CN104036128A - Battery SOC (State Of Charge) estimation method based on filter current - Google Patents

Abstract

The invention provides a battery SOC (State Of Charge) estimation method based on a filter current. Battery SOC can be estimated according to a battery current in 4tau and the current battery voltage, so that the battery SOC estimation method is suitable for the situation that the initial SOC cannot be obtained or the battery current cannot be continuously recorded for a long time. The method has the advantages of (1) not requiring an SOC initial value and historical SOC data, thereby having no cumulative errors; (2) not requiring a battery current value of the whole process and only requiring the battery current value in a shorter time (4tau) of the past and the current battery voltage to estimate the battery SOC; (3) being applicable to complex working conditions of battery current variation; (4) having universality.

Description

A kind of battery SOC method of estimation based on filtered circuit

Technical field

The present invention relates to battery management field, relate in particular to a kind of battery SOC method of estimation.

Background technology

In electric automobile, off-line type photovoltaic generating system, the moving equipment of Tong, boats and ships, aircraft and satellite, all need to use accumulator as power source.To battery charge state (state of charge, SOC) estimate the important evidence that user understands accumulator current state, the system spare working time is estimated, SOC estimates it is also an important research content in battery management field simultaneously.

Conventional SOC method of estimation has following several: (1) discharge test method.Calculate SOC by constant current continuous discharge, be usually used in battery capacity and demarcate, cannot be applied to the battery in work.(2) open-circuit voltage method.The method judges battery SOC according to open-circuit voltage, but that open-circuit voltage just can reach after battery standing a period of time is stable, therefore the method is not suitable for the battery in work equally.(3) Ah counting method.By SOC initial value and battery current, the integration of time is calculated to SOC, the method requires there is lasting current value, and SOC exists cumulative errors while calculating.(4) Kalman filtering method and neural network.The former adopts Kalman's optimal filtering algorithm to carry out recursion to SOC using the factor that affects SOC as system noise; The latter utilizes neural network to carry out SOC estimation to the strong capability of fitting of curve.These two kinds of methods all propose on Ah counting method basis, need equally initial SOC value, and the prior imformation of mistake easily causes algorithm not restrained.

In some applications, because replacing battery core, electromagnetic environment badly make the reasons such as integrated circuit reset, cause getting initial SOC and continuous recording battery current, therefore several conventional SOC method of estimation of the prior art is all inapplicable.

The present invention will refer to non-and sharp document 1:Tremblay O, Dessaint L A.Experimental validation of a battery dynamic model for EV applications[J] .World Electric Vehicle Journal, 2009,3:1-10.

Summary of the invention

In order to solve problem in prior art, the present invention proposes a kind of battery SOC method of estimation based on filtered circuit, do not need the battery current value of SOC initial value and overall process, can accurately estimate the current SOC of battery.

The present invention is achieved through the following technical solutions:

A battery state of charge SOC method of estimation based on filtered circuit, comprises the following steps:

S1: set up battery dynamic model, described dynamic model comprises firstorder filter, for asking for filtered circuit;

S2: obtain filtered circuit i ^*;

S3: utilize curved surface fitting method, obtain v according to the discharge curve of battery and internal resistance _batt=f (SOC, i, i ^*), wherein, i ^*be filtered circuit, i is the battery current in this moment;

S4: obtain current i, i ^*lower cell voltage v _battrelation with battery SOC

S5: dynamically generate current i, i ^*lower v _battthe tables of data changing with battery SOC;

S6: obtain not battery SOC in the same time by method of interpolation.

Further, the voltage v of described battery model in the time of electric discharge and charging _battfor:

Wherein, R is the internal resistance of cell, and Q is battery max cap., and it is battery discharge electric weight, relevant to SOC, E ₀, K, A, B be 4 undetermined coefficients.

Further, in described step S2, filtered circuit i ^*obtained by the current i in 4 τ.

Further, in described step S3, utilize curved surface fitting method rooting to obtain v according to discharge curve and the internal resistance of battery _batt=f (SOC, i, i ^*), be specially: utilize curved surface fitting method, try to achieve E ₀, K, A, B.

