TEI@I框架下的交通道路脆性预测模型研究

Transform brittle forecasting model based on the TEI@I research frame

交通道路的重要作用使得其成为受到风险影响后的关键环节,灾害事故的发生往往具有不可预测性,但若能及时监测道路的抗风险能力则能及时调度和部署资源在薄弱环节,则可能降低事故发生后造成的巨大损失。本研究以TEI@I系统分析方法为指导思想,从系统理论出发,提出测量道路交通抗风险能力的熵值监测法。在此基础上引入Hilbert-Huang变换、支持向量机的熵值预测法,研究结果表明该方法具有较好的应用前景,对于指导交通应急管理工作具有重要的参考意义。

The important role of traffic always makes it become key factor after break out disasters or accidents which occurrence tends to be probabilistic and unpredictable.However,the resources could be scheduling at the vulnerable spot and reduce the tremendous loss after the accident if the anti-risk ability of the road could be timely monitored.The existing research are mainly focused on the traffic monitoring algorithm,however there still have problems of higher miscarriage rate,long testing time,low robustness,etc.In addition,the existing classification methods are difficult to distinguish the traffic abnormal state and found the cause making gradual change of the traffic incident,which could not meet the exact requirements of emergency traffic management.In order to cope with the effects of disasters or accidents,monitoring and evaluating the withstand ability of road to disasters and accidents are necessary.The optimal method of real-time monitoring which needs more manpower,material resources and long observation time often focused on one of the key sections and has high cost,moreover,it always be restricted by its own special local conditions.Road system which capacity is influenced by many factors has the characteristics of typical complex system.A complete road system is composed of several subsystems.When a subsystem suffers certain damage,it will be broken and leading to the collapse of the whole system.This damage we called brittle events which have characteristics of potentiality,stealthiness,burstiness and complexity etc.It is difficult to predict the outbreak of brittle events which damage to the road systems,nevertheless,if we can detect the orderly degree of the whole system,the anti-risk ability of the road could be under detection and control.Therefore this research used TEI@I research frame as guiding ideology and introduced the concept of entropy used to measure the brittle degree based on the complex system theory.Based on this,support vector machine(SVM)and Hilbert-Huang transform method were introduced to forecast the brittleness of traffic road system from the angle of complex system crash theory,this presented model can be traffic monitoring system to evaluate the road disaster resistance ability.By comparing historical data it is founded that the entropy value will soar when meet with traffic accident or disaster.Therefore,rapid assessment should be carried out before a disaster or accident affect the road system and makes the corresponding emergency preparedness for the emergency management.Based on maximum entropy,the brittle entropy curve is put forwarded to measure the brittle degree of road system.In order to get the brittle entropy curve,the distribution form of multiple brittleness factor information should be confirmed by the historical data and constraint information,then obtain the brittle entropy curve according to the relationship and function of brittleness factors.Then forecasting model is proposed by using Hilbert-Huang transform.Firstly decompose the entropy curve into different components which have different time scale characteristics,then calculate the influence of random disturbance factors on the curve according to the Hilbert-Huang spectrum.For the different components we built different support vector machine with different parameters and overlay the final prediction results.In order to further verify the validity of the method of this research,the BP neural network,wavelet neural network,single SVM model,ARIMA model were introduced to forecast the curve of maximum entropy and comparing with the proposed methods in this study,the results showed the proposed prediction method has better prediction precision.By comparing historical data it is founded that the traffic accident were much more serious when the entropy value is bigger.Therefore,we could carry out rapid assessment of the road’s ability of resisting the disaster before the disaster and making the corresponding emergency preparedness.The model shows higher accuracy by comparing with several forecasting models.The results show that the method has good application prospect and important reference significance for traffic emergency management.