飞机飞行事故率预测建模与仿真研究

摘要

Study on the prediction problem of aircraft flight accidents. Flight accident rate changes with the influence factors such as natural resources, environment, economy, society, science, technology, and etc, the complex relationships among the factors lead to nonlinear changes, and traditional or single forecasting methods cannot get high predicting precision. In order to improve the prediction accuracy, a flight accident forecast model is put forward based on time series method and support vector support. Firstly, time series model and support vector machine (SVM) are used to predict the flight rule accidents and random changes, using linear regression forecasts to determine the weights of the two prediction results. The prediction results of flight accidents are calculated through the weights. The simulation results indicate that the combined model improves the flight accidents precision effectively, and provides an effective prediction method for aviation safety management.%研究飞机飞行事故率准确预测问题,飞行事故率的变化受到自然资源、环境、经济、社会和科技等多种因素影响,因素间关系复杂,导致飞行事故率的非线性变化,传统或单一预测方法难以获得较高的预测精度.为了提高飞行事故率预测精度,提出一种时间序列法和支持向量机组合的飞行事故率预测模型.模型首先分别用时间序列法和支持向量机对飞行事故率的规律发生部分和随机变化部分预测,采用线性回归确定两个预测结果的权值,通过权值计算获得飞行事故率预测结果.仿真结果表明,组合模型有效提高了飞行事故率预测精度,可为航空安全管理提供了有效的预测方法.