JP2014102111A - State estimation device for battery - Google Patents

Abstract

PROBLEM TO BE SOLVED: To quickly and accurately estimate the state of a battery.SOLUTION: A state estimation device includes: a frequency characteristic calculation part 11 for calculating the frequency characteristics of the AC impedance of a battery 1 and an estimation processing part 12 for estimating the state of the battery on the basis of the calculated frequency characteristics. The frequency characteristic calculation part 11 includes: storage means 11a for storing the impedance function of the battery 1; measurement means 11b for acquiring the actual measurement value data of the AC impedance of the battery 1; and arithmetic means 11c for performing processing (fitting processing) for adjusting the impedance function such that an error with respect to the measured actual measurement value data is reduced. The impedance function includes various parameters classified into a plurality of groups in accordance with the respective function blocks of an equivalent circuit 1E of the battery 1, and the arithmetic means 11c includes a function for adjusting a parameter independently of each group in the case of the fitting processing.

Description

  The present invention includes a signal generation unit capable of outputting a signal including a plurality of frequency components to a battery, a frequency characteristic calculation unit that calculates frequency characteristics of the AC impedance of the battery, and the frequency of the AC impedance calculated by the frequency characteristic calculation unit. The present invention relates to a battery state estimation device including an estimation processing unit that estimates a predetermined state amount related to a battery based on characteristics.

  In recent years, hybrid vehicles that travel using both an engine (internal combustion engine) and an electric motor, and electric vehicles that travel using only an electric motor are becoming widespread. Along with this, the importance of a battery (for example, a lithium ion battery) that can be charged with a large amount of electric power has been increasing, and various studies on this have been made.

  For example, Patent Document 1 below proposes a technique for accurately determining a capacity maintenance ratio (a ratio of a current battery capacity to a reference capacity), which is an indicator of a battery deterioration state. Specifically, in Patent Document 1, an AC signal having a plurality of frequency components is given to the battery from the signal generation unit, and the frequency characteristics of the AC impedance are obtained based on the battery response signal to the AC signal. Then, a feature frequency is determined from the calculated frequency characteristic, and a capacity maintenance rate is calculated based on the feature frequency (based on a relationship between the feature frequency stored in advance and the capacity maintenance rate).

JP 2011-38857 A

  Here, in Patent Document 1, in order to obtain the frequency characteristic of the AC impedance of the battery, it is necessary to plot the measured value of the impedance at each frequency on a complex plane and set a curve obtained by interpolating the discrete data obtained thereby. There is. In Patent Document 1, there is no specific disclosure about a method of interpolating discrete data (curve fitting), but generally, in curve fitting, a value (theoretical value) on a curve defined by a predetermined function is used. The process of adjusting various parameters included in the function is repeatedly performed so that the error from the actual measurement value is minimized.

  However, in the curve fitting described above, there are many cases in which a large number of parameter adjustments must be performed before the error converges. In particular, when the battery is an in-vehicle battery, the calculation capability is limited. Therefore, if parameter adjustment is attempted until the error converges, a large amount of calculation time is required. For this reason, there has been a demand for more efficient curve fitting and accurate battery state estimation based on the result.

  The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a battery state estimation device capable of accurately estimating the state of a battery in a short time.

  In order to solve the above problems, the present invention provides a signal generation unit capable of outputting a signal including a plurality of frequency components to a battery, a frequency characteristic calculation unit that calculates a frequency characteristic of an AC impedance of the battery, and a frequency characteristic A battery state estimation device comprising: an estimation processing unit configured to estimate a predetermined state quantity relating to the battery based on the frequency characteristic of the AC impedance calculated by the calculation unit, wherein the frequency characteristic calculation unit includes the frequency of the AC impedance A storage unit that stores an impedance function that defines characteristics, a measurement unit that obtains actual measurement data of the AC impedance based on a response signal output from a battery with respect to a signal from the signal generation unit, and a measurement unit Adjust the impedance function so that the error with respect to the actual measurement data is small, and adjust Computing means for performing a fitting process for storing the impedance function in the storage means, and the storage means has, as the impedance function, a plurality of terms respectively corresponding to a plurality of functional blocks included in an equivalent circuit of the battery. The impedance function includes various parameters classified into a plurality of groups corresponding to each functional block of the equivalent circuit, and the computing means performs the fitting process. A function of adjusting the parameter independently for each of the groups is provided (claim 1).

