柔性 Job Shop¶
柔性 Job Shop¶
同一工序可以选择多个候选机器，不同机器的加工时长可能不同；还可能存在
共享操作员或其他累计资源。flexible_job_shop_small.py 使用
SchedulingModel 的 task、alternative、exactly_one_alternative 和
chain 组织任务语义，再用 unary/cumulative resource 施加容量约束。
这类可选执行方式是调度 facade 的典型使用场景。
完整代码¶
examples/scheduling/flexible_job_shop_small.py
from __future__ import annotations
from pathlib import Path
import sys
from optagent import ModelBuilder, SchedulingModel, solve
sys.path.insert(0, str(Path(__file__).resolve().parents[1]))
from _common import print_solution
def build_model() -> object:
builder = ModelBuilder(metadata={"case": "flexible_job_shop_small"})
schedule = SchedulingModel(builder, horizon=30)
machine_a = schedule.unary_resource("machine_a")
machine_b = schedule.unary_resource("machine_b")
shared_operator = schedule.cumulative_resource("shared_operator", capacity=1)
cutting = schedule.task("cutting")
cutting.alternative(resource=machine_a, duration=3, requirements={shared_operator: 1})
cutting.alternative(resource=machine_b, duration=4, requirements={shared_operator: 1})
cutting.exactly_one_alternative()
drilling = schedule.task("drilling")
drilling.alternative(resource=machine_a, duration=4, requirements={shared_operator: 1})
drilling.alternative(resource=machine_b, duration=2, requirements={shared_operator: 1})
drilling.exactly_one_alternative()
packing = schedule.task("packing")
packing.alternative(resource=machine_b, duration=2, requirements={shared_operator: 1})
machine_a.no_overlap()
machine_b.no_overlap()
shared_operator.cumulative()
schedule.chain([cutting, drilling, packing])
builder.minimize(schedule.makespan(), name="makespan")
schedule.validate()
return builder.freeze()
def main() -> None:
solution = solve(build_model(), time_limit_s=10.0, seed=7, threads=1, log_level="on")
print_solution("flexible job shop solved by SchedulingModel and unified solve", solution)
if __name__ == "__main__":
main()