WO2021035500A1 - 用于48v轻混汽车锂离子电池的荷电状态在线估算系统 - Google Patents

用于48v轻混汽车锂离子电池的荷电状态在线估算系统 Download PDF

Info

Publication number
WO2021035500A1
WO2021035500A1 PCT/CN2019/102657 CN2019102657W WO2021035500A1 WO 2021035500 A1 WO2021035500 A1 WO 2021035500A1 CN 2019102657 W CN2019102657 W CN 2019102657W WO 2021035500 A1 WO2021035500 A1 WO 2021035500A1
Authority
WO
WIPO (PCT)
Prior art keywords
state
soc
estimation
covariance
time point
Prior art date
Application number
PCT/CN2019/102657
Other languages
English (en)
French (fr)
Inventor
王元
林双
冷枫
单颖会
尹求实
李开元
贺中捷
董书岭
Original Assignee
淄博火炬能源有限责任公司
全能质可科技有限责任公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 淄博火炬能源有限责任公司, 全能质可科技有限责任公司 filed Critical 淄博火炬能源有限责任公司
Priority to PCT/CN2019/102657 priority Critical patent/WO2021035500A1/zh
Priority to CN201980034166.6A priority patent/CN112601968A/zh
Publication of WO2021035500A1 publication Critical patent/WO2021035500A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01RMEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/378Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC] specially adapted for the type of battery or accumulator
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01RMEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01RMEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/367Software therefor, e.g. for battery testing using modelling or look-up tables
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01RMEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/382Arrangements for monitoring battery or accumulator variables, e.g. SoC
    • G01R31/3842Arrangements for monitoring battery or accumulator variables, e.g. SoC combining voltage and current measurements

