基于正态云模型和模糊层次分析法的列车通信网络性能评估方法

为保证高速列车安全、可靠运行，研究了列车通信网络性能评估方法；综合考虑列车通信网络的实时性、可靠性和服务质量，建立了合理的列车通信网络性能评价指标体系，采用模糊层次分析法确定列车通信网络性能评估指标的权重；考虑列车通信网络评估过程中具有不确定性，构建了基于正态云模型和模糊熵的二维评估模型；建立了基于交换式以太网的大容量和高可靠性列车通信网络仿真平台，获取各指标样本数据，运用二维评估模型计算各指标的隶属度，依据模糊理论最大隶属度法则确定列车通信网络性能等级。研究结果表明:在列车通信网络状态良好时，60%评估样本的网络性能等级为Ⅰ、Ⅱ级，在网络丢包率和误码率较大时，40%评估样本的评估等级为Ⅲ、Ⅳ级，表明二维评估模型能够有效地反映列车通信网络状态；与仅运用模糊综合评价法相比较，两者的评估结果基本一致，反映了二维评估模型的准确性；模糊综合评价法不能消除评估过程中不确定性因素的影响，从而导致评估结果缺乏精确度，因此，提出的方法更适合于列车通信网络性能评估。 

Evaluation method of train communication network performance based on normal cloud model and fuzzy analytic hierarchy process

To ensure the safety and reliability of high-speed trains, a method for evaluating the performance of train communication networks (TCNs) was studied. A suitable system of performance evaluation indexes was proposed by considering the stringent requirements for TCNs in terms of real-time responsiveness, reliability, and service quality. Fuzzy analytic hierarchy process (FAHP) was used to determine the weights of performance evaluation indexes of TCN. To address the uncertainty of TCN evaluation process, a two-dimensional (2D) evaluation model based on the normal cloud model and fuzzy entropy was constructed. A TCN simulation platform was constructed by using switched Ethernet with large capacity and high reliability, and then used to obtain sample data for each index. The membership degrees of each index were computed by using the 2D evaluation model, and the performance grade of the TCN was determined by the maximum membership degree (from fuzzy theory) principle. Research results show that 60% of the evaluated samples have network performance grades of Ⅰ and Ⅱ when the TCN is in a good state. When the network has high packet loss rate and bit error rate, 40% of the evaluated samples have performance grades of Ⅲ and Ⅳ. Therefore, the result of the 2D evaluation model accurately reflects the state of the TCN. The result is largely consistent with the result from the fuzzy comprehensive evaluation (FCE), indicating that the 2D evaluation model is accurate. However, as it is not possible for the FCE method to exclude the influence of uncertainty in the evaluation process, its result lacks precision. Hence, the proposed method is more suitable for the evaluation of TCN performance. 6 tabs, 15 figs, 32 refs.