基于ELM神经网络软件在线失效预测

Software Online Failure Prediction Based on ELM Neural Network

提出了一个软件在线失效预测模型,以Musa的数据为实验数据,基于ELM算法的神经网络构建模型,以部分Masu的数据集作为神经网络的训练数据,把余下部分数据作为对模型的测试数据。因此数据主要分为两部分,分别为训练数据集和测试数据集,比例为6:1。通过实验结果表明,提出失效预测模型的预测值与实际值的平均误差在14.03%,相对于BP神经网络模型具有很大的提升。

This paper presents a software online failure prediction model.The experimental data is Masu＇s Data in this paper.The model is build based on ELM algorithm of neural network,some Masu＇s dataset could be used as the neural network train-ing data,the rest of the dataset be used as the test data.So the data are divided into training data set and test data set, and here the ratio is 6：1.