In response to the increased screening costs associated with the reduction in size and increased complexity of integrated circuits, a new screening method for integrated circuits based on single-variable test parameters has been proposed. Initially, the merge sort algorithm is used to concatenate the parameter values with the integrated circuit numbers into an array, ensuring the accuracy of subsequent screening and sorting the data according to parameter values. Then, the K-means algorithm is employed to preprocess outliers in the test data, optimizing the test data preliminarily. Finally, an innovative algorithm called D-PSO is introduced, combining derivatives and particle swarm optimization. The D-PSO algorithm enhances the sensitivity and accuracy of locating inflection points, precisely identifying these points to directly screen integrated circuits with similar parameter data. Simulation results demonstrate that this algorithm converges much faster than other algorithms and can accurately and swiftly screen integrated circuits. While maintaining test accuracy, it optimizes the test set and performs similarity screening, effectively reducing the screening costs of integrated circuits.