Definitions

  • the invention relates to an online state-of-charge estimation system for a 48V light-hybrid vehicle lithium-ion battery, and belongs to the technical field of hybrid-power vehicle lithium-ion batteries.
  • 48V mild hybrid vehicles Compared with pure electric, plug-in hybrid and deep hybrid vehicles, mild hybrid vehicles equipped with 48V system (hereinafter referred to as 48V mild hybrid vehicles) have low technical complexity, relatively easy development and manufacturing, low cost, and low cost. The advantages of high efficiency of oil emission reduction are favored by the market and users.
  • the electrification scheme of 48V light-hybrid vehicles is realized by adding a 48V system to the traditional fuel vehicle powertrain-mainly including a starter/generator integrated motor, a power conversion unit and a 48V lithium-ion battery system-thus making light Hybrid vehicles can meet the emission and fuel consumption indicators required by national laws and regulations at a much lower cost than pure electric and other types of hybrid vehicles.
  • the estimation accuracy of the state of charge (SOC) of the lithium-ion battery will directly affect the performance and efficiency of the 48V battery system, and in turn affect the performance and fuel efficiency of the entire light-hybrid vehicle. Therefore, a battery management system (BMS, Battery Management System) is required to realize real-time, accurate, and highly robust estimation of the SOC.
  • BMS Battery Management System
  • the commonly used SOC estimation methods for hybrid electric vehicles on the market mainly include ampere-hour integration and extended Kalman filtering methods.
  • the ampere-hour integration calculates the SOC change by integrating the load current over a period of time. Because of its simple principle, it is easy to implement in the actual BMS system.
  • the ampere-hour integration method has problems such as the difficulty of determining the initial SOC value and the difficulty of eliminating the cumulative error caused by the current sensor acquisition error.
  • OCV Open Circuit Voltage
  • charge and discharge curve or other methods can be used for regular correction, it is not It is not suitable for the characteristics of the working conditions of the 48V system, and it is difficult to guarantee the accuracy of real-time online SOC estimation.
  • the extended Kalman method based on the equivalent circuit model and closed-loop control theory can effectively filter the sensor noise, but the use of the extended Kalman filter for SOC estimation depends on the open circuit voltage characteristics and the accuracy of the equivalent circuit model. And before actual deployment, a lot of trial and error debugging is required to determine accurate and reliable initial algorithm parameters, such as the covariance of measurement noise and process noise.
  • the problem to be solved by the present invention is to improve the accuracy and robustness of online SOC estimation in the 48V system through efficient real-time calculation in view of the difficulties and potential problems in the actual use of the extended Kalman filter. .
  • the present invention provides an online state-of-charge estimation system for 48V light-hybrid automobile lithium-ion batteries, including a parameter adjustment system and a SOC estimation system;
  • the parameter adjustment system includes an open circuit voltage characteristic adjustment system and an equivalent circuit model parameter calibration adjustment system.
  • the parameter adjustment system can accurately calibrate the value of the open circuit voltage and the parameters of the equivalent circuit model according to the ambient temperature, so as to reduce the temperature Model errors caused by changes;
  • the SOC estimation system includes noise statistical characteristics, filter gain adaptation, and a recursive state estimation system, and the SOC estimation system can improve the robustness and accuracy of SOC online estimation;
  • the input signals required by the parameter adjustment system and the SOC estimation system all come from the measured values of load current, terminal voltage and temperature collected by the BMS.
  • the open circuit voltage characteristic adjustment system includes a three-dimensional mapping relationship between OCV and SOC and ambient temperature to ensure the accuracy of real-time adjustment of the open circuit voltage characteristic; the three-dimensional mapping relationship can be expressed by the following formula
  • variable s and variable T respectively represent the value of SOC and the value of ambient temperature.
  • the establishment of the three-dimensional mapping relationship specifically includes the following steps:
  • OCV-SOC characteristic test of lithium battery at different temperatures constant control of the test temperature through the incubator, the tested temperature is -30°C to 60°C, and the test frequency is one test every 10°C; For each temperature test, a set of OCV measurement values are collected every 5% SOC during the test; the collected OCV measurement values are preprocessed and recorded as the corresponding temperature and OCV value under the corresponding SOC;
  • step 2) Fit the test data obtained in step 1) to the function OCV(s, T) to establish a three-dimensional mapping relationship that can be expressed by a binary polynomial regression model; the expression of the binary regression model is as follows
  • step 3 Store the coefficients of the binary regression model obtained in step 2) in the software of the BMS system; call the stored coefficients when the BMS system is running, in this way, reconstruct the regression model fitted in the previous step in real time, and use The three-dimensional mapping relationship represented by this model; and then according to the actual measured temperature and the real-time estimated SOC value obtained during the operation of the BMS system, the current OCV and the value of the derivative function of the SOC are jointly determined.
  • the equivalent circuit model parameter calibration and adjustment system performs real-time real-time on the values of the ohmic internal resistance R 0 , the polarization internal resistance R p , and the polarization capacitance C p in the equivalent circuit model according to changes in ambient temperature.
  • Calibration specifically includes the following steps:
  • the left term ⁇ (T) of the equation is a function of ambient temperature, which can represent any of the model parameters R 0 , R p and C p , and the coefficient of the right term of the equation ⁇ a 0 , a 1 ,..., a n ⁇ represent the coefficients of the regression model obtained by fitting;
  • step (3) Store the coefficients of the three regression models obtained in step (2) in the software of the BMS system; call the stored coefficients when the BMS system is running, in this way, reconstruct the previous step through the fitting method in real time.
  • the obtained regression model and then through the mapping relationship between the equivalent circuit model parameters R 0 , R p and C p represented by the regression model and the ambient temperature, the BMS system is running according to the actual measured ambient temperature, etc. Real-time calibration of the parameter values of the effective circuit model.
  • the SOC estimation system performs online real-time estimation of the SOC of a 48V lithium-ion battery according to the open circuit voltage characteristics and equivalent model parameters provided by the parameter adjustment system, and the SOC estimation system includes a recursive parallel calculation State estimation system and noise statistical characteristics and filter gain adaptive system.
  • the state space model used for SOC estimation in the recursive state estimation system is as follows:
  • the formula (5) and formula (6) are the system equation and the measurement equation of the state space model, respectively.
  • variable u k is the input signal of the state space model, and the true physical meaning of the u k is the working load current I L of the battery;
  • y k represents the observed value of the state variable, The true physical meaning of y k is the terminal voltage of the battery;
  • Ak represents the state transition matrix
  • B k is the input matrix
  • each element in the matrix Ak and B k can be determined by the battery equivalent circuit model parameters R 0 , R p and C p , calculated according to the following formula:
  • ⁇ and Q are the Coulomb efficiency and the current maximum available capacity of the battery, respectively;
  • the state variable x k is a two-dimensional vector, defined as follows
  • variable V p, k represents the voltage across the parallel RC circuit in the battery equivalent circuit model at the current time point, and the variable s k represents the SOC value at the current time point;
  • the value of the function OCV(s k , T k ) represents the open circuit voltage value when the ambient temperature is T k and the battery SOC is s k.
  • V p, 0 of V p, k the initial value s 0 of sk can be arbitrarily selected, or set to the last recorded SOC estimation stored in the corresponding BMS system at the end of the last vehicle operation Value; for any time point k>0, the following steps B to D are executed in a loop;
  • e k represents innovation, that is, the prediction error of the state space output; e k is calculated as follows:
  • the one-step prediction of the state variable is modified to obtain the state estimate at the kth time point vector
  • the second element of That is the estimated value of the updated SOC.
  • the noise statistical characteristics and filter gain adaptive system are performed in parallel with the recursive state estimation system; the noise statistical characteristics and filter gain adaptive system are used to update state prediction and state estimation
  • the covariance of the process noise and the covariance of the measurement noise are estimated in a signal adaptive manner, and the gain of the filter is updated in a signal adaptive manner.
  • the specific calculation steps of the noise statistical characteristics and filter gain adaptive system are as follows:
  • the symbol I 2 represents a two-dimensional unit matrix, and for any subsequent time point k>0, the following steps II to IV are cyclically executed;
  • the two-dimensional matrix Is the covariance of the one-step state prediction obtained at the current point in time, and P k-1 and Respectively represent the covariance of the state estimation obtained at the previous time point and the covariance of the process noise estimated at the previous time point;
  • M innovation data ⁇ e k-M+1 , e k-M+2 ..., e k ⁇ are in the form of a sliding window, from k-M+1 time point to current time point k
  • a total of M pieces of innovation data are recorded in the memory of the BMS system and updated continuously over time;
  • the symbol ⁇ represents the forgetting factor
  • the vector C k is defined as follows:
  • Noise statistical characteristics are highly adaptive; reduce the negative impact of uncertainty caused by model accuracy, measurement and process noise covariance over time, so as to improve the robustness and accuracy of SOC estimation.
  • Figure 1 is a schematic diagram of the architecture of the SOC state online estimation system involved in the present invention
  • Figure 2 is an operation flow of establishing the regulation system
  • Figure 3 is an example diagram of the fitted binary regression model function surface
  • Figure 4 is a schematic diagram of the first-order equivalent circuit model
  • FIG. 5 is a schematic diagram of HPPC data
  • Figure 6 shows the parallel logic calculation process in the SOC estimation system
  • Figure 7 shows the 10-hour cycle test results at 48°C
  • Figure 8 shows the results of the 10-hour cycle test at 25°C.
  • the online state-of-charge estimation system for 48V light-hybrid automobile lithium-ion batteries of the present invention includes two main components.
  • the first part is an open circuit voltage characteristic adjustment system and an equivalent circuit model parameter calibration adjustment system. According to the ambient temperature, the value of the open circuit voltage and the parameters of the equivalent circuit model are accurately calibrated to reduce the error of the model caused by temperature changes; the second part is the realization of SOC online estimation, including noise statistics and filter gain adaptation And the recursive state estimation system can improve the robustness and accuracy of SOC online estimation.
