Constrained Scheduling Optimization of Improved LLH Resources for Engineering Management

To investigate how to maximize the benefits of engineering projects through effective scheduling strategies in the context of limited resources, this study develops a high-level heuristic strategy that leverages deep reinforcement learning algorithms within a hyper-heuristic framework, complemented by enhancements to low-level heuristic methods. Experimental results demonstrate that the proposed super-heuristic algorithm achieves excellent convergence quality and efficiency, reaching a minimum convergence value of 0.31 and requiring only 32.16 seconds for computation. The super volume and inverse gen-eration distance produced by this algorithm are 0.906 and 0.254, respectively. Additionally, the algorithm exhibits strong performance in uniformity and breadth evaluations, making it adept at handling various scenarios in multi-skill, resource-constrained project scheduling problems. In the realm of engineering project scheduling, the proposed method yields the highest solution quality, achieving a coverage rate of up to 0.954. It also demonstrates low levels of constraint violation and significant solution set entropy during the solving process, satisfying the requirements for constrained scheduling optimiza-tion. Furthermore, the deviation between the optimized solution and the ideal shortest project duration is less than 0.15, indi-cating improved efficiency in project duration optimization. The algorithm's success rate in consistently finding feasible solutions across multiple runs exceeds 90%. This study provides effective scheduling strategies and optimization methods for engineering project managers, equipping them to tackle the challenges posed by resource constraints in engineering pro-jects.