Research On The Method Of Identifying And Warning Of Key Facilities Of LNG Receiving Station Based On Infrared Thermal Imaging Technology

Infrared thermal imaging monitoring of LNG receiving stations has the characteristics of large volume of key facilities and complex site conditions,etc.It has put forward higher requirements for data cleaning,abnormal identification and positioning of infrared thermal imaging.Aiming at the problems that may occur during the application of infrared thermal imaging monitoring technology in LNG receiving stations,this paper conducts research from three aspects: anomaly data cleaning,anomaly monitoring and identification,anomalous area location,and early warning.Corresponding solutions are proposed and based on cases The results of the analysis validate the feasibility and effectiveness of the method:(1)In response to the problem of false alarms caused by foreign objects passing through the front of the lens during the monitoring process,a method of cleaning infrared thermal imaging monitoring data based on HOG + SVM is proposed,which can accurately identify the video frames that foreign objects break into the field of view of the lens,and can be calibrated as abnormal.Reduce the interference of abnormal objects on the monitoring process,and the cleaning accuracy averaged up to 97.33%.(2)Aiming at the problems that abnormal data of key facilities of LNG receiving stations rarely cause misidentification and untimely identification of abnormalities,and that infrared monitoring is easily affected by the surrounding environment,an abnormality monitoring method based on convolutional neural networks is proposed.Select the storage tank as the research object,build a specific convolutional neural network,and then identify the abnormal time of the storage tank by training the historical data.The advantage is that there is good learning between different individuals in the same facility.Second,GPU-accelerated,one image recognition calculation time is 1 millisecond,and the recognition accuracy can reach 99%.(3)Aiming at the problem of difficult to lock the abnormal location of key facilities of LNG receiving stations and difficult maintenance decision-making,an abnormal area location and early warning method combining false color processing,model location matching and consequence simulation was proposed.Reduces the amount of work required to lock out abnormal locations,which helps field maintenance and decision making.