<p>Precise prediction of energy consumption offers essential data support for energy supplementation, economic and social development. Existence of nonlinear and time-varying characteristics in energy consumption poses significant challenges to its precise prediction. While previous study have explored these aspects, most focus on them individually, with limited attention to their integrated consideration. In view of this, an enhanced multivariable grey model is proposed to comprehensively account for both aspects. Additionally, the cross-validation method is implemented to enhance the novel model’s generalization capability and prevent overfitting. Furthermore, the particle swarm optimization algorithm is utilized to identify the optimal solutions for the proposed grey model. To validate the effectiveness of the proposed grey model, the proposed grey model is applied to predict the total energy consumption in China. The results indicate that, compared to the traditional multivariable grey model, the mean absolute percentage error of the proposed model improves by 53.89%. This demonstrates the superior performance of the novel grey model in predicting total energy consumption. Additionally, the total energy consumption is projected to reach 57.96 million tons of coal equivalent by 2025, representing an increase of 2.2 times compared to 2022. These findings demonstrate that the novel grey model proposed in this study provides an effective approach for energy consumption prediction and supplies robust data support for policy-making.</p>