  • the validity verification shows that the proposed method can achieve high-precision and high-robust online estimation of SOC under typical cycle conditions of light-hybrid vehicles.
  • the parameter adjustment system part improves the accuracy of SOC estimation by reducing the deviation of the equivalent circuit model parameters and the open circuit voltage characteristics due to temperature changes; and the SOC estimation system is used to achieve high-precision and high-robust SOC estimation algorithms Mathematical logic.
  • the input signals required for the two parts of the parameter adjustment system and SOC estimation come from the measured readings of load current, terminal voltage, and temperature collected by the BMS.
  • the present invention focuses on the open circuit voltage characteristics and the effect of temperature on the equivalent circuit model, so as to improve the applicability of the model in different environments.
  • the part of parameter adjustment is used to adjust the mathematical model that characterizes the mapping relationship between SOC and open circuit voltage, as well as the calibration of ohmic internal resistance, polarization internal resistance, and polarization capacitance values in the equivalent circuit model.
  • the internal characteristics of lithium-ion batteries will be greatly affected by the ambient temperature, especially high-power lithium-ion batteries used in 48V systems. Therefore, if only the open-circuit voltage characteristic curve at a single temperature is used (for example, considering only normal temperature), when the ambient temperature changes significantly, the previously calibrated open-circuit voltage characteristic will be larger than its actual temperature. deviation. This will cause a significant reduction in the accuracy of the SOC estimation.
  • the present invention establishes a three-dimensional mapping relationship between OCV, SOC and ambient temperature, so that the obtained open circuit voltage characteristics are more in line with the requirements of practical applications.
  • the mathematical nature of the three-dimensional mapping relationship is a binary function of SOC and ambient temperature, which can be expressed by the following formula
  • variable s and variable T respectively represent the value of SOC and the value of ambient temperature.
  • OCV-SOC characteristic test of lithium battery at different temperatures constant control of the test temperature through a thermostat, the tested temperature is -30°C to 60°C, and the test frequency is one test every 10°C; For each temperature test, collect a set of OCV measurement values every 5% SOC during the test; preprocess the collected OCV measurement values and record them as the corresponding temperature and the OCV value under the corresponding SOC;
  • step 3 Store the coefficients of the binary regression model obtained in step 2) in the software of the BMS system; call the stored coefficients when the BMS system is running to reconstruct the regression model fitted in the previous step in real time in this way, The three-dimensional mapping relationship represented by this model is also used; and the current OCV and the value of the derivative function with respect to SOC are jointly determined according to the actual measured temperature and the real-time estimated SOC value when the BMS system is running.
  • this embodiment uses a first-order RC, etc.
  • the effective circuit model is shown in Figure 4.
  • IL represents the load current
  • V t represents the terminal voltage of the battery
  • the first-order equivalent circuit model has simple and effective characteristics in actual BMS applications, it still needs other methods and techniques to compensate for the potential decrease in accuracy in order to reduce the error of SOC estimation.
  • the environmental temperature has the strongest influence. Therefore, in the present invention, the parameters are calibrated in time according to the change of the ambient temperature. In order to achieve this feature, you need to follow the steps below.
  • HPPC data used refers to the terminal voltage and load current data recorded when the SOC is in an interval of about 50% (starting at the start of the discharge pulse operation and ending at the end of the 40-second standstill). The time starting point and period length of data recording are shown in Figure 5.
  • the left term of the equation is a function of the ambient temperature, which can represent model parameters, and any one of them, and the coefficient of the right term of the equation represents the coefficient of the regression model obtained by fitting;
  • the function of this part of SOC estimation is to execute the logic operation of the SOC online real-time estimation algorithm of the 48V lithium-ion battery according to the OCV-SOC characteristics provided by the parameter adjustment function and the equivalent model parameters.
  • the SOC estimation proposed in the present invention is composed of two parallel computing systems: (1) recursive state estimation; (2) noise statistical characteristics and filter gain adaptation.
  • the logical operation procedures of these two systems are executed in parallel and complement each other. Specifically: the update of the innovation covariance depends on the value of the innovation variable obtained through output prediction; and the update of the state estimation You need to use the updated filter gain.
  • the meaning of the word update is: through iterative calculation, the value obtained at the current time point is substituted for the value obtained at the previous time point with the same meaning.
  • the formula (5) and formula (6) are the system equation and the measurement equation of the state space model, respectively.
  • variable u k is the input signal of the state space model, and the true physical meaning of the u k is the working load current I L of the battery;
  • y k represents the observed value of the state variable, The true physical meaning of y k is the terminal voltage of the battery;
  • formula (5) and formula (6) are the system equation and measurement equation of the state space model, respectively.
  • Ak represents the state transition matrix
  • B k is the input matrix.
  • Each element in the matrix A k and B k can be calculated from the battery equivalent circuit model parameters R 0 , R p and C p according to the following formula:
  • the state variable x k is a two-dimensional vector, defined as follows
  • V p,k is the voltage on the parallel RC circuit in the battery equivalent circuit model at the current time point
  • the variable sk represents the SOC value at the current time point.
  • the variable u k is the input signal of the state space model, and its true physical meaning is the working load current I L of the battery.
  • y k represents the observed value of the state variable, and its true physical meaning is the terminal voltage of the battery.
  • the value of the function OCV(s k , T k ) is the OCV value when the ambient temperature is T k and the battery SOC is s k.
  • the variables w k and v k represent the process noise and measurement noise of the state space model, respectively.
  • V p, 0 of V p, k the initial value s 0 of sk can be arbitrarily selected, or set to the last recorded SOC estimation stored in the corresponding BMS system at the end of the last vehicle operation Value; for any time point k>0, the following steps B to D are executed in a loop;
  • e k represents innovation, that is, the one-step prediction error of the output of the state space.
  • the calculation of e k is as follows:
  • the product of the filter gain L k and the innovation e k is used to modify the one-step state prediction to obtain the state estimate at the kth time point Further, the vector The second element of is the estimated value of the updated SOC In this step, the filter gain L k is obtained through a calculation process parallel to the recursive state estimation.
  • the symbol I 2 represents a two-dimensional identity matrix. For any subsequent time point k>0, the following steps II to IV are executed in a loop.
  • the two-dimensional matrix Is the covariance of the one-step state prediction obtained at the current point in time
  • M innovation data ⁇ e k-M+1 , e k-M+2 ..., e k ⁇ are in the form of a sliding window, from k-M+1 time point to current time point k
  • a total of M pieces of innovation data are recorded in the memory of the BMS system and updated continuously over time.
  • the symbol ⁇ represents the forgetting factor, and its value is a positive number much less than 1.
  • symbol Represents the covariance of the measurement noise estimated at the previous point in time.
  • the vector C k is defined as follows
  • the 8-Ah NMC lithium-ion battery designed for the 48V system was used, and the typical light-hybrid vehicle cycle test data was used at 48°C and 25°C (using a frequency of 1.0Hz, a total of 36,000 batteries in 10 hours). Point data) to perform simulation analysis to evaluate the accuracy of the invented SOC estimation method. All working condition cycles are obtained based on the WLTC cycle.
  • the time series of the SOC reference value is calculated based on the current collected by the test equipment and based on the ampere-hour integration method of the known initial SOC. In the technical field involved in the present invention, it is generally considered that the error of the current sensor in the experimental environment is sufficiently small, so the SOC reference value obtained by this method can be used to approximate the real value of the SOC.
  • Equation (23) The standard used to evaluate the accuracy of SOC estimation is the common Mean Absolute Error (MAE, Mean Absolute Error) and Root Mean Square Error (RMSE, Root Mean Square Error). The specific calculation method is shown in Equation (23):
  • s k represents the reference SOC value at the kth sampling time point
  • It is the estimated SOC value at the kth time point obtained by the SOC estimation system.
  • FIG. 7 and Figure 8 represent the simulation verification results of the SOC estimation at an ambient temperature of 48°C and 25°C, respectively.
  • the upper subgraph shows the comparison between the time series of SOC estimation value and the time series of SOC reference value.
  • the SOC estimation value is obtained by using the SOC online estimation system described in the present invention.
  • the subgraph below shows the evolution of the time series of the absolute error between the SOC estimation value and the SOC reference value.
  • the verification result shows that by using the SOC online estimation system described in the present invention, the SOC of a 48V lithium ion battery can be accurately estimated at different temperatures.
  • the specific numerical statistics of MAE and RMSE of the verification results are shown in Table 1.
  • the 48V system is a very economical and effective solution for the electrification of traditional fuel vehicles; it can effectively control CO 2 and pollutant emissions and fuel consumption at a relatively low cost, and has a high industrial promotion value and a broad market. potential.
  • some auto manufacturers have deployed 48V system-related technologies and corresponding 48V light-hybrid models.
  • the SOC online estimation system proposed by the present invention belongs to one of the core technologies of the BMS system of hybrid electric vehicles.
  • the present invention is particularly suitable for light-hybrid vehicles equipped with a 48V battery system. Because of its wide temperature range and frequent charging and discharging current changes under typical operating conditions, it poses a great challenge to online SOC estimation, and therefore requires additional measures to Ensure the accuracy and robustness of SOC estimation.
  • the invention aims to reduce the influence of temperature changes and model uncertainty on SOC estimation, and can realize high-precision and high-robust SOC estimation. Therefore, the present invention can become an excellent solution for SOC estimation of light hybrid vehicles. At the same time, it can also be used for online SOC estimation tasks of other types of hybrid vehicles equipped with lithium-ion batteries.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Secondary Cells (AREA)
  • Tests Of Electric Status Of Batteries (AREA)