Further, described method extends to lithium battery, lead-acid battery, nickel-cadmium battery and Ni-MH battery.

The invention has the beneficial effects as follows: the battery SOC method of estimation based on filtered circuit that the present invention proposes, can, according to the SOC of battery current in 4 τ and current cell voltage estimation battery, realize battery is carried out to SOC estimation.The method has following advantage: (1) does not need SOC initial value and historical SOC data, does not therefore have cumulative errors; (2) do not need the battery current value of overall process, battery current value and the current cell voltage that only need to pass by one period of short period (4 τ), just can estimate battery SOC; (3) be applicable to the complex working condition that battery current changes; (4) there is versatility.

Brief description of the drawings

Fig. 1 is battery model schematic diagram used in the present invention;

Fig. 2 is the discharge curve of battery;

Fig. 3 carries out surface fitting to discharge data in method of the present invention to ask parameter schematic diagram;

Fig. 4 be discharge current while being 0.2C, 0.5C, 1C the present invention discharge data is carried out surface fitting and is asked the method comparison diagram of parametric technique and document 1;

Fig. 5 be discharge current while being 0.35C, 0.85C the present invention discharge data is carried out surface fitting and is asked the method comparison diagram of parametric technique and document 1;

Fig. 6 asks filtered circuit i by firstorder filter in method of the present invention ^*schematic diagram;

Fig. 7 is the process flow diagram of battery SOC method of estimation of the present invention.

Embodiment

Below in conjunction with brief description of the drawings and embodiment, the present invention is further described.

A lot of battery models are only applicable to particular experiment object, but the battery model that non-patent literature 1 proposes is applicable to various types of batteries.It improves on the steady-state model that is applicable to constant current charge-discharge, has increased firstorder filter, considers the impact of size of current on cell voltage in the past period, the complex working condition when being applicable to electric current and changing.Wherein lithium ion battery can represent by following formula respectively in the time of electric discharge and charging:

Wherein, v _battfor cell voltage, R is the internal resistance of cell, and Q is battery max cap., and it is battery discharge electric weight, relevant to SOC, E ₀, K, A, B be 4 undetermined coefficients, i is battery current, i ^*for the electric current that battery current obtains after first-order filtering, filter time constant τ is 30s.

Filtered circuit i ^*the impact that reflection charging and discharging currents causes cell voltage while variation: with reflect respectively the impact of charging and discharging currents different when discharging and charging on cell voltage; the impact of reflection SOC on cell voltage; The indicial response of Aexp (Bit) reflection battery in the time approaching full voltage.

As shown in Figure 1: battery model is made up of resistance R and controlled source E, controlled source E is by discharge electricity amount it and filtered circuit i ^*try to achieve, and it and i ^*all can be tried to achieve through integration or filtering by battery current i.Model can be expressed as:

v _batt＝E-R·i＝f(it,i ^*)-R·i (2)

Yes should be noted that, this model is not considered the aging impact on capacity, internal resistance, does not consider that temperature is on the impact discharging and recharging yet.

In the battery model of accompanying drawing 1, there is E ₀, K, A, tetra-undetermined parameters of B.As shown in Figure 2, discharge curve people, for being divided into " index section ", " nominal section " and " descending branch " three sections, can be obtained to these undetermined parameters according to the value of each section of separation.But the shortcoming of the method is choosing with larger randomness of separation, the model parameter difference that different people obtains, model is easy to there is larger error with actual experimental data.Therefore, the present invention improves this: according to the model of formula (1), the curved surface of many discharge curve compositions is carried out to matching, obtain these undetermined parameters, experiment shows that model and actual battery discharge data error that the method is obtained are much smaller.

A lithium ion battery of producing taking company of Shenzhen below describes as example.The specification of this lithium ion battery is as shown in table 1.

