Improved genetic algorithm based intelligent scheduling model and efficient solution method for green financial resources

Green financial resource scheduling must reconcile heterogeneous project requirements, time-varying capital availability and tail risk exposure within a single allocation decision. Manual review and static rule-based procedures coordinate these dimensions poorly once the number of candidate projects and financial instruments grows. This research formulates the problem as a constrained multi-objective programme in which a continuous decision vector specifies the fraction of each project’s financing demand routed through each eligible instrument, and twelve instrument-by-quarter capital ceilings together with priority-dependent minimum funding requirements define the feasible set. Economic cost, environmental benefit and conditional value at risk at the 95 per cent level are optimised jointly. An improved multi-population genetic algorithm, denoted IMaPGA, integrates subpopulation co-evolution, elite migration, state-dependent crossover and mutation, crowding preservation, reference-direction selection and feasible neighbourhood refinement. Comparative experiments against NSGA-II, MOPSO, SPEA2 and NSGA-III at scales from 50 to 400 projects use an identical budget of one hundred thousand objective-function evaluations with three repetitions per instance and scale, while MOEA/D and an epsilon-constraint sweep are evaluated in a separate comparison under a reduced verification budget. IMaPGA attains the highest hypervolume and the lowest inverted generational distance throughout the tested scales, and paired significance testing identifies the comparisons in which that advantage is statistically reliable. The framework converts green financial scheduling into a quantitative multi-objective decision process.