用结构自适应神经网络预测航空发动机性能趋势

Forecasting Engine Performance Trend by Using Structure Self-Adaptive Neural Network

将航空发动机作为复杂非线性系统考虑,运用神经网络超强的非线性映射能力和非线性时间序列分析的相空间重构理论,建立航空发动机性能趋势预测的神经网络模型,同时,针对神经网络的结构设计困难问题,建立了基于遗传算法的结构自适应神经网络预测模型,实现了神经网络结构的优化。最后,利用三组民航飞机发动机的性能数据进行了预测分析,验证了利用结构自适应神经网络对航空发动机性能趋势进行预测的有效性。

In this paper, the aero-engine is considered as a complex non-linear system, and by using the strong nonlinear mapping ability of artificial neural network （ANN） and the phase space reconstruction theory, the ANN model of aero-engine performance trend forecasting is established. At the same time, aiming at the problem of ANN structure design, the structure self-adaptive ANN forecasting model is put forward, which can automatically realize structure optimizing by genetic algorithm （GA）. Finally, three groups of practical performance data from civil aviation engines are used as forecasting analysis, and the results verify fully the correctness of the method which is put forward in this paper.