CN113341330B - Lithium-sulfur power battery SOC estimation method based on OCV correction and Kalman filtering algorithm - Google Patents

Abstract

The invention discloses a lithium sulfur power battery SOC estimation method based on OCV correction and Kalman filtering algorithm, which comprises the following steps: the method comprises the following steps: OCV static correction; step two: and (5) Kalman filtering online estimation. The lithium-sulfur power battery SOC estimation method based on OCV correction and Kalman filtering algorithm combines the characteristics of the lithium-sulfur power battery, is based on ampere-hour integration, performs static correction according to the characteristics of an OCV-SOC curve and the battery standing time length during power-on, performs Kalman filtering dynamic correction on the SOC obtained by the ampere-hour integration according to the relation between dynamic voltage and SOC and transient current and the power-on time length during the charging and discharging process, and can effectively improve the SOC estimation precision of the lithium-sulfur power battery.

Description

Lithium-sulfur power battery SOC estimation method based on OCV correction and Kalman filtering algorithm

Technical Field

The invention belongs to the technical field of battery management systems, and particularly relates to a lithium-sulfur power battery SOC estimation method based on OCV correction and a Kalman filtering algorithm.

Background

An online estimation algorithm of SOC in a Battery Management System (BMS) is one of the core algorithms. The accuracy of SOC has a great influence on the energy utilization of the battery, the energy management strategy, the equalization algorithm, the service life, etc. Endurance mileage and charge-discharge efficiency have always been the most concerned issues for electric vehicles. For an electric vehicle with a driving range of 500km and an SOC online estimation accuracy of only 10%, under the condition that other factors are not considered, the driving range actually shown by the vehicle may be only 450km, and is less than 50km; in addition, the inaccurate SOC can erroneously limit the charging and discharging efficiency during the charging and discharging process of the battery, which affects the use experience, and more seriously, the battery has the risks of overcharge and overdischarge, greatly affects the service life of the battery, and even has the risks of causing fire, explosion and other safety accidents.

The current sensor precision that uses on whole car at present is not high, and the maximum precision of common sensor can only reach 1%, continues to promote the improvement development cost that the precision then can be very big under this magnitude, and is not practical. Therefore, when ampere-hour integration is performed, the accumulated integral error caused by current precision cannot be ignored, if correction is not performed in time, the practical performance of the battery is greatly influenced, and the battery can be overcharged or overdischarged in serious cases.

As shown in fig. 1, the OCV curve of the lithium-sulfur battery shows a large slope in both end regions of low SOC (0% to 10%) and high SOC (70% to 100%), and the SOC curve in the middle region (10% to 70%) is gentle and has almost no slope. In consideration of the error of cell voltage sampling, obtaining the SOC by observing the battery voltage or correcting the SOC has a good effect only in a low SOC region and a high SOC region, while in a common middle SOC region, it is difficult to achieve. If only the ampere-hour integration is carried out in the middle SOC region, the correction is not considered at all, and the correction is carried out suddenly when the high SOC or the low SOC region is entered, the SOC is likely to have the situation of sudden change of the SOC due to large accumulated error in the ampere-hour integration process, and the use experience is greatly influenced. By comprehensively considering the objective constraints, it is difficult to accurately estimate the SOC of the lithium-sulfur battery on line in real time.

Disclosure of Invention

In view of this, the present invention provides a lithium sulfur power battery SOC estimation method based on OCV correction and kalman filter algorithm, which can effectively improve SOC estimation accuracy.

In order to achieve the purpose, the invention provides the following technical scheme:

a lithium sulfur power battery SOC estimation method based on OCV correction and Kalman filtering algorithm comprises the following steps:

the method comprises the following steps: OCV static correction

11 After the whole vehicle is awakened, judging whether the power-on is the initial power-on; if so, adopting default values for the SOC value and the standing time; if not, reading the SOC value stored when the power is off last time, and acquiring the standing time;

12 Judging the validity of the cell voltage; if yes, executing step 13); if the effective cell voltage data are not obtained within the set time range, taking the SOC value obtained after the whole vehicle is awakened as the SOC (k 0) after the OCV static correction, and executing the step 14);

13 Calculating an SOC range correction coefficient and a standing time correction coefficient according to an SOC value and standing time obtained after the whole vehicle is awakened; calculating the SOC under the OCV curve, and modifying the SOC according to the SOC range correction coefficient and the standing time correction coefficient to obtain the SOC (k 0) after OCV static correction;

14 Output the SOC (k 0) after OCV static correction;

step two: kalman filtering on-line estimation

21 The OCV static corrected SOC (k 0) is taken as the initial SOC (k-1) estimated on line;

22 To determine the validity of the current measurement during the period from time k-1 to time k;

if not, making SOC (k) = SOC (k-1), and executing step 26);

if yes, executing step 23);

23 Estimate SOC by ampere-hour integration and obtain SOC _ah (k)；

24 Judging the validity of the voltage measurement value at the moment k;

if not, directly connecting the SOC _ah (k) Step 26) is executed as SOC (k) at time k;

if yes, executing step 25);

25 To SOC) _ah (k) Performing Kalman filtering processing to obtain SOC (k) at the moment k;

26 Taking SOC (k) as the SOC estimation value of the lithium sulfur power battery at the k moment, and outputting the SOC (k):

27 Judging whether the whole vehicle is powered off; if yes, storing SOC (k); if not, enabling k = k +1, and circulating the step 22) to the step 27) until the whole vehicle is powered off.

Further, in the step 13), the standing time correction coefficient is:

wt _sleeptm ＝func _sleeptm (T _sleep )

the SOC range correction coefficient is as follows:

wt _ocvsocrange ＝futnc _ocvsocrange (SOC _READ )

wherein, wt _sleeptm Correcting the coefficient for the standing time; t is a unit of _sleep The standing time is the time interval from last power-off of the whole vehicle to the awakening at this time; func _sleeptm () The mapping relation between the standing time and the standing time correction coefficient is obtained; wt. of _ocvsocrange Correcting coefficient for SOC range;SOC _READ the SOC value is stored for the last time of power down; func _socrange () And correcting the mapping relation between the read SOC and the SOC range.

Further, the SOC under the OCV curve is:

SOC _OCV ＝func _OCV (V)

the SOC (k 0) after OCV static correction is:

wherein V is the cell voltage; func _OCV () An OCV-SOC mapping curve relation is obtained; SOC (system on chip) _OCV The SOC value is obtained by the cell voltage through an OCV-SOC mapping curve relation.

Further, in the step 23),

n(k-1)＝func _eff (I(k-1)，SOC(k-1))

therein, SOC _ah (k) Obtaining estimated SOC according to ampere-hour integration at the moment k; SOC (k-1) is the SOC value at the k-1 moment; eta (k-1) is the coulombic efficiency at the time k-1, which is substantially constant during discharge and is affected by the SOC during charge; i (k-1) is the current from the moment k-1 to the moment k; Δ T is the time difference between time k and time k-1; c _rated Represents a rated capacity of the battery; func _eff () A representation represents a mapping of coulombic efficiency to current I and state of charge SOC.

