This study proposes a visual evaluation method based on virtual reality technology to address the problems of complex information and differences in evaluators’ cognition in the simulation evaluation of high-speed train bogies. Firstly, generate a dataset through Latin hypercube sampling and SIMPACK batch processing simulation. Based on the RBF neural network model, a proxy model is constructed to achieve rapid prediction of the dynamic performance of high-speed train bogies. Secondly, create a virtual operation scene for high-speed trains on the Unity3D development platform, achieve real-time interaction between dynamic data and the virtual scene through TCP protocol, and construct a multidimensional visual interface to display key performance indicators of the bogie. Then, the 2-tuple linguistic representation model and evaluation conflict resolution mechanism are introduced to quantify the conflict level of evaluators’ evaluation results for different schemes, and a conflict threshold determination method is proposed to coordinate the preferences of evaluators’ groups to achieve consensus. Finally, taking three sets of bogie design schemes from a certain host factory as examples, the feasibility of this method is verified. The experimental results show that scheme one has the best comprehensive performance, with consistent evaluation results and low evaluation conflict values, and the average evaluation time has decreased by 31.5%. This study provides an effective and intuitive visual evaluation method for the design and evaluation of rail transit equipment, promoting the application of virtual reality technology in product design.