As　one　of　the　most　fundamental　operations　in　mechanical　production,　hole-making　plays　a　crucial　role.　However,　existing　hole-making　sequence　optimization　models　are　not　suitable　for　workshops　with　variable　production　parameters.　To　address　this　issue,　a　new　model,　named　multi-objective　multi-tool　hole-making　sequence　optimization　with　precedence　constraints　(MO-MTpcHSO),　is　proposed　in　this　paper.　The　model　has　two　objectives:　spindle　travel　distance　and　tool　switching　time.　To　solve　MO-MTpcHSO,　a　customized　Q-learning　based　genetic　algorithm　(QLGA)　is　proposed.　The　adaptive　encoding　method　allows　chromosomes　to　express　feasible　solutions,　the　population　is　considered　as　the　agent,　and　the　states　are　intervals　of　the　diversity　coefficient.　Different　insertion　methods　in　the　crossover　operator　are　set　as　actions,　and　the　reward　is　related　to　the　diversity　and　values　of　objective　functions　of　the　population.　The　effectiveness　of　QLGA　is　validated　by　comparing　it　with　other　algorithms　in　practical　workpieces.　Moreover,　the　reasonability　of　actions　and　the　necessity　of　the　Q-learning　framework　in　QLGA　are　validated.