A bi-population hybrid genetic algorithm for the limited pre-emptive multimode resource-constrained project scheduling problem with fast tracking

The maximum number of pre-emptive multimode resource-constrained project scheduling problem with fast tracking (Maxnint_PMRCPSP-FT) is addressed by extending the traditional multimode resource-constrained project scheduling problem with activity pre-emption. Activities are divided into work packages with constraints on the maximum number of splits and minimum continuous execution workload to determine strategies to minimize the makespan using mode changes or fast-tracking subactivities after pre-emption. A mixed-integer programming model and bi-population hybrid genetic algorithm were developed for optimal solutions by adapting key heuristic elements (codification, serial schedule generation scheme and double justification) for pre-emption and fast tracking. This algorithm incorporates crossover and mutation operations, local search strategies, population cooperation and parameter optimization using an orthogonal experimental design. Validated through ablation studies and extensive comparisons on generated datasets, the proposed method significantly improves solution quality, particularly with increasing complexity. Real-world testing in a residential project confirmed the improved schedule quality through limited pre-emption and within-activity fast tracking.