CN109747654B - A working condition-oriented method for calibration of hybrid electric vehicle control parameters - Google Patents
A working condition-oriented method for calibration of hybrid electric vehicle control parameters Download PDFInfo
- Publication number
- CN109747654B CN109747654B CN201910025948.5A CN201910025948A CN109747654B CN 109747654 B CN109747654 B CN 109747654B CN 201910025948 A CN201910025948 A CN 201910025948A CN 109747654 B CN109747654 B CN 109747654B
- Authority
- CN
- China
- Prior art keywords
- working condition
- indicators
- control parameters
- characteristic
- optimal control
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 29
- 238000012417 linear regression Methods 0.000 claims abstract description 34
- 238000005457 optimization Methods 0.000 claims abstract description 25
- 238000004458 analytical method Methods 0.000 claims abstract description 15
- 239000002245 particle Substances 0.000 claims abstract description 13
- 238000012216 screening Methods 0.000 claims abstract description 9
- 238000012360 testing method Methods 0.000 claims description 25
- 239000000446 fuel Substances 0.000 claims description 19
- 238000004422 calculation algorithm Methods 0.000 claims description 14
- 230000001133 acceleration Effects 0.000 claims description 12
- 230000000717 retained effect Effects 0.000 claims description 6
- 230000002159 abnormal effect Effects 0.000 claims description 5
- 238000007726 management method Methods 0.000 description 5
- 238000011217 control strategy Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 125000004122 cyclic group Chemical group 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000002068 genetic effect Effects 0.000 description 2
- 238000007619 statistical method Methods 0.000 description 2
- 238000010219 correlation analysis Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000009396 hybridization Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Landscapes
- Electric Propulsion And Braking For Vehicles (AREA)
Abstract
本发明公开了一种面向工况的混合动力汽车控制参数标定方法，涉及混合动力汽车技术领域。方法主要包括建立工况样本、基于粒子群算法对各独立工况下的控制参数优化、基于相关性的工况特征指标筛选、多元线性回归分析、新工况的最佳控制参数标定等五个步骤。充分考虑工况特征与最佳控制参数之间的关系，建立最佳控制参数与工况特征指标之间的多元线性回归模型，对于不同工况都能迅速标定控制参数，一方面有助于理解工况对最佳控制参数的影响，一方面便于标定人员快速确定最佳控制参数，缩短标定周期。
The invention discloses a working condition-oriented method for calibrating control parameters of a hybrid electric vehicle and relates to the technical field of hybrid electric vehicles. The method mainly includes the establishment of working condition samples, the optimization of control parameters under each independent working condition based on particle swarm optimization, the screening of working condition characteristic indicators based on correlation, multiple linear regression analysis, and the calibration of optimal control parameters for new working conditions. step. Fully consider the relationship between the working condition characteristics and the optimal control parameters, establish a multiple linear regression model between the optimal control parameters and the working condition characteristic indicators, and quickly calibrate the control parameters for different working conditions. On the one hand, it is helpful to understand The influence of working conditions on the optimal control parameters, on the one hand, facilitates the calibration personnel to quickly determine the optimal control parameters and shorten the calibration cycle.
Description
技术领域technical field
本发明属于混合动力汽车技术领域，特别涉及一种混合动力汽车控制参数标定方法。The invention belongs to the technical field of hybrid electric vehicles, in particular to a method for calibrating control parameters of hybrid electric vehicles.
背景技术Background technique
节能是汽车混合动力化的主要目标之一，混合动力汽车由于涉及两个及以上的动力源，耦合关系复杂，能量管理控制策略(下称控制策略)及其中的关键控制参数对油耗有着重要影响，因此衍生出一系列以最小油耗为目标的控制参数优化方法，如动态规划算法、最小等效燃油消耗法、遗传算法、粒子群算法等。其中动态规划算法可以达到某特定工况下的经济性最优，但是需要进行复杂的控制规则提取才能应用到实际控制中，规则提取的精确程度也影响着实际油耗；最小等效燃油消耗法以瞬时燃油经济性最优为目标，无法达到整个工况的经济性最优；遗传算法、粒子群算法等智能优化算法既可以应用于全局最优，又可以应用于瞬时最优，且是对控制参数的直接优化，可以得到最佳控制参数的具体数值。Energy saving is one of the main goals of vehicle hybridization. Since hybrid vehicles involve two or more power sources, the coupling relationship is complex, and the energy management control strategy (hereinafter referred to as the control strategy) and its key control parameters have an important impact on fuel consumption. , so a series of control parameter optimization methods aiming at minimum fuel consumption are derived, such as dynamic programming algorithm, minimum equivalent fuel consumption method, genetic algorithm, particle swarm algorithm, etc. Among them, the dynamic programming algorithm can achieve the optimal economy under a specific working condition, but it needs complex control rule extraction before it can be applied to actual control, and the accuracy of rule extraction also affects the actual fuel consumption; the minimum equivalent fuel consumption method is based on The goal is to optimize the instantaneous fuel economy, and it is impossible to achieve the optimal economy of the entire working condition; intelligent optimization algorithms such as genetic algorithm and particle swarm optimization algorithm can be applied to both the global optimum and the instantaneous optimum. The direct optimization of parameters can obtain the specific values of the best control parameters.
