Abstract: ‍ ‍This study proposes a radar active interference intelligent recognition method based on multi-domain feature parameter correlation. This method aims to address the issues of large parameter variations and poor adaptability of decision tree features in complex interference scenarios. First， specific feature parameters are designed based on the characteristics of interference signals in the time， frequency， and pulse compression domains. These feature parameters can comprehensively describe the characteristics of the interference signals. Subsequently， the average information gain of the random forest is used to select feature parameters under different interference parameter conditions and associate them. Next， an improved ResNet18 network is proposed by combining multi-domain correlated features and a typical neural network. Using multi-domain correlated features， this network can learn interference signal features more accurately and achieve intelligent recognition. Finally， an improved ResNet18 network is trained with interference samples covering a wide parameter range to enhance its generalization performance. Through computer simulation experiments， the proposed method achieves an accuracy rate of over 98% in active interference recognition with a wide parameter range. Furthermore， the improved ResNet18 network has a parameter count of only 1/17 of the original network， significantly reducing its complexity. In summary， the proposed radar active interference intelligent recognition method based on multi-domain feature parameter correlation offers high recognition accuracy and low computational complexity， making it effective for complex interference scenarios.