Deep learning-based smoke fire-prevention monitoring method of rail-train equipment room

Abstract

The invention relates to a deep learning-based smoke fire-prevention monitoring method of a rail-train equipment room. The method includes: step S40, inputting smoke data-based test data samples intoan already trained CNN model, starting from an input layer of the CNN model to carry out forward propagation, and obtaining posterior probability of various smoke feature parameters at an output layer; and step S50, comparing the obtained posterior probability of the various smoke feature parameters and set probability threshold values of corresponding classes, and according to a comparison result, determining whether a fire of the rail-locomotive equipment room occurs. According to the method, accuracy of a fire judgment result can be improved, reaction duration from fire occurrence to monitoring realization can be reduced, and sensitivity of monitoring can be effectively improved.

Description

technical field [0001] The invention relates to the technical field of smog fire prevention monitoring, in particular to a method for monitoring smoke fire prevention in rail train equipment rooms based on deep learning. Background technique [0002] With the continuous speed up of my country's railways, safe operation is becoming more and more important, so higher requirements are put forward for the safety monitoring of locomotives. [0003] Existing fire monitoring methods for rail locomotive equipment rooms mainly monitor smoke characteristics, temperature characteristics, video characteristics, etc. through corresponding sensors to form a corresponding monitoring system, including fire monitoring based on the second bus of railway locomotives, multi-parameter real-time fire monitoring, etc. [0004] The fire monitoring based on the second bus of railway locomotives mainly uses a fire monitoring board and a second bus, wherein the fire monitoring board is electrically co...

AI Technical Summary

Problems solved by technology

[0006] Although the above-mentioned existing fire monitoring method for rail locomotive equipment room finally realizes fire monitoring, it has the following disadvantages: based on the second bus fire monitoring of railway locomotives, multi-parameter real-time fire monitoring, etc., all belong to the use of corresponding sensors such as smoke and temperature. , temperature and other parameters are monitored, the data analysis is not fine enough, and the corresponding monitoring results lack accuracy and sensitivity. Especially when the fire occurs in a small area of ​​the rail locomotive equipment room, the traditional method with the help of sensors It is difficult to detect quickly and accurately, so there is a lot of room for optimization from the perspective of monitoring response time and sensitivity

Images