以上对控制参数的优化均基于特定工况，即优化后得到的参数并不能使汽车在所有行驶工况下都有最佳的燃油经济性，工况不同，控制参数需要作出相应调整。对于标定人员来说，每个工况下都对控制参数进行优化会花费大量时间，延长开发周期；此外，汽车实际行驶工况与离线优化时使用的工况往往存在较大差别且经常变化，而控制器中的控制参数不能在线标定，导致混合动力汽车的燃油经济性不能充分发挥，目前常见的方法是先在大量工况下对控制参数进行离线优化，得到每个工况对应的最佳控制参数，然后进行工况识别和预测，再通过查表的方式在线标定控制参数。该方法的有效范围依赖于离线优化时的工况数量，当汽车行驶工况不在表格之中时，该方法失效，而增加离线工况数量会大大增加时间成本。因此，寻找工况特征与最佳控制参数之间的关系对于提高控制策略的离线标定效率和在线标定有效性具有重要意义。The above optimization of control parameters is based on specific working conditions, that is, the parameters obtained after optimization cannot make the car have the best fuel economy in all driving conditions, and the control parameters need to be adjusted accordingly for different working conditions. For calibration personnel, optimizing the control parameters under each working condition will take a lot of time and prolong the development cycle; in addition, there are often large differences and frequent changes between the actual driving conditions of the car and the working conditions used in offline optimization. However, the control parameters in the controller cannot be calibrated online, resulting in the fuel economy of the hybrid electric vehicle not being fully utilized. At present, the common method is to first optimize the control parameters offline under a large number of working conditions to obtain the best value corresponding to each working condition. Control parameters, then identify and predict the working conditions, and then calibrate the control parameters online by means of table lookup. The effective range of this method depends on the number of working conditions during offline optimization. When the driving conditions of the car are not in the table, the method will fail, and increasing the number of offline working conditions will greatly increase the time cost. Therefore, finding the relationship between the operating condition characteristics and the optimal control parameters is of great significance for improving the offline calibration efficiency and online calibration effectiveness of the control strategy.
中国专利公开号为CN104071161A，公开日为2014-10-01，公开了一种插电式混合动力汽车工况识别及能量管控的方法，首先使用支持向量机识别工况，将工况划分为若干特定种类，然后在不同种类工况下使用不同的模糊方法控制发动机扭矩，从而优化燃油经济性，该方法仅将工况分为有限的若干类，无法详细体现工况特征，模糊方法模拟人的判断，相当于根据经验制定好了工况特征与控制参数之间的关系，优化效果有限；中国专利公开号为CN102717797A，公开日为2012-10-10，公开了一种混合动力车辆能量管理方法及能量管理系统，该方法以燃油消耗、发动机排放、电池SOC为代价函数，以电机输出转矩为标定量，使用随机动态规划方法解决能量管理问题，与本专利优化方法不同，且并未探究最佳控制参数与工况特征的关系。The Chinese patent publication number is CN104071161A, and the publication date is 2014-10-01, which discloses a method for identification and energy management of plug-in hybrid electric vehicles. First, a support vector machine is used to identify the operating conditions, and the operating conditions are divided into several Specific types, and then use different fuzzy methods to control the engine torque under different types of working conditions, so as to optimize fuel economy. This method only divides the working conditions into a limited number of categories, and cannot reflect the characteristics of the working conditions in detail. The fuzzy method simulates human behavior. Judgment is equivalent to formulating the relationship between working condition characteristics and control parameters based on experience, and the optimization effect is limited; the Chinese patent publication number is CN102717797A, and the publication date is 2012-10-10, which discloses a hybrid vehicle energy management method And energy management system, this method uses fuel consumption, engine emission, battery SOC as the cost function, and the output torque of the motor as the calibration value, using the stochastic dynamic programming method to solve the energy management problem, which is different from the optimization method of this patent, and does not explore The relationship between optimal control parameters and operating condition characteristics.
发明内容Contents of the invention
为克服现有技术存在的不足，本发明提供一种面向工况的混合动力汽车控制参数标定方法，探究混合动力汽车行驶工况与使燃油经济性最佳的控制参数之间的关系及规律。在不同工况下，混合动力汽车达到最佳经济性的控制参数不同，说明最佳控制参数与工况的某些特征之间必然存在线性或非线性关系，而线性关系更容易被总结和使用。以动力学片段作为独立工况来扩充工况样本，用粒子群算法分别计算出各工况下的最佳控制参数，然后使用统计学方法分析与最佳控制参数相关性最高的工况特征指标以及这些指标与最佳控制参数之间的线性关系，并建立多元线性回归模型，根据回归模型，在得到新工况的相关特征指标后，即可计算在该工况下使汽车燃油经济性最佳的控制参数，完成控制参数的标定。In order to overcome the deficiencies in the prior art, the present invention provides a working condition-oriented method for calibrating control parameters of a hybrid electric vehicle, and explores the relationship and rules between the driving conditions of a hybrid electric vehicle and the control parameters for optimal fuel economy. Under different working conditions, the control parameters of hybrid electric vehicles to achieve the best economy are different, indicating that there must be a linear or nonlinear relationship between the optimal control parameters and certain characteristics of the working conditions, and the linear relationship is easier to summarize and use . Use the dynamic segment as an independent working condition to expand the working condition sample, use the particle swarm algorithm to calculate the optimal control parameters under each working condition, and then use statistical methods to analyze the working condition characteristic indicators with the highest correlation with the optimal control parameters And the linear relationship between these indicators and the optimal control parameters, and establish a multiple linear regression model. According to the regression model, after obtaining the relevant characteristic indicators of the new working condition, you can calculate the optimum fuel economy of the vehicle under this working condition. The optimal control parameters are used to complete the calibration of the control parameters.