Further, in the step 24), when the voltage measurement value at the time k is invalid, in order to avoid abrupt change of the SOC due to measurement error or correction process, a maximum value of increase or decrease of the SOC at each step is set, and then:

SOC(k)＝func _socramp (SOC _ah (k))

wherein, func _socramp () Is a function that limits the variation of the SOC.

Further, the steps25 In for SOC) _ah (k) The method for performing kalman filtering processing is as follows:

251 Based on SOC _ah (k) And the current excitation I (k) obtains a predicted instantaneous voltage corresponding to the predicted SOC:

V _est (k)＝func _Inst (SOC _ah (k)，I(k))

wherein, V _est (k) The predicted instantaneous voltage at the time k; func _Inst () The mapping relation of the function can be obtained by carrying out constant-current charge and discharge tests on batteries under different SOC under the experimental environment as the estimation function of the instantaneous voltage;

252 Computing kalman filter coefficients:

P-(k)＝G(k)P+(k-1)GT(k)+W

K(k)＝P-(k)HT(k)(H(k)P-(k)HT(k)+Q)-1

P+(k)＝(I-K(k)H(k))P-(k)

wherein G (k) is the derivation of the state estimation equation; GT (k) is the transposed vector of the G (k) vector; h (k) is the derivation of the observation prediction equation; p- (k) is an error covariance prior estimation at the moment k; p + (k) and P + (k-1) are respectively the error covariance posterior estimation at the k moment and the k-1 moment; they reflect the degree of difference between the estimated value and the true value of the state variable; w is the observed noise covariance; q is the system noise covariance; k (K) is a Kalman filter coefficient;

253 Because the error generated in the ampere-hour integration process is accumulated along with the power-on time, a power-on time correction coefficient is introduced when Kalman filtering is carried out:

wt _pwrontm (k)＝func _pwronwt (T _poweron (k))

T _poweron (k)＝T _poweron (k-1)+ΔT

wherein, T _poweron (k-1) and T _poweron (k) Respectively the power-on time at the k-1 moment and the k moment; func _pwronwt () Correcting the mapping relation between the power-on time correction coefficient and the power-on time; t is _poweron (k) The power-on time correction coefficient is in the range of 0-1; wt. of _pwrontm (k) Representing a correction factor based on the power-on time;

taking the characteristics of the lithium-sulfur battery into consideration, introducing an SOC range correction coefficient during filtering:

wt _socrange (k)＝func _socrange (SOC _ah (k))

wherein, func _socrange () Correcting coefficients and SOC for SOC range _ah (k) The mapping relationship between the two; wt. of _socrange (k) To estimate SOC;

the SOC at the k time obtained finally is:

SOC _kalm (k)＝SOC _ah (k)+K(k)*wt _pwrontm (k)*wt _socrange (k)*(V _sample (k)-V _est (k))

wherein, V _sample (k) Measuring the obtained cell voltage at the moment k; SOC _kalm (k) Obtaining SOC after Kalman filtering correction at the moment k; k (K) represents a kalman filter coefficient.

Further, in the step 253), in order to avoid abrupt change of the SOC caused by measurement error or correction process, a maximum value of increase or decrease of the SOC at each step is set, and then:

SOC(k)＝func _socramp (SOC _kalm (k))

wherein, func _socramp () Is a function that limits the SOC variation.

Further, in the step 251), the mapping relationships between the discharging and charging are respectively recorded as func _{Inst_dch} () And func _{Inst_chg} () Based on SOC _ah (k) The predicted instantaneous voltage corresponding to the predicted SOC obtained from the current excitation I (k) is:

wherein, func _{Inst_dch} () Indicating the predicted instantaneous voltage V during discharge _est (k) And SOC _ah (k) And I (k); func _{Inst_chg} () Indicating the predicted instantaneous voltage V during charging _est (k) And SOC _ah (k) And I (k).

The invention has the beneficial effects that:

the lithium-sulfur power battery SOC estimation method based on OCV correction and Kalman filtering algorithm combines the characteristics of the lithium-sulfur power battery, is based on ampere-hour integration, performs static correction according to the characteristics of an OCV-SOC curve and the battery standing time length during power-on, performs Kalman filtering dynamic correction on the SOC obtained by the ampere-hour integration according to the relation between dynamic voltage and SOC and transient current and the power-on time length during the charging and discharging process, and can effectively improve the SOC estimation precision of the lithium-sulfur power battery.

Drawings

In order to make the object, technical scheme and beneficial effect of the invention more clear, the invention provides the following drawings for explanation:

FIG. 1 is a plot of OCV-SOC for a lithium sulfur battery;

FIG. 2 is a flowchart of OCV static correction according to the present embodiment;

FIG. 3 is a flow chart of Kalman filtering on-line estimation in the embodiment;

fig. 4 is a schematic diagram illustrating a connection manner of battery cells of a battery pack;

FIG. 5 is a graph of an OCV-SOC mapping;

FIG. 6 is a graph of instantaneous voltage mapping relationship, (a) discharge process mapping relationship; (b) a charging process mapping relationship;

FIG. 7 is a OCV-SOC correction coefficient map, (a) sleep time (rest time) correction coefficient map; (b) SOC range correction factor mapping;

FIG. 8 is an EKF correction factor map, (a) run time correction factor map; (b) SOC range correction factor mapping;

FIG. 9 is a flowchart of a verification method for SOC online estimation algorithm;

FIG. 10 is a demanded power spectrum for UDDS, WLTC, HIWAY, NEDC operating conditions;

FIG. 11 is a combined operating mode power demand (Preq), (a) NEDC; (b) HIWAY; (c) WLTC; (d) mixed working condition;

FIG. 12 is a battery pack condition testing platform;

FIG. 13 shows measured current and voltage for continuous NEDC combined operation, (a) current; (b) a voltage;

FIG. 14 is a graph of measured voltage for continuous HIWAY combined conditions, (a) current; (b) a voltage;

FIG. 15 shows measured voltage (a) current for continuous WLTC combined conditions; (b) a voltage;

FIG. 16 is a graph of measured voltage, (a) current for a hybrid driving combination; (b) a voltage;

FIG. 17 is a combined operating mode true SOC (SOCreal), (a) NEDC; (b) HIWAY; (c) WLTC; (d) mixed working condition;

FIG. 18 is an algorithm validation test platform;

FIG. 19 is a continuous NEDC combined condition algorithm SOC test result;

FIG. 20 shows the SOC test results of the continuous HIWAY combined condition algorithm;

FIG. 21 shows SOC test results of continuous WLTC combined condition algorithm;

FIG. 22 is a hybrid driving condition algorithm SOC test result;

FIG. 23 illustrates combined condition algorithm SOC error, (a) NEDC; (b) HIWAY; (c) WLTC; (d) mixed working condition.

