基于BP神经网络的实验室计算机故障率预测模型

Computer Lab Failure Rate Prediction Model Based on BP Neural Network

随着信息化技术在各个学科领域的渗透,高校中越来越多的课程要求学生在计算机实验室完成相关操作,随着上机人次陡增,计算机的损耗也随之增大。为了更好地对实验室进行维护,以湖北大学知行学院计算机系2005年计算机实验室210台计算机的历史故障率为样本,采用JAVA语言,利用BP网络训练模型预测该批计算机的故障率,然后对照历史数据发现一定的误差,再利用增加动量项法对该BP算法进行改进,改进后的样本训练预测结果与历史数据基本保持一致。

As information technology penetrating in various subject areas,more and more courses re-quire students to complete the relevant operations in the computer lab. With increasing of computerusers,computer loss increase. For better maintainability of lab,takes the historical failure rate of210 computers in 2005 in computer lab of computer department of Zhixing college of Hubei Universi-ty as sample,uses JAVA language,BP network training computer model to predict the failure rate ofthe batch of computers,and then contrasts the historical data to find some errors,and then uses themethod of increasing momentum term to improve the BP algorithm,the improved training samplespredicted results are basically consistent with historical data.