CN103558556B - A kind of power battery SOH estimation method - Google Patents

Abstract

The present invention proposes a kind of lithium battery SOH evaluation method, and it is the actual capacity and the battery ohmic internal resistance that utilize systematic parameter evaluation method to estimate battery; The actual capacity utilizing redundancy processing method process to estimate and ohmic internal resistance; Utilize the health status of actual capacity and ohmic internal resistance difference estimating battery, and obtain the real health status of battery by the relation weighting process between them; Adopt the health status value of filtration combined weighted disposal methods battery.The present invention can reach the object more accurately reflecting the real health status of battery.

Description

A kind of power battery SOH estimation method

Technical field

The invention belongs to new energy car battery management system field, relate to a kind of electrokinetic cell health status (SOH) evaluation method.

Background technology

Efficiently, the attention obtaining people today that the electric automobile of energy-saving and environmental protection goes from bad to worse at environment, petroleum-based energy is day by day exhausted also achieves large development, therefore accurately estimate that the health status (SOH, StateofHealth) of electrokinetic cell seems particularly important for the further optimization of the whole control strategy of electric automobile and further developing of electric automobile.

Battery as time goes by with the use of battery, battery occurs aging, cause having different battery behaviors at different times, be specifically reflected in the size of the SOH value of battery, therefore SOH value reflects the overall performance of battery intuitively and discharges the ability of electric energy under certain condition.For one piece of new battery, its SOH value is more than or equal to 100% often, because of expendable physical chemical factor cause battery slowly the aging SOH value that causes decline gradually.General computing formula be there is no to the calculating of SOH value at present, specify when battery capacity drops to 80% of new battery capacity with SOH=Qnow/Qnew computing formula in such as ieee standard 1188-1996, namely, during SOH<80%, battery just should be replaced.

Due to health status and the internal resistance of cell of battery, charging and discharging capabilities, the many factors such as self discharge are relevant, therefore can obtain the SOH value of battery by test in testing laboratory, but in battery use procedure, there is no the various conditions of testing laboratory, therefore the SOH value of battery directly cannot be obtained by measuring method, if the battery recycling of electric automobile to be carried out the measurement of SOH value to testing laboratory, a large amount of manpower and financial resources will be wasted, therefore obtain by battery characteristics parameter identification the method that battery SOH value is a human-saving and financial resources, therefore a lot of battery management system developer is is researching and developing the evaluation method of SOH at present, but also do not have ripe method to be applied to product at present.

Summary of the invention

The present invention is directed to the deficiencies in the prior art, propose a kind of expanded Kalman filtration algorithm estimating battery internal resistance based on battery status spatial model and battery capacity thus estimate battery SOH evaluation method, reaching the object more accurately reflecting the real health status of battery.

The present invention proposes a kind of lithium battery SOH evaluation method, and it is the actual capacity and the battery ohmic internal resistance that utilize systematic parameter evaluation method to estimate battery; The actual capacity utilizing redundancy processing method process to estimate and ohmic internal resistance; Utilize the health status of actual capacity and ohmic internal resistance difference estimating battery, and obtain the real health status of battery by the relation weighting process between them; Adopt the health status value of filtration combined weighted disposal methods battery.

It is as follows that evaluation method of the present invention comprises step:

Step 1: set up following state-space model according to electrokinetic cell circuit model and use Kalman filtering algorithm identification system parameter:

V _B(k)＝OCV(k)-R ₀(k)I _B(k)-V _p(k)+v(k)(2)

V in formula (1) _pbattery polarization voltage; OCV is battery open circuit voltage, R _oit is battery ohmic internal resistance; C _bit is battery actual capacity; for state-noise.V in formula (2) _bit is the terminal voltage of battery; I _bbe the electric current of battery, v is measurement noises.

This step embodies funtcional relationship OCV (k)=F (SOC (k) that open-circuit voltage OCV is battery state of charge (SOC) and temperature (T), T), namely consider that battery temperature and state-of-charge are on the impact of battery open circuit voltage.

Step 2: gather the electric current of lithium battery, voltage and temperature value, and use Recursive Formulas of Kalman Filter appraising model parameter R _o, C _b.

This step introduces filter gain correction factor matrix ξ=[ξ in Recursive Formulas of Kalman Filter ₁ξ ₂ξ ₃ξ ₄ξ ₅ξ ₆] and dynamic noise correction to variances coefficient concrete introducing formula is as follows:

$K\left(k\right)=\mathrm{\&xi;}*P(k+1|k){\hat{C}}^T(k+1){[\hat{C}(k+1)P(k+1|k){\hat{C}}^T(k+1)+{\mathrm{\&xi;}}_{R_k}*R_k]}^{-1}$

$P(k+1|k)=\hat{A}\left(k\right)P\left(k\right){\hat{A}}^T\left(k\right){\mathrm{\&xi;}}_{Q_k}*Q_k.$

In above formula, K is Kalman filtering gain matrix; P varivance matrix; it is state-transition matrix; observing matrix; R is process noise; Q is observation noise.