Detailed Description

The present invention is further described below in conjunction with the drawings and the embodiments so that those skilled in the art can better understand the present invention and can implement the present invention, but the embodiments are not to be construed as limiting the present invention.

The lithium sulfur power battery SOC estimation method based on OCV correction and Kalman filtering algorithm of the embodiment comprises the following steps.

The method comprises the following steps: OCV static correction

As shown in fig. 2, which is a flowchart of the OCV static correction of the present embodiment, the OCV static correction of the present embodiment includes the following steps.

11 After the whole vehicle is awakened, whether the power-on is the initial power-on is judged; if so, adopting default values for the SOC value and the standing time; if not, reading the SOC value stored in the last power-off process, and acquiring the standing time through a built-in clock chip of the controller.

12 Judge the validity of the cell voltage;

if yes, executing step 13);

if the effective cell voltage data are not obtained within the set time range, the SOC value obtained after the whole vehicle is awakened is used as the SOC (k 0) after OCV static correction, namely:

SOC(k0)＝SOC _READ

therein, SOC _READ Reading the SOC value obtained from the storage space of the controller and stored when the controller enters sleep last time;

specifically, the set time range of the present embodiment is 500ms, that is, if valid data cannot be obtained even after 500ms, the SOC value obtained after the entire vehicle is waken up is taken as the SOC (k 0) after OCV static correction; step 14) is performed.

13 Calculating an SOC range correction coefficient and a standing time correction coefficient according to an SOC value and standing time obtained after the whole vehicle is awakened; specifically, the correction coefficient of the standing time is as follows:

wt _sleeptm ＝func _sleeptm (T _sleep )

the SOC range correction coefficient is:

wt _ocvsocrange ＝func _ocvsocrange (SOC _READ )

wherein, wt _sleeptm A correction coefficient for the standing time; t is _sleep The standing time is the time interval from last power-off of the whole vehicle to the awakening at this time; func _sleeptm () The mapping relation between the standing time and the standing time correction coefficient is obtained; wt. of _ocvsocrange Correcting coefficient for SOC range; SOC (system on chip) _READ The SOC value stored for the last power-off is read; func _socrange () And correcting the mapping relation between the read SOC and the SOC range.

Calculating the SOC under the OCV curve, and modifying the SOC according to the SOC range correction coefficient and the standing time correction coefficient to obtain the SOC (k 0) subjected to OCV static correction; specifically, the SOC under the OCV curve is:

SOC _OCV ＝func _OCV (V)

the SOC (k 0) after OCV static correction is:

wherein V is the cell voltage; func _OCV () An OCV-SOC mapping curve relation is obtained; SOC _OCV The SOC value is obtained by the cell voltage through an OCV-SOC mapping curve relation.

14 Outputs the SOC (k 0) statically corrected by the OCV.

Step two: kalman filtering on-line estimation

Fig. 3 is a flow chart of the kalman filter online estimation of the present embodiment. The online kalman filter estimation method of the present embodiment includes the following steps:

21 OCV statically corrected SOC (k 0) as an initial SOC (k-1) of the online estimation, that is:

SOC(k-1)＝SOC(k0)

where SOC (k-1) is the initial value for online estimation.

22 To determine the validity of the current measurement during the period from time k-1 to time k;

if not, making SOC (k) = SOC (k-1), executing step 26);

if so, go to step 23).

23 Estimate SOC by ampere-hour integration and obtain SOC _ah (k) Namely:

n(k-1)＝func _eff (I(k-1)，SOC(k-1))

therein, SOC _ah (k) Obtaining estimated SOC according to ampere-hour integration at the moment k; when SOC (k-1) is k-1The SOC value of the etching; eta (k-1) is the coulombic efficiency at the time k-1, which is substantially constant during discharge and is affected by the SOC during charge; i (k-1) is the current from the time k-1 to the time k; Δ T is the time difference between the time k and the time k-1; c _rated Represents the rated capacity of the battery; func _eff () A representation represents a mapping of coulombic efficiency to current I and state of charge SOC.

24 Judging the validity of the voltage measurement value at the moment k;

if not, directly connecting the SOC _ah (k) Step 26) is executed as SOC (k) at time k;

if so, go to step 25).

Preferably, when the voltage measurement value at the time k is invalid, in order to avoid sudden change of the SOC due to measurement error or correction process, the maximum value of the increase or decrease of the SOC at each step is set, and then:

SOC(k)＝func _socramp (SOC _ah (k))

wherein, func _socramp () Is a function that limits the SOC variation.

25 To SOC _ah (k) And performing Kalman filtering processing to obtain the SOC (k) at the k moment. The present embodiment is to SOC _ah (k) The method for performing kalman filtering processing is as follows:

251 Based on SOC _ah (k) And the current excitation I (k) obtains a predicted instantaneous voltage corresponding to the predicted SOC:

V _est (k)＝func _Inst (SOC _ah (k)，I(k))

wherein, V _est (k) The predicted instantaneous voltage at the time k; func _Inst () The mapping relation of the function can be obtained by performing constant-current charge and discharge tests on batteries under different SOC (state of charge) under an experimental environment as an estimation function of instantaneous voltage.

Specifically, the mapping relationships between discharge and charge are respectively denoted as func _{Inst_dch} () And func _{Inst_chg} () Based on SOC _ah (k) The predicted instantaneous voltage corresponding to the predicted SOC obtained from the current excitation I (k) is:

wherein, func _{Inst_dch} () Indicating the predicted instantaneous voltage V during discharge _est (k) And SOC _ah (k) And I (k); func _{Inst_chg} () Representing the predicted instantaneous voltage V during charging _est (k) And SOC _ah (k) And I (k).

252 Computing kalman filter coefficients:

P-(k)＝G(k)P+(k-1)GT(k)+W

K(k)＝P-(k)HT(k)(H(k)P-(k)HT(k)+Q)-1

P ⁺ (k)＝(I-K(k)H(k))P-(k)

g (k) is the derivation of a state estimation equation, GT (k) is the transposed vector of a G (k) vector, H (k) is the derivation of an observation estimation equation, P- (k) is the error covariance prior estimation of a k moment, P + (k) and P + (k-1) are the error covariance posterior estimation of the k moment and the k-1 moment respectively, and the difference degree between the estimation value and the true value of a state variable is reflected. W is the observed noise covariance and Q is the system noise covariance. K (K) is a Kalman filter coefficient.