为实现上述目的，根据本发明实施例的一种面向工况的混合动力汽车控制参数标定方法，包括以下内容：In order to achieve the above purpose, a method for calibrating control parameters of a hybrid electric vehicle oriented to working conditions according to an embodiment of the present invention includes the following contents:
第一，建立工况样本，具体包括以下步骤：First, establish a working condition sample, which specifically includes the following steps:
①首先选定若干个标准循环循环工况，然后提取每个标准循环工况的所有动力学片段，将每一动力学片段看作一个独立的工况，所述动力学片段的划分方法为以工况起始点作为第一个动力学片段的起始点，以起始点后车速经历大于零后的第一个零车速点作为第一个动力学片段的结束点，以上一动力学片段的结束点作为下一动力学片段的起始点，以此类推，若最后一个动力学片段起始点后车速始终为零，则舍弃该段，所有循环工况的动力学片段数之和即为样本数，计为N；① First select several standard cyclic working conditions, then extract all the dynamic segments of each standard cyclic working condition, and regard each dynamic segment as an independent working condition, and the division method of the dynamic segments is as follows: The starting point of the working condition is taken as the starting point of the first dynamic segment, and the first zero speed point after the starting point after the vehicle speed experience is greater than zero is used as the end point of the first dynamic segment, and the end point of the previous dynamic segment As the starting point of the next dynamic segment, and so on, if the vehicle speed is always zero after the starting point of the last dynamic segment, discard this segment, and the sum of the dynamic segment numbers of all cycle conditions is the number of samples. for N;
②计算各独立工况的20个特征指标C1,C2,C3......C20，所述20个特征指标依次指停车比例、停车次数、加速比例、减速比例、巡航比例、平均加速度、最大加速度、加速度标准差、平均减速度、最大减速度、减速度标准差、平均行驶车速、车速标准差、车速均方根、最高车速、0-20km/h车速比例、20-40km/h车速比例、40-60km/h车速比例、60-80km/h车速比例、80-100km/h车速比例，每个特征指标Cx为一元素数为N的向量，x为下角标，代表1,2,3......20；②Calculate 20 characteristic indexes C 1 , C 2 , C 3 ...... C 20 of each independent working condition. The 20 characteristic indexes refer to the stop ratio, stop times, acceleration ratio, deceleration ratio, and cruise ratio in sequence , average acceleration, maximum acceleration, standard deviation of acceleration, average deceleration, maximum deceleration, standard deviation of deceleration, average driving speed, standard deviation of vehicle speed, root mean square of vehicle speed, maximum speed, 0-20km/h speed ratio, 20- 40km/h vehicle speed ratio, 40-60km/h vehicle speed ratio, 60-80km/h vehicle speed ratio, 80-100km/h vehicle speed ratio, each characteristic index C x is a vector with a number of elements N, x is the subscript, Represents 1,2,3...20;
第二，各独立工况的控制参数优化，具体包括以下步骤：Second, the control parameter optimization of each independent working condition includes the following steps:
①确定要优化的控制参数P；① Determine the control parameter P to be optimized;
②使用粒子群算法在每一独立工况下对P进行优化，以燃油消耗量最小为适应度函数，以电池SOC平衡为约束条件，以P作为粒子位置，设定迭代次数K作为优化终止条件，最终得到各独立工况下使燃油消耗量最小的最佳控制参数Pb，Pb为一元素为N的向量；②Use the particle swarm optimization algorithm to optimize P under each independent working condition, take the minimum fuel consumption as the fitness function, take the battery SOC balance as the constraint condition, take P as the particle position, and set the number of iterations K as the optimization termination condition , and finally obtain the optimal control parameter P b that minimizes fuel consumption under each independent working condition, where P b is a vector whose element is N;
第三，基于相关性的工况特征指标筛选，具体包括以下步骤：The third is the correlation-based screening of working condition characteristic indicators, which specifically includes the following steps:
①使用线性回归分析中的相关性公式计算各特征指标之间的相关系数如式(1)所示，各特征指标与最佳控制参数之间的相关系数如式(2)所示：① Use the correlation formula in linear regression analysis to calculate the correlation coefficient between each characteristic index As shown in formula (1), the correlation coefficient between each characteristic index and the optimal control parameter As shown in formula (2):
式中下角标x、y均表示特征指标代号，为1,2,3......20；In the formula, the subscripts x and y both represent the characteristic index codes, which are 1, 2, 3...20;
②筛选出与最佳控制参数相关性较高的工况特征指标，设置阈值R1，保留所有的特征指标，其余的剔除；② Screen out the characteristic indicators of working conditions with high correlation with the optimal control parameters, set the threshold R 1 , and keep all The characteristic indicators of , and the rest are eliminated;
③依据各工况特征指标之间的相关性继续缩减特征指标数量，设置阈值R2，在步骤②中保留的工况特征指标中，寻找所有的特征指标对Cx、Cy，视为重复的工况特征指标，保留与最佳控制参数相关性较高的，即与中的较大值对应的工况特征指标，剔除较小值对应的工况特征指标；③ Continue to reduce the number of feature indicators according to the correlation between the feature indicators of each working condition, set the threshold R 2 , and find all the working condition characteristic indicators retained in step ② The characteristic indexes of C x and C y are regarded as repeated working condition characteristic indexes, and the ones with higher correlation with the optimal control parameters are reserved, namely and The operating condition characteristic index corresponding to the larger value in , and the operating condition characteristic index corresponding to the smaller value is eliminated;
