With the deep integration of distributed manufacturing and logistical transportation, how to collaboratively optimize multi-factory production scheduling and transport paths to reduce costs and improve efficiency has become a research hotspot. This paper first propose a reinforcement learning hyper-heuristic algorithm (LHHA) to solve the distributed production and transportation integrated scheduling problem with interval number transportation speed (DPTISP_INTS). The criterion is to minimize the total completion time. The proposed LHHA integrates a Q-learning framework with a dynamic probability distribution model, and constructs an adaptive knowledge learning library (KLL) to balance the exploration and exploitation in solution space. Meanwhile, the proposed algorithm also designs a four-stage encoding mechanism as well as several effective multi-strategy low-level heuristic rules (LLHs) and high-level heuristics to perform deep search for finding excellent solution or scheduling scheme. The findings of the experiment indicate that the LHHA demonstrates superior performance compared to conventional methods concerning both convergence speed and solution quality.