Abstract: There were fifty batches data got by SMI rapid fingerprint peak area and content of total saponins.Then it established mathematical model with twenty five batches data by partial least squares regression (PLSR) and non-negative correction (NNC)，and it proved the accuracy with the other twenty five batches data.As the result,it explained the coefficient of mathematical model reasonably.And the prediction deviation was less than 5%.The content results were accuracy. It could rapidly determine the total saponins content of SMI by the prediction model established with rapid fingerprint and NNC-PLSR.And It provided a new total saponins content rapidly determination method for SMI.

Key words: fingerprint, Shenmai injection, partial least squares regression, coefficient