Method Of Detecting Steel Plate Flaw Using Electromagnetic Pulsed Thermography

With the development of social economy, rolled steel and its components have beenwidely applied in transportation, aviation, nuclear energy and electric power fields, so thathigh requirements are proposed for their performance. However, the defects in rolled steel andits components can easily cause section stress concentration, so the performance deteriorationmay cause mechanical breakdown and other safety problems. To ensure the safe service ofrolled steel and its components, it is very important to conduct NDT study for the innershallow defects of rolled steel and its components.Based on the electromagnetic pulsed thermography detection method,45#steel plate asthe research object and recognition of its normal areas and defect areas as the goal. Firstly,analyzes the relationship between electromagnetic excitation parameters and the excitationeffect, and then designs the electromagnetic pulsed thermography experiment. Secondly,obtains the infrared sequential heat maps of normal surface during thermal excitation process,and acquired its initial characteristics through the time/frequency information processing.Thirdly, extracts the feature vector of normal areas and various kinds of defect areas byprincipal component analysis. Finally, the probabilistic neural network is employed toidentify the normal areas and defect areas with pixel as the unit.The experiment results show that:(1) the identification rates of normal areas is97%,indicate this method can obtain desirable recognition effect;(2) the identification rates for thedefects with the depth of2mm is92%.(3) With the increasing of the depth of defects, defectrecognition rate decreased, the experimental results demonstrate increasingly importantimpact on the recognition results of the skin effect. The identification results show undersatisfied the skin depth, the premise of the institute of the proposed method to the normalareas and defect areas can identify desirable effect.