基于神经网络和框架推理的风险应对专家系统

为了提高飞机装配项目风险应对的效率及可靠性，提出一种将神经网络和框架知识系统相结合的风险应对方法。该方法基于框架表示理论，以方案与风险因子之间存在映射为规则，将专家风险应对经验存储于专家系统知识库之中。使训练后的神经网络以连接权的形式获取知识。使用时，运用匹配算法对神经网络输出和知识库中的知识进行推理匹配，得到最终应对方案。该方法在某型飞机装配项目的风险管理中得到应用，验证了方法的有效性。

Risk Response Expert System Based on Neural Network and Framework Reasoning

The expert judgment risk response method used now in aircraft assembly project need to be improved for its subjectivity and complication. To make it more effective and reliable, this paper proposes an improved method that integrating neural network with framework reasoning. Based on frame theory, it stores experts’ experiences about risk response in the expert system repository by the rule that there are mapping relations between solutions and risk factors. The neural network is trained to obtain the knowledge in the form of weight. When using, it gets the response solution for the risk by reasoning and matching the output of the neural network with the knowledge in repository. It gives attention to the coupling of all risk factors to avoid the loss of information and also can simulate the expert thinking exactly to advance the project management ology level of aircraft assembly. This method has been replied in a certain airplane design project. Application illustrates the feasibility and validity of the method.