变更管制方式安全评估模型及软件设计

为确保因管制方式变更造成系统及运行环境变化带来的风险保持在可控范围，提出基于粗糙集模糊神经网络的安全评估模型。利用粗糙集在属性约简与规则提取方面的优势实现关键风险源提取，降低模糊神经网络数据输入维度、精简网络拓扑结构、缩短网络训练与学习时间；借助模糊神经网络在具有较强容差性和抗噪音能力前提下进行分类的能力，实现系统安全等级评定。为便于模型应用与推广，借助Visual Basic（VB）与MATLAB这2种语言将评估模型编译为可视化操作软件，该软件具有从指标体系中识别关键风险源、系统安全等级评估、评估信息汇总这3项功能。实验结果表明，该软件可从24个风险评估指标中提取出对系统安全影响最为关键的4项风险指标，系统安全等级评定为3级。该评估软件具有易于安装维护、操作简便、理论化程度高等优点，是对空管运行单位定量安全评估的一次创新尝试。

Safety evaluation model about air traffic control means change and its software design

Air traffic control means change brought risks to its system and operating environment. In order to make those risks controlled, the safety evaluation model based on rough sets and fuzzy neural network was proposed. Rough set, which was used to extract key risks, has the advantage for attributes reduction rule extraction and also provids the following functions to fuzzy neural network: decreasing samples dimension, simplifing the network topology and shortening the time of network training and learning. Fuzzy neural network, which was used to evaluate the safety level of system, has the capability of classification on the condition that fuzzy neural network can tolerate error and noise. In order to apply and promote the safety evaluation model, the model was compiled to visualized operation software by visual basic(VB) and MATLAB. The evaluation software had three functions that identified key risk indexes from the evaluation index system, evaluated the safety level of system and collected evaluation information. The experimental results showed that the software can extract 4 key risk indexes from 24 evaluation indexes and the safety level of system is 3. The evaluation software processed following advantages: easy installment and maintenance, straightforward operation and higher theoretical level. It was an innovative attempt that quantitative evaluation was used to operation units of air traffic control.