CN105226744A - A kind of power battery pack balance charge/discharge control method based on SOC and system - Google Patents

Abstract

The invention discloses a kind of power battery pack balance charge/discharge control method based on SOC and device, wherein, the method comprises: obtain the discharge and recharge data of electrokinetic cell, and discharge and recharge data comprise the terminal voltage of electrokinetic cell, electric current and temperature; Electrokinetic cell comprises battery cell; Calculate the Mathematical Modeling of electrokinetic cell SOC based on ELM method establishment, using the input of discharge and recharge data as Mathematical Modeling, and determine electrokinetic cell SOC data according to the output of Mathematical Modeling; According to electrokinetic cell SOC data, consistency analysis is carried out to power battery pack, when power battery pack exists inconsistent according to the on off operating mode of the switching tube of Balance route policy control equalizing circuit, realize electrokinetic cell energy trasfer.The method can predict the state-of-charge that battery is current more accurately and rapidly, effectively improves the accuracy of equalization and equalization efficiency.

Description

A kind of power battery pack balance charge/discharge control method based on SOC and system

Technical field

The present invention relates to battery pack balancing control technology field, particularly, relate to a kind of power battery pack balance charge/discharge control method based on SOC and system.

Background technology

Because lithium dynamical battery has the advantages such as energy density is high, self-discharge rate is low, have extended cycle life, be widely used in energy storage field at present.In electrokinetic cell use procedure, need more piece cell by providing energy for load after series connection.Now the difference of cell performance can cause battery pack to occur inconsistency phenomenon, and the inconsistency of battery pack can cause the decline of battery pack serviceability, cause active volume and the decay in useful life of battery pack, thus reduce the continual mileage of pure electric automobile, increase use cost.Therefore balancing technique is studied, the adverse effect that the inconsistency can effectively improving battery pack is brought, avoid the deterioration of in use battery pack inconsistency, increase the useful life of battery, reduce use cost, promote the memory property of battery.

According to different to the processing mode of transmitted energy, equalizing circuit can be divided into passive-type and active.The mode loss portion of energy that passive-type circuit utilizes its bypass resistance to carry out discharging mainly through making the higher battery of energy in battery pack, to reaching the consistent of battery pack power state.This method easily realizes, but causes a large amount of energy losses, and heat dissipation problem is difficult to solve.Active circuit is utilize energy-storage travelling wave tube and balanced bypass to build energy transferring passage in essence, and it is directly or indirectly transferred to the lower battery of energy from the higher battery of energy.Such as, speed-sensitive switch capacitance technology utilizes group capacitor transferring energy between the adjacent energy storage monomer of series connection accumulation power supply group, and efficiency is high, controls simple, but energy difference is apart from time less when between monomer, and balanced required time is longer.The energy transfer pathway of two-way One Buck-Boost converter body equalizing structure is two-way, can carry out energy transferring between adjacent two battery cells, but when battery cell is more in organizing, is difficult to the energy transferring realizing great-leap-forward, and equalization efficiency reduces.The distributed balancing technique of isolated DC/DC converter adopts isolated transformer configuration, and modal is inverse-excitation type structure.The method efficiency is higher, and control circuit is simple, but in circuit, number transformer is more, the equalizing system structure relative complex of structure.

In order to meet the power of electric automobile, the requirement of voltage, need more piece cell to be formed battery pack for its energy supply.Battery mode in groups has series, parallel and series-parallel connection three kinds, and wherein series system is the grouping method that current most of electric motor car adopts.When number of batteries is more, the balanced topology of single one deck, due to the restriction of self circuit, is difficult to design the equalizing system with superperformance.

Balanced variable is that equalizing system makes it to reach consistent battery characteristics parameter by controlling equalizing circuit, can the inconsistency state of characterizing battery, and has the feature such as real-time, high accuracy.According to the difference of balanced variable, equalization methods can be divided into capacity equilibrium method, chemical equalization, voltage balancing method, SOC (stateofcharge, state-of-charge) equalization.Early stage employing capacity equilibrium method, this method, for the purpose of battery actual capacity reaches unanimity, adopts float charge voltage to the continuous charging of battery pack, but causes battery to be in overcharge condition due to the method, shortens battery life, seldom uses at present.Chemistry equalization reaches portfolio effect by inside battery chemical reaction, adds the method that certain proportion oxidationreduction electricity is right in lithium battery electrolytes, suppresses anode current potential to raise, avoid electrode material and electrolyte oxidation, improves battery anti-over-charging ability.The method is also in theoretical research stage, in practical application, have a segment distance.A lot of Equilibrium Research is using voltage as balanced variable at present, in lead-acid battery, Ni-MH battery, effect is better, but due to the own characteristic of ferric phosphate lithium cell, voltage truly can not reflect the consistency of battery capacity state, and by inside battery various factors, portfolio effect is unstable, is easy to fluctuation.

SOC characterizes the ratio that present battery residual capacity accounts for maximum available, during using SOC as balanced variable, the difference of maximum available between cell in battery pack can be ignored, make all cells reach discharge and recharge cut-ff voltage simultaneously, battery capacity is utilized effectively.Meanwhile, when the SOC of battery is consistent, mean that all monomers all work in identical depth of discharge, avoid the difference of the cell degradation speed caused because depth of discharge is different.But the equalization methods based on SOC needs the state-of-charge predicting cell rapidly and accurately, prior art is difficult to real-time tracking lithium battery complexity internal-response fast.If the accuracy of SOC and real-time can not be guaranteed, balanced reliability can reduce greatly.

Summary of the invention

The present invention is to overcome in prior art the defect can not monitoring battery pack SOC change rapidly and accurately, according to an aspect of the present invention, proposing a kind of power battery pack balance charge/discharge control method based on SOC.

A kind of power battery pack balance charge/discharge control method based on SOC that the embodiment of the present invention provides, comprising:

Obtain the discharge and recharge data of electrokinetic cell, discharge and recharge data comprise the terminal voltage of electrokinetic cell, electric current and temperature; Electrokinetic cell comprises battery cell;

Calculate the Mathematical Modeling of electrokinetic cell SOC based on ELM method establishment, using the input of discharge and recharge data as Mathematical Modeling, and determine electrokinetic cell SOC data according to the output of Mathematical Modeling;

According to electrokinetic cell SOC data, consistency analysis is carried out to power battery pack, when power battery pack exists inconsistent according to the on off operating mode of the switching tube of Balance route policy control equalizing circuit, realize electrokinetic cell energy trasfer.