253 Because the error generated in the ampere-hour integration process is accumulated along with the power-on time, a power-on time correction coefficient is introduced when Kalman filtering is carried out:

wt _pwrontm (k)＝func _pwronwt (T _poweron (k))

T _poweron (k)＝T _poweron (k-1)+ΔT

wherein, T _poweron (k-1) and T _poweron (k) The power-on time of k-1 time and k time respectively；func _pwronwt () Correcting the mapping relation between the power-on time and the power-on time; t is _poweron (k) The power-on time correction coefficient is in the range of 0-1; wt. of _pwrontm (k) A correction coefficient based on the power-on time is represented;

taking the characteristics of the lithium-sulfur battery into consideration, introducing an SOC range correction coefficient during filtering:

wt _socrange (k)＝func _socrange (SOC _ah (k))

wherein, func _socrange () Correcting coefficients and SOC for SOC range _ah (k) The mapping relationship between the two; wt. of _socrange (k) To estimate the SOC;

the SOC at time k obtained finally is:

SOC _kalm (k)＝SOC _ah (k)+K(k)*wt _pwrontm (k)*wt _socrange (k)*(V _sample (k)-V _est (k))

wherein, V _sample (k) The cell voltage measured at the moment k; SOC _kalm (k) Obtaining SOC after Kalman filtering correction at the moment k; k (K) represents a kalman filter coefficient.

At this time, the SOC can be _kalm (k) As the SOC estimation value SOC (k) of the lithium sulfur power battery at the time k, but in order to avoid SOC abrupt change due to measurement error or correction process, the maximum value of SOC increase or decrease at each step is set, then:

SOC(k)＝func _socramp (SOC _kalm (k))

wherein, func _socramp () Is a function that limits the SOC variation.

26 The SOC (k) is taken as the SOC estimation value of the lithium sulfur power battery at the k moment, and the SOC (k) is output.

27 Judging whether the whole vehicle is powered off; if yes, storing SOC (k); if not, enabling k = k +1, and circulating the step 22) to the step 27) until the whole vehicle is powered off.

In the embodiment, the lithium sulfur power battery SOC estimation method based on OCV correction and Kalman filtering algorithm combines the characteristics of the lithium sulfur power battery, static correction is carried out according to the characteristics of an OCV-SOC curve and the standing time of the battery during power-on the basis of ampere-hour integration, kalman filtering dynamic correction is carried out on the SOC obtained by the ampere-hour integration according to the relation between dynamic voltage and SOC and transient current and the length of power-on time during charging and discharging, and the estimation precision of the SOC of the lithium sulfur power battery can be effectively improved.

In order to illustrate the technical effect of the method for estimating the SOC of the lithium sulfur power battery based on the OCV correction and the kalman filter algorithm according to the present embodiment, the following description is made in detail with reference to specific examples.

In consideration of the use and practical significance of the SOC, the online SOC estimation algorithm is formulated based on the following objectives:

a. the maximum application of the battery capacity is ensured as much as possible on the premise of ensuring that the battery is not overcharged and overdischarged;

b. the overall SOC accuracy is improved.

Four types of common online SOC estimation methods have respective advantages and disadvantages. On the whole, the OCV open-loop voltage method is only suitable for initialization correction when the battery starts to be used, ampere-hour integration can be used as a state equation and together with Kalman filtering, and good accuracy of SOC online estimation can be guaranteed. Neural network algorithms require a large amount of data support and are still in the research phase.

Considering the goal of the SOC online estimation algorithm, and integrating the above characteristics of the algorithm, the present embodiment provides a comprehensive algorithm:

a. considering the standing time and the current SOC range when the whole vehicle is started, and carrying out OCV weighted static correction on the initial SOC;

b. in the running process of the whole vehicle, ampere-hour integration is used as a state estimation equation, an observation estimation equation is established through instantaneous voltage obtained by instantaneous current, the running time of the whole vehicle and the current SOC range are considered, and the SOC is estimated on line by using a Kalman filtering algorithm.

The comprehensive SOC calculation method of the lithium-sulfur battery with the introduction of the open-loop voltage correction and the Kalman rate wave correction can be expressed by the following formula. And the finally output comprehensive SOC is a result calculated based on ampere-hour integral SOC, and is a result jointly corrected by open-loop voltage correction and Kalman filtering through a weight coefficient mechanism.

Wherein, W _OCV-time A weight coefficient representing OCV correction based on time; w _Kalm-time Representing time-based weight coefficients of Kalman filtering; w _Kalm-soc Representing a Kalman filtering SOC-based weight coefficient; w _OCV-soc The OCV correction is represented by a weight coefficient based on the SOC.

In actual conditions, in order to meet the requirements of use conditions, a plurality of battery cells are usually required to be connected together in a series-parallel connection mode to form a battery pack. Since the present embodiment focuses on studying the application of the SOC online algorithm of the lithium-sulfur battery, in order to simplify the description, the present embodiment only uses a single battery cell as a target to perform algorithm description, and simultaneously avoids the influence of temperature change on the strategy, and keeps the temperature at 25 ℃.

1. Battery cell parameter identification

1) Cell quantity and connection mode:

analyzing the required power spectrums of common UDDS, WLTC, high-speed working condition (HIWAY) and NEDC working condition, wherein the maximum discharge power requirement can reach 135kW, and the maximum charge power is close to 30KW; the voltage of the lithium-sulfur battery is about 2.1V, the rated voltage of a driving motor commonly used on an electric vehicle can reach more than 400V, the instantaneous discharge current can reach more than 300A and the instantaneous charging current can reach more than 70A corresponding to the maximum power requirement in working conditions.

In order to meet the requirement of the operating condition, in this embodiment, 44 battery cells are connected in parallel to form a battery pack so as to meet the requirement of the instantaneous current, and 196 battery packs are connected in series to form a battery pack so as to meet the requirement of the operating voltage of the motor, as shown in fig. 4, which is a schematic diagram of a connection mode of the battery cells of the battery pack.

The main point of the present embodiment is to study the effect of the SOC online estimation algorithm, so the present embodiment ignores the difference between the battery cells, and assumes that all the battery cells are completely consistent. Meanwhile, for the convenience of calculation, the battery pack formed by parallel connection of 44 battery cells is used as a single object for research in the algorithm of the embodiment, and the terminal voltage of the battery pack is consistent with that of the single battery cell, while the current capacity is 44 times that of the single battery cell.

2) OCV-SOC mapping relationship:

as can be seen from fig. 1, when the SOC is in the range of 10% to 70%, the open-loop voltage of the lithium-sulfur battery is always kept around 2.1V, and the whole curve is not monotonous, which means that the same voltage may correspond to a plurality of SOCs. To allow this curve to be used in the algorithm, this section adjusts the curve, preserving the OCV-SOC relationships in the SOC 0% to 10% and 70% to 100% regions, adjusting the OCV-SOC relationship for SOC in the 10% to 70% interval to monotonically increase and SOC increase by 5% for every 1mV increase in voltage. Finally obtaining func _OCV () The functional relationship is shown in fig. 5, and in order to avoid data deformation or data overflow, the function is calculated by linear interpolation.

3) Coulomb efficiency: in the present embodiment, the influence of SOC and current on the coulombic efficiency during charging is ignored, and the coulombic efficiency during charging and discharging is defined as 100%, that is, η (k-1) is constant as 1.