经过以上所述筛选步骤后保留下来的工况特征指标个数记为M，M≤20；The number of working condition characteristic indicators retained after the above screening steps is recorded as M, and M≤20;
第四，多元线性回归分析，具体包括以下步骤：Fourth, multiple linear regression analysis, specifically including the following steps:
①设置模型的显著性水平α，建立最佳控制参数Pb与M个工况特征指标之间的多元线性回归模型，模型如式(3)所示：① Set the significance level α of the model, and establish a multiple linear regression model between the optimal control parameter P b and M working condition characteristic indicators, the model is shown in formula (3):
Pb＝β0+β1·C1+β2·C2+......+βM·CM+ε (3)P b =β 0 +β 1 ·C 1 +β 2 ·C 2 +......+β M ·C M +ε (3)
式中，β0为常数项，β1、β2……βM为回归系数，ε为随机误差；In the formula, β 0 is a constant item, β 1 , β 2 ... β M is a regression coefficient, ε is a random error;
②残差分析，根据所述多元线性回归模型得到各样本的残差r1、r2……rN及对应的置信区间rint1、rint2……rintN，若某样本的置信区间不包含零点，则认为该样本数据异常，将其剔除，然后重新建立式(3)所示多元线性回归模型；② Residual error analysis, according to the multiple linear regression model to obtain the residual r 1 , r 2 ... r N of each sample and the corresponding confidence interval rint 1 , rint 2 ... rint N , if the confidence interval of a certain sample does not contain zero, the sample data is considered to be abnormal, and it is eliminated, and then the multiple linear regression model shown in formula (3) is re-established;
③对模型开展T检验，设进入该步骤时的工况特征指标个数为Me，先更新Me，若存在未通过T检验的回归系数βw，则剔除下标号最小的βw对应的Cw并进入步骤④，此时剩余工况指标个数为Mo，Mo＝Me-1；若不存在未通过T检验的回归系数，则剩余工况指标个数为Me，进入步骤⑤；③ Carry out T-test on the model, set the number of working condition characteristic indicators when entering this step as M e , first update M e , if there is a regression coefficient β w that fails the T-test, remove the one corresponding to β w with the smallest subscript C w and enter step ④, at this time, the number of remaining working condition indicators is M o , M o = M e -1; if there is no regression coefficient that fails the T test, the number of remaining working condition indicators is M e , enter Step ⑤;
④保持显著性水平α不变，建立最佳控制参数Pb与Mo个工况特征指标之间的多元线性回归模型并进行F检验，若F检验通过，则进入步骤③；若F检验未通过，则还原最新被剔除的一个工况特征指标并进入步骤⑤，此时剩余工况指标个数为Me；④Keep the significance level α unchanged, establish a multiple linear regression model between the optimal control parameter P b and M o working condition characteristic indicators, and perform the F test. If the F test passes, proceed to step ③; if the F test fails If it is passed, restore the newly eliminated feature index of a working condition and enter step ⑤, at this time, the number of remaining working condition indicators is M e ;
⑤以最终保留的Me个工况特征指标与Pb之间的多元线性回归模型作为最终模型，如式(4)所示：⑤The final model is the multiple linear regression model between the remaining M e operating condition characteristic indicators and P b as the final model, as shown in formula (4):
第五，对于要标定的新工况，计算所述最终保留的Me个工况特征指标，并将其代入式(4)，得到该工况对应的最佳控制参数。Fifth, for the new working condition to be calibrated, calculate the finally reserved M e working condition characteristic indicators and substitute them into formula (4) to obtain the optimal control parameters corresponding to the working condition.
本发明与现有技术相比，充分考虑工况特征与最佳控制参数之间的关系，对于不同工况都能迅速标定控制参数，优化汽车的燃油经济性。用动力学片段划分工况，将每一动力学片段看作一独立工况，扩充了样本数量，提高后续使用统计学方法时的精度；使用粒子群算法优化每个独立工况下的控制参数，优化效果较好且可直接得到最佳控制参数的具体数值；对工况特征指标与最佳控制参数以及各指标之间的相关性分析可以筛选出对最佳控制参数影响最大的指标，减少后续多元线性回归模型的自变量个数，简化计算；对建立的多元线性回归模型开展T检测有助于进一步缩减工况特征指标个数，提高模型精度。本发明一方面有助于理解工况对最佳控制参数的影响，一方面便于标定人员快速确定最佳控制参数，缩短标定周期。Compared with the prior art, the present invention fully considers the relationship between working condition characteristics and optimal control parameters, can quickly calibrate the control parameters for different working conditions, and optimizes the fuel economy of the vehicle. The working conditions are divided by dynamic segments, and each dynamic segment is regarded as an independent working condition, which expands the number of samples and improves the accuracy of subsequent statistical methods; the particle swarm algorithm is used to optimize the control parameters of each independent working condition , the optimization effect is better and the specific value of the optimal control parameter can be directly obtained; the correlation analysis between the working condition characteristic index and the optimal control parameter and each index can filter out the index that has the greatest influence on the optimal control parameter, reducing The number of independent variables in the subsequent multiple linear regression model simplifies the calculation; carrying out T-test on the established multiple linear regression model will help to further reduce the number of working condition characteristic indicators and improve the accuracy of the model. On the one hand, the invention helps to understand the influence of working conditions on the optimal control parameters, on the other hand, it facilitates the calibration personnel to quickly determine the optimal control parameters, and shortens the calibration period.