Step 3: when continuing N minute, start to calculate battery actual capacity C when electric current is between interval [-I, I] _bwith battery ohmic internal resistance R _oaccumulated value C=Σ C _b, R=Σ R _o.

Step 4: when electric current is between interval [-I, I] and when continuing N+M minute, then stop calculating battery actual capacity and battery ohmic internal resistance accumulated value, and calculate average cell actual capacity and average cell ohmic internal resistance $\stackrel{\&OverBar;}{C}=C/M,\stackrel{\&OverBar;}{R}=R/M,$ In formula, M is the duration.

Step 3,4 gets rid of the parameter transmitting case of identification algorithm estimation, ensures that the internal resistance of cell of estimation and battery capacity value converge on true value.

Step 5: utilize estimated value to calculate the health status (SOH) of battery:

${\mathrm{SOH}}_C=\frac{C_{\mathrm{New}}-\stackrel{\&OverBar;}{C}}{C_{\mathrm{New}}-C_{\mathrm{End}}},{\mathrm{SOH}}_R=1-\frac{\stackrel{\&OverBar;}{R}-R_{\mathrm{New}}}{{\stackrel{\&OverBar;}{R}}_{\mathrm{End}}-R_{\mathrm{New}}}$

C in formula _newnew electrokinetic cell actual capacity; C _endelectrokinetic cell is aged to performance and is discontented with actual capacity when car load needs; R _newnew electrokinetic cell internal resistance value; electrokinetic cell is aged to performance and is discontented with internal resistance value when car load needs;

Step 6: utilize method of weighting to obtain the SOH of battery, its weighted value is w, namely

SOH＝w*SOH _C+(1-w)*SOH _R

Step 5,6 utilizes weighting coefficient w to make battery capacity and internal resistance of cell combined reaction go out the real health status of battery.

Step 7: wavelet filteration method process is carried out to the SOH value of battery and meets car load application requirement, namely γ in formula _ibe wavelet filtering coefficient, m is wavelet coefficient number, SOH _iit is the health status in battery past.

Step 7 utilizes wavelet filteration method, and battery current health state is affected by health status in the past.

Compared with prior art, the invention has the advantages that:

1, the present invention utilizes System Discrimination algorithm to pick out the actual capacity of battery and the ohmic internal resistance of battery simultaneously, and utilizes these two important calculation of characteristic parameters to go out the SOH value of battery, more accurately can reflect the real health status of battery.

2, the present invention introduces dynamic noise variance and filter gain correction factor matrix makes identification algorithm possess stronger antijamming capability.

3, the present invention be applicable to various operating mode and estimation SOH value can reflect electrokinetic cell health status preferably.

The present invention is applicable to the estimation of various driven energy saving vapour Vehicular dynamic battery health status.

Accompanying drawing explanation

In order to be illustrated more clearly in example of the present invention or technical scheme of the prior art, be briefly described to the accompanying drawing used required in example or description of the prior art below.

Fig. 1 power battery SOH estimation method process flow diagram.

Embodiment

For the ease of the understanding of those skilled in the art, below in conjunction with accompanying drawing 1 and specific embodiment, enforcement journey mode of the present invention is described.

The method that the embodiment of the present invention proposes comprises the steps:

Step 1: set up following state-space model according to electrokinetic cell circuit model and use the Kalman filtering algorithm identification system parameter in system identifying method:

V _B(k)＝OCV(k)-R ₀(k)I _B(k)-V _p(k)+v(k)(2)

In formula (1) for state-noise; V is measurement noises; V _pbattery polarization voltage; OCV is battery open circuit voltage; R _oit is the internal resistance of cell; C _bit is battery actual capacity.V in formula (2) _bbe the terminal voltage of battery, open-circuit voltage OCV is funtcional relationship OCV (the k)=F (SOC (k), T) of battery state of charge and temperature.

Step 2: utilize the current/voltage temperature value gathered, and use Recursive Formulas of Kalman Filter appraising model parameter R _o, C _b.

Filter gain correction factor matrix ξ=[ξ is introduced in Recursive Formulas of Kalman Filter ₁ξ ₂ξ ₃ξ ₄ξ ₅ξ ₆] and dynamic noise correction to variances coefficient concrete introducing formula is as follows:

$K\left(k\right)=\mathrm{\&xi;}*P(k+1|k){\hat{C}}^T(k+1){[\hat{C}(k+1)P(k+1|k){\hat{C}}^T(k+1)+{\mathrm{\&xi;}}_{R_k}*R_k]}^{-1}$

$P(k+1|k)=\hat{A}\left(k\right)P\left(k\right){\hat{A}}^T\left(k\right){\mathrm{\&xi;}}_{Q_k}*Q_k$

Step 3: when continuing N minute, start to calculate battery actual capacity C when electric current is between interval [-I, I] _bwith battery ohmic internal resistance R _oaccumulated value C=Σ C _b, R=Σ R _o.

Step 4: when electric current is between interval [-I, I] and when continuing N+M minute, then stop calculating battery actual capacity and battery ohmic internal resistance accumulated value, and calculate average cell actual capacity and average cell ohmic internal resistance $\stackrel{\&OverBar;}{C}=C/M,\stackrel{\&OverBar;}{R}=R/M,$ In formula, M is the duration.