4) Instantaneous voltage mapping relation: respectively performing constant current discharge experiments and constant current charge experiments of different sizes on battery packs formed by connecting battery cells under different SOC in parallel for 1s, measuring the relation between instantaneous terminal voltage and SOC and current as shown in FIG. 6, and respectively recording the mapping relations of discharge and charge as func _{Inst_dch} () And func _{Inst_chg} () The function is calculated by linear interpolation and is based on SOC _ah (k) The predicted instantaneous voltage corresponding to the predicted SOC obtained for the current excitation I (k) can be rewritten as:

2. algorithm coefficient parameter identification

1) Calculating period: considering the available resources of the controller and the time interval of the condition data input to the algorithm, the present embodiment defines the calculation period of the algorithm to be 0.1 second, i.e. the time difference Δ T between the time k and the time k-1 is 0.1 second.

2) OCV correction coefficient: OCV standing time based on OCV characteristics of lithium-sulfur batteryCorrection coefficient wt _sleeptm And calculating the SOC range correction coefficient wt _ocvsocrange The mapping of (3) is shown in fig. 7, and the function is calculated by linear interpolation.

3) Correction coefficient of Kalman filtering: the accumulated error of ampere-hour integration is gradually increased along with the accumulation of the running time, so that the requirement of an algorithm on Kalman filtering is increased; as can be seen from fig. 1, the transient SOC is relatively flat in the SOC10% to 70%, so the kalman filter effect is not good at this stage. Based on the above reasons, the power-on time correction coefficient T of Kalman filtering _poweron (k) The mapping relation with the correction coefficient of the karl SOC range is shown in fig. 8, and the function is calculated by linear interpolation.

4) SOC change slope limit: considering the maximum charge and discharge power requirement in various test conditions and the number of battery cells and the serial-parallel connection relationship set in this chapter, this embodiment defines that the SOC does not rise more than 0.06% per second and does not fall more than 0.06% per second, that is, the formula SOC (k) = func _socramp (SOC _ah (k) And SOC (k) = func) _socramp (SOC _kalm (k) Can be rewritten as:

SOC(k)＝max(min(SOC _ah (k)，SOC(k-1)+0.02*ΔT)，SOC(k-1)-0.06*ΔT)

SOC(k)＝max(min(SOC _kalm (k)，SOC(k-1)+0.02*ΔT)，SOC(k-1)-0.06*ΔT)

3. experimental verification of SOC (System on chip) online estimation method

In order to verify the effect of the SOC comprehensive online algorithm provided by the present embodiment on the precision and robustness, the present embodiment creates a combined working condition based on NEDC, WLTC, UDDS and high-speed working condition data, acquires simulation data in a computer model environment, and verifies the algorithm in an HIL environment.

The specific steps of the verification method of the SOC online estimation algorithm are shown in FIG. 9:

1) Creating a combined working condition based on the common driving working condition and the common parking working condition, and outputting the awakening state (KL 15) and the power demand (P) of the whole vehicle _req )；

2) Testing a real battery core based on power requirements to obtain corresponding current (I) and voltage (V) data;

3) Integrating the current data in MATLAB to obtain the true SOC (SOC) _real ) (ii) a change;

4) In an HIL environment, a whole vehicle awakening state KL15, an excitation current (I) and a voltage state (V) are input, and an uncorrected SOC (state of charge) is obtained on a test board according to an ampere-hour integration algorithm and is used as an algorithm verification reference SOC (SOC) _Ahcalc ) Then calculating according to SOC online estimation algorithm to obtain calculated SOC (SOC) _calc ) And display SOC (SOC) _disp )；

5) Finally comparing the true SOC (SOC) _real ) Reference SOC (SOC) _Ahcalc ) Calculating the SOC (SOC) _calc ) And displaying SOC (SOC) _disp ) And judging the effect of the algorithm.

3.1 working condition data combination

In order to comprehensively verify the adaptability of the algorithm to different working conditions, the current chapter is used for common driving working conditions: the required power spectra of UDDS, WLTC, high speed operating mode (HIWAY) and NEDC were analyzed and their mean, absolute mean, maximum absolute value, root mean square error, absolute rate of change mean and duty ratio of discharge were compared, with the results shown in table 1.

TABLE 1 comparison of the required Power spectra for UDDS, WLTC, HIWAY, NEDC operating conditions

As shown in table 1 and fig. 10, the absolute average value and the maximum absolute value of UDDS and NEDC operating conditions are small, and there is no great power output requirement, while the absolute average value and the maximum absolute value of WLTC and HIWAY operating conditions are large, and the power output requirement is large; the average value of the HIWAY working condition is the largest, and the discharge working condition accounts for the highest ratio, so that the HIWAY working condition can consume energy more quickly; the root mean square errors of the NEDC working condition and the UDDS working condition are smaller, and the root mean square errors of the WLTC working condition and the HIWAY working condition are larger, which indicates that the NEDC working condition and the WLTC working condition are milder, and the WLTC working condition and the HIWAY working condition are more violent; the WLTC working condition is a part of the global unified light vehicle test regulation, and the covered working condition is more comprehensive no matter the vehicle speed range or the acceleration and deceleration working condition.

According to the data, the working conditions of the NEDC and the UDDS are the mildest, and one of the working conditions can be selected as a basic working condition target for judging the quality of the algorithm; the HIWAY has the most fierce working condition and the largest power requirement, and is suitable for being used as the effect of the verification algorithm under the condition of quick discharge; the WLTC working condition is considered most comprehensively and is suitable for being used as a working condition of the comprehensive judgment algorithm effect.

Because OCV-SOC static correction in the algorithm is influenced by the parking time, if the static correction algorithm needs to be researched, a section of parking condition (PARK) needs to be added to prepare for the static correction before the next driving condition is carried out. The present embodiment defines a period of one-half hour (1800 seconds) for a period of parking condition. In the engineering, the KL15 is usually used for representing the awakening state of the whole vehicle, namely that the KL15 is 1 for representing that the vehicle is awakened, otherwise, the vehicle is sleeping, so that the KL15 is 0 in the parking working condition, and the KL15 is 0 in other working conditions.

Because other energy strategies at the end of the whole vehicle are not considered in the verification process, if the initial SOC is defined to 100%, the risk of overcharging caused by energy recovery in the initial stage exists, and therefore the influence of the SOC range correction coefficient (including OCV static correction and kalman filter dynamic correction) is comprehensively considered, the initial SOC of the verification is set to be 90%, the cut-off SOC requirement is lower than 10%, and meanwhile, the battery is ensured not to be over-discharged.

