the model for heavy sulfur black smoke ships can reach 94.52%, which is better than the ordinary CNN network model and the VGG-16 network model.This method can quickly screen suspicious ships with excessive sulfur content, broaden the way for maritime supervision departments to detect ships under the background of “sulfur limit order”, and effectively improve the supervision efficiency of maritime personnel.