Stochastic Matrix Modelling and Scheduling Algorithm of Distributed Intelligent Computing System

Parallel and distributed processing has always been a hot field of scientific and technological research, development, and application. It is an important solution in the fields of scientific computing and data service processing, such as weather prediction, wind tunnel Reynolds numerical calculation, and financial services. Intelligent cloud computing has higher requirements for high-capacity and efficient computing. The ability of existing computing system has been difficult to meet its needs. It is necessary to establish an intelligent computing system with the self-organizing ability and realize efficient task scheduling. Since the coordination of computing and storage resource scheduling becomes the key to scheduling, this study designs scheduling tasks based on a large-scale multi-task distributed system, establishes the model of distributed intelligent computing system and the multi-objective optimization model of the task scheduling problem, and designs the IPSO algorithm combined with improved particle swarm optimization algorithm according to the model. First, the particle swarm optimization algorithm is used to generate the initial scheduling scheme, then the ant colony algorithm is initialized, and the final scheduling results are generated. Simulation results show that the performance of the algorithm has obvious performance advantages compared with the improved particle swarm optimization algorithm and the improved ant colony algorithm. In addition, this study presents the task migration conditions and optimization methods under the dual objectives of makespan and availability. This optimization operation increases the system availability without increasing the scheduling length. In the distributed system with heterogeneous availability, the algorithm is effective in the dual objective performance optimization of task completion time and system availability.