Infrared Temperature Field Modelling And Source Identification Of Leaking Steam

Steam leakage is easy to spread and does great harm,it is very important to identify early steam leakage in time.The key to the formation of steam is its above-ambient temperature characteristics.Infrared camera technology is used to capture and analyze the temperature field of steam,which is beneficial to grasp the essential characteristics of leakage steam and fundamentally solve the problem of leakage identification.In complex industrial field,the steam temperature field is easily disturbed by thermal noise,which makes it very difficult to obtain and identify steam characteristics.This paper studies the temperature field characteristics based on the mechanism of steam generation and development,and explores high-resolution characterization and identification methods of infrared temperature field.The paper contents include:(1)The steam leakage diffusion model is established to simulate the occurrence and development process of steam leakage,which revealed the static and dynamic hammer tail characteristic of the leakage of steam temperature field,including hierarchical diffusion characteristic,hammer tail characteristic,dynamic scattering characteristic and homologous heterogeneous characteristic.and an image representation method for each feature is proposed.The model also revealed the direction of leakage,the volume of leakage and leakage pressure’s influence on temperature field characteristics.(2)The dynamic characteristics of steam weak temperature field are studied,and the effective contrast enhancement and inter-frame differential multiplication are proposed to extract weak field.The characteristics of thermal-dryness interference in weak field were analyzed and the adaptive median filtering method was used for effective pretreatment.In this paper,the characteristics of the layered spindle tail in the temperature field are studied,and a variable scale extraction method of image features is proposed to realize the efficient extraction and high-resolution characterization of the details of the dynamic spindle tail distribution in the infrared temperature field.(3)Through measured sampling,multi-level hammer tail samples under different attitudes,pressures,flows and leakage angles,as well as data sets of interference samples were established,with a total sample number of 940.The Mask R-CNN network model was established,the hammer tail features of leakage steam were successfully learned and dynamically excavated,and an effective hammer-tail feature discriminator was established.The accuracy of the generalization test reached96.81%,and the identification efficiency was about 0.036 s for a single temperature layer image,which effectively improved the identification speed and accuracy.(4)Combined with the image representation method of temperature field and the discriminator based on Mask R-CNN network model,the identification algorithm of leakage steam was proposed.The experimental platform of steam leakage detection is designed and built.The detection algorithm proposed in this paper is used to detect and analyze the leaking steam with different leakage volume,different leakage pressure and different leakage direction.The experimental results show that the proposed algorithm can effectively detect the steam leakage and predict the direction of the leak source.The processing time of the algorithm is about 0.48 s for 5consecutive frames of images,and the overall recognition accuracy is 99.93%.In this paper,by studying the diffusion mechanism of leakage steam temperature field,the static and dynamic hammer tail distribution characteristics of steam temperature field are summarized,and the image characterization method of each characteristic is proposed,and the variable scale extraction method is proposed to achieve the high-resolution characterization of steam infrared temperature field.A Mask R-CNN network model was established to effectively learn and dynamically mine the hammer tail features of steam leakage,and a classifier with strong generalization ability was obtained to realize efficient identification of steam leakage sources.