A Two-Stage Optimization Approach Using Rounding-Off Particle Swarm Optimization for the Crowdsourced Resource Scheduling Problem

An application of the scheduling problem for projects with multiskill resource limits is the Crowdsourcing Resource Scheduling Problem (CRSP). Since the crowdsourcing issue with resource scheduling also involves resource constraints and skill matching problems, it is consistent with the core principles of Muti-skill Resource Constrained project Scheduling Problem (MS-RCPSP). Therefore, a two-stage optimization method (TS-RND-PSO) based on rounding-off particle swarm optimization (RND-PSO) was proposed to model the CRSP model by using MS-RCPSP framework. To begin, the task sequence vector is utilized to encode the solution, and a decoding technique incorporating resource limitations and talent matching is presented to provide a realistic scheduling scheme. Then, the twostage dynamic inertia weight adjustment strategy is adopted to expand the search area and enhance the widespread search ability by optimizing the high inertia weight in the early stage, to avoid falling into the local optimal. In the second stage, the inertia weight is gradually reduced, and the search area is aggregated to the local area near the optimal solution to increase the algorithm’s comprehension accuracy and convergence efficiency. Results from experiments indicate that when compared to conventional techniques, the proposed algorithm has significant advantages in scheduling accuracy and convergence speed.