In technique scheme, calculate the Mathematical Modeling of electrokinetic cell SOC based on ELM method establishment, comprising:

The input vector of Mathematical Modeling is x _j=[x _j1, x _j2..., x _jn] ^t∈ R ⁿ, output vector is t _j=[t _j1, t _j2..., t _jm] ^t∈ R ^m, wherein m, n represent the dimension of input and output respectively, R ⁿrepresent n-dimensional space, R ^mrepresent m-dimensional space, j=1 ..., N, N are sample number;

Mathematical Modeling is: H β=T, and in formula, H is the hidden layer output matrix of neural net:

$H({\mathrm{\ω}}_1,...,{\mathrm{\ω}}_{\tilde{N}},b_1,...,b_{\tilde{N}},x_1,...,x_N)={\left[\begin{array}{ccc}g({\mathrm{\ω}}_1\·x_1+b_1)& ...& g({\mathrm{\ω}}_{\tilde{N}}\·x_1+b_{\tilde{N}})\\ \begin{array}{c}.\\ .\\ .\end{array}& \begin{array}{c}\\ ...\\ \end{array}& \begin{array}{c}.\\ .\\ .\end{array}\\ g({\mathrm{\ω}}_1\·x_N+b_1)& ...& g({\mathrm{\ω}}_{\tilde{N}}\·x_N+b_{\tilde{N}})\end{array}\right]}_{N\×\tilde{N}},$

$\mathrm{\β}={\left[\begin{array}{c}{\mathrm{\β}}_1^T\\ \begin{array}{c}.\\ .\\ .\end{array}\\ {\mathrm{\β}}_{\tilde{N}}^T\end{array}\right]}_{\tilde{N}\×m},T={\left[\begin{array}{c}{t}_1^T\\ \begin{array}{c}.\\ .\\ .\end{array}\\ t_N^T\end{array}\right]}_{N\×m};$

Wherein, hidden node number is excitation function is g (x), ω _i=[ω _i1, ω _i2..., ω _in] be the weight vector of connection i-th hidden layer node and input node; β _i=[β _i1, β _i2..., β _im] ^tit is the weight vector of connection i-th hidden node and output node; b _iit is the threshold value of i-th hidden node.

In technique scheme, determine electrokinetic cell SOC data according to the output of Mathematical Modeling, comprising:

Another input vector x _jfor [U (k), I (k), C (k)], output vector t _jfor SOC (k)=t _j, wherein, k represents the label of the kth group training data collected, and U, I, C represent the terminal voltage of electrokinetic cell, electric current and temperature respectively; The training set of Mathematical Modeling is { (x _j, t _j) | x _j∈ R ⁿ, t _j∈ R ^m, j=1 ..., N} (n=3, m=1);

The least square solution determining linear system H β=T is equivalent to according to training set training SLFN

wherein, H ⁺for the Moore-Penrose generalized inverse of matrix H;

Obtaining least square solution after, according to least square solution determine electrokinetic cell SOC data.

In technique scheme, equalizing circuit comprises top layer circuit and bottom circuit; Balance route strategy comprises top layer control strategy and bottom control strategy, is respectively used to control top layer circuit and bottom circuit.

In technique scheme, electrokinetic cell also comprises battery module, and battery module is made up of multiple battery cell;

According to the on off operating mode of the switching tube of Balance route policy control equalizing circuit, comprising:

Obtain balanced variable, balanced variable comprises the difference of the difference of state-of-charge, the state-of-charge of certain battery module and battery module state-of-charge mean value between the mean square deviation of battery cell SOC in group, adjacent two battery cells;

When the mean square deviation of battery cell SOC is not less than the first predetermined threshold value in organizing, meet Balance route unlocking condition;

When between adjacent two battery cells, the difference of state-of-charge is greater than the second predetermined threshold value, according to bottom control policy control bottom circuit;

When the state-of-charge of certain battery module and the difference of battery module state-of-charge mean value are greater than the 3rd predetermined threshold value, control top layer circuit according to top layer control strategy.

In technique scheme, obtain balanced variable, comprising:

If altogether containing n battery cell in battery pack, be divided into m battery module, determine balanced variable according to following formula:

$\stackrel{\‾}{SOC}=\frac{1}{n}\underset{i=1}{\overset{n}{\Σ}}{\mathrm{SOC}}_i$

$\mathrm{\ϵ}=\sqrt{\frac{1}{n}\underset{i=1}{\overset{n}{\Σ}}({\mathrm{SOC}}_i-\stackrel{\‾}{SOC})}$

ΔSOC＝|SOC _i-SOC _i+1|

$\stackrel{\‾}{{\mathrm{SOC}}^{\′}}=\frac{1}{m}\underset{l=1}{\overset{m}{\Σ}}{\mathrm{SOC}}_l^{\′}$

${\mathrm{\ΔSOC}}^{\′}=|{\mathrm{SOC}}_l^{\′}-\stackrel{\‾}{{\mathrm{SOC}}^{\′}}|$

Wherein, for organizing the mean value of interior battery cell state-of-charge, the ε mean square deviation that is battery cell SOC in group, the Δ SOC difference that is state-of-charge between adjacent two battery cells, for the difference of the mean value of battery module state-of-charge, state-of-charge that Δ SOC ' is certain battery module and battery module state-of-charge mean value.

The power battery pack balance charge/discharge control method based on SOC that the embodiment of the present invention provides, compared with traditional neural net, have that structure is simple, pace of learning is fast, parameter easily adjusts and is not easily absorbed in the advantages such as Local Minimum, there is features such as approaching Multiinputoutput parametric function, non-linear, the robustness of height and fault-tolerance.The setting parameter of ELM is more simple, and operand is little, has the fast and advantage that Generalization Capability is good of pace of learning.Therefore, the SOC estimation method based on ELM is applied in the embodiment of the present invention, the state-of-charge that battery is current can be predicted more accurately and rapidly, effectively improve the accuracy of equalization and equalization efficiency.Be several little battery modules by battery components simultaneously, during Homogeneity between groups, can a module be regarded as a joint monomer, adopt double-deck topological structure to realize balanced.Based on the double-deck active equalization system of SOC, control the switching tube in top layer and bottom circuit respectively, the bi-directional of battery pack self-energy can be realized, realize the Balance route of battery pack fast and efficiently.

The embodiment of the present invention also provides a kind of power battery pack balance charge/discharge control system based on SOC, comprise: battery pack supervisory controller, SOC module, for the top layer circuit of Balance route battery pack and bottom circuit, altogether containing n battery cell in battery pack, be divided into m battery module, battery module is made up of multiple battery cell;

Bottom circuit comprises n two-way Buck-Boost circuit, and two-way Buck-Boost circuit and battery cell one_to_one corresponding;

Top layer circuit comprises m flyback transformer, and flyback transformer and battery module one_to_one corresponding;

Discharge and recharge data for gathering the discharge and recharge data of electrokinetic cell, and are sent to SOC module by battery pack supervisory controller, and discharge and recharge data comprise the terminal voltage of electrokinetic cell, electric current and temperature; Electrokinetic cell comprises battery cell and battery module;

SOC module is used for the Mathematical Modeling calculating electrokinetic cell SOC based on ELM method establishment, using the input of discharge and recharge data as Mathematical Modeling, and determines electrokinetic cell SOC data according to the output of Mathematical Modeling; According to electrokinetic cell SOC data, consistency analysis is carried out to power battery pack, when power battery pack exists inconsistent according to the on off operating mode of the switching tube of Balance route policy control equalizing circuit, realize electrokinetic cell energy trasfer.

In technique scheme, two-way Buck-Boost circuit comprises: the first switching tube, the first fly-wheel diode, energy storage inductor; First switching tube is connected with energy storage inductor with after the first fly-wheel diode parallel connection, and two-way Buck-Boost circuit is in parallel with battery cell, and two-way Buck-Boost circuit and adjacent two-way Buck-Boost circuit share energy storage inductor;

Flyback transformer comprises: second switch pipe, the second fly-wheel diode, transformer; The former limit winding of transformer is connected rear in parallel with power module with second switch pipe, the vice-side winding of transformer is connected with the second fly-wheel diode.

