<p>In this paper, a novel approach utilizing a double-layer deep neural network (DNN) model optimized by the red fox optimizer (RFO) to forecast the transportation energy demand of Jordan is presented. To address the early convergence limitations, the Levy flight method and a circular mapping-based chaos mechanism are implemented. The model is developed using a historical energy demand dataset over the period 1970–2022 and it comprises a main model and seven sub-models. The dataset is using socio-economic and transportation indicators based on seven key features: vehicles number, population, diesel price, gasoline price, gross national income, ownership level, and GDP in Jordan. The model was trained and validated and has shown very accurate results with an absolute percentage error of 0.26%. The <i>R</i>² was 0.999 for both training and validation stages. During testing on the unseen dataset, the model achieved an <i>R</i>² value of 0.998. The results show that the transportation energy demand has an upward trend from 4278 ktoe in 2023 to 5219 ktoe in 2047 with a compound annual growth rate of 21.98%. In addition, it is expected that the applicable results of this study are anticipated to offer valuable insights for Jordan’s planners to develop appropriate transport energy policies and legislation to ensure a secure future for this sector.</p>