  According to the present invention, the parameters included in the battery impedance function are classified into a plurality of groups corresponding to the functional blocks of the battery, and the fitting process for adjusting the parameters according to the actual measurement data is performed. Since the parameters are adjusted independently for each group, for example, even when the fitting accuracy of the impedance curve before adjustment (curve on the complex plane drawn based on the impedance function) varies greatly depending on the location, The parameter adjustment amount (change amount) can be set relatively freely for each group, and the optimum parameter can be obtained more efficiently. For this reason, unlike the case where all parameters can always be adjusted only simultaneously, high fitting accuracy can be obtained in a short time. And based on the impedance curve (impedance function) accurately fitted by such a procedure, the state of a battery can be estimated efficiently and appropriately.

  In the present invention, preferably, as a process of independently adjusting the parameters of the impedance function for each group, the parameters of the groups are sequentially adjusted so that the entire error of the impedance curve based on the impedance function is reduced. And a second process for sequentially adjusting the parameters of each group so that the error of each divided portion is individually reduced when the impedance curve is divided according to each functional block of the equivalent circuit. In addition, the calculation means selectively executes one of the first and second processes according to the variation in the amount of change during the previous parameter adjustment.

  According to this configuration, the parameters of each group can be adjusted in consideration of the error of the entire impedance curve, or can be adjusted in consideration of only the error of the related curve portion. Then, by selecting the adjustment according to any of these procedures according to the variation in the previous parameter adjustment amount, it is possible to perform an appropriate parameter adjustment according to the situation, and further improve the efficiency of the fitting process. .

  In the present invention, preferably, the estimation processing unit extracts a predetermined geometric feature amount from an impedance curve based on an impedance function after the fitting process is completed, and uses the geometric feature amount. Based on the predetermined prediction formula, at least one of the charged state and the deteriorated state of the battery is estimated as the state quantity.

  According to this configuration, the state of charge or deterioration of the battery can be accurately estimated based on the geometric characteristic amount of the impedance curve adjusted with high accuracy by the fitting process as described above. In addition, since it is only necessary to calculate a prediction formula using the geometric characteristic amount of the impedance curve, the state of charge or the deterioration state can be obtained easily and in a short time.

  The plurality of functional blocks included in the equivalent circuit of the battery include, for example, an inductance part derived from the battery cell structure, a positive electrode part derived from a reaction at the positive electrode, a negative electrode part derived from a reaction at the negative electrode, a positive electrode and a negative electrode A diffusion portion derived from the movement of ions can be assumed. In this case, the impedance function includes an inductance term, a positive term, a negative term, and a diffusion term corresponding to each functional block.

  As described above, according to the battery state estimation device of the present invention, the state of the battery can be estimated in a short time and with high accuracy.

1 is a diagram showing a schematic configuration of a vehicle equipped with a battery state estimation device according to an embodiment of the present invention. It is a circuit diagram which shows the equivalent circuit of the said battery. It is the graph which illustrated the impedance curve of the said battery. It is a main flowchart which shows the whole flow of the process which estimates the state of the said battery. It is a subroutine which shows the specific content of step S3 of the said FIG. It is a figure which shows an example of the geometric feature-value extracted from the said impedance curve. It is a figure which shows the result of the experiment conducted in order to confirm the effect of this invention.

(1) Overall Configuration FIG. 1 is a diagram showing a schematic configuration of a vehicle equipped with a battery state estimation device according to an embodiment of the present invention. The battery 1 targeted by the state estimation device shown in the figure is a rechargeable ion battery that can be repeatedly charged and discharged, and is an in-vehicle battery used for running a hybrid vehicle or an electric vehicle. The battery 1 is connected to a traveling motor 3 that drives a driving wheel 4 of the vehicle via an inverter 2. When the vehicle in which the drive wheels 4 are driven by the travel motor 3 is driven by power, the DC power stored in the battery 1 is converted into AC power by the inverter 2 and then supplied to the travel motor 3. On the other hand, when the vehicle that does not need to drive the drive wheels 4 is decelerated, AC power (regenerative power) generated by the traveling motor 3 is converted into DC power by the inverter 2 and then supplied to the battery 1.