In technique scheme, calculate the Mathematical Modeling of electrokinetic cell SOC based on ELM method establishment, comprising:

The input vector of Mathematical Modeling is x _j=[x _j1, x _j2..., x _jn] ^t∈ R ⁿ, output vector is t _j=[t _j1, t _j2..., t _jm] ^t∈ R ^m, wherein m, n represent the dimension of input and output respectively, R ⁿrepresent n-dimensional space, R ^mrepresent m-dimensional space, j=1 ..., N, N are sample number;

Mathematical Modeling is: H β=T, and in formula, H is the hidden layer output matrix of neural net:

$H({\mathrm{\ω}}_1,...,{\mathrm{\ω}}_{\tilde{N}},b_1,...,b_{\tilde{N}},x_1,...,x_N)={\left[\begin{array}{ccc}g({\mathrm{\ω}}_1\·x_1+b_1)& ...& g({\mathrm{\ω}}_{\tilde{N}}\·x_1+b_{\tilde{N}})\\ \begin{array}{c}.\\ .\\ .\end{array}& \begin{array}{c}\\ ...\\ \end{array}& \begin{array}{c}.\\ .\\ .\end{array}\\ g({\mathrm{\ω}}_1\·x_N+b_1)& ...& g({\mathrm{\ω}}_{\tilde{N}}\·x_N+b_{\tilde{N}})\end{array}\right]}_{N\×\tilde{N}},$

$\mathrm{\β}={\left[\begin{array}{c}{\mathrm{\β}}_1^T\\ \begin{array}{c}.\\ .\\ .\end{array}\\ {\mathrm{\β}}_{\tilde{N}}^T\end{array}\right]}_{\tilde{N}\×m},T={\left[\begin{array}{c}{t}_1^T\\ \begin{array}{c}.\\ .\\ .\end{array}\\ t_N^T\end{array}\right]}_{N\×m};$

Wherein, hidden node number is excitation function is g (x), ω _i=[ω _i1, ω _i2..., ω _in] be the weight vector of connection i-th hidden layer node and input node; β _i=[β _i1, β _i2..., β _im] ^tit is the weight vector of connection i-th hidden node and output node; b _iit is the threshold value of i-th hidden node.

In technique scheme, determine electrokinetic cell SOC data according to the output of Mathematical Modeling, comprising:

Another input vector x _jfor [U (k), I (k), C (k)], output vector t _jfor SOC (k)=t _j, wherein, k represents the label of the kth group training data collected, and U, I, C represent the terminal voltage of electrokinetic cell, electric current and temperature respectively; The training set of Mathematical Modeling is { (x _j, t _j) | x _j∈ R ⁿ, t _j∈ R ^m, j=1 ..., N} (n=3, m=1);

The least square solution determining linear system H β=T is equivalent to according to training set training SLFN

wherein, H ⁺for the Moore-Penrose generalized inverse of matrix H;

Obtaining least square solution after, according to least square solution determine electrokinetic cell SOC data.

In the embodiment of the present invention, voltage is there is high for top layer Balance route, the problems such as battery pack quantity is many, the embodiment of the present invention adopts the top layer circuit based on flyback transformer, the bidirectional equalization topological structure that the topological structure of this top layer circuit is made up of anti exciting converter and switch mosfet pipe, its balanced object is the battery module of some battery cells composition.This topological structure can realize the battery module energy back of overtension in integral battery door group.Fly-back converter circuit balancing speed is fast, cost is low, structure is simple, and circuit adopts isolated topological structure simultaneously, can effectively realize electrical isolation between battery pack, reduces electromagnetic compatibility impact.Top layer circuit adopts based on the equalizing circuit topological structure of inverse excitation type converter, by the mutual conversion of electric energy and magnetic energy, realizes the energy transferring between battery module.When batteries charging, when certain battery module energy is too high, using inverse excitation type converter as energy trasfer medium, give whole battery pack by energy transferring unnecessary for this battery module, prevent because energy overcharges the harm caused battery.

Other features and advantages of the present invention will be set forth in the following description, and, partly become apparent from specification, or understand by implementing the present invention.Object of the present invention and other advantages realize by structure specifically noted in write specification, claims and accompanying drawing and obtain.

Below by drawings and Examples, technical scheme of the present invention is described in further detail.

Accompanying drawing explanation

Accompanying drawing is used to provide a further understanding of the present invention, and forms a part for specification, together with embodiments of the present invention for explaining the present invention, is not construed as limiting the invention.In the accompanying drawings:

Fig. 1 is embodiment of the present invention medium power battery pack balancing charge/discharge control method flow chart;

Fig. 2 is battery SOC model schematic in the embodiment of the present invention;

Fig. 3 is the distribution map of battery cell in the embodiment of the present invention;

Fig. 4 is the control flow schematic diagram of equalizing system in the embodiment of the present invention;

Fig. 5 is embodiment of the present invention medium power battery pack balancing charge-discharge control system structure chart;

Fig. 6 is bottom circuit diagram in the embodiment of the present invention;

Fig. 7 is top layer circuit diagram in the embodiment of the present invention;

Fig. 8 is the battery monitor circuit theory diagrams based on LTC6804 in the embodiment of the present invention;

Fig. 9 is the top layer circuit theory diagrams based on LT8584 in the embodiment of the present invention;

Figure 10 is the workflow of embodiment of the present invention medium power battery pack balancing charge-discharge control system.

Embodiment

Below in conjunction with accompanying drawing, the specific embodiment of the present invention is described in detail, but is to be understood that protection scope of the present invention not by the restriction of embodiment.

According to the embodiment of the present invention, provide a kind of power battery pack balance charge/discharge control method based on SOC, shown in Figure 1, the method comprises:

Step 101: obtain the discharge and recharge data of electrokinetic cell, discharge and recharge data comprise the terminal voltage of electrokinetic cell, electric current and temperature; Electrokinetic cell comprises battery cell.

Step 102: the Mathematical Modeling calculating electrokinetic cell SOC based on ELM method establishment, using the input of discharge and recharge data as Mathematical Modeling, and determines electrokinetic cell SOC data according to the output of Mathematical Modeling.

Step 103: carry out consistency analysis to power battery pack according to electrokinetic cell SOC data, when power battery pack exists inconsistent according to the on off operating mode of the switching tube of Balance route policy control equalizing circuit, realizes electrokinetic cell energy trasfer.

Wherein, calculate the Mathematical Modeling of electrokinetic cell SOC in step 102 based on ELM method establishment, specifically comprise:

Suppose that known N number of sample is (x _j, t _j), the input vector of the Mathematical Modeling of this electrokinetic cell SOC is x _j=[x _j1, x _j2..., x _jn] ^t∈ R ⁿ, output vector is t _j=[t _j1, t _j2..., t _jm] ^t∈ R ^m, wherein m, n represent the dimension of input and output respectively, R ⁿrepresent n-dimensional space, R ^mrepresent m-dimensional space, j=1 ..., N, N are sample number.Then hidden node number is excitation function is the Mathematical Modeling of the standard Single hidden layer feedforward neural networks (SINGLE-HIDDENLAYERFEEDFORWARDNEURALNETWORK, SLFNs) of g (x):

$\underset{i=1}{\overset{\tilde{N}}{\Σ}}{\mathrm{\β}}_i{g}_i\left(x_j\right)=\underset{i=1}{\overset{\tilde{N}}{\Σ}}{\mathrm{\β}}_{i}g({\mathrm{\ω}}_i\·x_j+b_i)=o_j---\left(1\right)$

Wherein, ω _i=[ω _i1, ω _i2..., ω _in] be the weight vector of connection i-th hidden layer node and input node; β _i=[β _i1, β _i2..., β _im] ^tit is the weight vector of connection i-th hidden node and output node; b _ibe the threshold value of i-th hidden node, o _jthe output desired value that neural computing draws, namely desired output.

