摘要
LED寿命预测仍存在预测结果准确性不高、预测速度不足等问题。本文提出了一种基于门控循环单元神经网络的LED寿命预测方法,提出利用门控循环单元简化神经网络预测模型,并将模型输入数据的长度设置为变量,使预测过程能够考虑更多的LED历史退化信息;再采集LED流明维持数据作为训练样本,训练寿命预测模型;最后利用训练好的模型对LED进行寿命预测实验,并与其他三种方法进行对比。实验结果表明,本文方法具有更高的预测准确度和更快的预测速度,且具有良好的鲁棒性。
LED lifetime prediction still has problems such as low accuracy of prediction results and insufficient prediction speed.In this paper,an LED lifetime prediction method based on a gated recurrent unit(GRU)neural network is proposed.First,the neural network prediction model is simplified by using the GRU,and the length of the input data of the model is set as a variable,so that the prediction process can consider more historical LED degradation information.Then the collected LED lumen maintenance data is used as a training sample to train the lifetime prediction model.Finally,the trained model was used to conduct a lifetime prediction experiment and compared with the other three methods.The experimental results show that the proposed method has higher prediction accuracy and faster prediction speed,and the proposed method has good robustness and generalization performance.
作者
龚晓春
朱云
李晟
颜建堂
李玉晓
GONG Xiaochun;ZHU Yun;LI Sheng;YAN Jiantang;LI Yuxiao(School of Science,Jiangxi University of Science and Technology,Ganzhou 341000,China;Shenzhen Carol Lighting Co.,Ltd,Shenzhen 518012,China)
出处
《照明工程学报》
2022年第6期93-101,共9页
China Illuminating Engineering Journal
基金
国家自然科学基金(批准号:11875149)
江西省教育厅科技研究项目(GJJ210816)。
关键词
LED
门控循环单元
寿命预测
神经网络
LED
gate recurrent unit neural network
lifetime prediction
neural network