  Although detailed illustration is omitted, when the vehicle is a hybrid vehicle, a generator that generates power by obtaining the driving force of the engine is provided in the vehicle, and the electric power generated by the generator is charged in the battery 1. Is done. Further, when the vehicle is a so-called plug-in hybrid vehicle or a pure electric vehicle (a vehicle that runs only with the driving motor 3), a plug that receives power supplied from the charger is provided on the vehicle. Electric power supplied from the charger through the plug is charged in the battery 1.

  In the vehicle, a control unit 10 that comprehensively controls the operation of the traveling motor 3 and the charging / discharging operation of the battery 1, and a signal generation that can output a signal necessary to estimate the state of the battery 1 to the battery 1. A unit 13 and a voltage sensor 15 for detecting the output voltage of the battery 1 are provided.

  The control unit 10 is a microprocessor composed of a CPU, a ROM, a RAM, and the like, and has a function of estimating the state of the battery 1 in addition to the function of controlling the traveling motor 3 and the like described above. Specifically, the control unit 10 obtains a response signal output from the battery 1 as a functional element for estimating the state of the battery 1 from the voltage sensor 15 in response to a signal input from the signal generator 13, A frequency characteristic calculation unit 11 that calculates the frequency characteristic of the AC impedance of the battery 1 based on the acquired data, and a predetermined state quantity related to the battery 1 is estimated based on the frequency characteristic of the AC impedance calculated by the frequency characteristic calculation unit 11. And an estimation processing unit 12. Although details will be described later, in the present embodiment, a state of charge (SOC) and a state of health (SOH) are estimated as state quantities related to the battery 1. The state of charge SOC is a value (%) represented by “(remaining capacity) / (full charge capacity) × 100”. The deterioration state SOH is a value (%) represented by “(current full charge capacity) / (initial full charge capacity) × 100”, and is also a value called a capacity maintenance rate.

  In the present embodiment, the control unit 10 including the frequency characteristic calculation unit 11 and the estimation processing unit 12 described above, the signal generation unit 13, and the voltage sensor 15 constitute a state estimation device that estimates the state of the battery 1. Yes.

  The frequency characteristic calculation unit 11 includes a storage unit 11 a that stores an impedance function Z (jω) (the following equation (1)) that defines the frequency characteristic of the AC impedance of the battery 1, and a battery for an input signal from the signal generation unit 13. The measurement unit 11b that obtains the actual measurement value data of the AC impedance based on the response signal 1 and the impedance function Z (jω) are adjusted so that the error with respect to the actual measurement data measured by the measurement unit 11b is small, and the impedance after adjustment And an arithmetic unit 11c for performing a fitting process (so-called curve fitting described in detail later) for storing the function Z (jω) in the storage unit 11a.

  Here, the impedance function Z (jω) stored in the storage unit 11a is a polynomial function represented by the following equation (1).

In the above formula (1), R ₀ , R _S , R ₁ , R ₂ , R ₃ are resistors, LPE ₀ is an inductance, CPE _S , CPE ₁ , CPE ₂ , CPE ₃ are capacitances, Zw ₀ , Zw _S , Zw ₃ is the Warburg impedance.

  The impedance function Z (jω) in the above equation (1) is set based on the equivalent circuit 1E of the battery 1 shown in FIG. Specifically, the equivalent circuit 1E of FIG. 2 includes an inductance part (i) derived from the cell structure of the battery 1, a positive electrode part (ii) derived from the reaction at the positive electrode, and a negative electrode part (iii) derived from the reaction at the negative electrode. ), And divided into four functional blocks called diffusion parts (iv) derived from the movement of ions of the positive and negative electrodes. The AC impedances of these four functional blocks are expressed in terms (I) to (IV) in the impedance function Z (jω). That is, the inductance term (I), the positive term (II), the negative term (III), and the diffusion term (IV) of the impedance function Z (jω) are the inductance part (i) and the positive part (ii) of the equivalent circuit 1E, respectively. ), Negative electrode part (iii), and diffusion part (iv).

  The following Table 1 shows various parameters included in the impedance function Z (jω) as parameter group I included in the inductance term (I), parameter group II included in the positive term (II), and negative term ( They are classified into four groups: parameter group III included in III) and parameter group IV included in the diffusion term (IV). The storage unit 11a stores not only the impedance function Z (jω) and the values of all the parameters included therein, but also the attribute that each parameter belongs to which group of I to IV.