附图说明Description of drawings
本发明的上述和/或附加的方面和优点结合下面附图对实施例的描述中将变得明显和容易理解，其中：The above and/or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, wherein:
图1位根据本发明实施例的总体流程图；Fig. 1 is an overall flow chart according to an embodiment of the present invention;
图2为根据本发明实施例的按动力学片段划分工况示意图；Fig. 2 is a schematic diagram of working conditions divided by dynamic segments according to an embodiment of the present invention;
图3为根据本发明实施例的工况特征指标矩阵示意图；Fig. 3 is a schematic diagram of a working condition feature index matrix according to an embodiment of the present invention;
图4为根据本发明实施例的工况特征指标与最佳控制参数及工况特征指标之间相关系数表格示意图；Fig. 4 is a schematic diagram of a correlation coefficient table between working condition characteristic indexes and optimal control parameters and working condition characteristic indexes according to an embodiment of the present invention;
图5为根据本发明实施例的多元线性回归分析流程图。Fig. 5 is a flowchart of multiple linear regression analysis according to an embodiment of the present invention.
具体实施方式Detailed ways
下面通过参考附图描述的实施例是示例性的，仅用于解释本发明，而不能理解为对本发明的限制。The embodiments described below by referring to the figures are exemplary only for explaining the present invention and should not be construed as limiting the present invention.
由于20个工况特征指标的计算方法，粒子群优化算法，多元线性回归中的残差分析、F检验、T检验等为通用方法，因此在此不再赘述。Since the calculation method of the 20 operating condition characteristic indicators, the particle swarm optimization algorithm, the residual analysis in the multiple linear regression, the F test, the T test, etc. are common methods, so they will not be repeated here.
本发明所述的一种面向工况的混合动力汽车控制参数标定方法，包括以下内容：A method for calibrating control parameters of a working condition-oriented hybrid electric vehicle according to the present invention comprises the following contents:
第一，建立工况样本。由于本发明所述的一种面向工况的混合动力汽车控制参数标定方法涉及统计学中的多元线性回归分析，必须保证样本量充足，因此以划分动力学片段的方法扩充样本容量，工况样本的建立具体包括以下步骤：First, establish a working condition sample. Since the method for calibrating control parameters of a hybrid electric vehicle oriented to working conditions of the present invention involves multiple linear regression analysis in statistics, it is necessary to ensure that the sample size is sufficient, so the sample size is expanded by dividing the dynamics segments, and the working condition samples The establishment specifically includes the following steps:
①首先选定若干个标准循环工况，所述标准循环工况指法规规定的、汽车行业通用的典型行驶工况，如NEDC工况，具体的标准循环工况及个数由标定人员根据需要自行确定。然后提取每个标准循环工况的所有动力学片段，将每一动力学片段看作一个独立的工况，所述动力学片段的划分方法为以工况起始点作为第一个动力学片段的起始点，以起始点后车速经历大于零后的第一个零车速点作为第一个动力学片段的结束点，以上一动力学片段的结束点作为下一动力学片段的起始点，以此类推，若最后一个动力学片段起始点后车速始终为零，则舍弃该段，所有循环工况的动力学片段数之和即为样本数，计为N。① First select several standard cycle conditions, which refer to the typical driving conditions specified in the regulations and commonly used in the automotive industry, such as NEDC conditions. The specific standard cycle conditions and the number are determined by the calibration personnel according to the needs Determine for yourself. Then extract all the dynamic segments of each standard cycle working condition, and regard each dynamic segment as an independent working condition. The starting point, the first zero speed point after the starting point after the vehicle speed is greater than zero is used as the end point of the first dynamic segment, and the end point of the previous dynamic segment is used as the starting point of the next dynamic segment. By analogy, if the vehicle speed is always zero after the starting point of the last dynamic segment, this segment is discarded, and the sum of the dynamic segment numbers of all cycle conditions is the number of samples, which is counted as N.
参考附图2，以NEDC工况为例，按照所述动力学片段划分方法将NEDC工况划分为13个独立工况。Referring to Figure 2, taking the NEDC working condition as an example, the NEDC working condition is divided into 13 independent working conditions according to the dynamic segment division method.
②计算各独立工况的20个特征指标C1,C2,C3......C20。所述20个特征指标依次指停车比例、停车次数、加速比例、减速比例、巡航比例、平均加速度、最大加速度、加速度标准差、平均减速度、最大减速度、减速度标准差、平均行驶车速、车速标准差、车速均方根、最高车速、0-20km/h车速比例、20-40km/h车速比例、40-60km/h车速比例、60-80km/h车速比例、80-100km/h车速比例，每个特征指标Cx为一元素数为N的向量，x为下角标，代表1,2,3......20。②Calculate 20 characteristic indexes C 1 , C 2 , C 3 ... C 20 of each independent working condition. The 20 characteristic indicators refer to the parking ratio, parking times, acceleration ratio, deceleration ratio, cruising ratio, average acceleration, maximum acceleration, acceleration standard deviation, average deceleration, maximum deceleration, deceleration standard deviation, average driving speed, Speed standard deviation, root mean square speed, maximum speed, 0-20km/h speed ratio, 20-40km/h speed ratio, 40-60km/h speed ratio, 60-80km/h speed ratio, 80-100km/h speed ratio Ratio, each characteristic index C x is a vector with the number of elements N, and x is a subscript, representing 1, 2, 3...20.