By combining the characteristics of the above working conditions, this embodiment designs 4 groups of combined working conditions for verifying the effect of the algorithm, and the power requirements of the combined working conditions are as shown in fig. 11:

1) Continuous NEDC combined conditions: starting from S0C of 90%, circularly performing three NEDC working conditions, then performing one PARK working condition, and then circularly performing 10 times on a small circle consisting of the four working conditions to form a combined working condition comprising 30 NEDC working conditions;

2) Continuous HIWAY combined working condition: starting from S0C of 90%, circulating for three times of HIWAY working conditions, then performing PARK working conditions once, then circulating for 7 times the small circulation consisting of the four working conditions, and then performing HIWAY working conditions once again to form a combined working condition containing 22 HIWAY working conditions;

3) Continuous WLTC combined working condition: starting from S0C of 90%, circulating for three times WLTC working conditions, then performing PARK working conditions once, then circulating for 3 times a small cycle consisting of the four working conditions, and then performing WLTC working conditions once again to form a combined working condition comprising 10 WLTC working conditions;

4) Mixed driving combined working condition: starting from S0C of 90%, sequentially performing WLTC working condition, HIWAY working condition, NEDC working condition, UDDS working condition and PARK working condition, circulating the small cycle consisting of the five working conditions for 4 times, and then respectively performing the NEDC working condition and the UDDS working condition once to finally form a combined working condition comprising four working conditions.

3.2 testing working conditions of battery pack

In order to obtain the real terminal voltage change of the battery pack under different combination working conditions, the present embodiment builds a battery pack working condition test platform as shown in fig. 12. The platform consists of an upper computer, a charging and discharging machine, a tested battery pack and a constant temperature cabinet.

Placing a battery pack consisting of 44 battery cells with the initial SOC of 90% in parallel in a thermostat at the temperature of 25 ℃; inputting power requirements corresponding to the battery pack under different working conditions into the charge and discharge machine through the upper computer; the charging and discharging machine inputs program-controlled required power to the battery pack, and simultaneously acquires terminal voltage and current data of the battery pack and uploads the terminal voltage and current data to the upper computer for storage. The resulting battery current and voltage data for different combination conditions are shown in fig. 13-16.

3.3 true SOC off-line calculation

In the Matlab environment, ampere-hour integration is performed on the current acquired in the battery pack working condition test, and a real SOC curve is finally obtained, as shown in fig. 17. The SOC of the continuous NEDC combined working condition, the SOC of the continuous HIWAY combined working condition, the SOC of the continuous WLTC combined working condition and the SOC of the hybrid driving combined working condition are finally respectively reduced to 8.989%, 0.739%, 2.888% and 7.618%, the SOC is lower than 10%, and no overdischarge occurs, which shows that the four combined working conditions all meet the target of working condition design and can be used for verifying the estimation effect of the algorithm.

3.4, SOC Algorithm on-line estimation

In order to obtain the SOC calculated by the SOC online estimation algorithm, an algorithm verification test platform is set up in this section as shown in fig. 18. The platform consists of an upper computer, a MicroAutobox and an HIL simulation test system.

Downloading a program into a MicroAutoBox through an upper computer; inputting the acquisition signals corresponding to the excitation current under different working conditions, the measurement voltage and the low-voltage wake-up signal into an HIL simulation test system through an upper computer; the test system inputs the program-controlled signals to the MicroAutobox; and uploading the SOC obtained by online estimation to an upper computer for storage by the MicroAutobox for subsequent algorithm effect analysis.

In order to verify the optimization effect of the SOC online estimation algorithm provided by the embodiment, the embodiment adopts the SOC online estimation algorithm which only retains the ampere-hour integral and has no correction process as the reference group, and tests the four combined working conditions in the algorithm verification test environment to finally obtain the reference group uncorrected SOC curve (SOC) _unFilted ) (ii) a In the same test environment, the SOC online estimation algorithm provided by the embodiment is tested, and finally the corrected algorithm SOC (SOC) is obtained _Filted ). True SOC and SOC of four combined working conditions _unFilted And SOC _Filted As shown in fig. 19-22.

3.5 analysis of algorithm results

Compare FIG. 23 true SOC (SOC) for four sets of operating conditions _real ) And a reference set SOC (SOC) without correction _Unfilted ) When the working condition of the latter is over, compared with the SOC of the former, the SOC of the latter is respectively reduced by 8.066% (continuous NEDC combined working condition), 8.9% (continuous HIWAY combined working condition), 8.7% (continuous WLTC combined working condition) and 8.23% (mixed driving combined working condition), and is averagely reduced by 8.47%, which shows that if ampere-hour integration is carried out only without correcting the data in the working condition circulation process that the SOC is reduced from 90% to below 10%, the obtained SOC can be lost by about 8.5% because of the accumulated error in the integration process.

In the continuous HIWAY combined working condition, the continuous WLTC combined working condition and the mixed driving combined working condition, when the SOC is in _Unfilted When 0% is reached, SOC _real 8.15%, 8.163% and 8.167% respectively, which indicates that the uncorrected SOC algorithm has the situations of insufficient energy utilization and shortened driving range in the using process.

Correction algorithm for comparing four groups of combined working conditionsSOC(SOC _Filted ) And true SOC (SOC) _real ) Difference value of (Delta _ SOC) _Filted ) And reference set no correction SOC (SOC) _Unfilted ) And true SOC (SOC) _real ) Difference value of (Delta _ SOC) _Unfilted ) The obtained results are shown in fig. 23.

Delta_SOC _Unfilted Gradually increases along with the increase of time, and other three groups of working conditions Delta _ SOC except for continuous HIWAY combined working conditions _Filted When the SOC is larger than 70%, the oscillation is reduced, but when the SOC reaches 70%, all working conditions Delta _ SOC including continuous HIWAY combined working conditions _Filted Compare Delta _ SOC _Unfilted Are all reduced by the value of Delta _ SOC _Unfilted The ratios of (A) to (B) are 84.81% (continuous NEDC combined working condition), 11.39% (continuous HIWAY combined working condition), 43.03% (continuous WLTC combined working condition) and 79.85% (mixed driving combined working condition) in sequence, and are reduced by 54.77% on average. The correction algorithm is shown to play a role in optimizing the SOC when the SOC is more than 70%.

Due to the optimization before the SOC reaches 70%, even though the SOC is not excessively corrected during the process of reducing the SOC from 70% to 10%, when the SOC reaches 10%, the Delta _ SOC of all the working conditions _Filted Compare Delta _ SOC _Unfilted Still small, its reduced value and Delta _ SOC _Unfilted The ratios of (A) to (B) are 25.38% (continuous NEDC combined working condition), 3.33% (continuous HIWAY combined working condition), 12.54% (continuous WLTC combined working condition) and 22.47% (mixed driving combined working condition) in sequence, and are reduced by 15.93% on average. It is shown that the SOC is also optimized in the range from 70% to 10% in SOC due to the influence of the algorithm.

