Study on the aviation accidents due to human factors based on improved Support Vector Machine

With the development of technology and equipment´s reliability, more and more accidents are caused by human error, so it is very important to do some research for analyzing the aviation accidents due to human factors. The appearance of Support Vector Machines(SVM) provides a new way to solve the problem. However, human factors accidents have great uncertainty in aviation, traditional SVM showed some limitations in solving the aviation accidents due to human factors. So, this paper proposes improved SVM based on kernel density estimation, it can estimate the whole distribution of probability density using samples when the whole data distribution is unknown. After investigating pilot, air traffic control, repair personnel, we analysis it carefully and get index system including 22 index layers. Firstly, using AHP to calculate weights, the importance of each index can be known. What´s more, we give the model of calculating human factors accidents based on improved SVM in kernel density estimation. Finally, the model was verified by examples.