Abstract

一种用于48V轻混汽车锂离子电池的荷电状态在线估算系统,属于混合动力汽车锂离子电池技术领域。本系统包括两个主要组成部分,第一部分为开路电压特性调节系统和等效电路模型参数校准调节系统,这两个系统可以根据环境温度对开路电压的数值与等效电路模型的参数进行精确校准,以降低由温度变化引起的模型的误差;第二部分为SOC在线估算的实现,噪声统计特性、滤波器增益自适应和递归式状态估计系统,这两个系统能够提高SOC在线估算的鲁棒性与精度。该系统能够在轻混汽车典型循环工况下对SOC实现高精度,高鲁棒性的在线估算。

Description

用于48V轻混汽车锂离子电池的荷电状态在线估算系统 技术领域
本发明涉及一种用于48V轻混汽车锂离子电池的荷电状态在线估算系统,属于混合动力汽车锂离子电池技术领域。
背景技术
相较于纯电动、插电式混合动力以及深度混合动力汽车,搭载48V系统的轻度混合动力汽车(下称48V轻混汽车)以其技术复杂度低、开发制造相对容易、成本低、节油减排效果高效等优势备受市场以及用户的青睐。48V轻混汽车的电气化方案,是通过在传统燃油汽车动力总成的基础上添加48V系统实现的——主要包括启动/发电一体式电机、电力转换单元以及48V锂离子电池系统——从而使得轻混汽车能以相较纯电动和其他类型混合动力汽车低得多的成本达到各国法规规定的排放与油耗指标。在48V系统中,锂离子电池荷电状态(SOC,State of Charge)的估算精度将会直接影响48V电池系统的性能和效率,进而影响整个轻混汽车的性能和燃油效率。因此需要电池管理系统(BMS,Battery Management System)对SOC实现实时、精确、高鲁棒性的估算。
目前市场上混合动力汽车常用的SOC估算方法主要包括安时积分与扩展卡尔曼滤波方法。安时积分通过对一段时间内的负载电流进行积分以实现对SOC变化量的计算,因其原理简单所以在实际BMS系统中很容易实现。但是安时积分法存在SOC初值难以确定以及由电流传感器采集误差造成的累计误差难以消除等问题,虽然可以使用开路电压(OCV,Open Circuit Voltage)、充放电曲线或其他方式定期修正,但并不适合48V系统的工况特点,难以保证实时在线SOC估算的精度。基于等效电路模型以及闭环控制理论的扩展卡尔曼方法能够对传感器噪声进行有效滤波,但运用扩展卡尔曼滤波进行SOC估算依赖于开路电压特性和等效电路模型的精确性。而且在实际部署之前,需要大量试错调试才能确定准确可靠的算法初始参数,例如量测噪声和过程噪声的协方差。
另外,其他的一些方法例如无迹卡尔曼滤波滤波以及粒子滤波虽然在理论上比扩展卡尔曼滤波具有更高的估算精度以及鲁棒性,但是这些方法复杂度高,需要依赖更多的计算资源。受制于成本等因素,产品化的车载BMS系统上的单片机通常仅具备有限的处理能力和内存,因而这些方法在实际应用中较难保障混合动力汽车SOC估算对于实时性的要求,难以进行产业化应用。
发明内容
根据以上现有技术中的不足,本发明要解决的问题是:针对扩展卡尔曼滤波在实际使用中的难点和潜在问题,通过高效的实时计算提高48V系统中SOC在线估算的精度以及鲁棒性。
本发明解决其技术问题所采用的技术方案是:本发明提供一种用于48V轻混汽车锂离子电池的荷电状态在线估算系统,包括参数调节系统和SOC估算系统;
所述参数调节系统包括开路电压特性调节系统和等效电路模型参数校准调节系统,所述参数调节系统可以根据环境温度对开路电压的数值与等效电路模型的参数进行精确校准,以降低由温度变化引起的模型误差;
所述SOC估算系统包括噪声统计特性、滤波器增益自适应和递归式状态估计系统,所述SOC估算系统能够提高SOC在线估算的鲁棒性与精度;
所述参数调节系统与SOC估算系统所需输入信号均来自于BMS所采集的负载电流、端电压和温度的测量数值。
作为一种优选的方案,所述开路电压特性调节系统包括OCV与SOC以及环境温度的三维映射关系,以保证开路电压特性实时调节的精确度;所述三维映射关系可通过下式表达
OCV(s,T)      (1)
其中变量s与变量T分别代表SOC的值和环境温度的值。
作为一种优选的方案,所述三维映射关系的建立具体包括以下步骤:
1)在不同温度下对锂电池进行OCV-SOC特性测试,通过温箱对测试温度实施恒定控制,所测试的温度为-30℃到60℃,测试频率为每隔10℃进行一次测试;对于每一个温度值下的测试,在测试过程中每隔5%SOC收集一组OCV测量值;对所收集到的OCV测量值进行预处理,并记录为对应温度和对应SOC下的OCV值;
2)将步骤1)所得的测试数据拟合到函数OCV(s,T),建立一个可用二元多项式回归模型表达的三维映射关系;该二元回归模型表达式如下
Figure PCTCN2019102657-appb-000001
同时,OCV关于SOC的导函数、及该导函数对应的回归模型也可以通过公式(2)以如下方式获得:
Figure PCTCN2019102657-appb-000002
3)把步骤2)获得的二元回归模型的系数储存在BMS系统的软件中;在BMS系统运行时调用所储存的系数,以此方式实时地重构前一步所拟合的回归模型,并用此模型所代表的三维映射关系;进而在BMS系统运行时根据实际测得的温度与实时获得的SOC估算数值,共同确定当前的OCV及其关于SOC的导函数的数值。
作为一种优选的方案,所述等效电路模型参数校准调节系统根据环境温度的变化对等效 电路模型中欧姆内阻R 0、极化内阻R p、和极化电容C p数值进行实时校准,具体的包括以下步骤:
(1)在不同温度下对锂电池进行混合脉冲功率特性测试,通过温箱对测试温度实施恒定控制,所测试的温度为-30℃到60℃,测试频率为每隔10℃进行一次测试;对于-30℃到60℃所获得的每一组数据进行预处理并进行参数辨识,提取电池等效电路模型的参数R 0,R p与C p,所述R 0为欧姆内阻,R p为极化内阻,C p为极化电容;
(2)把步骤(1)中不同温度下获得的模型参数R 0,R p与C p的数值,分别拟合到三个多项式回归模型中,以此建立对应的模型参数关于温度的映射关系;所述三个多项式回归模型具有相同的数学形式,并可通过下式表达:
Figure PCTCN2019102657-appb-000003
公式(4)中,等式的左端项ψ(T)是环境温度的函数,它可代表模型参数R 0,R p与C p中的任意一个,等式右端项的系数{a 0,a 1,...,a n}代表通过拟合得到的回归模型的系数;
(3)把通过步骤(2)所获得的三个回归模型的系数储存在BMS系统的软件中;在BMS系统运行时调用所储存的系数,以此方式实时地重构前一步通过拟合所得到的回归模型,进而通过该回归模型所代表的等效电路模型参数R 0,R p与C p与环境温度之间的映射关系,在BMS系统运行时根据实际测得的环境温度,实现等效电路模型参数数值的实时校准。
作为一种优选的方案,所述SOC估算系统根据参数调节系统所提供的开路电压特性以及等效模型参数,执行48V锂离子电池的SOC在线实时估算,所述SOC估算系统包括并行计算的递归式状态估计系统和噪声统计特性与滤波器增益自适应系统。
作为一种优选的方案,递归式状态估计系统中用于SOC估算的状态空间模型如下所示:
x k=A kx k-1+B ku k+w k     (5)
y k=h(x k,u k)+v k    (6)
所述公式(5)和公式(6)分别为状态空间模型的系统方程和量测方程,公式(5)和公式(6)中下标字母k代表离散系统中时间点的索引;所述k指代当前采样时间点t k,k-1指代前一采样时间点t k-1,并有t k=t k-1+Δt,其中Δt即为BMS系统中的采样时间间隔;变量w k与v k分别代表状态空间模型的过程噪声和量测噪声;
所述公式(5)和公式(6)中,变量u k是状态空间模型的输入信号,所述u k的真实物理意义为电池的工作负载电流I L;y k代表状态变量的观测值,所述y k的真实物理意义为电池的端电压;
所述公式(5)和公式(6)中,A k代表状态转移矩阵,B k为输入矩阵,矩阵A k和B k中的各个元素可由电池等效电路模型参数R 0,R p和C p,按下列公式计算而得:
Figure PCTCN2019102657-appb-000004
Figure PCTCN2019102657-appb-000005
其中符号η和Q所代表的物理意义,分别为库伦效率和电池的当前最大可用容量;
在公式(5)中,状态变量x k是一个二维向量,定义如下
x k=[V p,k s k] T  (9)
其中变量V p,k代表当前时间点下电池等效电路模型中并联RC电路两端的电压,变量s k代表当前时间点下的SOC值;
所述公式(6)中,非线性函数h(x k,u k)的具体表达形式如下:
h(x k,u k)=OCV(s k,T k)-I LR 0-V p,k   (10)
其中函数OCV(s k,T k)的值即代表当环境温度为T k且电池SOC为s k时的开路电压数值。
作为一种优选的方案,所述递归式状态估计系统的具体计算步骤如下:
A状态估计初始化
将V p,k的初值V p,0设为零;s k的初始值s 0可任意选取,或设置为在相应BMS系统中所储存的上一次车辆运行结束时所最后记录的SOC估算值;对于任意时间点k>0,循环执行以下步骤B至步骤D;
B一步状态预测:
按下式进行状态变量一步预测的计算
Figure PCTCN2019102657-appb-000006
其中
Figure PCTCN2019102657-appb-000007
代表在前一时间点所得的状态估计的值,
Figure PCTCN2019102657-appb-000008
表示在当前时间点所得的一步状态 预测,并且有
Figure PCTCN2019102657-appb-000009
C输出预测和新息的计算:
按下式计算输出量的预测值
Figure PCTCN2019102657-appb-000010
以符号e k代表新息,即状态空间输出量的预测误差;e k的计算如下:
Figure PCTCN2019102657-appb-000011
D状态估计更新:
按下式计算并更新当前时间点下的状态估计
Figure PCTCN2019102657-appb-000012
通过滤波器增益L k与新息e k的乘积,修正状态变量的一步预测,从而得到第k个时间点下的状态估计
Figure PCTCN2019102657-appb-000013