When the SOC is less than 10%, the Delta _ SOC expands along with the expansion of the influence factor of the correction _Filted Rapidly decreasing, delta _ SOC at the end of the final operating mode _Filted Compare Delta _ SOC _Unfilted Respectively reducing 96 percent (continuous NEDC combined working condition), 94.78 percent (continuous HIWAY combined working condition), 94.67 percent (continuous WLTC combined working condition) and 91.51 percent (mixed driving combined working condition), averagely reducing 94.24 percent, and only having 0.48 percent difference between the final SOC and the real SOC. This indicates that SOC is less than 10% during SOC _Filted In the role of correction algorithmLower fast heading SOC _real The convergence and SOC optimization effects are obvious, and specific data are shown in a table 2.

TABLE 2 Combined behavior Algorithm error analysis

In order to more intuitively analyze the optimization effect of the algorithm proposed herein at different SOC stages, the embodiment adopts three indexes, namely Max Absolute Error (Max-AE), mean Absolute Error (MAE) and Root Mean Square Error (RMSE), as indexes for judging the optimization effect of the correction algorithm under different working conditions, and detailed index analysis results are shown in table 3.

At the stage that the SOC is larger than 70%, due to the fact that the running time is short and the accumulated error is small, max-AE, MAE and RMSE of all combined working conditions are not large under the condition of an uncorrected SOC algorithm. After the correction algorithm, the continuous NEDC combined working condition and the mixed driving combined working condition are reduced; in addition to the continuous HIWAY combined condition, MAE and RMSE of other conditions are reduced to a certain extent, and although Max-AE of WLTC is increased, the reduction of MAE and RMSE shows that the correction process is effective in the SOC range. Unexpected fluctuation occurs in the SOC correction process due to overlarge power change in the optimization process of the HIWAY working condition, the Max-AE, MAE and RMSE data are slightly increased, but when the SOC reaches 70%, the Delta _ SOC is _Filted Ratio Delta _ SOC _Unfilted Small, so overall the goal of SOC optimization is achieved. Therefore, the correction algorithm acts to optimize the SOC at the stage when the SOC is greater than 70%.

TABLE 3 analysis of reference indexes of errors of algorithm for combined operating conditions

The SOC is in the range of 70% to 10%, and due to the optimization result when the SOC reaches 70%, the Max-AE, MAE and RMSE data of the four working conditions are smaller in the result obtained by correcting the SOC algorithm.

When the SOC is less than 10%, the SOC is quickly towards the SOC under the help of the correction algorithm _real And convergence is carried out, so that MAE and RMSE of the four working conditions are reduced, wherein the continuous HIWAY combined working condition is reduced most, because the optimization effect of the algorithm is not obvious when the SOC is more than 10%, and the power requirement of the working condition is larger, so that the SOC convergence speed is higher and the effect is more obvious in the stage.

In the whole process, max-AE, MAE and RMSE of four working conditions are reduced with the help of a correction algorithm, the Max-AE is reduced by 2.2 on average and accounts for 26.03% of the total without correction algorithm, the MAE is reduced by 1.27 on average and accounts for 30.94% of the total without correction algorithm, the RMSE is reduced by 1.35 on average and accounts for 28.46% of the total without correction algorithm, and the difference value between the final SOC and the real SOC is reduced to 0.48% from 8.47% on average, so that the algorithm has a remarkable effect on the following condition of the real SOC and the overall energy utilization rate of the battery.

The accuracy of SOC estimation is directly related to the whole vehicle energy management strategy, the endurance mileage and the like. After analyzing the principle and the applicable scene of the traditional SOC online estimation algorithm, the embodiment combines the advantages and disadvantages of each algorithm and the characteristics of the lithium-sulfur battery to aim at improving the SOC precision and the utilization rate of the battery capacity, provides a set of comprehensive SOC online estimation algorithm and analyzes the parameter selection of the algorithm: the algorithm is based on ampere-hour integration, static correction is carried out according to the characteristics of an OCV-SOC curve and the standing time of a battery when the battery is electrified, and Kalman filtering dynamic correction is carried out on the SOC obtained by the ampere-hour integration according to the relation between dynamic voltage and SOC and transient current and the electrifying time length in the charging and discharging processes.

After analyzing the power demand data of the NEDC working condition, the UDDS working condition, the high-speed working condition and the WLTC working condition, the embodiment designs four groups of combined working conditions as the input of the optimization effect of the research algorithm according to the parameter characteristics of the battery pack; the battery pack is tested on the battery pack test platform to obtain the current and voltage data of the battery under the four groups of combined working conditions; calculating a real SOC curve through ampere-hour integration in an MATLAB environment based on the data; meanwhile, an algorithm SOC curve is obtained through testing in an HIL testing environment.

Experimental results show that compared with an online SOC algorithm which only carries out ampere-hour integration without any correction process, the maximum absolute error, the average absolute error and the root mean square error of the comprehensive estimation algorithm provided by the embodiment are smaller, the SOC precision is higher, and the SOC can be quickly converged to a difference level which is only 0.5% of the real SOC after being lower than 10%, so that the battery capacity can be fully used, and the endurance mileage is ensured. Therefore, the SOC online estimation algorithm designed by the embodiment achieves the expected target of the algorithm. The verification result shows that the SOC obtained by the comprehensive SOC online estimation algorithm developed for the LIS battery is closer to the real SOC, compared with the traditional ampere-hour integral online estimation algorithm, the SOC precision is averagely improved by 1.27%, when the test working condition is finished, the SOC error is averagely reduced from 8.47% to 0.48%, and the utilization rate of the battery capacity is improved by 8%.

The above-mentioned embodiments are merely preferred embodiments for fully illustrating the present invention, and the scope of the present invention is not limited thereto. The equivalent substitutions or changes made by the person skilled in the art on the basis of the present invention are all within the protection scope of the present invention. The protection scope of the invention is subject to the claims.

Claims (6)

1. A lithium sulfur power battery SOC estimation method based on OCV correction and Kalman filtering algorithm is characterized in that: the method comprises the following steps:

the method comprises the following steps: OCV static correction

11 After the whole vehicle is awakened, whether the power-on is the initial power-on is judged; if so, adopting default values for both the SOC value and the standing time; if not, reading the SOC value stored in the last power-off process, and acquiring standing time;

12 Judging the validity of the cell voltage; if yes, executing step 13); if the effective cell voltage data are not obtained within the set time range, taking the SOC value obtained after the whole vehicle is awakened as the SOC (k 0) after OCV static correction, and executing the step 14);

13 Calculating an SOC range correction coefficient and a standing time correction coefficient according to the SOC value and the standing time obtained after the whole vehicle is awakened; calculating the SOC under the OCV curve, and modifying the SOC according to the SOC range correction coefficient and the standing time correction coefficient to obtain the SOC (k 0) subjected to OCV static correction;

the correction coefficient of the standing time is as follows:

wt _sleeptm ＝func _sleeptm (T _sleep )

the SOC range correction coefficient is as follows:

wt _ocvsocrange ＝func _ocvsocrang (SOC _READ )

wherein, wt _sleeptm A correction coefficient for the standing time; t is _sleep The standing time is the time interval from last power-off of the whole vehicle to the awakening at this time; func _sleeptm () The mapping relation between the standing time and the standing time correction coefficient is obtained; wt. of _ocvsocrange Correcting coefficient for SOC range; SOC (system on chip) _READ The SOC value stored for the last power-off is read; func _socrange () The mapping relation between the read SOC and the SOC range correction coefficient is obtained;

the SOC under the OCV curve is:

