依据直升机空气动力学理论建立起来的旋翼 /机身气动干扰模型 ,至今尚不能达到满意的准确度 ,而且它的计算工作量相当大。文中采用多层前向通道神经网络建立旋翼 /机身气动干扰模型 ,探索一种新的建模方法。在试验结果的基础上 ,对机身的空气动力随前进比和桨盘载荷变化情况进行了分析 ,然后提出了一个二输入 /三输出的旋翼 /机身气动干扰神经网络模型。网络训练样本直接来源于试验数据。训练好的干扰模型清楚地反映了机身阻力与前进比 μ和桨盘载荷p的变化都有着密切关系 ;对于机身升力和俯仰力矩 ,桨盘载荷 p的变化起着主导作用。给定状态点的实际测量与采用神经网络模型计算结果的比较 ,进一步验证了旋翼 /机身气动干扰神经网络模型的合理性与有效性。由该方法建立起来的干扰神经网络模型可直接用于直升机空气动力特性的研究及实时仿真 According to helicopter aerodynamics theory, the rotor / fuselage aerodynamic interference model has not been able to achieve satisfactory accuracy, and its computational workload is quite large. In this paper, a multi-layer forward channel neural network is used to establish a rotor / fuselage aerodynamic interference model and a new modeling method is explored. Based on the experimental results, the aerodynamic performance of the fuselage is analyzed with the advance ratio and the change of propeller load. Then a two-input / three-output aerodynamic model of rotor / fuselage aerodynamic disturbance is proposed. Network training samples come directly from the experimental data. The well-trained interference model clearly shows that there is a close relationship between the resistance of the fuselage and the variation of the forward ratio μ and the paddle load p; the change of the paddle load p plays a leading role in the lift and pitching moments of the fuselage. The actual measurement of a given state point and the comparison with the results of neural network model further verify the rationality and effectiveness of the rotor / fuselage aerodynamic disturbance neural network model. The disturbance neural network model established by this method can be directly applied to the helicopter aerodynamics research and real-time simulation