Time, cost, and quality are the main factors affecting industrial production, and balancing various constraints to make the production parameters reach the optimal value is an NP-hard problem. With the progress of process production, the production process of various products is more refined, and in the actual production process, the production process has the characteristics of multi-stage and parallel proceeding, therefore, for the difficult problem of multi-stage and non-linear production process optimization, this paper proposes a workflow multi-objective optimization algorithm ( DVSP) based on dynamic virtual phased pruning strategy, which is divided into three stages, firstly, eliminating the indirect constraint relationship between tasks by pruning strategy and simplifying directed acyclic graph (DAG), secondly, dividing the task node set by virtual hierarchical strategy and generating virtual nodes by using virtualization technology, and finally, calculating the stage in stages by an inverse reductive contracting optimal set of service nodes, forward scheduling generates the optimal path, and the global optimal solution is sought through algorithmic integration. After experimental comparison, the algorithm has a more obvious optimization effect than the traditional critical path algorithm and mass maximum algorithm, which meets the actual production needs.