Export zero error to make network and approach desired output, mean again because w and b is constant value, so g _i(x _j) g (ω can be expressed as _ix _j+ b _i) function, that is:

$\underset{i=1}{\overset{\tilde{N}}{\Σ}}{\mathrm{\β}}_{i}g({\mathrm{\ω}}_i\·x_j+b_i)=t_j,j=1,...,N---\left(2\right)$

Formula (2) can be abbreviated as following equation:

Hβ＝T(3)

In formula (3), H is the hidden layer output matrix of neural net:

$H({\mathrm{\ω}}_1,...,{\mathrm{\ω}}_{\tilde{N}},b_1,...,b_{\tilde{N}},x_1,...,x_N)={\left[\begin{array}{ccc}g({\mathrm{\ω}}_1\·x_1+b_1)& ...& g({\mathrm{\ω}}_{\tilde{N}}\·x_1+b_{\tilde{N}})\\ \begin{array}{c}.\\ .\\ .\end{array}& \begin{array}{c}\\ ...\\ \end{array}& \begin{array}{c}.\\ .\\ .\end{array}\\ g({\mathrm{\ω}}_1\·x_N+b_1)& ...& g({\mathrm{\ω}}_{\tilde{N}}\·x_N+b_{\tilde{N}})\end{array}\right]}_{N\×\tilde{N}}---\left(4\right)$

$\mathrm{\β}={\left[\begin{array}{c}{\mathrm{\β}}_1^T\\ \begin{array}{c}.\\ .\\ .\end{array}\\ {\mathrm{\β}}_{\tilde{N}}^T\end{array}\right]}_{\tilde{N}\×m},T={\left[\begin{array}{c}{t}_1^T\\ \begin{array}{c}.\\ .\\ .\end{array}\\ t_N^T\end{array}\right]}_{N\×m}---\left(5\right)$

Wherein, hidden node number is excitation function is g (x), ω _i=[ω _i1, ω _i2..., ω _in] be the weight vector of connection i-th hidden layer node and input node; β _i=[β _i1, β _i2..., β _im] ^tit is the weight vector of connection i-th hidden node and output node; b _iit is the threshold value of i-th hidden node.

Compared with conventional function approximation theory, ELM algorithm produces the threshold value of connection weights between input layer and hidden layer and hidden layer neuron at random, and in the training process without the need to adjustment, only need arrange the number of hidden layer neuron, just can obtain good output.

Input weights ω _iwith hidden layer threshold value b _iafter determining, training SLFN is just equivalent to the least square solution finding linear system H β=T

$\left|\right|H({\mathrm{\ω}}_1,...,{\mathrm{\ω}}_{\tilde{N}},b_1,...,b_{\tilde{N}})\widehat{\mathrm{\β}}-T\left|\right|=\underset{\mathrm{\β}}{\min }\left|\right|H({\mathrm{\ω}}_1,...,{\mathrm{\ω}}_{\tilde{N}},b_1,...,b_{\tilde{N}})\mathrm{\β}-T\left|\right|---\left(6\right)$

From the generalized inverse definition of Moore-Penrose.The LS solution of the least norm of above-mentioned linear system is:

$\widehat{\mathrm{\β}}=H^{+}T---\left(7\right)$

H in formula ⁺for the Moore-Penrose generalized inverse of matrix H.

Using the input as model of the terminal voltage of electrokinetic cell, electric current and temperature in the embodiment of the present invention, using SOC data as the output of model.Namely

p(k)＝[U(k),I(k),C(k)](8)

In formula, k represents the label of the kth group training data collected, and U, I, C represent the terminal voltage of battery, electric current and temperature respectively.Make p (k)=x _j, SOC (k)=t _j, then training set is just { (x _j, t _j) | x _j∈ R ⁿ, t _j∈ R ^m, j=1 ..., N} (n=3, m=1).The battery SOC model set up as shown in Figure 2.

The power battery pack balance charge/discharge control method based on SOC that the embodiment of the present invention provides, compared with traditional neural net, have that structure is simple, pace of learning is fast, parameter easily adjusts and is not easily absorbed in the advantages such as Local Minimum, there is features such as approaching Multiinputoutput parametric function, non-linear, the robustness of height and fault-tolerance.The setting parameter of ELM is more simple, and operand is little, has the fast and advantage that Generalization Capability is good of pace of learning.Therefore, the SOC estimation method based on ELM is applied in the embodiment of the present invention, the state-of-charge that battery is current can be predicted more accurately and rapidly, effectively improve the accuracy of equalization and equalization efficiency.

Preferably, this equalizing circuit comprises top layer circuit and bottom circuit; Accordingly, Balance route strategy comprises top layer control strategy and bottom control strategy, is respectively used to control top layer circuit and bottom circuit.

Meanwhile, electrokinetic cell also comprises battery module, and battery module is made up of multiple battery cell.Namely while obtaining the discharge and recharge data of cell in a step 101, also obtain the discharge and recharge data of battery module, and then the SOC data of this battery module can be determined according to the SOC Mathematical Modeling in step 102.

Now, according to the on off operating mode of the switching tube of Balance route policy control equalizing circuit, specifically comprise: obtain balanced variable, balanced variable comprises the difference of the difference of state-of-charge, the state-of-charge of certain battery module and battery module state-of-charge mean value between the mean square deviation of battery cell SOC in group, adjacent two battery cells; When the mean square deviation of battery cell SOC is not less than the first predetermined threshold value in organizing, meet Balance route unlocking condition; When between adjacent two battery cells, the difference of state-of-charge is greater than the second predetermined threshold value, according to bottom control policy control bottom circuit; When the state-of-charge of certain battery module and the difference of battery module state-of-charge mean value are greater than the 3rd predetermined threshold value, control top layer circuit according to top layer control strategy.

Concrete, if altogether containing n battery cell in battery pack, be divided into m battery module, the distribution map of battery cell can be shown in Figure 3, wherein Fig. 3 is only a kind of execution mode, other distribution maps (such as battery pack comprises altogether n battery cell, and this n battery cell is divided into m battery module) can be adopted, do not describe in detail in literary composition.In embodiments of the present invention, balanced variable is determined according to following formula:

$\stackrel{\‾}{SOC}=\frac{1}{n}\underset{i=1}{\overset{n}{\Σ}}{\mathrm{SOC}}_i---\left(9\right)$

$\mathrm{\ϵ}=\sqrt{\frac{1}{n}\underset{i=1}{\overset{n}{\Σ}}({\mathrm{SOC}}_i-\stackrel{\‾}{SOC})}---\left(10\right)$

ΔSOC＝|SOC _i-SOC _i+1|(11)

$\stackrel{\‾}{{\mathrm{SOC}}^{\′}}=\frac{1}{m}\underset{l=1}{\overset{m}{\Σ}}{\mathrm{SOC}}_l^{\′}---\left(12\right)$

${\mathrm{\ΔSOC}}^{\′}=|{\mathrm{SOC}}_l^{\′}-\stackrel{\‾}{{\mathrm{SOC}}^{\′}}|---\left(13\right)$

Wherein, for organizing the mean value of interior battery cell state-of-charge, the ε mean square deviation that is battery cell SOC in group, the Δ SOC difference that is state-of-charge between adjacent two battery cells, for the difference of the mean value of battery module state-of-charge, state-of-charge that Δ SOC ' is certain battery module and battery module state-of-charge mean value.

