中文摘要：

为了研究某型陀螺仪在长期存储期间的可用性,需要对其存储寿命进行预测和评估,确定其存储寿命和可靠度。对谐振陀螺仪在高温试验箱中进行672 h的热应力试验,记录其性能参数退化数据;利用最小二乘支持向量机（LS-SVM）在处理多元小样本数据方面的优点,在对热应力性能退化试验数据进行SVM训练学习的基础上,拟合回归出高温条件下陀螺仪的寿命模型;将室温储存条件下的参数作为输入,得到陀螺对应此参数的可靠度,此可靠度所对应的时间就是陀螺在正常储存条件下的寿命。根据以上方法预测出在20℃的条件下该陀螺仪的储存平均无故障工作时间等于5.4×104h,并与用LS估计的结果进行了对比,证明了用LS-SVM预测陀螺寿命的可行性。

英文摘要：

In order to study the availability of a certaintype of gyroscope during long storage period, it needs prediction and assessment on gyro storage life to determine its life and storage reliability. Thermal stress test on resonant gyroscope is carried out to record its performance parameters degradation datas. Taking advantages of least square support vector machine （LS-SVM）for fitting the gyroscope＇s life model. Through training on the degradation data regresses of the gyroscope＇ s life model and using parameters at room temperature as inputs that obtains the gyro corresponding reliability of this parameter. The reliability corresponding time is gyro＇ s life under normal storage conditions. According to the above method, under the conditions of 20 ℃, the storage MTBF of the gyroscope is predicted equal to 5.4 × 104 h, the results is compared with LS estimation and the feasibility of gyro life prediction using LS-SVM is proved.