The invention discloses a rule fusion deep reinforcement learning energy management method based on working condition identification. A plug-in hybrid electric vehicle is taken as an object, and a parallel structure is adopted as a connection mode of an engine and a motor to establish a hybrid power system model. Eight standard working conditions are selected to establish a working condition library, kinematic segmentation is carried out on the working condition library, vehicle working conditions are classified and identified by comparing nine representative parameters according to segmented kinematic segments, and then states, actions, agents and punishment functions in a deep Q learning algorithm are designed. Then the designed rule-fused deep reinforcement learning algorithm is trained and distributed under three different training conditions, so that the purpose of efficient distribution and utilization of energy is achieved, fewer poor samples exist in the training process of the algorithm, the training efficiency is high, and the comprehensive performance of the hybrid power system is high.
本发明公开了一种基于工况辨识的规则融合深度强化学习能量管理方法。本发明以插电式混合动力汽车为对象建立,采用并联式结构作为发动机和电动机的连接方式建立混合动力系统模型,通过选取了8种标准工况建立了工况库,并对其进行了运动学分割,根据所分割的运动学片段对比9个代表性的参数对车辆工况进行分类与辨识,然后对深度Q学习算法中状态、动作、agent和惩罚函数进行设计,之后在三种不同训练工况下对所设计的规则融合的深度强化学习算法进行训练、分配,从而达到能量高效分配和利用的目的,并且算法在训练过程中存在较少的较差样本,训练效率高,混合动力系统综合性能高。
Rule fusion deep reinforcement learning energy management method based on working condition identification
一种基于工况辨识的规则融合深度强化学习能量管理方法
2021-11-30
Patent
Elektronische Ressource
Chinesisch
IPC: | B60W CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION , Gemeinsame Steuerung oder Regelung von Fahrzeug-Unteraggregaten verschiedenen Typs oder verschiedener Funktion |
Europäisches Patentamt | 2024
|Automatic driving lane changing decision control method based on rule fusion reinforcement learning
Europäisches Patentamt | 2022
|HEV energy management method based on deep reinforcement learning A3C algorithm
Europäisches Patentamt | 2020
|