基于改进LS-SVM算法的列车通信网络时延预测方法

由于通信网络诱导时延的存在会对列车牵引制动系统造成影响,因此对时延精准预测并实现补偿十分重要。提出了一种基于改进粒子群(PSO)算法优化的最小二乘法支持向量机(LS-SVM)算法对列车通信网络时延进行预测,搭建了列车网络控制系统半实物平台,使数据通过多功能车辆总线(MVB)进行传输,分别改变车辆控制单元(VCU)特征周期及负端口数量大小,以获取大量不同特性的时延数据。将数据分组后利用改进的PSO算法优化LS-SVM算法进行预测仿真。仿真结果表明,与传统的LS-SVM算法及Elman神经网络算法的预测方法相比,所提出的方法在列车通信网络的时延预测方面具有更好的快速性和准确性。

Time Delay Prediction Method for Train Communication Network Based on Improved LS-SVM Algorithm

Communication network induced delay will affect the train traction braking system,therefore it is very important to accurately predict and compensate the delay.A least squares support vector machine(LS-SVM)algorithm optimized on the basis of improved particle swarm optimization(PSO)is proposed to predict the train communication network delay.A semi-physical platform of the train network control system is constructed for data to be transmitted through Multi-function Vehicle Bus(MVB),and the vehicle control unit(VCU)characteristic period and number of negative ports are respectively changed to obtain a large number of delay data of different characteristics.After the data is grouped,the improved PSO is used to optimize LS-SVM method and the prediction simulation is conducted.The simulation results show that compared with conventional LS-SVM algorithm and Elman neural network algorithm prediction methods,the proposed method has better prediction of the train communication network time delay in terms of speed and accuracy.