  FIG. 3 is a graph representing an impedance function Z (jω) composed of a polynomial function including the various parameters on a complex plane. As shown in the figure, the impedance curve Y defined by the impedance function Z (jω) is a first curve portion y1 mainly reflecting the inductance term (I) from the low frequency side (left side) to the high frequency side (right side). The second curve portion y2 mainly reflecting the positive electrode term (II), the third curve portion y3 mainly reflecting the negative electrode term (III), and the diffusion term (IV) of the impedance function Z (jω). Can be divided into the fourth curved line portion y4 reflected in FIG. The fourth curved line portion y4 and the third curved line portion y3 can be divided by a downwardly convex valley portion (a portion having the maximum curvature), and the third curved portion y3 and the second curved portion y2 are the same. The downward convex valley can be divided into boundaries, and the second curved line y2 and the first curved line y1 can be separated from the point where the horizontal axis (real axis component) is the smallest.

(2) Specific Processing for Estimating Battery State Next, a specific procedure for estimating the state (SOC, SOH) of the battery 1 will be described. FIG. 4 is a main flowchart showing the overall flow of state estimation, and FIG. 5 is a subroutine showing in detail the process of step S3 of FIG.

  When the process shown in the flowchart of FIG. 4 starts, the control unit 10 executes a process of determining whether or not a measurement condition permitting measurement of the impedance of the battery 1 is satisfied (step S1). Specifically, here, when the ignition is OFF (the vehicle is not in use), or when the current of the battery 1 is stable at almost zero even when the ignition is ON (the vehicle is in use). Then, it is determined that the measurement condition is satisfied. Note that the state in which the ignition is ON and the battery current is almost zero as in the latter is realized when the vehicle is stopped due to a signal or the like and an electric load such as an air conditioner is not operating.

  When it is confirmed in step S1 that the measurement condition is satisfied, the control unit 10, more specifically, the measurement unit 11b of the frequency characteristic calculation unit 11 executes a process of measuring the AC impedance of the battery 1 (step S2).

  Specifically, in step S <b> 2, the measuring unit 11 b sequentially generates a plurality of AC signals (sine waves) having different frequencies from the signal generator 13 to the battery 1. And the alternating current impedance of the battery 1 is specified for every frequency by acquiring the output voltage of the battery 1 with respect to the input signal from the voltage sensor 15. The data obtained by this is the impedance of the jumping frequency corresponding to the frequency of the input signal from the signal generator 13 to the battery 1 and is discrete data of the frequency characteristic of the impedance.

  In the measurement process in step S2, it is necessary to obtain a certain amount of actually measured value data in order to ensure the accuracy of curve fitting (S4) described later. For this reason, when the measurement is performed while the vehicle is stopped, a sufficient amount of data may not be obtained until the vehicle starts. In such a case, the data is discarded and the process waits until the next measurement condition (S1) is satisfied.

  When the measurement process in step S2 is completed, the control unit 10, more specifically, the calculation means 11c of the frequency characteristic calculation unit 11 performs so-called curve fitting that adjusts the impedance function Z (jω) in accordance with the measured value data ( Step S3). That is, the calculation unit 11c reduces an error generated between the impedance curve Y (FIG. 3) defined by the impedance function Z (jω) and the AC impedance measurement data measured by the measurement unit 11b. Then, a process of changing the value of the impedance function Z (jω) stored in the storage unit 11a (more specifically, various parameters included in the impedance function Z (jω)) is executed.

  Details of the curve fitting performed in step S3 will be specifically described using the subroutine of FIG. When the curve fitting process is started, the calculation unit 11c executes a process of determining whether or not the number of loops of the subroutine of FIG. 5 (that is, the number of parameter adjustments) has reached a predetermined number (step S11). At the start of curve fitting, the number of loops is naturally zero, and the determination in step S11 is NO. Then, the calculation means 11c uses all the parameters included in the impedance function Z (jω), that is, the parameter groups I to IV shown in Table 3, as the weighted error of the entire impedance curve Y based on the impedance function Z (jω). A process of adjusting so as to be reduced is executed (step S12).