Mean square deviation ε can represent the inconsistency of battery pack state-of-charge.In the embodiment of the present invention, the condition that equilibrium is opened is set as: mean square deviation ε >=γ.After system judges that battery pack meets balanced condition of opening, equalization operation will be carried out to battery pack.When top controller monitors Δ SOC ' > β of certain module, then corresponding top layer circuit is started working; When bottom controller monitors the Δ SOC > η in certain module between adjacent two monomers, then open corresponding bottom circuit.Wherein γ, β, η are three threshold values (i.e. the first predetermined threshold value, the second predetermined threshold value, the 3rd predetermined threshold value) set in Balance route strategy respectively.The control flow block diagram of equalizing system as shown in Figure 4.

A kind of power battery pack balance charge/discharge control method based on SOC that the embodiment of the present invention provides, compared with traditional neural net, have that structure is simple, pace of learning is fast, parameter easily adjusts and is not easily absorbed in the advantages such as Local Minimum, there is features such as approaching Multiinputoutput parametric function, non-linear, the robustness of height and fault-tolerance.The setting parameter of ELM is more simple, and operand is little, has the fast and advantage that Generalization Capability is good of pace of learning.Therefore, the SOC estimation method based on ELM is applied in the embodiment of the present invention, the state-of-charge that battery is current can be predicted more accurately and rapidly, effectively improve the accuracy of equalization and equalization efficiency.Be several little battery modules by battery components simultaneously, during Homogeneity between groups, can a module be regarded as a joint monomer, adopt double-deck topological structure to realize balanced.Based on the double-deck active equalization system of SOC, control the switching tube in top layer and bottom circuit respectively, the bi-directional of battery pack self-energy can be realized, realize the Balance route of battery pack fast and efficiently.

Based on same inventive concept, the embodiment of the present invention also provides a kind of power battery pack balance charge/discharge control system based on SOC, shown in Figure 5, comprise: battery pack supervisory controller, SOC module, for the top layer circuit of Balance route battery pack and bottom circuit, altogether containing n battery cell in battery pack, be divided into m battery module, battery module is made up of multiple battery cell.Wherein, not shown battery pack supervisory controller in Fig. 5, only flows to " data acquisition " and illustrates.

Concrete, bottom circuit comprises n two-way Buck-Boost circuit, and two-way Buck-Boost circuit and battery cell one_to_one corresponding; Top layer circuit comprises m flyback transformer, and flyback transformer and battery module one_to_one corresponding.

Discharge and recharge data for gathering the discharge and recharge data of electrokinetic cell, and are sent to SOC module by battery pack supervisory controller, and discharge and recharge data comprise the terminal voltage of electrokinetic cell, electric current and temperature; Electrokinetic cell comprises battery cell and battery module; SOC module is used for the Mathematical Modeling calculating electrokinetic cell SOC based on ELM method establishment, using the input of discharge and recharge data as Mathematical Modeling, and determines electrokinetic cell SOC data according to the output of Mathematical Modeling; According to electrokinetic cell SOC data, consistency analysis is carried out to power battery pack, when power battery pack exists inconsistent according to the on off operating mode of the switching tube of Balance route policy control equalizing circuit, realize electrokinetic cell energy trasfer.

Wherein, SOC module calculates the Mathematical Modeling of electrokinetic cell SOC based on ELM method establishment, comprising:

The input vector of Mathematical Modeling is x _j=[x _j1, x _j2..., x _jn] ^t∈ R ⁿ, output vector is t _j=[t _j1, t _j2..., t _jm] ^t∈ R ^m, wherein m, n represent the dimension of input and output respectively, R ⁿrepresent n-dimensional space, R ^mrepresent m-dimensional space, j=1 ..., N, N are sample number;

Mathematical Modeling is: H β=T, and in formula, H is the hidden layer output matrix of neural net:

$H({\mathrm{\ω}}_1,...,{\mathrm{\ω}}_{\tilde{N}},b_1,...,b_{\tilde{N}},x_1,...,x_N)={\left[\begin{array}{ccc}g({\mathrm{\ω}}_1\·x_1+b_1)& ...& g({\mathrm{\ω}}_{\tilde{N}}\·x_1+b_{\tilde{N}})\\ \begin{array}{c}.\\ .\\ .\end{array}& \begin{array}{c}\\ ...\\ \end{array}& \begin{array}{c}.\\ .\\ .\end{array}\\ g({\mathrm{\ω}}_1\·x_N+b_1)& ...& g({\mathrm{\ω}}_{\tilde{N}}\·x_N+b_{\tilde{N}})\end{array}\right]}_{N\×\tilde{N}},$

$\mathrm{\β}={\left[\begin{array}{c}{\mathrm{\β}}_1^T\\ \begin{array}{c}.\\ .\\ .\end{array}\\ {\mathrm{\β}}_{\tilde{N}}^T\end{array}\right]}_{\tilde{N}\×m},T={\left[\begin{array}{c}{t}_1^T\\ \begin{array}{c}.\\ .\\ .\end{array}\\ t_N^T\end{array}\right]}_{N\×m};$

Wherein, hidden node number is excitation function is g (x), ω _i=[ω _i1, ω _i2..., ω _in] be the weight vector of connection i-th hidden layer node and input node; β _i=[β _i1, β _i2..., β _im] ^tit is the weight vector of connection i-th hidden node and output node; b _iit is the threshold value of i-th hidden node.

Preferably, determine electrokinetic cell SOC data according to the output of Mathematical Modeling, comprising:

Another input vector x _jfor [U (k), I (k), C (k)], output vector t _jfor SOC (k)=t _j, wherein, k represents the label of the kth group training data collected, and U, I, C represent the terminal voltage of electrokinetic cell, electric current and temperature respectively; The training set of Mathematical Modeling is { (x _j, t _j) | x _j∈ R ⁿ, t _j∈ R ^m, j=1 ..., N} (n=3, m=1);

The least square solution determining linear system H β=T is equivalent to according to training set training SLFN

wherein, H ⁺for the Moore-Penrose generalized inverse of matrix H;

Obtaining least square solution after, according to least square solution determine electrokinetic cell SOC data.

Concrete, shown in Figure 6, for the battery cell B1 in Fig. 6, the two-way Buck-Boost circuit in bottom circuit comprises: the first switching tube Q1, the first sustained diode 1, energy storage inductor L1; Connect with energy storage inductor L1 after first switching tube Q1 is in parallel with the first sustained diode 1, two-way Buck-Boost circuit is in parallel with battery cell B1, and two-way Buck-Boost circuit and adjacent two-way Buck-Boost circuit (namely controlling the two-way Buck-Boost circuit of battery cell B2) share energy storage inductor L1.

Shown in Figure 7, for the battery module B2 shown in Fig. 7, flyback transformer comprises: second switch pipe Q, the second sustained diode, transformer; The former limit winding of transformer is connected in parallel with power module B2 after Q with second switch pipe, the vice-side winding of transformer is connected with the second sustained diode.In the figure 7, B ₁, B ₂, B ₃, B ₄the battery module be made up of some battery cells, DC-DC represents flyback transformer (flyback transformer is also a kind of DC-DC circuit).

