摘要
为了适应智慧交通与智能网联汽车的快速发展,依据北京部分地图建立了智慧交通模拟数据集,结合因果膨胀卷积和残差连接的网络结构,设计并验证了新型基于时间卷积的数据驱动速度预测模型.为了提升燃料电池汽车的节能性,建立了等效氢耗的多目标优化函数及动力源寿命的约束条件.为了提高策略的实时性,建立了能量管理策略凸优化数学模型,使用了OSQP(operator splitting quadratic program)求解算法,在满足需求功率和计算实时性的前提下,通过合理分配动力源的输出功率,减少了动力源的寿命衰减.结果表明,提出的基于智慧交通的凸优化燃料电池汽车能量管理策略,相比于动态规划计算时间减少了90％以上,并保持等效氢耗基本一致.
Abstract
To suit the rapid development of intelligent transportation and internet connected vehicles,this paper builds an intelligent transportation simulation dataset based on some partial maps of Beijing. A new time-convolution-based data-driven speed prediction model is designed and validated by combining the network structure of causal inflationary convolution and residual connection. To improve the energy efficiency of fuel cell vehicles, the multi-objective optimization function of equivalent hydrogen consumption and the constraints of power source life are established. To boost the real-time performance of the strategy, a convex optimization mathematical model of the energy management strategy is built, and the OSQP ( operator splitting quadratic program) solution algorithm is employed, reducing the lifetime decay of the power source by properly allocating the output power of the power source under the premise of satisfying the demanded power and the computational real-time performance. Our results show the proposed convex-optimized fuel cell vehicle energy management strategy based on intelligent transportation reduces the computation time by over 90％ compared with the dynamic planning and keeps the equivalent hydrogen consumption basically at the same level.
国家科技期刊平台
NSTL
万方数据