We have developed an algorithm that defines position correspondence, which is the most important tuning factor in basis weight profile control for paper machines, automatically through results of step response tests. The algorithm is composed of two parts. One is the part that analyzes the result of step response test of each slice by using wavelet transform. And the other is the part that compensates position correspondence of the rest slices to which step response tests have not been applied, and defines total position correspondence by using back propagation in neural networks. We introduce an example where position correspondence was corrected by actually using this algorithm and show the effect of this algorithm.