Dataset for Training and Validation of a High-Performance Liquid Chromatography Signal Denoising Model

It was used for model training, performance evaluation, and real-sample validation. The dataset includes 500 pairs of clean and noisy simulated high-performance liquid chromatography signals, covering a broad signal-to-noise ratio range of 1–35 dB.

Each signal contains 8,000 sampling points over a time range of 0–20 min. The dataset was divided into training, validation, and test subsets at a ratio of 6:2:2 for training and evaluating a wavelet-domain multi-scale convolutional neural network (MSCNN). In addition, real chromatographic signals were included to validate the denoising performance of the proposed model under instrumental noise and complex sample-matrix conditions.