The Meridian Project, as an important space environment monitoring infrastructure in China, is of great significance in space physics research, especially providing key support for the observation of Ultra-Low-Frequency (ULF) waves. Based on the second-sampled data of the horizontal component (H) and declination component (D) of the geomagnetic field collected by multi-station fluxgate magnetometers of the Meridian Project from 2011 to 2021, this paper constructs a high-quality and long-time-series dataset of geomagnetic ULF waves through data cleaning, noise reduction preprocessing, and combined with time-frequency analysis (Morlet wavelet transform) and statistical methods. The dataset covers 5 stations at high, middle and low latitudes, includes event annotations and original sequences of Pc3-Pc5 regular pulsations, and is stored in JSON format, supporting reading by mainstream scientific research software. This dataset fills the gap in regional ULF wave observations, provides basic data for research such as magnetospheric fluctuations and space weather modeling, helps improve China's space environment monitoring capabilities, and can be used for automatic ULF wave identification and coupling mechanism research.