混合动力公交车能量控制策略的优化模型

针对混合动力公交车在循环工况内功率需求的特点，建立了未来功率需求贝叶斯预测模型；利用2-阶段随机动态规划模型将大规模的随机动态规划问题简化为多个小规模的随机动态规划问题和一个确定型动态规划问题；对于随机动态规划模型的求解，给出了稀疏表示的降维方法，将复杂的泛函极值问题转化为常规的随机动态优化问题，并采用分布估计算法和计算资源最优配置算法的计算机仿真优化算法对随机动态优化问题进行求解；给出了基于查表的在线控制策略，为模型的实际应用进行了有益的探索。

Power Allocation Strategy of Hybrid Electric Bus

The strategy of power allocation is one of the key techniques in the designing of a Hybrid Electrical Bus (HEB). In this article, a Bayesian forecasting model is firstly proposed to predict the future power demand based on the historical data of driving circles. Then a two-phase model is developed so that the original large-scale Stochastic Dynamic Program (SDP) can be decomposed to several small-scale SDPs and a Deterministic Dynamic Program (DDP). Since it is time-consuming to solve an SDP by the traditional iterative method, the sparse representation of an SDP solution and the Stochastic Simulation Optimization (SSO) are implemented to accelerate the SDP solver. Further, an Estimation of Distribution Algorithm (EDA) is applied to reduce the searching scope, and an Optimal Computing Budget Allocation (OCBA) method is also used for the purpose of improving the precision and decreasing the computational cost. Finally, a table-based strategy is given as the solution of the online power allocation problem.