The dual probabilistic linguistic term set (DPLTS) is able to express the evaluation information of decision makers in both membership degree and non-membership degree, which is more effective in dealing with multi-attribute group decision-making (MAGDM) problems. First, to address the shortcomings of the current research on distance measures of DPLTSs, this paper proposes a new integrated distance measure that can accurately characterize the differences between DPLTSs without extending the number of probabilistic linguistic term elements. Second, the method of determining expert weights is given based on the similarity of evaluation and expert trustworthiness; then, the specific steps of reaching group consensus are constructed and the required decision matrix is obtained. Third, the attribute weights are calculated based on the deviation maximization method, and the MAGDM decision method based on regret theory is constructed. Finally, the numerical analysis of the heavy-duty truck research and development strategy selection for a new energy vehicle enterprise was carried out as an example to verify the applicability and effectiveness of the method; and the stability and rationality were further verified through sensitivity analysis and comparative analysis.