3 corrosion damage forecasting methods were put forward based on accelerated corrosion test data of LD2 structural material of aircraft, which were data fitting, neural network, and time series methods. Basic forecasting principle, forecasting accuracy, and forecasting extension of the 3 prediction methods were compared and analyzed. The result showed that the prediction accuracy of neural network and time series method are higher than data fitting method; the prediction extensionality of time series method is the best, which can predict the value and development trend of aluminum alloy corrosion depths in a future period with higher precision. It was suggested that appropriate forecasting method should be selected based on the need of prediction research.%依据飞机LD2结构材料的加速腐蚀试验数据,提出了数据拟合、神经网络、时间序列3种腐蚀损伤预测方法,并对3种预测方法的基本原理、预测精度及预测外延性进行了对比分析.研究发现神经网络和时间序列方法预测精度较高;时间序列方法的预测外延性较好,能够以较高精度预测未来一段时期腐蚀损伤的发展趋势及损伤程度.因此,应根据研究需要选用合适的方法进行腐蚀损伤预测.
展开▼