Study of quantitative identification of infrared thermal wave testing based on BP neural networks

In order to resolve the problem of quantitative identifying, in pulsed thermography, taking the highest temperature difference and the best testing time as input, and taking defect depth and diameter as output, made use of BP Neural Networks to achieve it, and it was done. According to result, when testing value was in area of swatch, identifying precision was high, and error is less than 3.5%. The feasibility of BP Neural Networks was validated, and it has very important meaning to quantitative identifying of factual application.