In the embodiment of the present invention, using balanced variable as the criterion realizing Balance route, concrete, balanced variable comprises the difference of the difference of state-of-charge, the state-of-charge of certain battery module and battery module state-of-charge mean value between the mean square deviation of battery cell SOC in group, adjacent two battery cells; When the mean square deviation of battery cell SOC is not less than the first predetermined threshold value in organizing, meet Balance route unlocking condition; When between adjacent two battery cells, the difference of state-of-charge is greater than the second predetermined threshold value, according to bottom control policy control bottom circuit; When the state-of-charge of certain battery module and the difference of battery module state-of-charge mean value are greater than the 3rd predetermined threshold value, control top layer circuit according to top layer control strategy.

Mean square deviation ε can represent the inconsistency of battery pack state-of-charge.In the embodiment of the present invention, the condition that equilibrium is opened is set as: mean square deviation ε >=γ.After system judges that battery pack meets balanced condition of opening, equalization operation will be carried out to battery pack.When top controller monitors Δ SOC ' > β of certain module, then corresponding top layer circuit is started working; When bottom controller monitors the Δ SOC > η in certain module between adjacent two monomers, then open corresponding bottom circuit.Wherein γ, β, η are the threshold value set in Balance route strategy respectively.

Because bottom module monomer voltage is low, quantity is few, disturbs little, and the embodiment of the present invention adopts two-way Buck-Boost circuit as the balanced topological structure of bottom, processor controls according to balance policy switch tube, shifts the energy be stored in inductance successively between adjacent monomer.This circuit can provide larger euqalizing current, and energy transferring does not limit by voltage, can realize bidirectional equalization.Required element cost is low, and expansion is convenient, and when the monomer joint number of battery module changes, circuit need not larger change.Two-way Buck-Boost circuit topology as shown in Figure 6.

For first and second battery cell, suppose that equalization algorithm module detects that the inconsistency ε of battery pack meets balanced unlocking condition under current state, and battery cell B ₁and B ₂between Δ SOC exceed the threshold value (SOC of setting ₁> SOC ₂), then need to open B ₁and B ₂between equalization operation.Whole balancing procedure is divided into B ₁electric discharge and B ₂charge two stages:

B ₁discharge regime: as t=0, makes switching tube Q by controlling pwm signal ₁conducting, by B ₁in unnecessary stored energy in inductance L ₁in, inductive current i _llinear increase from 0, expression formula is:

wherein for battery B ₁terminal voltage.Work as t=t ₁time, i _larrive maximum i _maxif inductive drop is now V _l.In this stage, i _lincrement can be expressed as

B ₂charging stage: work as t=t ₁moment, Q ₁turn off, i _lby diode D ₂afterflow, inductance L ₁the energy of middle storage is to B ₂transfer, i _llinear decline, can be expressed as work as t=t ₂moment, i _lbe reduced to 0, now a balanced end cycle.

In like manner known, work as SOC ₁< SOC ₂time, B ₂also can by the energy trasfer that overcharges to B ₁, realize the bi-directional of energy.

After each balanced end cycle, if the inconsistency ε of battery pack does not meet balanced unlocking condition under monitoring current state, then stop equilibrium; Otherwise, then continue to carry out equalization operation to the adjacent monomer of Δ SOC >=η, until the inconsistency ε of battery pack is less than the threshold gamma set in balance policy, stop balanced.

In the embodiment of the present invention, voltage is there is high for top layer Balance route, the problems such as battery pack quantity is many, the embodiment of the present invention adopts the top layer circuit based on flyback transformer, the bidirectional equalization topological structure that the topological structure of this top layer circuit is made up of anti exciting converter and switch mosfet pipe, its balanced object is the battery module of some battery cells composition.This topological structure can realize the battery module energy back of overtension in integral battery door group.Fly-back converter circuit balancing speed is fast, cost is low, structure is simple, and circuit adopts isolated topological structure simultaneously, can effectively realize electrical isolation between battery pack, reduces electromagnetic compatibility impact.

Top layer circuit adopts based on the equalizing circuit topological structure of inverse excitation type converter, by the mutual conversion of electric energy and magnetic energy, realizes the energy transferring between battery module.When batteries charging, when certain battery module energy is too high, using inverse excitation type converter as energy trasfer medium, give whole battery pack by energy transferring unnecessary for this battery module, prevent because energy overcharges the harm caused battery.

Concrete, in the embodiment of the present invention, for organizing the mean value of interior battery cell state-of-charge, the ε mean square deviation that is battery cell SOC in group, the Δ SOC difference that is state-of-charge between adjacent two battery cells, for the difference of the mean value of battery module state-of-charge, state-of-charge that Δ SOC ' is certain battery module and battery module state-of-charge mean value.Suppose that the inconsistency ε of battery pack meets balanced unlocking condition under current state, need to open top layer equalizing circuit.And top layer circuit monitoring is to B ₁Δ SOC ' > β of battery module, then carry out equalization operation to this module.The whole top layer balanced cycle is divided into B ₁electric discharge and two stages of whole batteries charging:

(1) B ₁discharge process: as t=0, makes switching tube Q close, by B by controlling pwm signal ₁in overcharge stored energy in the former limit winding of transformer.Work as t=t ₁time, primary side current of transformer i _larrive maximum i _max.

(2) whole batteries charging process: work as t=t ₁in the moment, control switch pipe Q disconnects, and the magnetization energy in the winding of former limit transfers to battery pack by vice-side winding, until secondary current is reduced to 0.

After each balanced end cycle, if the inconsistency ε of battery pack does not meet balanced unlocking condition under monitoring current state, then stop equilibrium; Otherwise, then continue to carry out equalization operation to the battery module of Δ SOC ' > β, until the inconsistency ε of battery pack is less than the threshold gamma set in balance policy, stop balanced.

In the embodiment of the present invention, devise the battery pack supervisory controller based on LTC6804, the voltage of nearly 12 Stringing cells can be measured and overall measurement error lower than 1.2mv, the voltage of all 12 batteries can complete measurement within 290 μ s, and lower data acquisition rate can be selected to suppress to realize strong noise.Multiple LTC6804 device can be connected in series, thus can realize the real time monitoring of every batteries in a lot of high-voltage battery.Use LTC6804 can complete the synchro measure of terminal voltage to cell, electric current, required input data can be provided rapidly and accurately for SOC module.Based on LTC6804 battery monitor circuit theory diagrams as shown in Figure 8.

Simultaneously, embodiments provide the anti exciting converter formula top layer circuit based on the 2.5A one chip active cell charge balance device LT8584 with telemetry interface, it is a one chip flyback DC/DC transducer, aims at the active balancing design of high-voltage battery group.The high efficiency of switching regulator significantly increases attainable balanced balanced current, decreases caloric value simultaneously.LT8584 includes integrated 6A, 50V mains switch, reduces the design complexity of application circuit.This device relies on its battery discharged to run completely, thus need not set up complicated biasing circuit required when adopting external power source.The enable pin DIN of this device is designed for and battery pack monitoring IC coordinate operation, when with LTC680x family device with the use of time, the system telemetry function comprising electric current and temperature monitoring also can be provided.Based on LT8584 top layer circuit theory diagrams as shown in Figure 9.

The workflow of this system is introduced in detail below by an embodiment.

Embodiment one

In embodiment one, shown in Figure 10, the workflow of this power battery pack balance charge/discharge control system is specific as follows:

Step 201: temperature monitoring module, based on the terminal voltage detecting module of LTC6804, temperature, the voltage and current data of current monitoring module Real-time Obtaining battery, by SPI serial ports, data are passed to SOC estimation block.

Step 202:SOC module, according to the SOC forecast model trained, is estimated the state-of-charge of each battery cell, and the cell SOC data calculated pass to top controller and corresponding bottom controller.

