CN103135065A - Iron phosphate lithium battery electric quantity detecting method based on feature points - Google Patents

Abstract

The invention relates to an iron phosphate lithium battery electric quantity detecting method based on feature points. Sectional type Kalman filtering is carried out through selecting the feature points. Kalman prediction is adopted at the feature points and an ampere-hour method is utilized to achieve integrating processing among the feature points, so that the problem that a system on chip (SOC) prediction error in a flat area is large due to the fact that voltage collecting accuracy and current collecting accuracy are low in iron phosphate lithium battery system on chip (SOC) measuring is solved, and the accuracy and the stability in iron phosphate lithium battery SOC estimation are improved. The iron phosphate lithium battery electric quantity detecting method based on the feature points has certain application value, and can accurately provides the cruising ability of the battery.

Description

Ferric phosphate lithium cell detection method of quantity of electricity based on unique point

Technical field

The present invention relates to a kind of battery detecting technology, particularly a kind of ferric phosphate lithium cell detection method of quantity of electricity based on unique point.

Background technology

Ferric phosphate lithium cell is long because of its life-span, security performance good, low cost and other advantages more and more becomes a kind of desirable electrokinetic cell.The SOC value of battery electric quantity (State of Charge, state of charge) as the main characteristic parameter of battery, is one of Focal point and difficult point of battery management system research in recent years.Traditional battery electric quantity detection technique mainly comprises: open-circuit voltage method, ampere-hour method, look-up table etc., but the problem that these method ubiquity errors are large, reliability is lower.For these deficiencies, the ferric phosphate lithium cell SOC of employing Kalman filtering estimates to become gradually the focus of this area research.In the ferric phosphate lithium cell discharge process, in voltage-SOC curve, most of zone is too smooth, and change in voltage is limited by the acquisition precision of voltage all less than 0.3V, and traditional Kalman filtering algorithm error is still larger.

The technology similar to the present invention has at present: 1. Cheng Yan green grass or young crops is waited the paper " based on the electric automobile remaining capacity estimation of Kalman filtering " that is published in " Electronic University Of Science ﹠ Technology Of Hangzhou's journal "; 2. the paper that is published in " Tsing-Hua University's journal " such as woods Cheng Tao " is estimated electric automobile power battery SOC with improved Ah counting method "; 3. Electronic University Of Science ﹠ Technology Of Hangzhou, the patent of invention of He Zhiwei etc. " a kind of estimation method of battery dump energy of sample-based point Kalman filtering " (CN101598769B) and " a kind of estimation method of battery dump energy based on combined sampling point Kalman filtering " (CN101604005B); 4. University Of Chongqing, the patent of invention of Deng Li etc. " evaluation method of residual capacity of iron-lithium phosphate power cell " (CN101629992B).Above-mentioned these technology all do not have the problem for the valuation of SOC curve flat site, propose solution targetedly.

Summary of the invention

For the problems referred to above, the present invention proposes a kind of ferric phosphate lithium cell detection method of quantity of electricity based on unique point, improve in ferric phosphate lithium cell low discharging current situation, accurately detect the problem of the larger error of SOC curve flat region method existence.

A kind of ferric phosphate lithium cell detection method of quantity of electricity based on unique point of the present invention specifically comprises the steps:

Step 1, measure to obtain associated data table embryo between battery electric quantity SOC, load voltage U and electrode temperature T by experiment

Wherein $M=\left[\begin{array}{c}{M}_{T_0}\\ M_{T_1}\\ M\\ M_{T_k}\end{array}\right],$ $M_{T_k}=\left[\begin{array}{c}0.0\%\\ 0.1\%\\ M\\ 100.0\%\end{array}\right.,\left.\begin{array}{c}{U}_0\\ U_1\\ M\\ U_n\end{array}\right]$

T _kBe the battery electrode temperature,

Expression battery electrode temperature T _kUnder battery electric quantity SOC and the associated data table of load voltage U;

Step 2, each SOC value x and the voltage y, current i relation constantly that represents battery with battery state model and observation model:

State model: $x_k=x_{k-1}+\frac{i_k\mathrm{\Δt}}{{\mathrm{\η}}_i{\mathrm{\η}}_{T}Q}+w_k---\left(1\right)$

Observation model: y _k=E ₀+ g (x _k)-R ( _k) i _k+ v _k(2)

