纯电动汽车剩余续驶里程估算研究

发布时间：2019-01-11 09:35

【摘要】：目前纯电动汽车剩余续驶里程估算方法较为简单,其估算结果与真实行驶里程存在较大差距,这样直接造成用户时刻担心现有电量不能支持车辆到达预期地点,从而导致“里程焦虑”,降低了用户购买电动汽车的信心。因此,为了提高电动汽车普及率及使用方便性,在一定程度上提高剩余续驶里程估算精度是解决该问题重要方法之一,也是本课题研究的目的,其核心内容如下:(1)基于MATLAB/Simulink建立整车动力学、驾驶员、电机、电池以及整车能耗模型,该模型的建立是为后续章节对剩余续驶里程估算模型仿真作铺垫。(2)详细研究了三种剩余续驶里程估算方法:首先将传统燃油车估算剩余续驶里程方法应用在纯电动汽车上,即平均能耗法,该方法靠获取电池电压、电流来进行能耗计算,比较粗糙;然后引出工况识别法,该方法根据当前行驶工况自动判别当前能耗,且在传统工况识别方法基础上新增下述三点:(1)建立工况特征参数与能耗之间模糊规则库,实现能耗预测。(2)单位能耗行驶里程优化,根据实际经验建立单位能耗与剩余能耗之间的线性关系,使其剩余里程呈递减优化趋势。(3)对输出的剩余续驶里程进行卡尔曼滤波,使其更加真实;最后在工况识别基础上新增工况预测法,该方法将BP神经网络和马尔科夫结合在一起,实现在一定时间范围的工况预测,从而再次提高剩余续驶里程估算精度。(3)因空调对剩余续驶里程影响较大,针对空调的开启与关闭又单独采用了另一种平均能耗估算方案:基于空调功率及开启时间的平均能耗法。(4)为了能详细对比三种方案的效果,分别从已行驶里程和能量消耗两方面进行剩余续驶里程优劣评估,仿真结果验证了工况预测方法的优势。(5)实验论证部分采用快速原型开发平台,首先基于D2P-Motohawk及HIL平台搭建硬件在环仿真测试环境,论证所提出算法模型的可行性及实时性;然后在转鼓实验台进行循环跟踪NEDC工况测试剩余续驶里程;最后通过已行驶里程和能量消耗对比三种方案的优劣性,实验结果表明将工况识别与预测相结合的方法满足实际控制需求,并提高剩余续驶里程估算精度。
[Abstract]:At present, the method of estimating the remaining driving range of pure electric vehicle is relatively simple, and there is a big gap between the estimation result and the real driving mileage, which directly causes the user to worry that the existing electric quantity can not support the vehicle to reach the expected location. This led to "mileage anxiety", reducing the confidence of users to buy electric cars. Therefore, in order to improve the popularity and ease of use of electric vehicles, to a certain extent, improving the estimation accuracy of the remaining driving range is one of the important methods to solve this problem, and is also the purpose of this research. The main contents are as follows: (1) based on MATLAB/Simulink, the vehicle dynamics, driver, motor, battery and vehicle energy consumption model are established. The establishment of the model is to pave the way for the simulation of the residual range estimation model in the following chapters. (2) three methods for estimating the residual continuous mileage are studied in detail: firstly, the traditional fuel vehicle is applied to estimate the residual continued mileage. In pure electric cars, That is, the average energy consumption method, which obtains the battery voltage and current to calculate the energy consumption, is relatively rough; Then the working condition identification method is introduced. The method automatically discriminates the current energy consumption according to the current driving condition, and adds the following three points on the basis of the traditional working condition identification method: (1) establishing the fuzzy rule base between the operating condition characteristic parameters and the energy consumption, (2) Optimization of driving mileage per unit energy consumption. The linear relationship between unit energy consumption and residual energy consumption is established according to practical experience. The residual mileage is reduced and optimized. (3) Kalman filter is applied to the output residual driving mileage to make it more real; Finally, a new condition prediction method is added on the basis of condition recognition. The method combines BP neural network with Markov to realize the condition prediction in a certain time range. Therefore, the estimation accuracy of residual range is improved again. (3) because the air conditioning has a great influence on the residual range, In order to compare the effects of the three schemes in detail, another method of estimating the average energy consumption is used separately: the average energy consumption method based on the air conditioning power and the opening time. (4) in order to compare the effects of the three schemes in detail, The advantages and disadvantages of the remaining driving range are evaluated from the aspects of driving mileage and energy consumption respectively. The simulation results verify the advantages of the working condition prediction method. (5) in the experimental demonstration part, the rapid prototyping development platform is used. Firstly, based on D2P-Motohawk and HIL platform, the hardware in loop test environment is built to demonstrate the feasibility and real-time performance of the proposed algorithm model. Then the remaining driving mileage is measured on the rotary drum test bench by circulating tracking the NEDC working condition. Finally, by comparing the advantages and disadvantages of the three schemes, the experimental results show that the combined method of condition identification and prediction can meet the actual control requirements and improve the estimation accuracy of the remaining driving range.
【学位授予单位】：江苏大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：U469.72

【相似文献】