摘要
在对高可靠性产品进行寿命预测时,传统的加速模型拟合精度低,且预测误差大,难以对产品做出较为准确的寿命预测.针对此问题,提出了一种基于优化灰色预测模型的方法对传统加速模型进行参数修正,提高加速模型的显著性和预测精度.首先采用动态寻优法确定最优背景值,减小灰色预测模型的拟合误差,再利用优化后的灰色模型预测出产品在接近正常使用温度下的寿命,将失效数据进行有效等维扩充,利用扩充数据对传统加速模型进行参数修正并检验其显著性.最后通过实例分析表明:修正模型的显著性更高、预测相对误差更小,且在越接近产品正常使用的温度下,其预测精度越高,说明此方法具有“数据量少、预测精度高、实用性强”的特点,因此采用该方法对高可靠性产品进行寿命预测是可行且更加有效的.
In the life prediction of high reliability products,the traditional acceleration model is blamed for its low fitting accuracy and large prediction error,so it is rahter difficult to make a more accurate life prediction for products.In response to this,a new method based on optimized grey prediction model is proposed to modify the parameters of the traditional acceleration model so as to improve the significance and prediction accuracy of the acceleration model.Firstly,the dynamic optimization method is used to determine the optimal background value and reduce the fitting error of the grey prediction model.Then,the optimized grey model is used to predict the life of the product at near normal service temperature,and the failure data is effectively expanded in the same dimension.The expanded data is used to correct the parameters of the traditional acceleration model and test its significance.Finally,the example analysis shows that the modified model has higher significance and smaller prediction relative error,and the prediction accuracy is higher at the temperature closer to the normal use of the product,indicating that this method has the characteristics of less data,high prediction accuracy and strong practicability.Therefore,it is feasible and more effective to use this method to predict the life of high reliability products.
作者
沈杰
李云飞
SHEN Jie;LI Yunfei(School of Mathematics and Information,China West Normal University,Nanchong,Sichuan 637002,China)
出处
《内江师范学院学报》
CAS
2024年第2期17-23,共7页
Journal of Neijiang Normal University
基金
国家自然科学基金项目(72101172)
西华师范大学英才科研基金项目(17YC381)。
关键词
加速寿命试验
灰色预测模型
寿命预测
动态寻优法
高可靠性
accelerated life test
grey prediction model
life prediction
dynamic optimization method
high reliability