| 姚松涛,崔洁,赵河明,彭志凌,孔德景.引信步进应力加速试验贮存寿命预测研究[J].装备环境工程,2024,21(2):51-58. YAO Songtao,CUI Jie,ZHAO Heming,PENG Zhiling,KONG Dejing.Storage Life Prediction of Fuze under Step Stress Accelerated Test[J].Equipment Environmental Engineering,2024,21(2):51-58. |
| 引信步进应力加速试验贮存寿命预测研究 |
| Storage Life Prediction of Fuze under Step Stress Accelerated Test |
| 投稿时间：2023-12-17 修订日期：2024-02-03 |
| DOI：10.7643/issn.1672-9242.2024.02.007 |
| 中文关键词: 步进应力加速寿命试验 BP神经网络 引信 改进粒子群优化算法 Bayes理论 环境因子中图分类号:TJ430 文献标志码:A 文章编号:1672-9242(2024)02-0051-08 |
| 英文关键词:step stress accelerated life test BP neural network fuze improved particle swarm optimization algorithm Bayes theory environmental factor |
| 基金项目: |
|
|
| Author | Institution |
| YAO Songtao | Changzhi Industrial Technology Research Academy, North University of China, Shanxi Changzhi 046012, China |
| CUI Jie | Changzhi Industrial Technology Research Academy, North University of China, Shanxi Changzhi 046012, China |
| ZHAO Heming | Changzhi Industrial Technology Research Academy, North University of China, Shanxi Changzhi 046012, China |
| PENG Zhiling | Changzhi Industrial Technology Research Academy, North University of China, Shanxi Changzhi 046012, China |
| KONG Dejing | The 714th Research Institute of China Shipbuilding Industry Corporation, Beijing 100101, China |
|
| 摘要点击次数: |
| 全文下载次数: |
| 中文摘要: |
| 目的 针对某机电引信加速寿命试验数据，采用传统统计分析方法存在计算量大、寿命预测精度难以保证的问题，开展与智能算法相结合的引信贮存寿命预测研究。方法 针对步进应力加速寿命试验数据，采用贝叶斯理论的环境因子法，对各级应力下的贮存时间进行折合计算。利用进化策略对粒子群算法进行改进，进而对所建立的BP神经网络预测模型的全局参数进行调整和优化，突破传统方法的局限。将折合后的试验时间、样本量、应力水平作为网络输入，失效数作为输出，来预测引信贮存寿命。结果 利用训练好的 BP神经网络预测引信在正常应力水平下的失效数，计算其贮存可靠度。在迭代402次后，模型找到最优解，且预测误差在1%以内。结论 步进应力加速寿命试验与智能算法相结合的方法计算过程简单，预测精度较高，可有效提高引信贮存寿命的预测精度。 |
| 英文摘要: |
| The work aims to study the storage life prediction of fuze combined with the intelligent algorithm against the problem that the traditional statistical analysis method adopted for accelerated test data of fuze in a certain motor has high computational complexity and cannot guarantee the storage life prediction accuracy. For the step stress accelerated life test data, the environmental factor method based on Bayesian theory was adopted to convert the storage time at different stress levels. The particle swarm algorithm was improved by evolutionary strategy to adjust and optimize the global parameters of the BP neural network, breaking through the limitations of the traditional method. The converted test time, sample size, and stress level were used as inputs to the network, and the failure count was used as the output to predict the fuze storage life. The trained BP neural network was used to predict the failure count of the fuze under normal stress levels, and then calculate its storage reliability. After 402 iterations, the model found the optimal solution with a prediction error within 1%. Therefore, the combination of step stress accelerated life test and intelligent algorithm can effectively improve the prediction accuracy of fuze storage life. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