向量
Figure PCTCN2019102657-appb-000014
的第二个元素
Figure PCTCN2019102657-appb-000015
即为更新后的SOC的估算值。
作为一种优选的方案,所述噪声统计特性和滤波器增益自适应系统与所述递归式状态估计系统并行进行;所述噪声统计特性和滤波器增益自适应系统用于更新状态预测和状态估计的协方差、以信号自适应的方式估算过程噪声和量测噪声的协方差、以信号自适应的方式更新滤波器的增益。
作为一种优选的方案,所述噪声统计特性和滤波器增益自适应系统的具体计算步骤如下:
I参数矩阵初始化:
Figure PCTCN2019102657-appb-000016
以及
Figure PCTCN2019102657-appb-000017
分别设置状态估计的协方差的初值P 0、过程噪声协方差的初值
Figure PCTCN2019102657-appb-000018
以及量测噪声协方差的初值
Figure PCTCN2019102657-appb-000019
其中符号I 2代表二维单位矩阵,对于任意后续时间点k>0,循环执行以下步骤II至步骤IV;
II进行状态预测协方差和新息协方差估算的更新:
首先,按下式计算并更新状态预测的协方差矩阵
Figure PCTCN2019102657-appb-000020
其中二维矩阵
Figure PCTCN2019102657-appb-000021
是在当前时间点所得的一步状态预测的协方差,而P k-1
Figure PCTCN2019102657-appb-000022
分别代表在前一个时间点所得的状态估计的协方差、和在前一个时间点所估算出的过程噪声的协方 差;
进一步地,新息的协方差通过下式以自适应的方式进行估算:
Figure PCTCN2019102657-appb-000023
其中,M个新息数据{e k-M+1,e k-M+2...,e k}是通过滑动窗口的形式,将从k-M+1时间点到当前时间点k的共计M个新息数据记录在BMS系统的内存当中,并随时间的递进不断更新;
III噪声统计特性和滤波器增益自适应:
首先,按照下式以滑动条平均方法预测量测噪声的协方差:
Figure PCTCN2019102657-appb-000024
其中,符号α代表遗忘因子;
Figure PCTCN2019102657-appb-000025
代表在前一个时间点估算出的量测噪声的协方差;公式(18)中,向量C k定义如下:
Figure PCTCN2019102657-appb-000026
进一步地,按照下式更新当前时间点的滤波器增益:
Figure PCTCN2019102657-appb-000027
最后,按照下式估算当前时间点下的过程噪声的协方差:
Figure PCTCN2019102657-appb-000028
IV更新状态估计的协方差:
在这一步,按照以下公式计算并更新当前时间点下的状态估计的协方差矩阵:
Figure PCTCN2019102657-appb-000029
本发明的有益效果
本发明具有如下优点:
1)高精度的等效电路模型;通过对温度因素的考虑,减少温度对模型参数偏差的影响。
2)高精度的OCV映射表;通过对温度因素的考虑,提高查表精度,保证等效电路模型中端电压估算精度,从而提高SOC估算精度;
3)噪声统计特性具有高度自适应性;减少由于模型精度、量测和过程噪声协方差随时间变化引起的不确定性带来的负面影响,以提高SOC估算的鲁棒性和精度。
4)降低了算法参数调试和难度和所需次数,增强了实际应用的便利性。
5)即使在SOC初始条件不确定的情况下,SOC的在线估算能够很快收敛到真实值附近。
附图说明
图1为本发明所涉及的SOC状态在线估算系统的架构示意图;
图2为建立所述调节系统的操作流程;
图3为所拟合的二元回归模型函数曲面的示例图;
图4为一阶等效电路模型示意;
图5为HPPC数据示意图;
图6为SOC估算系统中的并行逻辑计算流程;
图7为48℃下的10小时工况循环测试结果;
图8为25℃下的10小时工况循环测试结果。
具体实施方式
实施例1:
为使本领域技术人员更好的理解本发明的技术方案,下面参照说明书附图,对本发明进行详细描述。
本发明所述的一种用于48V轻混汽车锂离子电池的荷电状态在线估算系统,包括两个主要组成部分,第一部分为开路电压特性调节系统和等效电路模型参数校准调节系统,可以根据环境温度对开路电压的数值与等效电路模型的参数进行精确校准,以降低由温度变化引起的模型的误差;第二部分为SOC在线估算的实现,包括噪声统计特性、滤波器增益自适应和递归式状态估计系统,能够提高SOC在线估算的鲁棒性与精度。通过有效性验证表明,所提出的方法能够在轻混汽车典型循环工况下对SOC实现高精度,高鲁棒性的在线估算。
本实施例中提出的发明基于如下四个系统实现对SOC的在线估算:
(1)开路电压特性调节系统;(2)等效电路模型参数校准调节系统;(3)噪声统计特性、滤波器增益自适应系统;(4)递归式状态估计系统。
以上四个系统根据功能和用途可以分为参数调节系统和SOC估算系统部分;两个系统之间的关联以及本发明的总体架构视图如图1所示。
其中参数调节系统部分通过减小等效电路模型参数以及开路电压特性由于温度变化引起的偏差,以提高SOC估算的精度;而SOC估算系统用于实现高精度、高鲁棒性的SOC估算的算法数学逻辑。如图1所示,参数调节系统与SOC估算这两部分所需输入信号均来自于BMS所采集的负载电流、端电压、和温度等的测量读数。
为了解决由环境温度变化对参数的标定值所造成的偏差,本发明着重考虑了开路电压特性以及等效电路模型受温度的影响,以提高模型在不同环境下的适用能力。具体而言,参数 调节这一部分功能是用于调节表征SOC与开路电压之间映射关系的数学模型,以及等效电路模型中欧姆内阻、极化内阻、和极化电容数值的校准。
在图2中,居于左侧纵向连接的框图显示了在离线阶段建立参数调节系统所需的操作流程,而右边水平连接的框图则显示了所建立的系统在进行在线SOC估算时的应用方式。
锂离子电池的内部特性会受到环境温度的较大影响,特别是48V系统所用的高功率型锂离子电池。因此,如果仅使用单一温度下的开路电压特性曲线(比如只考虑常温下),则当环境温度发生较大幅度变化时,先前所标定开路电压特性与其实际温度下的相比会发生较大的偏差。这将会引起SOC估算精度的大幅度降低。为解决这一问题,本发明中建立了OCV与SOC以及环境温度的三维映射关系,使所得的开路电压特性更符合实际应用的要求。该三维映射关系的数学本质,是一个关于SOC和环境温度的二元函数,并可通过下式表达
OCV(s,T)      (1)
其中变量s与变量T分别代表SOC的值和环境温度的值。为实现上述三维映射关系,需要按照以下步骤进行实施:
(1)在不同温度下对锂电池进行OCV-SOC特性测试,通过温箱对测试温度实施恒定控制,所测试的温度为-30℃到60℃,测试频率为每隔10℃进行一次测试;对于每一个温度值下的测试,在测试过程中每隔5%SOC收集一组OCV测量值;对所收集到的OCV测量值进行预处理,并记录为对应温度和对应SOC下的OCV值;
(2)将步骤1)所得的测试数据拟合到函数OCV(s,T),建立一个可用二元多项式回归模型表达的三维映射关系;该二元回归模型表达式如下
Figure PCTCN2019102657-appb-000030
通过拟合得到的二元回归模型的三维曲面如图3所示。
同时,OCV关于SOC的导函数、及该导函数对应的回归模型也可以通过公式(2)以如下方式获得:
Figure PCTCN2019102657-appb-000031
(3)把步骤2)获得的二元回归模型的系数储存在BMS系统的软件中;在BMS系统运行时调用所储存的系数,以此方式实时地重构前一步所拟合的回归模型,并用此模型所代表的三维映射关系;进而在BMS系统运行时根据实际测得的温度与实时获得的SOC估算数值,共同确定当前的OCV及其关于SOC的导函数的数值。
为了平衡等效电路模型的复杂度与其能够实现的对电池特性还原的精度,并且考虑到在48V系统的实际工况中负载电流常常为小于30秒的脉冲形式,本实施例选用一阶RC等效电路模型,如图4所示。
在图4中,I L表示负载电流,V t表示电池的端电压;在一阶等效电路模型中,共有三个重要的模型参数,其分别是:欧姆内阻R 0;极化内阻R p;极化电容C p
尽管一阶等效电路模型在实际BMS应用中具有简单有效的特点,但其依然需要其他方法和技术来弥补潜在的精度的下降问题,以便降低SOC估算的误差。在各种可能引起模型参数的实际数值发生快速改变的因素中,环境温度的影响最为强烈。因此,在本发明根据环境温度的变化对参数进行及时校准。为实现这一功能特点,需要按照以下步骤进行实施。
(1)在不同温度下对锂电池进行混合脉冲功率特性测试(即HPPC测试)。测试的温度为-30℃到60℃(每隔10℃进行一次测试),通过温箱对测试温度实施恒定控制。
(2)对于-30℃到60℃(每隔10℃)所获得的每一组HPPC数据进行分析并进行参数辨识,提取电池等效电路模型的参数R 0,R p与C p。所运用的HPPC数据,是指当SOC处于约50%的区间内所记录的端电压和负载电流数据(以放电脉冲操作开始的时刻为起始,以40秒静置结束的时刻为终止)。数据记录的时间起始点和周期长度如图5所示。
(3)把前一步中不同温度下获得的模型参数R 0,R p与C p的数值,分别拟合到三个多项式回归模型中,以此建立模型参数关于温度的映射关系。此处所指三个多项式回归模型具有相同的数学形式,并可通过下式表达。
Figure PCTCN2019102657-appb-000032
公式(4)中,等式的左端项是环境温度的函数,它可代表模型参数,与中的任意一个,等式右端项的系数代表通过拟合得到的回归模型的系数;
(4)把通过前一步获得的三个回归模型的系数储存在BMS系统的软件中。在BMS系统运行时调用所储存的系数,以此方式实时地重构前一步通过拟合所得到的回归模型,进而通过该回归模型所代表的等效电路模型参数R 0,R p与C p与环境温度之间的映射关系,在BMS系统运行时根据实际测得的环境温度,实现等效电路模型参数数值的实时校准。
SOC估算这一部分的作用是根据参数调节功能所提供的OCV-SOC特性,以及等效模型参数,执行48V锂离子电池的SOC在线实时估算算法的逻辑运算。本发明中所提出的的SOC估算是由两个并行计算的系统组成:(1)递归式状态估计;(2)噪声统计特性和滤波器增益 自适应。
