Study On Discriminant Algorithm Of Network Traffic Congestion Status Based On Cloud Model

In intelligent transport systems, advanced traffic management system has beenwidespread concern. On this basis, the algorithm of the automatic congestionidentification was generated. It was referred to as ACI algorithm. Previous ACIalgorithms do not involve the processing of qualitative information. Therefore, thispaper proposes an ACI algorithm based on cloud model.In the process of traffic data preprocessing, the identification and reparation offault data based on cloud model was introduced into this paper. Then we can convertthe qualitative concept into the quantitative data by comparing the degree of similarityabout concept, and repair the fault by using the quantitative data. Compared withtraditional data recovery methods, its average relative error was reduced by nearly fourpercentage points.A three-level indicator system of road network traffic status was established Theunderlying indicator was common parameter of the traffic flow, the indicator of themiddle layer was indicator of road section traffic running state, and the top-levelindicator was indictor of network traffic running state. The traffic running state wasdivided into five grades: unimpeded, basically unimpeded, slight congestion, moderatecongestion and severe congestion.By using peak cloud transform and concept promotion of cloud transformalgorithm system, we get a qualitative concept of cloud reflecting the five kinds oftraffic running state. It was the numerical characteristics. We can achieve conversionbetween qualitative concept and quantitative data through discriminate the degree ofmembership. At the same time, we can select corresponding traffic condition of themaximum about the degree of membership by using the discrimination of cloud dropto particular parameter data of traffic flow. The final result of discrimination wasobtained. We can output discriminate result in accordance with a set of qualitativestatements.Compared with the classic ACI algorithm and the new ACI algorithm, theidentification rate of ACI algorithm based on cloud model was lower than DS-ANNalgorithm by0.19%. The false identification rate of ACI algorithm based on cloudmodel was higher than DS-ANN algorithm by0.02%. However, these differences werenegligible. At the same time, the mean time to identification of ACI algorithm based oncloud model was the shortest. The algorithm basically meets expectations.