通过分析回归分析法,BP神经网络和加速遗传算法原理,针对BP神经网络的变异特性,运用遗传算法对其进一步修正,最终建立了适用于航空装备维修保障费用的BP网络组合预测时间序列输出模型,结合某型航空装备维修保障费用数据进行预测系统仿真,得出结果实际数据相比误差较小。结论说明,此方法具有更高的预测精度,有较好的实际应用价值。
Through the analysis return analytic method, the BP nerve network and the acceleration heredity algorithm principle, in view of the BP nerve network variation characteristic, further revises using the heredity algorithm to it, finally established has been suitable for the aircraft equipment service safeguard expense BP network combination forecast time series output model, unified some aircraft equipment service safeguard cost data to carry on the forecast system simulation, obtained the result actual data to compare the error to be smaller. The conclusion showed that, this method has the higher forecast precision, has the better actual value.