A Research On Real Time Algorithm For Static Asset-Task Assignment Problems

The Assignment of Assets to Tasks(AATT)problem is to consider the problem of optimal assignment of tasks under limited resources.The methods that can solve the problem of asset-task assignment have important application value in many kinds of scenarios,such as goods distribution,advertising,personnel scheduling,weapon target assignment and so on.In the context of the Internet and internationalization,we will face larger problems,the problem of advertising in the media industry is more complicated,and the problem of weapons target assignment needs to be dealt with on a larger scale.According to the feature of AATT,this thesis designs a multi-layer assignment algorithm,a random assignment algorithm,two exchange algorithms,and a composite algorithm,which can solve the asset-task assignment problem in real time while ensuring a higher degree of accuracy.Our study mainly includes the following:Firstly,according to asset-task assignment problem,we designed six near-optimal algorithms that can be solved in real time and a synthetic algorithm.The near-optimal algorithm includes multi-layer assignment algorithm,random algorithm and exchange improvement algorithm based on these two algorithms.The synthetic algorithm is a combination of two algorithms from six near-optimal algorithms.According to the test results,the real-time solution algorithm proposed in this thesis can generally exceed the international level in terms of accuracy.Secondly,To address the asset-task assignment problem,we have collected 15 instance data based on related domestic and foreign research,and established a instance library containing 604 cases to serve as the test object of the algorithm model in this thesis.In this thesis,the minimum number of weapons and targets in the instance library are 4 and the maximum number of weapons and targets are 400.Compared with tests in the previous literature,the case database established in this paper is the largest,which is characterized by large quantity and scale.Thirdly,we build an algorithm testing system using C ++,and the testing system is includes input and output modules,algorithm modules,and auxiliary modules.In the input and output modules,we can complete calculations,load instances,data visualization,and calculation reports by setting parameters.The algorithm module can deploy the seven algorithms designed in this thesis to solve the calculation instances.The auxiliary module can complete the change of the reading path and clear the dialogue screen.Finally,we selected a computer test environment,and set the performance test indicators of the algorithm according to the asset-task assignment problem,including the accuracy of the algorithm calculation results and the calculation time.Through the use of an algorithm test system,six real-time algorithm tests are completed based on the established instance library.This thesis uses the algorithm test system to complete six real-time algorithms on the computer,the calculation instance library and the test system.In terms of algorithm test results,the synthetic algorithm can complete the calculation of the case with a size not greater than 100 within 1s.In terms of algorithm calculation accuracy,the statistical ratio of obtaining accurate solutions reaches 68.62%,and the maximum error of the combined survival target value of cases where no accurate solutions are obtained and the exact solutions does not exceed 9%.The calculation accuracy of the research results in real-time solution generally exceeds the results published by internationally renowned research institutions in 2019.