X wherein _kBe battery k SOC value constantly, i _kBe k moment electric current, Q is the battery rated capacity, η _iAnd η _TBe respectively the discharge and recharge coefficient relevant with electric current and temperature, Δ t is measuring intervals of TIME, w _kState-noise, y _kK cell load voltage estimated value constantly, E ₀The constant relevant with cell voltage potential, R(i _k) be the internal resistance of cell, with current i _kRelevant, " the HPPC method of testing in FreedomCAR battery testing handbook is calculated, v to adopt the U.S. _kFor measuring noise, g(x _k) be and x _kRelevant variable is by N(x, the g of tabling look-up) obtain;

This N(x, g) by measuring, assay method is: measure k battery electrode temperature T constantly _k, load voltage y _kWith load current i _k, obtain this moment battery y by the associated data table M of step 1 _kCorresponding SOC value x _k, by the model of formula (2) and ignore and measure noise v _k, calculate g(x _k):

g(x _k)=y _k+R(i _k)i _k-E ₀ （3）

Result is charged to table N(x, g)

$N(x,g)=\left[\begin{array}{c}{x}_0\\ x_1\\ M\\ x_n\end{array}\right.,\left.\begin{array}{c}g\left(x_0\right)\\ g\left(x_1\right)\\ M\\ g\left(x_n\right)\end{array}\right]$

Step 3, operation self-check program are completed the comparison of internal reference magnitude of voltage, temperature correction parameter initialization, the initialization of Model Parameter, the operation of correlated variables initial assignment;

Step 4, detect the battery electrode temperature T with temperature sensing circuit ₀, according to temperature T ₀, find table

Measure the cell load current i with current detection circuit ₀, measure the cell load voltage U with voltage detecting circuit ₀, according to load voltage U ₀At table

In check in this moment battery the SOC value, with SOC this moment ₀As initial value Namely

Be battery 0 SOC value constantly, load voltage U ₀The detection variance

As the Kalman Filter Estimation initial variance;

Step 5, employing Fast Kalman filtering iteration recursive operation calculate k battery SOC constantly _kValue:

(1) measure the cell load voltage U constantly at k _k, the cell load current i _k, and temperature T, k=1,2,3,

(2) each that represents battery with battery state model and observation model SOC value and voltage, current relationship constantly:

State model: $x_k=x_{k-1}+\frac{{\mathrm{\η}}_i{i}_k\mathrm{\Δt}}{{\mathrm{\η}}_{T}Q}+w_k$

Observation model: y _k=E ₀+ g (x _k)-R (i) i _k+ v _k

X wherein _kBe battery k SOC value constantly, Q is the battery rated capacity, η _iAnd η _TBe respectively the discharge and recharge coefficient relevant with electric current and temperature, Δ t is measuring intervals of TIME, w _kState-noise, y _kK cell load voltage estimated value constantly, E ₀The constant relevant with cell voltage potential, g(x _k) be and x _kRelevant variable is by the M(x that tables look-up _k) obtain; R (i) is the internal resistance coefficient of battery, and relevant with discharge current, and R (i) is by measuring, v _kFor measuring noise;

(3) K moment battery SOC predicted value

K is cell load voltage estimated value y constantly _k:

Battery SOC estimated value after K filtering constantly

K=1,2,3…

Wherein, kalman gain $L_k=\frac{P_k^{-}{C}_k}{P_k^{-}{C}_k^2+D_v},$ $P_k^{-}=P_{k-1}^{-}+D_w,$

The Kalman filtering variance, C _kBe and x _kRelevant parameter is by the M(C that tables look-up _k) obtain D _vAnd D _wBe respectively the state-noise variance and measure noise variance;

Step 6, the SOC curve that will divide ferric phosphate lithium cell SOC curve that looks records and step 5 Kalman filtering method to calculate compare, and the intersection point place of two curves is defined as unique point;

Step 7, judgement battery SOC at this moment _kWhether be near unique point, if not near unique point, utilize the Kalman filtering method of step 5 to calculate the SOC value, then get back to step 7;

If step 8 battery SOC at this moment _kBe near unique point, begin to adopt the current integration method to calculate the SOC value, and judge whether to arrive next unique point;

If step 9 does not arrive next unique point, continue to adopt the current integration method to calculate the SOC value, and judge whether to arrive unique point;

If step 10 arrives next unique point, utilize eigenwert that the current integration value is revised, and get back to step 8.