如图6所示,这两个系统的逻辑运算流程是并行执行、相辅相成的,具体而言:新息协方差的更新依赖于通过输出预测所得到的新息变量的数值;而状态估计的更新则需要利用更新后的滤波器增益。在此处以及后文中,更新一词的含义是:通过迭代计算,用当前时间点下所得的数值替代前一时间点下所得的代表相同意义的数值。下面就图6所展示的计算流程中每一步所涉及的具体原理和有关计算公式进行详细的介绍。
递归式状态估计系统中用于SOC估算的状态空间模型如下列方程所示:
x k=A kx k-1+B ku k+w k     (5)
y k=h(x k,u k)+v k   (6)
所述公式(5)和公式(6)分别为状态空间模型的系统方程和量测方程,在公式(5)和公式(6)中,下标字母代表离散系统中时间点的索引。具体而言,k指代当前采样时间点t k,k-1则指代前一个采样时间点t k-1,并且有t k=t k-1+Δt。其中Δt即为BMS系统中的采样时间间隔;变量w k与v k分别代表状态空间模型的过程噪声和量测噪声。
所述公式(5)和公式(6)中,变量u k是状态空间模型的输入信号,所述u k的真实物理意义为电池的工作负载电流I L;y k代表状态变量的观测值,所述y k的真实物理意义为电池的端电压;
另外,公式(5)和公式(6)分别为状态空间模型的系统方程和量测方程。在公式(5)中,A k代表状态转移矩阵,B k为输入矩阵。
矩阵A k和B k中的各个元素可由电池等效电路模型参数R 0,R p和C p,按下列公式计算而得:
Figure PCTCN2019102657-appb-000033
Figure PCTCN2019102657-appb-000034
其中符号η和Q所代表的物理意义,分别为库伦效率和电池的当前最大可用容量。在公式(5)中,状态变量x k是一个二维向量,定义如下
x k=[V p,k s k] T   (9)
其中V p,k为当前时间点下电池等效电路模型中并联RC电路上的电压,变量s k代表当前时间点下的SOC值。在公式(5)和公式(6)中,变量u k是状态空间模型的输入信号,其真实的物理意义为电池的工作负载电流I L。在公式(6)中,y k代表状态变量的观测值,其真实的物理意义为电池的端电压。非线性观测函数h(x k,u k)的具体表达形式如下:
h(x k,u k)=OCV(s k,T k)-I LR 0-V p,k    (10)
其中函数OCV(s k,T k)的值即为当环境温度为T k且电池SOC为s k时的OCV数值。最后,变量w k与v k分别代表状态空间模型的过程噪声和量测噪声。递归式状态估计所涉及的算法的具体计算步骤如下:
A初始化
将V p,k的初值V p,0设为零;s k的初始值s 0可任意选取,或设置为在相应BMS系统中所储存的上一次车辆运行结束时所最后记录的SOC估算值;对于任意时间点k>0,循环执行以下步骤B至步骤D;
B一步状态预测:
按下式进行状态变量一步预测的计算
Figure PCTCN2019102657-appb-000035
其中
Figure PCTCN2019102657-appb-000036
代表在前一时间点所得的状态估计的值,
Figure PCTCN2019102657-appb-000037
表示在当前时间点所得的一步状态预测,并且有
Figure PCTCN2019102657-appb-000038
C输出预测和新息的计算:
按下式计算输出量的预测值
Figure PCTCN2019102657-appb-000039
以符号e k代表新息,即状态空间输出量的一步预测误差。e k的计算如下:
Figure PCTCN2019102657-appb-000040
D状态估计更新:
按下式计算并更新当前时间点下的状态估计。
Figure PCTCN2019102657-appb-000041
最后,通过滤波器增益L k与新息e k的乘积,来修正一步状态预测从而得到第k个时间点下的状态估计
Figure PCTCN2019102657-appb-000042
进一步地,向量
Figure PCTCN2019102657-appb-000043
的第二个元素即为更新后的SOC的估算值
Figure PCTCN2019102657-appb-000044
在这一步中,滤波器增益L k是通过与递归状态估计并行的计算流程所得。
噪声统计特性和滤波器增益自适应系统所包含的计算流程与所述递归式状态估计系统并行进行,其所涉及的运算有三个主要用途:
(1)更新状态预测和状态估计的协方差;(2)以信号自适应的方式估算过程噪声和量测噪声的协方差;(3)以信号自适应的方式更新滤波器的增益。
以上三点也是本发明中SOC在线估算所具有的高鲁棒性以及对算法参数自适应能力的来源,其所涉及的计算步骤和计算描述如下:
I参数矩阵初始化:
Figure PCTCN2019102657-appb-000045
以及
Figure PCTCN2019102657-appb-000046
分别设置状态估计的协方差的初值P 0,过程噪声协方差的初值
Figure PCTCN2019102657-appb-000047
以及量测噪声协方差的初值
Figure PCTCN2019102657-appb-000048
其中符号I 2代表二维单位矩阵。对于任意后续时间点k>0,循环执行以下步骤II至步骤IV。
II进行状态预测协方差和新息协方差估算的更新:
首先,按下式计算并更新状态预测的协方差矩阵
Figure PCTCN2019102657-appb-000049
其中二维矩阵
Figure PCTCN2019102657-appb-000050
是在当前时间点所得的一步状态预测的协方差,而P k-1
Figure PCTCN2019102657-appb-000051
分别代表在前一个时间点所得的状态估计的协方差、和在前一个时间点所估算出的过程噪声的协方差。
进一步地,新息的协方差通过下式以自适应的方式进行估算
Figure PCTCN2019102657-appb-000052
其中,M个新息数据{e k-M+1,e k-M+2...,e k}是通过滑动窗口的形式,将从k-M+1时间点到当前时间点k的共计M个新息数据记录在BMS系统的内存当中,并随时间的递进而不断更新。
III噪声统计特性和滤波器增益自适应:
首先,按照下式以滑动条平均方法预测量测噪声的协方差:
Figure PCTCN2019102657-appb-000053
其中,符号α代表遗忘因子,其取值是一个远小于1的正数。符号
Figure PCTCN2019102657-appb-000054
代表在前一个时间点估算出的量测噪声的协方差。另外,向量C k定义如下
Figure PCTCN2019102657-appb-000055
进一步地,按照下式计算当前时间点的滤波器增益
Figure PCTCN2019102657-appb-000056
最后,按照下式进行估算当前时间点下的过程噪声的协方差
Figure PCTCN2019102657-appb-000057
IV更新状态估计的协方差:
在这一步,按照以下公式计算并更新当前时间点下的状态估计的协方差矩阵:
Figure PCTCN2019102657-appb-000058
在评估验证中,采用为48V系统设计的8-Ah NMC锂离子电池,分别在48℃以及25℃下通过运用轻混汽车典型工况循环测试数据(采用频率1.0Hz,共10个小时36,000个点数据)进行仿真分析,以评估所发明的SOC估算方法的精确性。所用工况循环均以WLTC循环为基础获得。
SOC参考值的时间序列是根据试验设备所采集的电流,基于已知初始SOC的安时积分方法进行计算而得。在本发明所涉及的技术领域中,通常认为实验环境下的电流传感器的误差足够小,因此该方法的所获得SOC参考值可用来近似替代SOC的真实值。
用于评价SOC估算精度的标准为常见的平均绝对误差(MAE,Mean Absolute Error)和均方根误差(RMSE,Root Mean Square Error),具体计算方法如式(23)所示:
Figure PCTCN2019102657-appb-000059
其中s k表示在第k个采样时间点的参考SOC值,
Figure PCTCN2019102657-appb-000060
则是根据SOC估算系统得出的在第k个时间点的SOC估算值。另外,N代表所使用的数据点的总数,在本实施例所述的评估验证中其数值为N=36,000。
验证的结果如图7和图8所示,其分别表示环境温度为48℃与25℃下的SOC估算仿真验 证结果。在图7和图8中,位于上方的子图显示了SOC估算值的时间序列与SOC参考值的时间序列之间的对比,SOC估算值是通过运用本发明中所述的SOC在线估算系统所得的;位于下方的子图展示了SOC估算值和SOC参考值之间的绝对误差的时间序列的变化演进。验证结果表明,通过运用本发明中所述的SOC在线估算系统,能够在不同的温度下对48V锂离子电池的SOC进行精确的估算。验证结果的MAE和RMSE具体数值统计如表1。
表1 SOC估算评估验证结果
测试场景 MAE RMSE
48℃工况循环 0.9% 0.016
25℃工况循环 0.8% 0.012
48V系统是传统燃油汽车电气化的一个十分经济有效的解决方案;其能够以相对较低的成本,有效的控制CO 2和污染物的排放以及燃油消耗,具有较高的产业推广价值和广阔的市场潜力。目前已经有一些汽车厂家布局了48V系统相关技术和相应的48V轻混车型。
本发明所提出的SOC在线估算系统,属于混合动力汽车BMS系统的核心技术之一。本发明尤其适用于搭载48V电池系统的轻混汽车,由于其典型工况下的温度变化范围宽、充放电电流交替变化频繁,对在线SOC估算提出了很大的挑战,因此需要额外的措施来保证SOC估算的精度和鲁棒性。本发明旨在降低温度变化和模型不确定性对SOC估算造成的影响,能够实现高精度、高鲁棒性的SOC估算。因此本发明能够成为轻混汽车SOC估算的优良解决方案。同时也可以用于搭载锂离子电池的其他类型的混合动力汽车的SOC在线估算任务。
以上所述仅为本发明的实施例,并非因此限制本发明的专利范围,凡是利用本发明说明书及附图内容所作的等效结构或等效流程变换,或直接或间接运用在其他相关的技术领域,均同理包括在本发明的专利保护范围内。

