Study Of Heuristic Algorithm For Expert Assignment Problem

Peer review is the main method in project review of project management. Expert assignment is chief and basic work of peer review. With the increasing of the project amount and the widening of the research scope, traditional assignment method and manual operation has not satisfied with the need of project management. So for solve current expert assignment status, applying optimization technique to study a more efficient and more convenient method becomes important approach and certain measure.According to the process and method of modeling, this paper studies the expert assignment problem, then set up the math model. Based on the property of model, this paper studies the solutions of non-linear programming and combinatorial optimization, analyses and contrasts the common heuristic algorithms, proposes the idea that solving expert assignment problem with heuristic algorithms.First, this paper analyses the property and performance of genetic algorithm, validates the advantages of genetic algorithm on solving high dimension and multimodal problem. Combining with the property of the expert assignment problem, this paper designs the genetic operator of selection, cross, mutation, proposes the method of solving expert assignment problem with genetic algorithm.Secondly, this paper analyses the property of ant algorithm, studies the solutions of resource assignment problem with ant algorithm, proposed the method of solving expert assignment problem with ant algorithm and solves the key problems: heuristic information select, valid solution construction and pheromone updating.Finally, the method solving expert assignment problem based on GA and AA is applied in actual data as experiment. The experiment result shows that both of them get better result than rough assignment and the GA result is better than manually modified assignment. The experiment result indicates that the solution is efficiency in practical application, which proofs validity of the model and feasibility of the method.