Method for linearize power function with three parameters of incomplete accelerated degradation data

A method was provided to linearize the power function with three parameters. It was in view of the nonlinear prediction with the incompletely accelerated degradation paths of null bias and scale factor of accelerometer. The method was based on the assumption that the exponential parameter of power function is independent with stresses. Due to the degradation rate was faster at the higher stress than that at the lower, and its degradation amount was close to the failure threshold, while the lower were far away from the failure threshold. Therefore, the degradation trajectory at the maximum stress level can better reflect the law of the degradation models. Accordingly, a principle called the maximum average correlation coefficient at the maximum stress level to determine the exponential parameter was put forward. Through this principle, the power function with three parameters is linearized, which overcomes the poor robustness of the model prediction due to the sensitive initial value and provides an engineering method for the prediction of nonlinear accelerated degradation paths.