参考附图3，每个独立工况都有所述20个特征指标，亦即每个特征指标对应N个值，所有独立工况的特征指标用一个20行N列的矩阵表示，矩阵的第x列表示N个独立工况的第x个特征指标的值。With reference to accompanying drawing 3, each independent working condition has described 20 characteristic indexes, that is to say each characteristic index corresponds to N values, the characteristic indexes of all independent working conditions are represented by a matrix of 20 rows and N columns, the first of the matrix The x column represents the value of the xth characteristic index of N independent working conditions.
第二，各独立工况的控制参数优化，具体包括以下步骤：Second, the control parameter optimization of each independent working condition includes the following steps:
①确定要优化的控制参数P。本发明所述的一种面向工况的混合动力汽车控制参数标定方法具有通用性，可以标定不同控制参数，因此需要标定的控制参数的由标定人员自行选取。① Determine the control parameter P to be optimized. The method for calibrating control parameters of a working condition-oriented hybrid electric vehicle described in the present invention is versatile and can calibrate different control parameters, so the calibration personnel can select the control parameters that need to be calibrated.
②使用粒子群算法在每一独立工况下对P进行优化。以燃油消耗量最小为适应度函数，以电池SOC平衡为约束条件，以P作为粒子位置，设定迭代次数K作为优化终止条件，由粒子群算法的优化原理可知，当设定的K足够大时，总能得到各独立工况下使燃油消耗量最小的最佳控制参数Pb，Pb为一元素为N的向量，即每一独立工况对应一个最佳的控制参数。② Use particle swarm optimization algorithm to optimize P under each independent working condition. Taking the minimum fuel consumption as the fitness function, taking the battery SOC balance as the constraint condition, taking P as the particle position, and setting the number of iterations K as the optimization termination condition, it can be seen from the optimization principle of the particle swarm optimization algorithm that when the set K is large enough , the optimal control parameter P b that minimizes fuel consumption under each independent working condition can always be obtained, and P b is a vector with elements N, that is, each independent working condition corresponds to an optimal control parameter.
第三，基于相关性的工况特征指标筛选。所述的20个工况特征指标并不是全部与最佳控制参数呈显著的相关性，需要筛选出相关性较高的指标，剔除相关性不高的指标；此外，某些工况特征指标之间可能存在着程度较高的相关性，这意味着可以用一个指标来代替其余与其相关性较高的指标。工况特征指标的筛选能够减少指标的个数，保留与最佳控制参数相关性较高的工况特征指标，简化后续的多元线性回归模型及其可信度，筛选过程具体包括以下步骤：The third is the screening of working condition characteristic indicators based on correlation. Not all of the 20 operating condition characteristic indicators have a significant correlation with the optimal control parameters, and it is necessary to screen out the indicators with high correlation and eliminate the indicators with low correlation; in addition, some of the operating condition characteristic indicators There may be a high degree of correlation among them, which means that one indicator can be used to replace other indicators with high correlation with it. The screening of working condition characteristic indicators can reduce the number of indicators, retain the working condition characteristic indicators with high correlation with the optimal control parameters, and simplify the subsequent multiple linear regression model and its reliability. The screening process specifically includes the following steps:
①使用线性回归分析中的相关性公式计算各特征指标之间的相关系数如式(1)所示，各特征指标与最佳控制参数之间的相关系数如式(2)所示：① Use the correlation formula in linear regression analysis to calculate the correlation coefficient between each characteristic index As shown in formula (1), the correlation coefficient between each characteristic index and the optimal control parameter As shown in formula (2):
式中下角标x、y均表示特征指标代号，为1,2,3......20；In the formula, the subscripts x and y both represent the characteristic index codes, which are 1, 2, 3...20;
②筛选出与最佳控制参数相关性较高的工况特征指标，设置阈值R1，保留所有的特征指标，其余的剔除② Screen out the characteristic indicators of working conditions with high correlation with the optimal control parameters, set the threshold R 1 , and keep all feature indicators, and the rest are eliminated
③依据各工况特征指标之间的相关性继续缩减特征指标数量，设置阈值R2，在步骤②中保留的工况特征指标中，寻找所有的特征指标对Cx、Cy，视为重复的工况特征指标，保留与最佳控制参数相关性较高的，即与中的较大值对应的工况特征指标，剔除较小值对应的工况特征指标；③ Continue to reduce the number of feature indicators according to the correlation between the feature indicators of each working condition, set the threshold R 2 , and find all the working condition characteristic indicators retained in step ② The characteristic indexes of C x and C y are regarded as repeated working condition characteristic indexes, and the ones with higher correlation with the optimal control parameters are reserved, namely and The operating condition characteristic index corresponding to the larger value in , and the operating condition characteristic index corresponding to the smaller value is eliminated;
经过以上所述筛选步骤后保留下来的工况特征指标个数记为M，M≤20。The number of working condition characteristic indicators retained after the above screening steps is recorded as M, and M≤20.
参考附图4，单元格中数据是根据式(1)、式(2)计算得到的相关系数，如第2行第1列表示工况特征指标C1与最佳控制参数Pb之间的相关系数。根据式(1)、式(2)可知，变量与自身的相关系数为1，故表格对角线元素均为1；对角线上方的元素与其关于对角线对称的元素相等，如故对角线上方元素值不再计算。Referring to accompanying drawing 4, the data in the cell is the correlation coefficient calculated according to formula (1) and formula (2), such as row 2, column 1 Indicates the correlation coefficient between the working condition characteristic index C 1 and the optimal control parameter P b . According to formula (1) and formula (2), it can be seen that the correlation coefficient between the variable and itself is 1, so the diagonal elements of the table are all 1; the elements above the diagonal are equal to the symmetrical elements about the diagonal, as Therefore, the values of the elements above the diagonal are no longer calculated.