SOC _OCV ＝func _OCV (V)

the SOC (k 0) after OCV static correction is:

wherein V is the cell voltage; func _OCV () An OCV-SOC mapping curve relation is obtained; SOC _OCV The SOC value is obtained by the cell voltage through an OCV-SOC mapping curve relation;

14 Output the SOC (k 0) after OCV static correction;

step two: kalman filtering online estimation

21 The OCV statically corrected SOC (k 0) is taken as the initial SOC (k-1) of online estimation;

22 To determine the validity of the current measurement during the period from time k-1 to time k;

if not, making SOC (k) = SOC (k-1), and executing step 26);

if yes, executing step 23);

23 Estimate SOC by ampere-hour integration and obtain SOC _ah (k)；

24 Judging the validity of the voltage measurement value at the moment k;

if not, directly connecting the SOC _ah (k) Step 26 is executed as the SOC (k) at time k);

if yes, executing step 25);

25 To SOC _ah (k) Performing Kalman filtering to obtain SOC (k) at the moment k;

26 Taking SOC (k) as an SOC estimation value of the lithium sulfur power battery at the k moment and outputting the SOC (k);

27 Judging whether the whole vehicle is powered off; if yes, storing the SOC (k); if not, enabling k = k +1, and circulating the step 22) to the step 27) until the whole vehicle is powered off;

therein, SOC _ah (k) Obtaining estimated SOC according to ampere-hour integration at the moment k; SOC (k-1) is the SOC value at time k-1.

2. The lithium sulfur power battery SOC estimation method based on OCV correction and Kalman filtering algorithm according to claim 1, characterized in that: in the step 23), the first step is executed,

η(k-1)＝func _eff (I(k-1)，SOC(k-1))

wherein, SOC _ah (k) Obtaining estimated SOC according to ampere-hour integration at the moment k; SOC (k-1) is the SOC value at the k-1 moment; η (k-1) is the coulombic efficiency at time k-1, which is substantially constant during discharge and affected by SOC during charge; i (k-1) is the current from the time k-1 to the time k; Δ T is the time difference between the time k and the time k-1; c _rated Represents the rated capacity of the battery; func _eff () And representing the mapping relation of coulombic efficiency to current I and state of charge SOC.

3. The lithium sulfur power battery SOC estimation method based on OCV correction and Kalman filtering algorithm according to claim 1, characterized in that: in the step 24), when the voltage measurement value at the time k is invalid, in order to avoid abrupt change of the SOC due to measurement errors or a correction process, a maximum value of increase or decrease of the SOC at each step is set, and then:

SOC(k)＝func _socramp (SOC _ah (k))

wherein, func _socramp () Is a function that limits the variation of the SOC.

4. The lithium sulfur power battery SOC estimation method based on OCV correction and Kalman filtering algorithm according to claim 1, characterized in that: in the step 25), for SOC _ah (k) The kalman filtering process is performed as follows:

251 Based on SOC _ah (k) And obtaining a predicted instantaneous voltage corresponding to the predicted SOC with the current excitation I (k):

V _est (k)＝func _Inst (SO _ah (k)，I(k))

wherein, V _est (k) The predicted instantaneous voltage at time k; fu _Inst () The function is an estimated function of instantaneous voltage, and the mapping relation of the function can be obtained by carrying out specific current charge and discharge test on the battery under a specific SOC under an experimental environment;

252 Computing kalman filter coefficients:

P-(k)＝G(k)P+(k-1)GT(k)+W

K(k)＝P-(k)HT(k)(H(k)P-(k)HT(k)+Q)-1

P+(k)＝(I-K(k)H(k))P-(k)

wherein G (k) is the derivation of the state estimation equation; GT (k) is the transposed vector of the G (k) vector; h (k) is the derivation of the observation prediction equation; p- (k) is an error covariance prior estimation at the moment k; p + (k) and P + (k-1) are respectively the error covariance posterior estimation at the k moment and the k-1 moment; they reflect the degree of difference between the estimated value and the true value of the state variable; w is the observed noise covariance; q is the system noise covariance; k (K) is a Kalman filtering coefficient;

253 Because the error generated in the ampere-hour integration process is accumulated along with the power-on time, a power-on time correction coefficient is introduced when Kalman filtering is carried out:

wt _pwrontm (k)＝func _pwronwt (T _poweron (k))

T _poweron (k)＝T _poweron (k-1)+ΔT

wherein, T _poweron (k-1) and T _poweron (k) Respectively the power-on time at the k-1 moment and the k moment; func _pwronwt () Correcting the mapping relation between the power-on time and the power-on time; t is _poweron (k) The power-on time correction coefficient is in the range of 0-1; wt. of _pwrontm (k) Representing a correction factor based on the power-on time;

taking the characteristics of the lithium-sulfur battery into consideration, introducing an SOC range correction coefficient during filtering:

wt _socrange (k)＝func _socrange (SOC _ah (k))

wherein, func _socrange () Correcting the coefficient and SOC for SOC range _ah (k) The mapping relationship between the two; wt. of _socrange (k) Is a correction coefficient based on the SOC range;

the SOC at the k time obtained finally is:

SOC _kalm (k)＝SOC _ah (k)+K(k)*wt _pwrontm (k)*wt _socrange (k)*(V _sample (k)-V _est (k))

wherein, V _sample (k) The cell voltage measured at the moment k; SOC (system on chip) _kalm (k) Obtaining SOC after Kalman filtering correction at the moment k; k (K) represents a kalman filter coefficient.

5. The lithium sulfur power battery SOC estimation method based on OCV correction and Kalman filtering algorithm according to claim 4, characterized in that: in the step 253), in order to avoid abrupt change of the SOC caused by measurement errors or correction processes, a maximum value of increase or decrease of the SOC at each step is set, and then:

SOC(k)＝func _socramp (SOC _kalm (k))

wherein, func _socramp () Is a function that limits the variation of the SOC.

6. The lithium sulfur power battery SOC estimation method based on OCV correction and Kalman filtering algorithm according to claim 4, characterized in that: in the step 251), the mapping relationships of discharging and charging are respectively recorded as func _{Inst_dc} () And fu _{Inst_chg} () Based on SOC _ah (k) The predicted instantaneous voltage corresponding to the predicted SOC obtained from the current excitation I (k) is:

wherein, func _{Inst_dch} () Indicating the predicted instantaneous voltage V during discharge _est (k) And SOC _ah (k) And I (k); func _{Inst_chg} () Indicating the predicted instantaneous voltage V during charging _est (k) And SOC _ah (k) And I (k).

Images