Aiming at reducing logistics cost and carbon emission, a low-carbon location-routing problem considering simultaneous pickup and delivery and heterogeneous fleet(LCLRPSPDHF) is proposed, and a three-index exponential-size MIP model is defined. Aiming at this project, an evolutionary-heuristic(HH) algorithm is developed by utilizing evolutionary mechanisms as high level learning strategies to improve the performance of hyper-heuristic framework to monitor the performance information of low-level heuristics(LLH) timely and rapidly, and judge for choosing the most appropriate heuristic rightly, including quantum-inspired selection(QS), ant-based selection(AS), shuffled frog selectionn(LS) and nature-competition selection(NCS). Meanwhile, two adaptive acceptance criteria are developed by mining information of LLHs for realizing global search and improving convergence. Simulation results and comparisons show that the proposed algorithms are effective and robust, providing high quality solution for different scales instances within reasonable computing time.