航空型号项目风险预测的BP神经网络模型及应用

Application and Model of the Aviation Model Project Risk Predicting Based on BP Neural Network

本研究克服了单纯采用专家风险因子测度方法主观性较强的缺点,以及单纯采用人工神经网络评估模型模糊性的缺点,结合两种方法的优点,利用基于BP神经网络算法的Microsoft Visual C++程序,在专家风险因子测度基础上,通过大量风险评估成功案例数据的训练,成功建立了航空型号项目风险预测模型。该模型可以较为精确、客观地预测型号项目的失败概率、成功概率、失败后果等风险因子,为型号项目管理决策提供更可靠的理论指导。

The shortcoming of both expert＇s risk measurement method,which is artificial and subjective,and artificial neural network method with pure subjectivity are overcome in this paper.Combined with the advantages of both expert＇s risk measurement method and artificial neural network method,using Microsoft Visual C＋＋ neural network based on BP algorithm procedures,through a large number of successful risk assessment cases of training data which estimated by experts,the aviation model project risk prediction model has been successfully established.The model can accurately and objectively predict the aviation model project＇s failure probability,success probability and failure consequences risk factors etc.The model can provide the project management and decision with more reliable theory and guidance.