Skip to Main Content
In the alternative current field measurement (ACFM) technique, the nonzero value of liftoff distance for the magnetic sensor acts as a low-pass filter on surface crack signals, causing errors in crack detection and sizing. We present a blind deconvolution algorithm for liftoff evaluation and surface crack signal restoration. The algorithm employs the available closed-form expressions for the distribution of electromagnetic fields at the metal surface in the vicinity of a crack. To examine the accuracy of the algorithm, we use the original and the restored signals for crack sizing by a wavelet network inversion method. We present simulated and experimental results to demonstrate the role of the proposed algorithm in improving the inversion process.