Abstract: We selected 20 cases of two types of weather processes with different precipitation properties in Liaoning province from June to September in 2017 and 2018.These cases are used to evaluate the forecast capabilities of the meso- and small-scale systems in Liaoning province with the East China Model, the North China Model, the Global/Regional Assimilation and Prediction System (GRAPES)_3km Model, and the Northeast China Weather Research and Forecasting (WRF) Model, based on the Fraction Skill Score (FSS) in the fuzzy test neighborhood method.The results showed that for the regional precipitation process and local precipitation process, the smaller the radar echo intensity is, the larger the neighborhood radius is, and the better the prediction skill of the high-resolution model is.When the radar echo is greater than 30 dBz, the FSS scores of the radar echo prediction for localized precipitation are higher in all high-resolution models.When the neighborhood radius is 3 km, the forecasting skills of the North China Model for all-level radar echo during regional precipitation are better than other models, with a maximum FSS difference being 0.031.In the process of localized precipitation, the East China Model performs better, with a maximum FSS score of 0.127, indicating that the East China Model has a stronger capability to predict small and medium-scale convective systems.For the local precipitation process, among the forecast periods between 08:00 and 23:00 in the Northeast China WRF_3km model, the forecast performs better in the "middle" period than the "beginning and ending" period, with their maximum FSS difference of 0.121.