Research On Cloud Software Aging Prediction Based On Artificial Bee Colony Algorithm And BP Neural Network Model

Software aging refers to the phenomenon of abnormal software status,performance degradation,system downtime or even failure caused by memory leaks,fragmentation problems,and numerical accumulation errors on long-running software systems.With the continuous development of cloud computing technology,the software aging phenomenon under the cloud platform continues to occur.To address these software aging phenomena,there are currently three main software aging prediction methods,namely the model-based methods,the measurement-based methods,and the mixture of the two methods.Existing research mainly uses the measurement-based methods,that is,using machine learning algorithms to predict the collected data,so as to determine the age of the software and perform rejuvenation on it.Among many machine learning algorithms,the BP neural network can achieve satisfactory prediction results,but the weights and thresholds of the BP neural network are randomly initialized,and there are problems that it is easy to fall into the local optimal solution and slow convergence speed,so Using BP neural network for software aging prediction still has room for improvement in accuracy.This thesis proposes a prediction method based on artificial bee colony algorithmto optimize BP neural network.Firstly,a software aging prediction model based on artificial bee colony algorithm to optimize BP neural network is designed and implemented,and the effectiveness of the experimental method in this thesis is verified through the data set collected by Alibaba Cloud.Then the method proposed in this thesis is tested on the large-scale data set of the Google cloud platform.By standardizing and normalizing the collected real data set,the single BP neural network and the experimental method used in this thesis are compared.The prediction accuracy illustrates the effectiveness of the method in this thesis in real-time software aging prediction,and the superiority of the experimental method in this thesis through comparison with other machine learning algorithms.Finally,the data set under the Google cloud platform is used to set a fixed threshold value on the parameter index to estimate the time to perform the rejuvenation operation,based on the actual data set to complete the analysis and confirmation of rejuvenation strategy formulation.Experimental results show that the method proposed in this thesis has faster convergence speed and higher prediction accuracy in the software aging prediction problem under the cloud platform,and can well complete the software aging prediction task under the cloud platform.