The present invention carries out the sectional type Kalman filtering by the selected characteristic point, the unique point place adopts and adopts ampere-hour method Integral Processing between Kalman Prediction, unique point, voltage in ferric phosphate lithium cell SOC measurement, the low larger problem of flat region SOC predicated error that causes of current acquisition precision have been made up, improved the precision and stability during ferric phosphate lithium cell SOC estimates, have certain using value, can provide more accurately the flying power of battery.

Description of drawings

Fig. 1 is voltage-SOC curvilinear characteristic point schematic diagram;

Fig. 2 is voltage in the present invention-SOC curvilinear characteristic point choosing method schematic diagram;

Fig. 3 is the inventive method schematic flow sheet.

The invention will be further described below in conjunction with specific embodiment.

Embodiment

Ultimate principle of the present invention is: according to the SOC curvilinear characteristic of ferric phosphate lithium cell discharge, the charging and discharging curve of the ferric phosphate lithium cell of a 2000mAH as shown in Figure 1 is divided into several sections to the flat region, and the starting point of every section and terminal point are defined as unique point.Near unique point SOC value adopts Kalman Filter Estimation, and the SOC value between unique point adopts electric current ampere-hour integral method to estimate.

As shown in Figure 3, a kind of ferric phosphate lithium cell detection method of quantity of electricity based on unique point of the present invention specifically comprises the steps:

Step 1, measure to obtain associated data table M between battery electric quantity SOC, load voltage U and electrode temperature T by experiment:

Wherein $M=\left[\begin{array}{c}{M}_{T_0}\\ M_{T_1}\\ M\\ M_{T_k}\end{array}\right],$ $M_{T_k}=\left[\begin{array}{c}0.0\%\\ 0.1\%\\ M\\ 100.0\%\end{array}\right.,\left.\begin{array}{c}{U}_0\\ U_1\\ M\\ U_n\end{array}\right]$

T _kBe the battery electrode temperature,

Expression battery electrode temperature T _kUnder battery electric quantity SOC and the associated data table of load voltage U;

Step 2, each SOC value x and the voltage y, current i relation constantly that represents battery with battery state model and observation model:

State model: $x_k=x_{k-1}+\frac{i_k\mathrm{\Δt}}{{\mathrm{\η}}_i{\mathrm{\η}}_{T}Q}+w_k---\left(1\right)$

Observation model: y _k=E ₀+ g (x _k)-R ( _k) i _k+ v _k(2)

X wherein _kBe battery k SOC value constantly, i _kBe k moment electric current, Q is the battery rated capacity, η _iAnd η _TBe respectively the discharge and recharge coefficient relevant with electric current and temperature, Δ t is measuring intervals of TIME, w _kState-noise, y _kK cell load voltage estimated value constantly, E ₀The constant relevant with cell voltage potential, R(i _k) be the internal resistance of cell, with current i _kRelevant, " the HPPC method of testing in FreedomCAR battery testing handbook is calculated, v to adopt the U.S. _kFor measuring noise, g(x _k) be and x _kRelevant variable is by N(x, the g of tabling look-up) obtain;

This N(x, g) by measuring, assay method is: measure k battery electrode temperature T constantly _k, load voltage y _kWith load current i _k, obtain this moment battery y by the associated data table M of step 1 _kCorresponding SOC value x _k, by the model of formula (2) and ignore and measure noise v _k, calculate g(x _k):

g(x _k)=y _k+R( _k)i _k-E ₀ （3）

Result is charged to table N(x, g)

$N(x,g)=\left[\begin{array}{c}{x}_0\\ x_1\\ M\\ x_n\end{array}\right.,\left.\begin{array}{c}g\left(x_0\right)\\ g\left(x_1\right)\\ M\\ g\left(x_n\right)\end{array}\right]$

Step 3, operation self-check program are completed the comparison of internal reference magnitude of voltage, temperature correction parameter initialization, the initialization of Model Parameter, the operation of correlated variables initial assignment;

Step 4, detect the battery electrode temperature T with temperature sensing circuit ₀, according to temperature T ₀, find table

Measure the cell load current i with current detection circuit ₀, measure the cell load voltage U with voltage detecting circuit ₀, according to load voltage U ₀At table

