飞行事故率是表征飞行安全水平的重要指标,其预测是典型的小样本问题。针对飞行事故率预测建模难的问题,采用加权LS-SVM方法,建立了飞行事故率的时间序列预测模型。仿真实验表明,加权LS-SVM具有很高的建模精度和较强的泛化能力,从而验证了该方法的有效性和先进性。 Flight accident rate is an important indicator of flight safety level, and its prediction is a typical small sample problem. Aiming at the difficulty of predicting flight accident rate, a weighted LS-SVM method is used to establish the time series forecasting model of flight accident rate. The simulation results show that the weighted LS-SVM has high modeling accuracy and strong generalization ability, which verifies the effectiveness and advancedness of the proposed method.