复杂快变不确定非线性系统的定性/定量混合故障预测方法研究

项目名称： 复杂快变不确定非线性系统的定性/定量混合故障预测方法研究

项目编号： No.60874117

项目类型： 面上项目

立项/批准年度： 2009

项目学科： 轻工业、手工业

项目作者： 任章

作者单位： 北京航空航天大学

项目金额： 36万元

中文摘要： 针对飞行控制系统复杂快变的特性与恶劣工作环境的特点，深入研究了飞控系统故障产生机理和征兆传播机理。以此为基础，在深入研究了基于定量模型的故障预测方法和基于定性模型的故障预测方法后，发现定量预测方法能够较为精确的预测出系统状态的变化趋势，而定性预测方法对系统状态的转折或突变比较敏感。因此，以解决此类复杂快变不确定非线性系统的故障预测问题为出发点，将定量、定性故障预测方法有机结合，建立了适合于复杂快变不确定非线性系统故障预测的定性/定量混合模型，并用更为先进的方法对其进行改进。主要研究内容包括以径向基神经网络（RBF）为代表的定量时间序列预测方法；以模糊推理为代表的定性时间序列预测方法，并与定量方法进行了预测效果对比；根据分析，提出以模糊神经网络这种混合方式作为预测模型，并运用云模型方法对其优化改进。仿真实验结果表明，这种定性/定量混合预测模型具有很好的预测性能，并能够用于滚动预测，为在线故障预测提供了可能性，也为飞机等系统的健康管理提供了理论依据。

中文关键词： 故障预测；神经网络；模糊推理；云模型；定性/定量

英文摘要： In view of the complicated characteristics and hash work environment characteristics of the flight control system, we studied the mechanism of fault generation and symptom transmission. According to this, we found that the method of quantitative forecasting can predict the trend of system states more accurately while the method of qualitative forecasting is more sensitive to the transition or mutation of system states after in-depth study about the failure prediction based on quantitative and qualitative methods. Therefore, in order to solve the problems of fault prediction of complex fast-varying uncertain nonlinear system, we establish the qualitative/quantitative hybrid model and use more advanced methods to improve it. The main contents include the radial basis function neural network(RBF) as the representative of quantitative time series forecasting methods, fuzzy reasoning which is compared with quantitative method as the representative of qualitative time series forecasting methods, and the hybrid approach of fuzzy neural network as a predictive model which is optimized using the method of cloud model. The simulation results show that the qualitative/quantitative hybrid forecasting model has good predictive performance and it can be used for rolling prediction. This method offers a possibility for online fault prediction and provides a theoretical basis for health management of aircraft.

英文关键词： Fault Prediction;Neural Networks;Fuzzy Inference;Cloud Model;Qualitative/Quantitative