Step 5: utilize estimated value to calculate the health status (SOH) of battery:

${\mathrm{SOH}}_C=\frac{C_{\mathrm{New}}-\stackrel{\&OverBar;}{C}}{C_{\mathrm{New}}-C_{\mathrm{End}}},{\mathrm{SOH}}_R=1-\frac{\stackrel{\&OverBar;}{R}-R_{\mathrm{New}}}{{\stackrel{\&OverBar;}{R}}_{\mathrm{End}}-R_{\mathrm{New}}}$

C in formula _newnew electrokinetic cell actual capacity; C _endelectrokinetic cell is aged to performance and is discontented with actual capacity when car load needs; R _newnew electrokinetic cell internal resistance value; electrokinetic cell is aged to performance and is discontented with internal resistance value when car load needs;

Step 6: utilize method of weighting to obtain the SOH of battery, its weighted value is w, namely

SOH＝w*SOH _C+(1-w)*SOH _R

Step 7: wavelet filteration method process is carried out to the SOH value of battery and meets car load application requirement, namely γ in formula _ibe wavelet filtering coefficient, m is wavelet coefficient number, SOH _iit is the health status in battery past.

Claims (1)

1. a lithium battery health status evaluation method, is characterized in that, described method is the actual capacity and the battery ohmic internal resistance that utilize systematic parameter evaluation method to estimate battery; The actual capacity utilizing redundancy processing method process to estimate and battery ohmic internal resistance; Utilize the health status of actual capacity and battery ohmic internal resistance difference estimating battery, and obtain the real health status of battery by the relation weighting process between them; Finally adopt the health status value of filtration combined weighted disposal methods battery;

It is as follows that described evaluation method comprises step:

Step 1: set up following state-space model according to electrokinetic cell circuit model and use Kalman filtering algorithm identification system parameter:

V _B(k)＝OCV(k)-R ₀(k)I _B(k)-V _p(k)+v(k)(2)

V in formula (1) _pbattery polarization voltage; OCV is battery open circuit voltage, R _oit is battery ohmic internal resistance; C _bit is battery actual capacity; for state-noise; V in formula (2) _bit is the terminal voltage of battery; I _bbe the electric current of battery, v is measurement noises; Open-circuit voltage OCV is funtcional relationship OCV (the k)=F (SOC (k), T) of battery state of charge SOC and temperature T;

Step 2: detect lithium battery electric current, voltage and temperature value, and use Recursive Formulas of Kalman Filter appraising model parameter R _o, C _b;

Filter gain correction factor matrix ξ=[ξ is introduced in Recursive Formulas of Kalman Filter ₁ξ ₂ξ ₃ξ ₄ξ ₅ξ ₆] and dynamic noise correction to variances coefficient concrete introducing formula is as follows:

$K\left(k\right)=\mathrm{\&xi;}*P(k+1|k){\hat{C}}^T(k+1){\&lsqb;\hat{C}(k+1)P(k+1|k){\hat{C}}^T(k+1)+{\mathrm{\&xi;}}_{R_k}*R_k\&rsqb;}^{-1}$

$P\left(k+1|k\right)=\hat{A}\left(k\right)P\left(k\right){\hat{A}}^T\left(k\right)+{\mathrm{\&xi;}}_{Q_k}*Q_k$

In above formula, K is Kalman filtering gain matrix; P varivance matrix; it is state-transition matrix; observing matrix; R is process noise; Q is observation noise;

Step 3: when continuing N minute, start to calculate battery actual capacity C when electric current is between interval [-I, I] _bwith battery ohmic internal resistance R _oaccumulated value C=Σ C _b, R=Σ R _o;

Step 4: when electric current is between interval [-I, I] and when continuing N+M minute, then stop calculating battery actual capacity and battery ohmic internal resistance accumulated value, and calculate average cell actual capacity and average cell ohmic internal resistance in formula, M is the duration;

Step 5: utilize estimated value to calculate the health status SOH of battery:

${\mathrm{SOH}}_C=\frac{C_{New}-\stackrel{\&OverBar;}{C}}{C_{New}-C_{End}},{\mathrm{SOH}}_R=1-\frac{\stackrel{\&OverBar;}{R}-R_{New}}{{\stackrel{\&OverBar;}{R}}_{End}-R_{New}}$

C in formula _newnew electrokinetic cell actual capacity; C _endelectrokinetic cell is aged to performance and is discontented with actual capacity when car load needs; R _newnew electrokinetic cell internal resistance value; electrokinetic cell is aged to performance and is discontented with internal resistance value when car load needs;

Step 6: utilize method of weighting to obtain the SOH of battery, its weighted value is w, namely

SOH＝w*SOH _C+(1-w)*SOH _R

Step 7: wavelet filteration method process is carried out to the SOH value of battery, namely γ in formula _ibe wavelet filtering coefficient, m is wavelet coefficient number, SOH _iit is the health status in battery past.