第四，多元线性回归分析。以数学模型的形式表达以上步骤确定的与最佳控制参数相关性较高的M个工况特征指标与最佳控制参数之间的多元线性回归关系，分析模型的合理性并做出相应调整。参考图5，多元线性回归分析具体包括以下步骤：Fourth, multiple linear regression analysis. In the form of a mathematical model, express the multiple linear regression relationship between the M operating condition characteristic indicators with high correlation with the optimal control parameters determined in the above steps and the optimal control parameters, analyze the rationality of the model and make corresponding adjustments. Referring to Figure 5, multiple linear regression analysis specifically includes the following steps:
①设置模型的显著性水平α，建立最佳控制参数Pb与M个工况特征指标之间的多元线性回归模型，如式(3)所示：① Set the significance level α of the model, and establish a multiple linear regression model between the optimal control parameter P b and M working condition characteristic indicators, as shown in formula (3):
Pb＝β0+β1·C1+β2·C2+......+βM·CM+ε (3)P b =β 0 +β 1 ·C 1 +β 2 ·C 2 +......+β M ·C M +ε (3)
式中，β0为常数项，β1、β2……βM为回归系数，ε为随机误差。In the formula, β 0 is a constant item, β 1 , β 2 ... β M are regression coefficients, and ε is a random error.
②残差分析，对式(3)所示模型进行残差分析的目的是找到数据异常的样本，删掉异常样本后，模型精度更高。对于每一个样本，均可得到其残差及对应的置信区间，根据各样本的残差r1、r2……rN及对应的置信区间rint1、rint2……rintN，根据多元线性回归相关理论，若某样本的置信区间不包含零点，则认为该样本数据异常，将其剔除，然后重新建立式(3)所示多元线性回归模型。② Residual error analysis. The purpose of residual analysis for the model shown in formula (3) is to find samples with abnormal data. After removing abnormal samples, the accuracy of the model is higher. For each sample, its residuals and corresponding confidence intervals can be obtained. According to the residuals r 1 , r 2 ... r N of each sample and the corresponding confidence intervals rint 1 , rint 2 ... rint N , according to the multivariate linear Regression-related theory, if the confidence interval of a sample does not contain zero, the sample data is considered abnormal, and it is eliminated, and then the multiple linear regression model shown in formula (3) is re-established.
③对模型开展T检验，T检验的目的是找到式(3)中对Pb解释能力较弱的工况特征指标并剔除，剔除一个指标后，多元线性回归模型变化，再次进行T检验的结果会变化，故每次只剔除一个工况特征指标。③ Carry out T-test on the model. The purpose of T-test is to find and eliminate the working condition characteristic indicators that are weak in explaining Pb in formula (3). After one indicator is eliminated, the multiple linear regression model changes, and the result of T-test will change, so only one working condition characteristic index is eliminated each time.
设进入该步骤时的工况特征指标个数为Me，先更新Me，Me≤M。若存在未通过T检验的回归系数βw，则其对应的工况特征指标Cw对Pb解释能力较弱，剔除下标号最小的Cw并进入步骤④，此时剩余工况指标个数为Mo，Mo＝Me-1；若不存在未通过T检验的回归系数，则剩余工况指标个数为Me，进入步骤⑤。Assuming that the number of operating condition characteristic indicators when entering this step is M e , update M e first, and M e ≤ M. If there is a regression coefficient β w that does not pass the T test, its corresponding working condition characteristic index C w is weak in explaining P b , so remove the C w with the smallest subscript and enter step ④. At this time, the number of remaining working condition indicators is M o , M o = M e -1; if there is no regression coefficient that fails the T-test, then the number of remaining working condition indicators is M e , and proceed to step ⑤.
④保持显著性水平α不变，建立最佳控制参数Pb与Mo个工况特征指标之间的多元线性回归模型并进行F检验，若F检验通过，则进入步骤③；若F检验未通过，则还原最新被剔除的一个工况特征指标并进入步骤⑤，此时剩余工况指标个数为Me。④Keep the significance level α unchanged, establish a multiple linear regression model between the optimal control parameter P b and M o working condition characteristic indicators, and perform the F test. If the F test passes, proceed to step ③; if the F test fails If it passes, restore the latest removed working condition characteristic index and enter step ⑤, at this time, the number of remaining working condition indicators is M e .
F检验是验证所建立多元线性回归模型中最佳控制参数与各工况特征指标的线性关系在整体上是否显著成立，若不通过，则该模型不能使用，故每剔除一个工况特征指标就要进行一次F检验。The F test is to verify whether the linear relationship between the optimal control parameters and the characteristic indicators of each working condition in the established multiple linear regression model is significantly established as a whole. If it does not pass, the model cannot be used. An F test is to be performed.
⑤以最终保留的Me个工况特征指标与Pb之间的多元线性回归模型作为最终模型，如式(4)所示：⑤The final model is the multiple linear regression model between the remaining M e operating condition characteristic indicators and P b as the final model, as shown in formula (4):
第五，对于要标定的新工况，计算所述最终保留的Me个工况特征指标，并将其代入式(4)，得到该工况对应的最佳控制参数。Fifth, for the new working condition to be calibrated, calculate the finally reserved M e working condition characteristic indicators and substitute them into formula (4) to obtain the optimal control parameters corresponding to the working condition.
