The Application Of Tensor Decomposition Algorithm Under Speed Effect In Infrared Nondestructive Testing

In industrial production,rails,engines,gears or other parts are prone to fatigue damage and internal defects over time,which can easily lead to cracks and resulting in incalculable catastrophic accidents and huge economics losses.Therefore,in order to ensure the security of equipment,defects and cracks detection of equipment are essential.Nondestructive Testing(NDT)technology can be used to judge whether the material have defects or non-defects.Meanwhile,NDT technology also has the advantages of non-contact,non-destructive and high detection efficiency.Infrared thermal imaging detection is a non-destructive,non-invasive and non-contact detection method.The infrared thermal imaging system can capture the thermal information of the object surface,and then we can get the position of defects after processing the thermal data.Most of the existing infrared thermal imaging nondestructive testing systems can only detect defects specimen under static conditions.If there are multiple defects in a specimen,it will take a lot of time to detect them individually.However,by means of moving,the continuous use of coils to stimulate the specimen will greatly improve the detection efficiency.In this paper,by using the infrared thermal imaging nondestructive testing technology,a method based on scanning specimen at speed and feature extraction by tensor decomposition and background difference is proposed to reconstruct the morphology of cracks and defects,enhance the image contrast of cracks and defect areas,and complete the detection of cracks and defects.The specific research contents of this paper are as follows: 1)Establish an infrared thermal imaging nondestructive testing system for experiments,collect and sort out experimental data by control variable method;2)Explore the detection results of various specimens by the infrared thermal imaging system,and verify them through different metal materials and crack types;3)Research the commonly target detection,feature extraction algorithm in the field of infrared thermal imaging nondestructive testing and tensor decomposition algorithm in other fields;4)Research the cracks detection problems under speed effect,then combining the advantage of tensor decomposition algorithms in target detection and background subtraction with the characteristics of the nondestructive testing to propose a tensor decomposition algorithm for defect detection,and verifying the proposed algorithm with the existing algorithms on several evaluation criteria,such as the time and space complexity of these algorithm,finally,we can the signal-to-noise ratio(SNR)to judge which algorithms is the best.Also,we try to solve the traditional nondestructive testing methods' disadvantages such as the low SNR and poor performance,and the difficulty of detecting the complex structural specimens.By comparing the different experiment results,the final results show that the proposed algorithm can solve the problem of crack detection under speed effect,and compared to the traditional nondestructive testing methods and other areas of tensor decomposition algorithm,the algorithm is very fast in processing,and it shows its outstanding robustness and high signal-to-noise ratio,also,the proposed algorithm can enhance the contrast between defect areas and non-defect areas,and it is possible for real-time automatic detection of cracks in the future.