基于工况识别与预测的纯电动汽车剩余里程估算研究

Estimation of Remaining Mileage for Pure Electric Vehicle Based on Condition Identification and Prediction

为了提升纯电动汽车剩余里程估算方法的准确度,在汽车行驶工况识别与预测的基础上,提出了一种新的剩余里程估算模型。通过采集实际汽车行驶工况数据,利用模糊聚类等方法对工况进行状态识别和分析,并建立了车辆能耗与工况特征参数之间的模糊规则库。同时,应用隐马尔可夫模型对行驶工况进行预测,通过将工况识别与工况预测相结合的方法,构建了基于工况识别与预测的纯电动汽车剩余里程估算方法。在AVL CRUISE中对纯电动汽车进行整车剩余里程仿真,选取了综合CLTC与WLTC的混合工况,以更贴近实际汽车行驶情况,详细比较了工况识别和工况识别与预测两种估算方法。研究结果表明,对于基于工况识别的剩余里程方法,其估算值随着时间的增加与仿真值间的误差逐渐增大,而工况预测方法的引入能有效减小误差。通过对比发现,工况识别与预测方法的最大绝对误差和绝对误差平均值相比仅采用工况识别的方法分别降低了34.04%和55.79%,标准偏差也减小至1.44 km,表明其具有更好的估算精度,为纯电动汽车续驶里程的预测提供了一种新途径。

To enhance the accuracy of remaining mileage estimation method for pure electric vehicles,a novel model centered on the identification and prediction of vehicle driving conditions was put forward.By gathering real-world vehicle driving condition data,the techniques such as fuzzy clustering were used to discern and analyze condition states.Furthermore,a fuzzy rule base correlating vehicle energy consumption with condition-specific parameters was established.Concurrently,the Hidden Markov model was employed to forecast driving conditions.By combining condition identification and prediction,a methodology for estimating the remaining mileage of pure electric vehicles was developed.During the simulation of entire vehicle remaining mileage in AVL CRUISE,the hybrid working conditions which integrate CLTC and WLTC were used to closely mimic real-world driving conditions.Two estimation methods of condition identification alone and condition identification with prediction were compared in detail.The results reveal that the estimation error increases gradually with the increase of time based on the identification of working conditions,and decreases effectively after the introduction of working condition prediction method.Specifically,the maximum absolute error and mean absolute error for the condition identification and prediction methods reduce by 34.04%and 55.79%respectively.Moreover,the standard deviation decreases to 1.44 km.These results prove the superior accuracy of the proposed method,which presents a new perspective on forecasting the range of pure electric vehicles.