Claims (1)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910025948.5A CN109747654B (en) | 2019-01-11 | 2019-01-11 | A working condition-oriented method for calibration of hybrid electric vehicle control parameters |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910025948.5A CN109747654B (en) | 2019-01-11 | 2019-01-11 | A working condition-oriented method for calibration of hybrid electric vehicle control parameters |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109747654A CN109747654A (en) | 2019-05-14 |
| CN109747654B true CN109747654B (en) | 2019-08-09 |
Family
ID=66405487
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910025948.5A Expired - Fee Related CN109747654B (en) | 2019-01-11 | 2019-01-11 | A working condition-oriented method for calibration of hybrid electric vehicle control parameters |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109747654B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111348029B (en) * | 2020-03-16 | 2021-04-06 | 吉林大学 | Method for determining optimal value of calibration parameter of hybrid electric vehicle by considering working condition |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11338816B2 (en) * | 2019-02-02 | 2022-05-24 | Ford Global Technologies, Llc | Over-the-air flashing and reproduction of calibration data using data regression techniques |
| CN110816291B (en) * | 2019-11-11 | 2021-05-11 | 常熟理工学院 | A distributed driving vehicle energy efficiency optimization control method based on second-order oscillating particle swarm |
| CN110926827A (en) * | 2019-11-30 | 2020-03-27 | 河南科技大学 | Automatic optimization and calibration system for vehicle control parameters |
| CN111835237B (en) * | 2020-07-17 | 2022-10-14 | 中车永济电机有限公司 | Traction motor control method and device, electronic equipment and storable medium |
| CN112508317A (en) * | 2020-09-01 | 2021-03-16 | 中国汽车技术研究中心有限公司 | Subjective and objective relevance scoring method based on multi-source power assembly vehicle type drivability |
| CN112415892B (en) * | 2020-11-09 | 2022-05-03 | 东风汽车集团有限公司 | Gasoline engine starting calibration control parameter optimization method |
| CN114537416B (en) * | 2022-02-18 | 2025-05-09 | 希迪智驾科技股份有限公司 | A method, device, electronic device and storage medium for generating a calibration table |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102717797B (en) * | 2012-06-14 | 2014-03-12 | 北京理工大学 | Energy management method and system of hybrid vehicle |
| CN103112450B (en) * | 2013-02-27 | 2015-06-24 | 清华大学 | Real-time optimized control method for plug-in parallel hybrid electric vehicle |
| CN104071161B (en) * | 2014-04-29 | 2016-06-01 | 福州大学 | A kind of method of plug-in hybrid-power automobile operating mode's switch and energy management and control |
| CN106004865B (en) * | 2016-05-30 | 2019-05-10 | 福州大学 | Range-adaptive hybrid electric vehicle energy management method based on operating condition identification |
| KR101901801B1 (en) * | 2016-12-29 | 2018-09-27 | 현대자동차주식회사 | Hybrid vehicle and method of predicting driving pattern |
-
2019
- 2019-01-11 CN CN201910025948.5A patent/CN109747654B/en not_active Expired - Fee Related
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111348029B (en) * | 2020-03-16 | 2021-04-06 | 吉林大学 | Method for determining optimal value of calibration parameter of hybrid electric vehicle by considering working condition |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109747654A (en) | 2019-05-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109747654B (en) | A working condition-oriented method for calibration of hybrid electric vehicle control parameters | |
| CN107544290B (en) | New energy automobile performance evaluation analysis and optimization system and method | |
| CN103091112B (en) | Method and device of car emission fault detection and diagnosis based on fuzzy reasoning and self-learning | |
| CN105468850B (en) | Electronic product degradation trend prediction technique based on more residual error regression forecasting algorithms | |
| JP6968282B2 (en) | Causal analysis of powertrain management | |
| CN111348029B (en) | Method for determining optimal value of calibration parameter of hybrid electric vehicle by considering working condition | |
| CN114036647A (en) | Power battery safety risk assessment method based on real vehicle data | |
| CN108804800A (en) | Lithium ion battery SOC on-line prediction methods based on echo state network | |
| CN113705615A (en) | Neural network-based electric vehicle charging process multistage equipment fault diagnosis method and system | |
| CN103593519B (en) | A kind of carrier rocket optimization of Overall Parameters of Muffler method based on experimental design | |
| CN117147166A (en) | Engine calibration method and device, electronic equipment and storage medium | |
| CN112993344A (en) | Neural network-based fuel cell system output performance prediction method and device | |
| CN117828413A (en) | Transformer oil temperature prediction method and system based on LSTM neural network | |
| CN113297685A (en) | Vehicle operation condition mode identification method | |
| CN117734526A (en) | New energy automobile battery pack temperature control method and system | |
| CN116992247A (en) | Abnormal data detection method of tail gas analyzer | |
| CN118478695B (en) | A safety warning method, device and electronic equipment for power battery | |
| CN114021465A (en) | Robust state estimation method and system for power system based on deep learning | |
| CN113313406A (en) | Power battery safety risk assessment method for big data of electric vehicle operation | |
| CN112748331A (en) | Circuit breaker mechanical fault identification method and device based on DS evidence fusion | |
| CN116910499A (en) | System state monitoring method and device, electronic equipment and readable storage medium | |
| CN115840891A (en) | Engine air inflow prediction method and device and storage medium | |
| CN115169630A (en) | A kind of electric vehicle charging load prediction method and device | |
| CN113111588A (en) | NO of gas turbineXEmission concentration prediction method and device | |
| CN116572928B (en) | Control method, device and system of hybrid vehicle and hybrid vehicle |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190809 |