This lithium ion battery is carried out to discharge test 3 times, before electric discharge, first need battery to charge, adopt 0.2C constant current to turn constant-voltage charge, after being full of electricity, carry out discharge test, discharge into final discharging voltage 2.75V, discharge current is respectively 0.2C (430mA), 0.5C (1075mA), 1C (2150mA), can obtain cell voltage under these three kinds of discharge currents in time/data that discharge electricity amount changes.According to the known max cap. Q=2.275Ah of discharge data.Can record internal resistance R=0.065 Ω by internal resistance test device.

Table 1: experiment lithium ion battery specification

According to these data, parameter is carried out to matching below:

When constant-current discharge, i=i ^*, according to formula (1), have:

$v_{\mathrm{batt}}=-R\·i+E_0-K\frac{Q}{Q-\mathrm{it}}(i+\mathrm{it})+\mathrm{Aexp}(-B\·\mathrm{it})---\left(3\right)$

Make v _batt=z, it=x, i=y, and R, Q substitution above formula, can obtain

$\begin{array}{c}z=f(x,y)=-0.065\·y+E_0-K\frac{2.257}{2.257-x}(x+y)\\ +\mathrm{Aexp}(-B\·x)\end{array}---\left(4\right)$

In formula, x is discharge electricity amount, and y is three discharge currents; Z is the magnitude of voltage that (x, y) is corresponding.Taking formula (4) as self-defining function is to data x, y, z carries out matching.Can obtain parameter is: E ₀=3.47, K=0.004209, A=0.7333, B=1.102.Fit procedure as shown in Figure 3, is actually three discharge curve data is carried out to surface fitting.

The required discharge curve of method when accompanying drawing 4 is 0.2C, 0.5C, 1C for discharge current in the method applied in the present invention and document 1 and the contrast of testing discharge curve.As can be seen from the figure, use the required discharge curve of the method applied in the present invention and experimental data to coincide better.

In order further to verify the accuracy of required parameter under different discharge currents, this lithium ion battery is carried out to the discharge test of 0.35C (0.753A) and 0.85C (1.828A).Accompanying drawing 5 is contrasts of the required discharge curve of method in the method applied in the present invention and document 1 and experiment discharge curve under these two kinds of discharge currents.As can be seen from the figure, use the method applied in the present invention required discharge curve to coincide better with experimental data mid-term at electric discharge initial stage and electric discharge, in the time discharging latter stage and experimental data have certain error.Note: for the discharge curve under different discharge currents is more easily distinguished, the horizontal ordinate of Fig. 4 and Fig. 5 does not use discharge electricity amount, but changed into the time (hour).

Filtered circuit i ^*add and make model become dynamic model, go for the situation of curent change.As shown in Figure 6, filtered circuit i ^*asked for through firstorder filter by battery current i.The timeconstantτ of this firstorder filter is 30s, and broken line slope is-20dB/10dec.Normal conditions, i ^*need to ask for according to all historical datas of i, but due in the application the present invention is directed to, can not obtain long, continuous battery current data, therefore cannot obtain all historical current data.For firstorder filter, can ask for i according to the data of current i in short time period ^*.The method of calculation of filtered electric current is as follows:

$i^{*}\left(t\right)={\∫}_0^t\left(\frac{1}{\mathrm{\τ}}{e}^{-\frac{x}{\mathrm{\τ}}}\right)\×i(t-x)\mathrm{dx}$

This time period t is got 3 τ conventionally to 5 τ, adopts the required i of i of different time sections ^*error can be referring to table 2.The present invention has adopted the current data of 4 τ (120s) times, and error is 1.8%.

Table 2: the not error between step response value and final value in the same time of firstorder circuit

Time period	3τ(90s)	4τ(120s)	5τ(150s)
				Error	5％	1.8％	0.7％

Cell voltage v _battwith the variation formula of discharge electricity amount it, battery current i, and determine parameter, can abstractly be

v _batt＝f(it,i,i ^*) (5)

Wherein i ^*obtained through firstorder filter by i.

