Rail Transit Early Warning Method Based on SVM

Aiming at the low precision of existing rail transit early warning methods, an early warning method based on SVM is proposed. Among them, the grid search method is used to optimize SVM parameters, and then the early warning model based on SVM is built. The simulation results show that, by optimizing the parameters of SVM model, the penalty coefficient C and g are set to 85.125 and 0.0613 respectively, and the corresponding accuracy can reach 92.112%. Among the influencing factors of accuracy, when the data dimension is 6, the data volume is 1000, and the noise ratio is 0.5, the accuracy of the early warning model keeps a high level. Compared with the early warning model based on BP neural network, the accuracy of model based on SVM can reach 65%, while the accuracy of model based on BP is only 45%. Therefore, the model based on SVM can effectively improve the accuracy of early warning, and has certain engineering reference and application value.