论文部分内容阅读
针对小功率LED失效时间预测需要大量失效数据、预测成本较高的情况,从工程实践的实际需求出发,提出了基于改进蛙跳算法(ISFLA)优化支持向量机(SVM)的小功率LED寿命预测模型。SVM使用径向基核函数(RBF),以小功率LED的可靠度作为SVM输入,失效时间作为输出。为了减小蛙跳算法的计算成本,现引入进化阶段指标T加快收敛速度,并利用ISFLA优化核函数参数的选取,使得仅需少量训练样本即可建立SVM。该模型不仅能预测加速应力下的小功率LED失效时间,也可根据小功率LED在加速应力下的失效时间预测正常应力水平下的寿命。实验证明,在小样本条件下,该方法得到训练结果的相关系数为0.998,检验组误差小于3%,快速简洁地实现了小功率LED高精度寿命预测。
According to the actual demand of engineering practice, a small power LED lifetime prediction based on Improved Frog Leaping Algorithm (ISFLA) optimized support vector machine (SVM) is proposed in order to predict the failure time of low power LED which needs a large amount of failure data. model. The SVM uses a Radial Basis Function (RBF), which takes the reliability of a low-power LED as the SVM input and the dead time as the output. In order to reduce the computational cost of frog leaping algorithm, the index of evolution phase T is introduced to speed up the convergence, and the selection of kernel function parameters is optimized by using ISFLA so that the SVM can be established with only a few training samples. The model not only predicts the failure time of low power LED under accelerated stress, but also predicts the life under normal stress level according to the failure time of low power LED under accelerating stress. Experiments show that under the condition of small sample, the correlation coefficient of the training result obtained by the method is 0.998, the error of the test group is less than 3%, and the high-precision life prediction of small power LED can be realized quickly and concisely.