Highway Engineering Premium Rate Study

China is now in the rapid development of highway construction period.Investment in highway construction increase every year. Because of our country are a vast country,and have a lot of domestic natural disasters and geological environment. So the construction of a highway will face many risks. Even though our country highway investment are very large, but the investment in highway engineering insurance are very small. For one thing road construction company do not have enough risk awareness, for another the road construction insurance in China is not mature. There are many problems, such as insurance price is very high but the scope of protection are small. While some of the insurance company in order to gain more share of the highway project insurance,they often do not consider seriously of the insurance risk. Because of the low price,all of them are in hopes of risk does not occur. If a claim happened them will not deal with it according to the contract terms. If we can find a method for determining the premium rate which can reasonable reflect a highway risk will help our country to improve insurance status of highway projects.View of this,this paper in order to determined a highway premium rate method which based on neural network and analytic hierarchy process. This method can reflect the risk of the highway construction. And combined with previous highway projects of the insurance rate and project information, which based on their risk level by their respective insurance rates for highway works. This method can not only risk their own reflect to the level of each project, but also based on the experience of previous projects insurance rates.This article focuses on three parts. First is risk analysis of highway engineering insurance. The appropriate risk analyse is the premise of the next step. Followed by the use of expert scoring method to determine the overall risk level of the highway. Accurate determination of the level of risk is a key factor in determining insurance rates. And finally we use the RBF neural network to simulate. The simulation with the RBF network is to pay special attention to the selection of training samples. Training samples directly affect the output value of network simulation. Finally we put the front the risk profile into the RBF neural network, get the final output value-- highway insurance rates.