Engine dense smoke and light smoke automatic detection method based on improved NanoDet deep network

Abstract

The invention discloses an engine dense smoke and light smoke automatic detection method based on an improved NanoDet deep network. The method comprises the steps that engine smoke pictures are collected to form a data set; by utilizing the improved NanoDet deep network, under the condition that the same receptive field is ensured, only a C5 feature layer is used, and network parameters are reduced; frame prediction is carried out on each pixel of the feature layer; positive and negative samples are screened by using an adaptive training sample selection algorithm, a detection head is composed of a classification branch, a frame regression branch and an implicit unsupervised target predictor branch, and the detection precision is improved. And whether the engine generates smoke and the type of the smoke is judged according to a network output result. If the smoke is detected, the type of the detected smoke is continuously judged according to the chromaticity difference between the smoke area and the background area, if light smoke is generated, an alarm is immediately given to prevent the light smoke, and if the detection result is dense smoke, emergency measures except the alarm are automatically started; and if no smoke is detected, continuously detecting. According to the method, dense smoke and light smoke detection of automobiles and aero-engines can be realized.

Description

technical field [0001] The invention belongs to the technical field of automatic control, relates to the automatic detection of engine smoke, in particular to an automatic detection method of engine thick smoke and light smoke based on the improved NanoDet deep network. Background technique [0002] In order to meet the functions of high efficiency and small size, the structure of the engine is becoming more and more complex, and it works for a long time under harsh working conditions such as high temperature, high pressure, and high speed, and the probability of failure increases accordingly. On the one hand, smoke is one of the important image features that appear at the initial stage of anomalies such as fire and explosion. Therefore, accurate and real-time smoke detection of engine test videos in various environments is important to reduce economic losses and ensure user safety. means; on the other hand, with the birth and development of deep learning technology, video u...

AI Technical Summary

Problems solved by technology

The sensor has the advantages of low price and easy installation, but the smoke sensor requires a closed space, and it is in contact with the air in a large area to cause oxidation, so the sensitivity is not high; the traditional algorithm will change with the color, texture, and shape characteristics of the smoke image as the lighting conditions and other factors change. acting unusually unstable

Images