A Relationship-map Matching Based Accident Prediction Method And System Implementation In Highway

In the recent years, to build more highways is one of the primary developmentareas in traffic area of china, the miles of highway increased quickly. At the sametime, accompany with the economic development of China, the amount of cars alsoincreased fastly. Under such circumstance, the traffic safety of highway becomes aproblem, many traffic accidents with serious aftereffect happened frequently. Toadapt to the new situations and new challenges in highway management, it isnecessary to adopt information technologies to assist the management of highway.Consequently, researching methods for highway accident prediction, designing andimplementing the highway accidents prediction and management system, havebecome the important approach for decreasing or avoiding highway accidents, andensuring the safety of highway.During the running period of highway, one of the most effect method to ensurethe safety of driving in highway is, monitoring all kinds of driving factors, andpredicting and managing the accidents. This theses begins its research aiming at thisgoal, and obtains the following results:1) We have surveyed and reviewed deeply on the theory and angering practiceof related areas. Currently, the accident probing method for highway mainly dividedinto two kinds such as by manpower or automatically, and there are many kinds ofaccident prediction methods. This theses classified these methods, and compared andanalyzed these methods with probing factors, decision algorithms, suitable situations,etc. The performance metrics are also discussed in this theses.2)We designed and implemented an highway accident prediction andmanagement system. Predicting the accident on highway are important for bothdrivers and management department of highway. This theses, firstly, analyzed thedemands of system, and proposed the functional division of system, outl ined thewhole design framework; then we discussed the implementation of system, anddescribed the whole implementation framework of system and the design andimplementation of function modules.3)We proposed a matching based method for highway accidents prediction.Prediction methods are the basis for predicting and managing highway accidents, themethod given in this theses is based on case reasoning process of human kind when resolve a problem. It monitors the factors during driving, and constricts the relationmap of driving factors, then matches it with the danger accident relation maps storedin the database of system, and calculates a similarity value, to decide whether anaccident would happen. We also analyzed the affection of this method by severalcases.The results of this theses are mainly useful in accidents prediction andmanagement of highway, and it provides guarantees for improving the safety ofdriving on the highway.