针对加速度计零偏和标度因数不完全加速退化数据的非线性预测问题,提出一种加速度计加速退化的三参数幂函数模型线性化方法.该方法假设三参数幂函数退化轨迹模型的幂指数与加速应力无关,考虑到不同应力水平的试验时间相当时,较高应力下的参数退化速率较快,其退化轨迹与较低应力水平的相比,退化量更加接近性能参数的失效阈值,而低应力水平下退化量与失效阈值差距较大,因此最高应力水平下的退化轨迹更能反映加速度计性能参数的长期非线性退化规律.据此进一步提出了加速退化数据三参数幂函数模型幂指数确定的均方相关系数最大化原则,该原则通过高应力水平试验数据确定模型的幂指数,由此实现三参数幂函数模型的线性化,克服由于初值敏感而导致模型预测时稳健性较差的问题.为加速退化轨迹非线性模型的预测提供了一种工程化方法. Aiming at the problem of non-linear prediction of accelerometer zero-bias and scale factor incompletely accelerated degradation data, a linearization method of accelerometer acceleration degradation based on three-parameter power function model is proposed. This method assumes that the power- Regardless of accelerating stress, when the test time of different stress levels is equal, the parameter degradation rate under higher stress is faster, and its degradation trajectory is closer to the failure threshold of performance parameter than the lower stress level Therefore, the degradation trajectory under the highest stress level can better reflect the long-term nonlinear degradation of accelerometer performance parameters.Accordingly, the power-of-three exponent index of accelerating degeneration data is further proposed The principle of maximizing the mean-square correlation coefficient was established. The principle was to determine the power exponent of the model through the test data of high stress level, so as to realize the linearization of the power model with three parameters and to overcome the poor robustness when the model was predicted by the initial value sensitivity Which provides an engineering method for accelerating the prediction of nonlinear models of degenerate trajectories.