Life-prediction of accelerated life testing based on support vector machine

It is difficult for parametric methods to be used to analyze data from accelerated life testing when the type of life distributions is unknown. A practical model is proposed in this paper to predict the life of the items in accelerated life testing based on support vector machine. Simulation and experimental analysis show that, this approach has an advantage of high precision. The model can reflect the relationship between the stress levels, the reliability, and the life of the items each other. By using this model, a useful way to predict the life in accelerated life testing is presented in this paper.