Research On Resource Allocation Optimization Method Based On Time Series

With the widespread use of information technology in people’s daily grind, all kinds of information system has accumulated a large amount of business data with time series properties, which provides convenient conditions for the data science research work. That communication link and the terminal equipment are not stable in the process of data acquisition would lead to data exception or data missing, and seriously affect the integrity and availability of data. Research on time series prediction and restora-tion technique, can effectively improve the quality of data, provide data support for the follow-up production and business decisions. And when the data integrity is ensured, we can do further work on data mining and analysis. This paper matches individual and overall business data, proposes adjustment scheme in order to realize the optimization of resource allocation. This paper mainly completed restoration and prediction of time-series data (based on electrical load) and optimization algorithm of resource allocation.(1) Restoration and prediction of time-series data (based on electrical load).In order to satisfy the requirements of fast restoration, a fast linear restoration al-gorithm is proposed according to the characteristic of electrical load which is sequential and periodical.In order to satisfy the requirement of prediction, power load data is divided ac-cording to its characteristics, and ARIMA model parallelization based on multi-thread technology in Java is presented. Finally through the comparison experiments, the ad-vantages and disadvantages of the linear restoration method and the ARIMA model are pointed out. After comparing with the serial and parallel algorithm, we verify the usefulness and applicability of the parallel work.2) Research on optimization method of resource allocationThis part, we first abstract and define the problem of resources allocation math-ematically, and after analyzing the problems, we pointed out that the allocation of re-sources is a nonlinear programming problem with equality constraints. Next, we an-alyze this problem in analytical method and pointed out that the analytical solution in polynomial time is very difficult, therefore a better solution is need to be found.In order to seek numerical solution, we first propose a traversal method. But through computing the complexity of this method, we point out that the solution is not suitable for the problem. And then we put forward global approximate algorithm to solve the problem. Moreover, we analyze the accuracy and complexity of the approx-imate method.Next, considering some situations, we point out the local approximate algorithm.Finally, with the use of bank business data, tweo approximate algorithms and was tested, After comparing with the global and local algorithm,the result shows that tweo approximate algorithms are efficient to solve the configuration problem.