In order to improve the sorting accuracy of radar modulated signals in the electronic countermeasure environment, based on the partial connection fuzzy C-means (PCFCM) algorithm and the teaching-learning random forest (TLRF) algorithm, a radar modulated signal sorting model PCFCM-TLRF is proposed. In this model, we introduce the partial connection number (PCN) to improve the K-means clustering algorithm and optimize the fuzzy C-means (FCM) algorithm. Then the signal sample is pre-processed with the improved FCM algorithm. The teaching-learning-based optimization (TLBO) algorithm is used to optimize the random forest (RF) algorithm, so that the optimized RF algorithm can form a better classifier with much lower complexity. The pre-processed sample is used as the training sample in the TLRF algorithm to realize the sorting of radar signals. The research results show that the sorting accuracy of the PCFCM-TLRF model is higher than those of other sorting models. This model can realize the effective sorting of radar modulated signals.