The increasingly serious energy and environmental problems have forced the power industry to transform to low-carbon, and power grid companies are facing major challenges. Due to the characteristics of power transmission projects, such as long line span, wide influence area, and difficult vegetation protection. It is necessary to optimize the line, improve energy efficiency, and strengthen ecological protection in design. Therefore, this paper proposes a low-carbon transmission engineering design method based on the Q-learning algorithm to assist route planning decision-making. To improve the low carbon of the scheme, this study improves the Q-learning reward function and introduces the route weight setting and corner incentive setting. The route weight is graded according to topographic, geological, humanistic, and ecological factors to avoid ecologically sensitive areas, bad geological areas and densely populated areas and achieve ecological protection. The corner incentive reduces line angle changes, reduces project costs and material use, and reduces carbon emissions. Practical examples show that Q-learning can not only converge in given rounds, but also output the optimal and sub-optimal solutions with reference value. Compared with the manual decision-making scheme, the route length is shorter, the construction cost is lower, and it is more in line with the needs of low-carbon power grid construction.