In check in this moment battery the SOC value, with SOC this moment ₀As initial value Namely

Be battery 0 SOC value constantly, load voltage U ₀The detection variance

As the Kalman Filter Estimation initial variance;

Step 5, employing Fast Kalman filtering iteration recursive operation calculate k battery SOC constantly _kValue:

(1) measure the cell load voltage U constantly at k _k, the cell load current i _k, and temperature T, k=1,2,3,

(2) each that represents battery with battery state model and observation model SOC value and voltage, current relationship constantly:

State model: $x_k=x_{k-1}+\frac{{\mathrm{\η}}_i{i}_k\mathrm{\Δt}}{{\mathrm{\η}}_{T}Q}+w_k$

Observation model: y _k=E ₀+ g (x _k)-R (i) i _k+ v _k

X wherein _kBe battery k SOC value constantly, Q is the battery rated capacity, η _iAnd η _TBe respectively the discharge and recharge coefficient relevant with electric current and temperature, Δ t is measuring intervals of TIME, w _kState-noise, y _kK cell load voltage estimated value constantly, E ₀The constant relevant with cell voltage potential, g(x _k) be and x _kRelevant variable is by the M(x that tables look-up _k) obtain; R (i) is the internal resistance coefficient of battery, and relevant with discharge current, and R (i) is by measuring, v _kFor measuring noise;

(3) K moment battery SOC predicted value

K is cell load voltage estimated value y constantly _k:

Battery SOC estimated value after K filtering constantly

K=1,2,3…

Wherein, kalman gain $L_k=\frac{P_k^{-}{C}_k}{P_k^{-}{C}_k^2+D_v},$ $P_k^{-}=P_{k-1}^{-}+D_w,$

The Kalman filtering variance, C _kBe and x _kRelevant parameter is by the M(C that tables look-up _k) obtain D _vAnd D _wBe respectively the state-noise variance and measure noise variance;

Step 6, the SOC curve that will divide ferric phosphate lithium cell SOC curve that looks records and step 5 Kalman filtering method to calculate compare, and as shown in Figure 2, the intersection point place of two curves are defined as unique point;

Step 7, judgement battery SOC at this moment _kWhether be near unique point, if not near unique point, utilize the Kalman filtering method of step 5 to calculate the SOC value, then get back to step 7;

If step 8 battery SOC at this moment _kBe near unique point, begin to adopt the current integration method to calculate the SOC value, and judge whether to arrive next unique point;

If step 9 does not arrive next unique point, continue to adopt the current integration method to calculate the SOC value, and judge whether to arrive unique point;

If step 10 arrives next unique point, utilize eigenwert that the current integration value is revised, and get back to step 8.

The above, it is only preferred embodiment of the present invention, be not that technical scope of the present invention is imposed any restrictions, therefore every foundation technical spirit of the present invention all still belongs in the scope of technical solution of the present invention any trickle modification, equivalent variations and modification that above embodiment does.

Claims (1)

1. the ferric phosphate lithium cell detection method of quantity of electricity based on unique point, is characterized in that comprising the steps:

Step 1, measure to obtain associated data table M between battery electric quantity SOC, load voltage U and electrode temperature T by experiment:

Wherein $M=\left[\begin{array}{c}{M}_{T_0}\\ M_{T_1}\\ M\\ M_{T_k}\end{array}\right],$ $M_{T_k}=\left[\begin{array}{c}0.0\%\\ 0.1\%\\ M\\ 100.0\%\end{array}\right.,\left.\begin{array}{c}{U}_0\\ U_1\\ M\\ U_n\end{array}\right]$

T _kBe the battery electrode temperature, Expression battery electrode temperature T _kUnder battery electric quantity SOC and the associated data table of load voltage U;

Step 2, each SOC value x and the voltage y, current i relation constantly that represents battery with battery state model and observation model:

State model: $x_k=x_{k-1}+\frac{i_k\mathrm{\Δt}}{{\mathrm{\η}}_i{\mathrm{\η}}_{T}Q}+w_k---\left(1\right)$

Observation model: y _k=E ₀+ g (x _k)-R ( _k) i _k+ v _k(2)