Step 203: top controller, according to balance policy, calculates the inconsistency of battery pack, analyzes battery pack the need of equalization operation.

Step 204: when the inconsistency monitoring battery pack exceedes the threshold value of setting, then need to open equalization operation.

Now, top controller according to balance policy, can carry out equalization operation to the battery module of Δ SOC ' > β, and bottom controller can carry out equilibrium by the battery excessive to adjacent monomer difference simultaneously.

Step 205: the energy monitoring certain battery module when top controller is too high, during with the difference of battery module average higher than the threshold value set, top controller sends balanced enable signal by LTC6804 to LT8584.

Namely the signal that the DIN pin of LT8584 receives can become low level from high level, and when it monitors the trailing edge of this signal, top layer anti exciting converter equalizing circuit is started working.One-period will after LT8584 receives enable signal after about 2 μ s, the switch lock storage of LT8584 is set, inner NPN switch conduction, exert pressure in the two ends of the battery module overcharged to armature winding, electric current in primary coil linearly rises, but due to the diode reverse biased of primary side series connection, prevent the current flowing in secondary winding, now energy is stored in the magnetic core of transformer.When reaching current limit 6A, current limit comparator makes switch latch reset, and inner NPN switch disconnects, and the energy be stored in magnetic core of transformer applies forward bias to secondary tandem diode, and electric current flows into whole battery pack.When SW pin voltage is reduced to VVIN+95mV, DCM comparator is by configuration switch latch and start a new switch periods.Repeat above-mentioned equalization operation, until top controller monitors the difference of battery pack inconsistency or this battery module and battery module average lower than the threshold value set, balanced stop signal is sent to LT8584 by LTC6804, DIN pin monitors rising edge signal, now device is deactivated, and top layer balance module quits work.

Step 206: when certain bottom controller state-of-charge difference monitored between two adjacent cell monomers exceedes the threshold value of balance policy setting, bottom controller will drive the equalizing circuit that this two batteries are corresponding.

Bottom controller sends the pwm control signal with suitable duty ratio by drive circuit to the MOSFET that the battery cell that SOC is higher is corresponding.In the bottom balanced cycle, when control signal is high level, MOSFET pipe is closed, and the battery cell that now energy overcharges fills to power inductance can; When control signal is low level, MOSFET pipe disconnects, and the electric current in power inductance is flowed in this battery by the fly-wheel diode that battery cell that energy is lower is corresponding, realizes the reasonable transfer of energy.Bottom balance module repeats the above-mentioned balanced cycle, until top controller monitors the threshold value of the SOC difference between battery pack inconsistency or two monomers lower than setting, bottom controller stops the drive singal sending corresponding MOSFET pipe, and this bottom balance module quits work.

The present invention can have multiple multi-form embodiment; above for Fig. 1-Figure 10 by reference to the accompanying drawings to technical scheme of the present invention explanation for example; this does not also mean that the instantiation that the present invention applies can only be confined in specific flow process or example structure; those of ordinary skill in the art should understand; specific embodiments provided above is some examples in multiple its preferred usage, and the execution mode of any embodiment the claims in the present invention all should within technical solution of the present invention scope required for protection.

Last it is noted that the foregoing is only the preferred embodiments of the present invention, be not limited to the present invention, although with reference to previous embodiment to invention has been detailed description, for a person skilled in the art, it still can be modified to the technical scheme described in foregoing embodiments, or carries out equivalent replacement to wherein portion of techniques feature.Within the spirit and principles in the present invention all, any amendment done, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (10)

1., based on a power battery pack balance charge/discharge control method of SOC, it is characterized in that, comprising:

Obtain the discharge and recharge data of electrokinetic cell, described discharge and recharge data comprise the terminal voltage of electrokinetic cell, electric current and temperature; Described electrokinetic cell comprises battery cell;

Calculate the Mathematical Modeling of electrokinetic cell SOC based on ELM method establishment, using the input of described discharge and recharge data as described Mathematical Modeling, and determine described electrokinetic cell SOC data according to the output of described Mathematical Modeling;

According to described electrokinetic cell SOC data, consistency analysis is carried out to power battery pack, when power battery pack exists inconsistent according to the on off operating mode of the switching tube of Balance route policy control equalizing circuit, realize electrokinetic cell energy trasfer.

2. method according to claim 1, is characterized in that, the described Mathematical Modeling calculating electrokinetic cell SOC based on ELM method establishment, comprising:

The input vector of described Mathematical Modeling is x _j=[x _j1, x _j2..., x _jn] ^t∈ R ⁿ, output vector is t _j=[t _j1, t _j2..., t _jm] ^t∈ R ^m, wherein m, n represent the dimension of input and output respectively, R ⁿrepresent n-dimensional space, R ^mrepresent m-dimensional space, j=1 ..., N, N are sample number;

Described Mathematical Modeling is: H β=T, and in formula, H is the hidden layer output matrix of neural net:

$H({\mathrm{\&omega;}}_1,...,{\mathrm{\&omega;}}_{\tilde{N}},b_1,...,b_{\tilde{N}},x_1,...,x_N)={\left[\begin{array}{ccc}g({\mathrm{\&omega;}}_1\&CenterDot;x_1+b_1)& ...& g({\mathrm{\&omega;}}_{\tilde{N}}\&CenterDot;x_1+b_{\tilde{N}})\\ \begin{array}{c}.\\ .\\ .\end{array}& \begin{array}{c}\\ ...\\ \end{array}& \begin{array}{c}.\\ .\\ .\end{array}\\ g({\mathrm{\&omega;}}_1\&CenterDot;x_N+b_1)& ...& g({\mathrm{\&omega;}}_{\tilde{N}}\&CenterDot;x_N+b_{\tilde{N}})\end{array}\right]}_{N\&times;\tilde{N}},$ $\mathrm{\&beta;}={\left[\begin{array}{c}{\mathrm{\&beta;}}_1^T\\ \begin{array}{c}.\\ .\\ .\end{array}\\ {\mathrm{\&beta;}}_{\tilde{N}}^T\end{array}\right]}_{\tilde{N}\&times;m},$ $T={\left[\begin{array}{c}{t}_1^T\\ \begin{array}{c}.\\ .\\ .\end{array}\\ t_N^T\end{array}\right]}_{N\&times;m};$

Wherein, hidden node number is excitation function is g (x), ω _i=[ω _i1, ω _i2..., ω _in] be the weight vector of connection i-th hidden layer node and input node; β _i=[β _i1, β _i2..., β _im] ^tit is the weight vector of connection i-th hidden node and output node; b _iit is the threshold value of i-th hidden node.

3. method according to claim 2, is characterized in that, described electrokinetic cell SOC data are determined in the described output according to described Mathematical Modeling, comprising:

Another input vector x _jfor [U (k), I (k), C (k)], output vector t _jfor SOC (k)=t _j, wherein, k represents the label of the kth group training data collected, and U, I, C represent the terminal voltage of electrokinetic cell, electric current and temperature respectively; The training set of described Mathematical Modeling is { (x _j, t _j) | x _j∈ R ⁿ, t _j∈ R ^m, j=1 ..., N} (n=3, m=1);

The least square solution determining linear system H β=T is equivalent to according to described training set training SLFN

wherein, H ⁺for the Moore-Penrose generalized inverse of matrix H;

Obtaining described least square solution after, according to described least square solution determine described electrokinetic cell SOC data.

