Task Scheduling Algorithm of Administrative Management System Based on Ant Colony Optimization Neural Network

In order to better solve the problem of large amount of interactive data in cloud computing task scheduling in administrative system, a hybrid genetic annealing algorithm is proposed based on the advantages of genetic algorithm and annealing algorithm. According to the characteristics of the task scheduling data relationship in the administrative system, the algorithm encodes the execution task in the cloud computing environment, searches the global optimal solution of the task data through the fitness function of the genetic algorithm, and uses the annealing algorithm to adjust the fitness function to speed up the completion time of the task scheduling in the system virtual machine during the coding and cross reorganization of the task data. Compared with genetic algorithm, simulated annealing algorithm and BP algorithm with momentum term, the simulation results show that the neural network trained by ACO algorithm has a faster convergence speed and can reach a smaller mean square error. The system realizes the free customization of assessment indicators, the flexible setting of assessment indicators’ weights and assessment scores, and the storage, tracking and management of assessment results over the years. Scheduling is a NP-hard problem, and many heuristic methods at present still have defects in optimization ability. With the development and maturity of artificial intelligence, one of the important branches, the theory of computational intelligence, has been widely used. The dial test system simulates the behavior of terminal users, executes dial test services, and obtains service usage by analyzing service performance indicators that users care about, and provides services for terminal service testing of mobile networks. This paper mainly studies the task resource scheduling model of the dial test system, and proposes an autonomous task scheduling algorithm suitable for the dial test system environment.