X wherein _kBe battery k SOC value constantly, i _kBe k moment electric current, Q is the battery rated capacity, η _iAnd η _TBe respectively the discharge and recharge coefficient relevant with electric current and temperature, Δ t is measuring intervals of TIME, w _kState-noise, y _kK cell load voltage estimated value constantly, E ₀The constant relevant with cell voltage potential, R(i _k) be the internal resistance of cell, with current i _kRelevant, " the HPPC method of testing in FreedomCAR battery testing handbook is calculated, v to adopt the U.S. _kFor measuring noise, g(x _k) be and x _kRelevant variable is by N(x, the g of tabling look-up) obtain;

This N(x, g) by measuring, assay method is: measure k battery electrode temperature T constantly _k, load voltage y _kWith load current i _k, obtain this moment battery y by the associated data table M of step 1 _kCorresponding SOC value x _k, by the model of formula (2) and ignore and measure noise v _k, calculate g(x _k):

g(x _k)=y _k+R(i _k)i _k-E ₀ （3）

Result is charged to table N(x, g)

$N(x,g)=\left[\begin{array}{c}{x}_0\\ x_1\\ M\\ x_n\end{array}\right.,\left.\begin{array}{c}g\left(x_0\right)\\ g\left(x_1\right)\\ M\\ g\left(x_n\right)\end{array}\right]$

Step 3, operation self-check program are completed the comparison of internal reference magnitude of voltage, temperature correction parameter initialization, the initialization of Model Parameter, the operation of correlated variables initial assignment;

Step 4, detect the battery electrode temperature T with temperature sensing circuit ₀, according to temperature T ₀, find table Measure the cell load current i with current detection circuit ₀, measure the cell load voltage U with voltage detecting circuit ₀, according to load voltage U ₀At table

In check in this moment battery the SOC value, with SOC this moment ₀As initial value

Namely

Be battery 0 SOC value constantly, load voltage U ₀The detection variance

As the Kalman Filter Estimation initial variance;

Step 5, employing Fast Kalman filtering iteration recursive operation calculate k battery SOC constantly _kValue:

(1) measure the cell load voltage U constantly at k _k, the cell load current i _k, and temperature T, k=1,2,3,

(2) each that represents battery with battery state model and observation model SOC value and voltage, current relationship constantly:

State model: $x_k=x_{k-1}+\frac{{\mathrm{\η}}_i{i}_k\mathrm{\Δt}}{{\mathrm{\η}}_{T}Q}+w_k$

Observation model: y _k=E ₀+ g (x _k)-R (i) _ik+v _k

X wherein _kBe battery k SOC value constantly, Q is the battery rated capacity, η _iAnd η _TBe respectively the discharge and recharge coefficient relevant with electric current and temperature, Δ t is measuring intervals of TIME, w _kState-noise, y _kK cell load voltage estimated value constantly, E ₀The constant relevant with cell voltage potential, g(x _k) be and x _kRelevant variable is by the M(x that tables look-up _k) obtain; R (i) is the internal resistance coefficient of battery, and relevant with discharge current, and R (i) is by measuring, v _kFor measuring noise;

(3) K moment battery SOC predicted value

K is cell load voltage estimated value y constantly _k:

Battery SOC estimated value after K filtering constantly

K=1,2,3…

Wherein, kalman gain $L_k=\frac{P_k^{-}{C}_k}{P_k^{-}{C}_k^2+D_v},$ $P_k^{-}=P_{k-1}^{-}+D_w,$

The Kalman filtering variance, C _kBe and x _kRelevant parameter is by the M(C that tables look-up _k) obtain D _vAnd D _wBe respectively the state-noise variance and measure noise variance;

Step 6, the SOC curve that will divide ferric phosphate lithium cell SOC curve that looks records and step 5 Kalman filtering method to calculate compare, and the intersection point place of two curves is defined as unique point;

Step 7, judgement battery SOC at this moment _kWhether be near unique point, if not near unique point, utilize the Kalman filtering method of step 5 to calculate the SOC value, then get back to step 7;

If step 8 battery SOC at this moment _kBe near unique point, begin to adopt the current integration method to calculate the SOC value, and judge whether to arrive next unique point;

If step 9 does not arrive next unique point, continue to adopt the current integration method to calculate the SOC value, and judge whether to arrive unique point;

If step 10 arrives next unique point, utilize eigenwert that the current integration value is revised, and get back to step 8.

Images