在交通碳达峰和碳中和的背景下，高精度、细粒度、可实施性强的机动车能耗实时预测方法成为交通减碳关键组成之一。针对传统基于回归的车辆能耗模型普适性较差的问题，提出了一种基于径向基函数神经网络（radial basis function neural network, RBFNN）的车辆能耗预测模型。首先分析车辆能耗影响因素并基于Min-Max标准化方法对影响因素矩阵进行归一化处理，然后基于灰狼算法（grey wolf optimization, GWO）优化RBFNN算法隐藏层中心点、高斯函数的宽度和隐含层与输出层连接的权值的训练，最后从横向模型对比和实车实测数据进行模型预测准确度分析。测试结果表明：RBFNN算法预测准确度较传统回归模型提高约12%，整体准确度达到90%以上，能够很好地对城市机动车能耗进行预测。

In the context of achieving carbon peak and carbon neutrality in transportation, high-precision, fine-grained, and highly feasible real-time prediction methods for motor vehicle energy consumption have become key components in reducing carbon emissions. Addressing the issue of limited universality in traditional regression-based vehicle energy consumption models, a prediction model based on the radial basis function neural network (RBFNN) has been developed. Firstly, the influencing factors of vehicle energy consumption were analyzed, and the influence factor matrix was normalized using the Min-Max standardization method. Then, the grey wolf optimization (GWO) algorithm was employed to optimize the training of the centers of the hidden layer, the width of the Gaussian function, and the weights connecting the hidden layer to the output layer in the RBFNN algorithm. Finally, a comprehensive analysis of the model's prediction accuracy was conducted through horizontal model comparisons and real-world vehicle measurements. The test results demonstrate that the RBFNN algorithm improves prediction accuracy by approximately 12% compared to traditional regression models, achieving an overall accuracy of over 90%. This makes it highly effective in accurately predicting the energy consumption of urban motor vehicles.