Claims (9)

  1. 一种用于48V轻混汽车锂离子电池的荷电状态在线估算系统,其特征在于,包括参数调节系统和SOC估算系统;
    所述参数调节系统包括开路电压特性调节系统和等效电路模型参数校准调节系统,所述参数调节系统可以根据环境温度对开路电压的数值与等效电路模型的参数进行精确校准,以降低由温度变化引起的模型误差;
    所述SOC估算系统包括噪声统计特性、滤波器增益自适应和递归式状态估计系统,所述SOC估算系统能够提高SOC在线估算的鲁棒性与精度;
    所述参数调节系统与SOC估算系统所需输入信号均来自于BMS所采集的负载电流、端电压和温度的测量数值。
  2. 根据权利要求1所述的一种用于48V轻混汽车锂离子电池的荷电状态在线估算系统,其特征在于,所述开路电压特性调节系统包括OCV与SOC以及环境温度的三维映射关系,以保证开路电压特性实时调节的精确度;所述三维映射关系可通过下式表达OCV(s,T)          (1)
    其中变量s与变量T分别代表SOC的值和环境温度的值。
  3. 根据权利要求2所述的一种用于48V轻混汽车锂离子电池的荷电状态在线估算系统,其特征在于,所述三维映射关系的建立具体包括以下步骤:
    1)在不同温度下对锂电池进行OCV-SOC特性测试,通过温箱对测试温度实施恒定控制,所测试的温度为-30℃到60℃,测试频率为每隔10℃进行一次测试;对于每一个温度值下的测试,在测试过程中每隔5%SOC收集一组OCV测量值;对所收集到的OCV测量值进行预处理,并记录为对应温度和对应SOC下的OCV值;
    2)将步骤1)所得的测试数据拟合到函数OCV(s,T),建立一个可用二元多项式回归模型表达的三维映射关系;该二元回归模型表达式如下
    Figure PCTCN2019102657-appb-100001
    同时,OCV关于SOC的导函数、及该导函数对应的回归模型也可以通过公式(2)以如下方式获得:
    Figure PCTCN2019102657-appb-100002
    3)把步骤2)获得的二元回归模型的系数储存在BMS系统的软件中;在BMS系统运行时调用所储存的系数,以此方式实时地重构前一步所拟合的回归模型,并用此模型 所代表的三维映射关系;进而在BMS系统运行时根据实际测得的温度与实时获得的SOC估算数值,共同确定当前的0CV及其关于SOC的导函数的数值。
  4. 根据权利要求1所述的一种用于48V轻混汽车锂离子电池的荷电状态在线估算系统,其特征在于,所述等效电路模型参数校准调节系统根据环境温度的变化对等效电路模型中欧姆内阻R 0、极化内阻R p、和极化电容C p数值进行实时校准,具体的包括以下步骤:
    (1)在不同温度下对锂电池进行混合脉冲功率特性测试,通过温箱对测试温度实施恒定控制,所测试的温度为-30℃到60℃,测试频率为每隔10℃进行一次测试;对于-30℃到60℃所获得的每一组数据进行预处理并进行参数辨识,提取电池等效电路模型的参数R 0,R p与C p,所述R 0为欧姆内阻,R p为极化内阻,C p为极化电容;
    (2)把步骤(1)中不同温度下获得的模型参数R 0,R p与C p的数值,分别拟合到三个多项式回归模型中,以此建立对应的模型参数关于温度的映射关系;所述三个多项式回归模型具有相同的数学形式,并可通过下式表达:
    Figure PCTCN2019102657-appb-100003
    公式(4)中,等式的左端项ψ(T)是环境温度的函数,它可代表模型参数R 0,R p与C p中的任意一个,等式右端项的系数{a 0,a 1,…,a n}代表通过拟合得到的回归模型的系数;
    (3)把通过步骤(2)所获得的三个回归模型的系数储存在BMS系统的软件中;在BMS系统运行时调用所储存的系数,以此方式实时地重构前一步通过拟合所得到的回归模型,进而通过该回归模型所代表的等效电路模型参数R 0,R p与C p与环境温度之间的映射关系,在BMS系统运行时根据实际测得的环境温度,实现等效电路模型参数数值的实时校准。
  5. 根据权利要求2所述的一种用于48V轻混汽车锂离子电池的荷电状态在线估算系统,其特征在于,所述SOC估算系统根据参数调节系统所提供的开路电压特性以及等效模型参数,执行48V锂离子电池的SOC在线实时估算,所述SOC估算系统包括并行计算的递归式状态估计系统和噪声统计特性与滤波器增益自适应系统。
  6. 根据权利要求5所述的一种用于48V轻混汽车锂离子电池的荷电状态在线估算系统,其特征在于,递归式状态估计系统中用于SOC估算的状态空间模型如下所示:
    x k=A kx k-1+B ku k+w k          (5)
    y k=h(x k,u k)+v k      (6)
    所述公式(5)和公式(6)分别为状态空间模型的系统方程和量测方程,公式(5)和公式(6)中下标字母k代表离散系统中时间点的索引;所述k指代当前采样时间点t k,k-1指代前一采样时间点t k-1,并有t k=t k-1+Δt,其中Δt即为BMS系统中的采样时间间隔;变量w k与v k分别代表状态空间模型的过程噪声和量测噪声;
    所述公式(5)和公式(6)中,变量u k是状态空间模型的输入信号,所述u k的真实物理意义为电池的工作负载电流I L;y k代表状态变量的观测值,所述y k的真实物理意义为电池的端电压;
    所述公式(5)和公式(6)中,A k代表状态转移矩阵,B k为输入矩阵,矩阵A k和B k中的各个元素可由电池等效电路模型参数R 0,R p和C p,按下列公式计算而得:
    Figure PCTCN2019102657-appb-100004
    Figure PCTCN2019102657-appb-100005
    其中符号η和Q所代表的物理意义,分别为库伦效率和电池的当前最大可用容量;
    在公式(5)中,状态变量x k是一个二维向量,定义如下
    x k=[V p,k s k] T      (9)
    其中变量V p,k代表当前时间点下电池等效电路模型中并联RC电路两端的电压,变量s k代表当前时间点下的SOC值;
    所述公式(6)中,非线性函数h(x k,u k)的具体表达形式如下:
    h(x k,u k)=OCV(s k,T k)-I LR 0-V p,k     (10)
    其中函数OCV(s k,T k)的值即代表当环境温度为T k且电池SOC为s k时的开路电压数值。
  7. 根据权利要求5所述的一种用于48V轻混汽车锂离子电池的荷电状态在线估算系统,其特征在于,所述递归式状态估计系统的具体计算步骤如下:
    A状态估计初始化
    将V p,k的初值V p,0设为零;s k的初始值s 0可任意选取,或设置为在相应BMS系统中所储存的上一次车辆运行结束时所最后记录的SOC估算值;对于任意时间点k>0,循 环执行以下步骤B至步骤D;
    B一步状态预测:
    按下式进行状态变量一步预测的计算
    Figure PCTCN2019102657-appb-100006
    其中
    Figure PCTCN2019102657-appb-100007
    代表在前一时间点所得的状态估计的值,
    Figure PCTCN2019102657-appb-100008
    表示在当前时间点所得的一步状态预测,并且有
    Figure PCTCN2019102657-appb-100009
    C输出预测和新息的计算:
    按下式计算输出量的预测值
    Figure PCTCN2019102657-appb-100010
    以符号e k代表新息,即状态空间输出量的预测误差;e k的计算如下:
    Figure PCTCN2019102657-appb-100011
    D状态估计更新:
    按下式计算并更新当前时间点下的状态估计
    Figure PCTCN2019102657-appb-100012
    通过滤波器增益L k与新息e k的乘积,修正状态变量的一步预测,从而得到第k个时间点下的状态估计
    Figure PCTCN2019102657-appb-100013
    向量
    Figure PCTCN2019102657-appb-100014
    的第二个元素
    Figure PCTCN2019102657-appb-100015
    即为更新后的SOC的估算值。
  8. 根据权利要求6所述的一种用于48V轻混汽车锂离子电池的荷电状态在线估算系统,其特征在于,所述噪声统计特性和滤波器增益自适应系统与所述递归式状态估计系统并行进行;所述噪声统计特性和滤波器增益自适应系统用于更新状态预测和状态估计的协方差、以信号自适应的方式估算过程噪声和量测噪声的协方差、以信号自适应的方式更新滤波器的增益。
  9. 根据权利要求8所述的一种用于48V轻混汽车锂离子电池的荷电状态在线估算系统,其特征在于,所述噪声统计特性和滤波器增益自适应系统的具体计算步骤如下:
    I参数矩阵初始化:
    按P 0=θ 1·I 2
    Figure PCTCN2019102657-appb-100016
    以及
    Figure PCTCN2019102657-appb-100017
    分别设置状态估计的协方差的初值P 0、过程噪声协方差的初值
    Figure PCTCN2019102657-appb-100018
    以及量测噪声协方差的初值
    Figure PCTCN2019102657-appb-100019
    其中符号I 2代表二维单 位矩阵,对于任意后续时间点k>0,循环执行以下步骤II至步骤IV;
    II进行状态预测协方差和新息协方差估算的更新:
    首先,按下式计算并更新状态预测的协方差矩阵
    Figure PCTCN2019102657-appb-100020
    其中二维矩阵
    Figure PCTCN2019102657-appb-100021
    是在当前时间点所得的一步状态预测的协方差,而P k-1
    Figure PCTCN2019102657-appb-100022
    分别代表在前一个时间点所得的状态估计的协方差、和在前一个时间点所估算出的过程噪声的协方差;
    进一步地,新息的协方差通过下式以自适应的方式进行估算:
    Figure PCTCN2019102657-appb-100023
    其中,M个新息数据{e k-M+1,e k-M+2…,e k}是通过滑动窗口的形式,将从k-M+1时间点到当前时间点k的共计M个新息数据记录在BMS系统的内存当中,并随时间的递进不断更新;
    III噪声统计特性和滤波器增益自适应:
    首先,按照下式以滑动条平均方法预测量测噪声的协方差:
    Figure PCTCN2019102657-appb-100024
    其中,符号α代表遗忘因子;
    Figure PCTCN2019102657-appb-100025
    代表在前一个时间点估算出的量测噪声的协方差;公式(18)中,向量C k定义如下:
    Figure PCTCN2019102657-appb-100026
    进一步地,按照下式更新当前时间点的滤波器增益:
    Figure PCTCN2019102657-appb-100027
    最后,按照下式估算当前时间点下的过程噪声的协方差:
    Figure PCTCN2019102657-appb-100028
    IV更新状态估计的协方差:
    在这一步,按照以下公式计算并更新当前时间点下的状态估计的协方差矩阵:
    Figure PCTCN2019102657-appb-100029
PCT/CN2019/102657 2019-08-27 2019-08-27 用于48v轻混汽车锂离子电池的荷电状态在线估算系统 WO2021035500A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/CN2019/102657 WO2021035500A1 (zh) 2019-08-27 2019-08-27 用于48v轻混汽车锂离子电池的荷电状态在线估算系统
CN201980034166.6A CN112601968A (zh) 2019-08-27 2019-08-27 用于48v轻混汽车锂离子电池的荷电状态在线估算系统

