飞机完好率预测仿真研究

Simulation of Aircraft Readiness Rate Prediction

利用飞机完好率时间序列特性,建立了NAR神经网络模型和基于不同核函数的3种支持向量机模型对平时状态下的飞机完好率变化趋势进行建模、训练和预测;运用Matlab仿真软件进行试验验证,结果表明:支持向量机模型具有较好的拟合效果,预测精度优于NAR神经网络模型,基于RBF核函数的支持向量机预测准确率相对较高。两种预测模型相比于部队现行的预测方法均具有更高的准确度和可靠度。

Nonlinear Auto-regressive（ NAR） Neural network Model and three SVM models with different kernels were built. Then the trends in peacetime aircraft readiness rate ware modeled,trained and predicted. Finally,the simulation result was presented. The results show that Support Vector Machine（ SVM） model has better fit and higher predicting accuracy than NAR Neural network Model,and prediction with the method of RBF-Kernel SVM has the highest accuracy. All the models mentioned above are superior to the forces current prediction method.