Battery SOC can represent with formula (6):

$\mathrm{SOC}=(1-\frac{\mathrm{it}}{Q})\×100\%---\left(6\right)$

By formula (5) and formula (6), can know cell voltage v by inference _battvariation formula with SOC, battery current i:

v _batt＝f(SOC,i,i ^*) (7)

Due to the complicacy of equation, can not be explicit by formula (7) obtain

SOC＝f(v _batt,i,i ^*) (8)

The present invention takes following methods to obtain SOC:

At a time, i determines, i ^*can be obtained by the i in 4 τ (120s), also determine, can obtain

$v_{\mathrm{batt}}={f\left(\mathrm{SOC}\right)|}_{i=C1,i^{*}=C2}---\left(9\right)$

Can obtain i, i by formula (9) ^*v while determining _battcurve while variation with SOC, because the discharge curve shown in accompanying drawing 2 and formula (6) they are dull, the v therefore finally obtaining _battwith SOC be one to one, so capable of dynamic generates current i, i ^*lower v _battthe tables of data changing with SOC, tries to achieve current v by method of interpolation _battthe SOC value of unique correspondence.The overall flow of the battery SOC method of estimation based on filtered circuit of the present invention as shown in Figure 7.

Below by the validity of utilizing current cell voltage and the interior battery current data of 4 τ (120s) to carry out battery SOC estimation of verifying that the present invention proposes.First the validity of the method while verifying constant-current discharge, the validity of the method when secondly checking complex working condition discharges.

Show by battery constant current charging-discharging and complex working condition being discharged and recharged to experiment, SOC and actual SOC absolute error that method of the present invention is estimated are less than 5%.Therefore, the battery SOC method of estimation based on filtered circuit that the present invention proposes, according to the SOC of battery current in 4 τ and current cell voltage estimation battery, has realized battery has been carried out to SOC estimation.The method has following advantage: (1) does not need SOC initial value and historical SOC data, does not therefore have cumulative errors; (2) do not need the battery current value of overall process, battery current value and the current cell voltage that only need to pass by one period of short period (4 τ), just can estimate battery SOC; (3) be applicable to the complex working condition that battery current changes; (4) there is versatility.

Above content is in conjunction with concrete preferred implementation further description made for the present invention, can not assert that specific embodiment of the invention is confined to these explanations.For general technical staff of the technical field of the invention, without departing from the inventive concept of the premise, can also make some simple deduction or replace, all should be considered as belonging to protection scope of the present invention.

Claims (5)

1. the battery state of charge SOC method of estimation based on filtered circuit, its spy is: said method comprising the steps of:

S1: set up battery dynamic model, described dynamic model comprises firstorder filter, for asking for filtered circuit;

S2: obtain filtered circuit i ^*;

S3: utilize curved surface fitting method, obtain v according to the discharge curve of battery and internal resistance _batt=f (SOC, i, i ^*), wherein, i ^*be filtered circuit, i is the battery current in this moment;

S4: obtain current i, i ^*lower cell voltage v _battrelation with battery SOC

S5: dynamically generate current i, i ^*lower v _battthe tables of data changing with battery SOC;

S6: obtain not battery SOC in the same time by method of interpolation.

2. battery SOC method of estimation according to claim 1, is characterized in that: the voltage v of described battery model in the time of electric discharge and charging _battfor:

Wherein, R is the internal resistance of cell, and Q is battery max cap., and it is battery discharge electric weight, relevant to SOC, E ₀, K, A, B be 4 undetermined coefficients.

3. battery SOC method of estimation according to claim 1, is characterized in that: in described step S2, and filtered circuit i ^*obtained by the current i in 4 τ.

4. battery SOC method of estimation according to claim 1, is characterized in that: in described step S3, utilize curved surface fitting method rooting to obtain v according to discharge curve and the internal resistance of battery _batt=f (SOC, i, i ^*), be specially: utilize curved surface fitting method, try to achieve E ₀, K, A, B.

5. battery SOC method of estimation according to claim 1, is characterized in that: described method extends to lithium battery, lead-acid battery, nickel-cadmium battery and Ni-MH battery.