4., according to the arbitrary described method of claim 1-3, it is characterized in that, described equalizing circuit comprises top layer circuit and bottom circuit; Described Balance route strategy comprises top layer control strategy and bottom control strategy, is respectively used to control described top layer circuit and bottom circuit.

5. method according to claim 4, is characterized in that, described electrokinetic cell also comprises battery module, and described battery module is made up of multiple battery cell;

The on off operating mode of the described switching tube according to Balance route policy control equalizing circuit, comprising:

Obtain balanced variable, described balanced variable comprises the difference of the difference of state-of-charge, the state-of-charge of certain battery module and battery module state-of-charge mean value between the mean square deviation of battery cell SOC in group, adjacent two battery cells;

When the mean square deviation of battery cell SOC is not less than the first predetermined threshold value in described group, meet Balance route unlocking condition;

When between described adjacent two battery cells, the difference of state-of-charge is greater than the second predetermined threshold value, bottom circuit according to described bottom control policy control;

When the state-of-charge of certain battery module described and the difference of battery module state-of-charge mean value are greater than the 3rd predetermined threshold value, control described top layer circuit according to described top layer control strategy.

6. method according to claim 5, is characterized in that, the balanced variable of described acquisition, comprising:

If altogether containing n battery cell in battery pack, be divided into m battery module, determine balanced variable according to following formula:

$\stackrel{\&OverBar;}{SOC}=\frac{1}{n}\underset{i=1}{\overset{n}{\&Sigma;}}{\mathrm{SOC}}_i$

$\mathrm{\&epsiv;}=\sqrt{\frac{1}{n}\underset{i=1}{\overset{n}{\&Sigma;}}({\mathrm{SOC}}_i-\stackrel{\&OverBar;}{SOC})}$

ΔSOC＝|SOC _i-SOC _i+1|

$\stackrel{\&OverBar;}{{\mathrm{SOC}}^{\&prime;}}=\frac{1}{m}\underset{l=1}{\overset{m}{\&Sigma;}}{\mathrm{SOC}}_l^{\&prime;}$

Wherein, for organizing the mean value of interior battery cell state-of-charge, the ε mean square deviation that is battery cell SOC in group, the Δ SOC difference that is state-of-charge between adjacent two battery cells, for the difference of the mean value of battery module state-of-charge, state-of-charge that Δ SOC ' is certain battery module and battery module state-of-charge mean value.

7. the power battery pack balance charge/discharge control system based on SOC, it is characterized in that, comprise: battery pack supervisory controller, SOC module, for the top layer circuit of Balance route battery pack and bottom circuit, altogether containing n battery cell in described battery pack, be divided into m battery module, described battery module is made up of multiple battery cell;

Described bottom circuit comprises n two-way Buck-Boost circuit, and described two-way Buck-Boost circuit and described battery cell one_to_one corresponding;

Described top layer circuit comprises m flyback transformer, and described flyback transformer and described battery module one_to_one corresponding;

Described discharge and recharge data for gathering the discharge and recharge data of electrokinetic cell, and are sent to SOC module by described battery pack supervisory controller, and described discharge and recharge data comprise the terminal voltage of electrokinetic cell, electric current and temperature; Described electrokinetic cell comprises battery cell and battery module;

Described SOC module is used for the Mathematical Modeling calculating electrokinetic cell SOC based on ELM method establishment, using the input of described discharge and recharge data as described Mathematical Modeling, and determines described electrokinetic cell SOC data according to the output of described Mathematical Modeling; According to described electrokinetic cell SOC data, consistency analysis is carried out to power battery pack, when power battery pack exists inconsistent according to the on off operating mode of the switching tube of Balance route policy control equalizing circuit, realize electrokinetic cell energy trasfer.

8. system according to claim 7, is characterized in that, described two-way Buck-Boost circuit comprises: the first switching tube, the first fly-wheel diode, energy storage inductor; Described first switching tube is connected with described energy storage inductor with after described first fly-wheel diode parallel connection, described two-way Buck-Boost circuit is in parallel with battery cell, and described two-way Buck-Boost circuit and adjacent two-way Buck-Boost circuit share described energy storage inductor;

Described flyback transformer comprises: second switch pipe, the second fly-wheel diode, transformer; The former limit winding of described transformer is connected rear in parallel with power module with described second switch pipe, the vice-side winding of described transformer is connected with described second fly-wheel diode.

9. the system according to claim 7 or 8, is characterized in that, the described Mathematical Modeling calculating electrokinetic cell SOC based on ELM method establishment, comprising:

The input vector of described Mathematical Modeling is x _j=[x _j1, x _j2..., x _jn] ^t∈ R ⁿ, output vector is t _j=[t _j1, t _j2..., t _jm] ^t∈ R ^m, wherein m, n represent the dimension of input and output respectively, R ⁿrepresent n-dimensional space, R ^mrepresent m-dimensional space, j=1 ..., N, N are sample number;

Described Mathematical Modeling is: H β=T, and in formula, H is the hidden layer output matrix of neural net:

$H({\mathrm{\&omega;}}_1,...,{\mathrm{\&omega;}}_{\tilde{N}},b_1,...,b_{\tilde{N}},x_1,...,x_N)={\left[\begin{array}{ccc}g({\mathrm{\&omega;}}_1\&CenterDot;x_1+b_1)& ...& g({\mathrm{\&omega;}}_{\tilde{N}}\&CenterDot;x_1+b_{\tilde{N}})\\ \begin{array}{c}.\\ .\\ .\end{array}& \begin{array}{c}\\ ...\\ \end{array}& \begin{array}{c}.\\ .\\ .\end{array}\\ g({\mathrm{\&omega;}}_1\&CenterDot;x_N+b_1)& ...& g({\mathrm{\&omega;}}_{\tilde{N}}\&CenterDot;x_N+b_{\tilde{N}})\end{array}\right]}_{N\&times;\tilde{N}},$ $\mathrm{\&beta;}={\left[\begin{array}{c}{\mathrm{\&beta;}}_1^T\\ \begin{array}{c}.\\ .\\ .\end{array}\\ {\mathrm{\&beta;}}_{\tilde{N}}^T\end{array}\right]}_{\tilde{N}\&times;m},$ $T={\left[\begin{array}{c}{t}_1^T\\ \begin{array}{c}.\\ .\\ .\end{array}\\ t_N^T\end{array}\right]}_{N\&times;m};$

Wherein, hidden node number is excitation function is g (x), ω _i=[ω _i1, ω _i2..., ω _in] be the weight vector of connection i-th hidden layer node and input node; β _i=[β _i1, β _i2..., β _im] ^tit is the weight vector of connection i-th hidden node and output node; b _iit is the threshold value of i-th hidden node.

10. system according to claim 9, is characterized in that, described electrokinetic cell SOC data are determined in the described output according to described Mathematical Modeling, comprising:

Another input vector x _jfor [U (k), I (k), C (k)], output vector t _jfor SOC (k)=t _j, wherein, k represents the label of the kth group training data collected, and U, I, C represent the terminal voltage of electrokinetic cell, electric current and temperature respectively; The training set of described Mathematical Modeling is { (x _j, t _j) | x _j∈ R ⁿ, t _j∈ R ^m, j=1 ..., N} (n=3, m=1);

The least square solution determining linear system H β=T is equivalent to according to described training set training SLFN

wherein, H ⁺for the Moore-Penrose generalized inverse of matrix H;

Obtaining described least square solution after, according to described least square solution determine described electrokinetic cell SOC data.