Prediction Of Remaining Life Of Web System Based On Recurrent Neural Network

With the development of the network information age,the level of technology has increased significantly,and various fields are increasingly dependent on computers and software systems such as the Web.Especially in the fields of finance and military,people have higher requirements for the security and reliability of software systems.However,a large number of software aging problems still exist in systems that run uninterruptedly for a long time,making the security and reliability of the system under serious threat and causing incalculable losses.For software systems that have already undergone aging,it is possible to prevent the occurrence or failure of software by proactively performing anti-degradation operations.If we can accurately predict the remaining service life of the software,and take reasonable software rejuvenation operations at the right moment before system failure to solve the problem of system performance decline,It can not only reduce the loss caused by taking software anti-decay operation,but also help to maintain the high reliability of the software system.Therefore,it is crucial to accurately predict the remaining lifetime of Web software systems.In this paper,based on the Keras framework,the remaining service life prediction of Web software systems is implemented on the basis of recurrent neural networks(RNN),and the specific work is as follows.(1)A real-time residual life prediction method of the Web software system based on the long short-term memory(LSTM)network has been proposed.Firstly,an accelerated life test platform has been built to collect the performance indicators reflecting the aging trend of the Web software system.Then,according to the timing characteristics of the indicated data,a real-time residual life prediction model of the Web software system based on LSTM has been constructed and trained.The experimental results show that the prediction model can effectively predict the remaining life of the Web software system in real time,with good accuracy and applicability.Applying the proposed model to the life prediction of the Web software system can effectively complete the prediction.This method provides technical support for optimizing the anti-aging decision-making of the system.(2)A real-time remaining life prediction method of web software system based on Self-Attention Long Short Term Memory(LSTM)was proposed.In order to predict the Remaining Useful Life(RUL)of web software system in real time and accurately,taking into consideration the time series characteristics of the web system health status performance indicators and the interdependence between indicators,By assigning coefficients with different attention to the output results of the LSTM hidden layer at different moments,a real-time Web software system residual life prediction model based on self-attention long short-term memory networks was proposed to distinguish the importance of local residual life information features more comprehensively.A comparison was made with the BP network and the conventional recurrent neural network on the three test sets.The experimental results show that the Mean Absolute Error(MAE)of the model is 16.92% lower than that of LSTM,and the relative accuracy(Accuracy)is 5.53% higher than that of LSTM,which verifies the validity of the residual life prediction model of Self-Attention-LSTM network.This method provides technical support for optimizing the software rejuvenation decision of the web system.