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2019/102657 WO2021035500A1 (zh) 2019-08-27 2019-08-27 用于48v轻混汽车锂离子电池的荷电状态在线估算系统

Publications (1)

Publication Number Publication Date
WO2021035500A1 true WO2021035500A1 (zh) 2021-03-04

Family

ID=74684897

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/102657 WO2021035500A1 (zh) 2019-08-27 2019-08-27 用于48v轻混汽车锂离子电池的荷电状态在线估算系统

Country Status (2)

Country Link
CN (1) CN112601968A (zh)
WO (1) WO2021035500A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113496009A (zh) * 2021-06-28 2021-10-12 北京控制工程研究所 卫星太阳光压力矩高精度在线估计方法

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113495214B (zh) * 2021-05-25 2023-07-07 四川轻化工大学 一种基于温度变化模型的超级电容荷电状态估计方法

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103675706A (zh) * 2013-12-13 2014-03-26 桂林电子科技大学 一种动力电池电荷量估算方法
CN104569835A (zh) * 2014-12-16 2015-04-29 北京理工大学 一种估计电动汽车的动力电池的荷电状态的方法
CN106054081A (zh) * 2016-06-17 2016-10-26 合肥工业大学智能制造技术研究院 一种用于电动汽车动力电池soc估计的锂电池建模方法
CN107589379A (zh) * 2017-08-30 2018-01-16 电子科技大学 一种在线估计锂电池soc和阻抗的方法
CN108594135A (zh) * 2018-06-28 2018-09-28 南京理工大学 一种用于锂电池均衡充放电控制的soc估算方法
CN108761340A (zh) * 2018-05-31 2018-11-06 天津工业大学 基于噪声干扰的强跟踪容积卡尔曼滤波的电池估算方法

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106324521B (zh) * 2016-09-05 2018-09-11 北京理工大学 一种联合估计动力电池系统参数与荷电状态的方法
KR102160274B1 (ko) * 2017-09-07 2020-09-25 주식회사 엘지화학 배터리 충전 상태 추정 장치 및 방법
CN108445408A (zh) * 2018-03-20 2018-08-24 重庆大学 一种基于参数估计ocv的全温度soc估计方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103675706A (zh) * 2013-12-13 2014-03-26 桂林电子科技大学 一种动力电池电荷量估算方法
CN104569835A (zh) * 2014-12-16 2015-04-29 北京理工大学 一种估计电动汽车的动力电池的荷电状态的方法
CN106054081A (zh) * 2016-06-17 2016-10-26 合肥工业大学智能制造技术研究院 一种用于电动汽车动力电池soc估计的锂电池建模方法
CN107589379A (zh) * 2017-08-30 2018-01-16 电子科技大学 一种在线估计锂电池soc和阻抗的方法
CN108761340A (zh) * 2018-05-31 2018-11-06 天津工业大学 基于噪声干扰的强跟踪容积卡尔曼滤波的电池估算方法
CN108594135A (zh) * 2018-06-28 2018-09-28 南京理工大学 一种用于锂电池均衡充放电控制的soc估算方法

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113496009A (zh) * 2021-06-28 2021-10-12 北京控制工程研究所 卫星太阳光压力矩高精度在线估计方法
CN113496009B (zh) * 2021-06-28 2023-07-28 北京控制工程研究所 卫星太阳光压力矩高精度在线估计方法

Also Published As

Publication number Publication date
CN112601968A (zh) 2021-04-02

Similar Documents

Publication Publication Date Title
Xu et al. State of charge estimation for lithium-ion batteries based on adaptive dual Kalman filter
CN111581904B (zh) 考虑循环次数影响的锂电池soc及soh协同估算方法
Xiong et al. A robust state-of-charge estimator for multiple types of lithium-ion batteries using adaptive extended Kalman filter
CN109188293B (zh) 基于新息协方差带渐消因子的ekf锂离子电池soc估算方法
CN110196393B (zh) 一种锂电池荷电状态、能量状态和功率状态的联合在线估计方法
CN106842060A (zh) 一种基于动态参数的动力电池soc估算方法及系统
CN111679199B (zh) 一种锂离子电池soc估计方法及装置
CN110824363B (zh) 一种基于改进ckf的锂电池soc和soe联合估算方法
CN108828448B (zh) 基于充电电压曲线融合卡尔曼滤波的电池荷电状态在线估算方法
CN108445422B (zh) 基于极化电压恢复特性的电池荷电状态估算方法
CN112595979B (zh) 一种考虑非充分激励的锂电池参数在线辨识方法及系统
CN112285570B (zh) 一种基于衰减记忆滤波器的电动汽车故障诊断方法
CN105223487A (zh) 一种锂离子电池的多状态解耦估计方法
WO2018188321A1 (zh) 一种增强电池状态估计鲁棒性的方法
CN111142025A (zh) 一种电池soc估算方法、装置、存储介质及电动汽车
CN112528472A (zh) 一种多新息混合Kalman滤波和H无穷滤波算法
CN111751750B (zh) 基于模糊ekf的多阶段闭环锂电池soc估算方法
WO2021035500A1 (zh) 用于48v轻混汽车锂离子电池的荷电状态在线估算系统
CN113190969A (zh) 一种基于信息评估机制的锂电池模型参数辨识方法
CN115327415A (zh) 基于限定记忆递推最小二乘算法的锂电池soc估算方法
CN113608126B (zh) 一种不同温度下的锂电池soc在线预估方法
CN113625174B (zh) 一种锂离子电池soc与容量联合估计方法
CN112415412A (zh) 估算电池soc值的方法和装置及车辆、存储介质
Chen et al. Estimation of state of charge for lithium-ion battery considering effect of aging and temperature
CN113759258A (zh) 一种动力电池soc估计方法、装置及纯电动汽车

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19943288

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19943288

Country of ref document: EP

Kind code of ref document: A1