Here, “to reduce the error” means that the AC impedance measured values (discrete data) at a plurality of frequencies measured in step S2 and the calculated AC impedance at each frequency (in FIG. 3). It means that the sum of squares of errors with respect to the value on the impedance curve Y is small. Further, the weighted error is an error included in the impedance function Z (jω), that is, the error of the first curve portion y1, the error of the second curve portion y2, and the error of the third curve portion y3 shown in FIG. And the error of the fourth curve portion y4, and the sum of squares of the error in each portion is multiplied by a weighting factor (w ₁ to w _{4 to be} described later) that determines the degree of influence of the error.

  In step S12, all parameter groups included in the impedance function Z (jω) are set so that the weighted error of the entire impedance curve Y, that is, the sum of the weighted errors of all the first curve portions y1 to y4 is reduced. Adjust I-IV. As the initial values of the parameter groups I to IV, when parameter adjustment (curve fitting) has been completed in the past, values set in the latest past are set, and curve fitting has been performed once in the past. If not performed (that is, the first fitting after the vehicle is sold), a default value set at the time of manufacture of the vehicle is set.

  When the process of step S12 is completed, the calculation unit 11c executes a process of determining whether or not the weighted error of the entire impedance curve Y has decreased as compared with the weighted error before adjusting the parameters (step S12). S13).

  When it is determined as YES in Step S13 and it is confirmed that the weighted error of the entire curve is reduced, the calculation unit 11c stores the parameter adjusted in Step S12 in the storage unit 11a. Processing for updating the parameter of the impedance function Z (jω) to the adjusted value is executed (step S14).

Subsequently, the operating unit 11c executes a process of updating the weighting coefficients w ₁ to w ₄ error (step S15). For example determined and the mean square of the error of the curved portion yi (i = 1~4) in the impedance curve Y after parameter adjustment and e _i, a new weighting coefficients by w _i × (e _i) ^1/2 Is done. As a result, the weight coefficient w _i is set to be larger as the curve portion having a larger error among the first to fourth curve portions y1 to y4.

  On the other hand, if it is determined NO in step S13, that is, if it is confirmed that the weighted error of the entire curve has not been reduced, the computing unit 11c skips the processing of steps S14 and S15, and sets the parameters and weights. Do not update the coefficient.

  Next, the computing unit 11c executes a process of determining whether or not the correlation coefficient between the calculated value of the AC impedance and the actually measured value is smaller than 0.999 (step S16). Here, the correlation coefficient is an index indicating how much the calculated frequency characteristic (impedance curve Y in FIG. 3) obtained using the impedance function Z (jω) has a correlation with the measured data. The relationship number = 1 corresponds to the fact that the two are completely positively correlated.

  When it is determined as NO in step S16 and it is confirmed that the correlation coefficient is equal to or greater than 0.999 (that is, the correlation with the actually measured value is quite high), the calculation unit 11c returns to step S11 described above, The subsequent processing is repeated.

  On the other hand, if it is determined as YES in step S16 and it is confirmed that the correlation coefficient is <0.999, the calculation unit 11c changes all the parameters (parameter groups I to IV) of the impedance function Z (jω). A process of determining whether or not the amount is equal to or less than a predetermined threshold value is executed (step S17). That is, before and after adjusting the parameter groups I to IV in step S12, the amount of change of all parameters belonging to each parameter group I to IV is specified, and the amount of change of each parameter is determined for each parameter. Compare with the set threshold value. The fact that the amount of change in all parameters is equal to or less than the threshold means that the fitting accuracy of the impedance curve Y (FIG. 3) is almost equal in any curve portion y1 to y4, and the amount of change in all parameters is larger than the threshold. The fitting accuracy differs depending on the curved parts y1 to y4 (the accuracy varies).

When it is determined as NO in step S17 and it is confirmed that the amount of change of all parameters is larger than the threshold value (fitting accuracy varies), the calculation unit 11c calculates all parameters included in the impedance function Z (jω). Of these, only the parameters belonging to the parameter group I (L ₀ , q ₀ , R ₀ ,..., Σ _s ) are adjusted so as to reduce the weighted error of the first curve portion y1 of the impedance curve Y ( Step S18).

Similarly, the calculation means 11c adjusts only the parameters (C ₁ , p ₁ , R ₁ ) belonging to the parameter group II so that the weighted error of the second curve portion y2 is reduced (step S19), the parameter A process (step S20) for adjusting only the parameters (C ₂ , p ₂ ) belonging to the group III so that the weighted error of the third curve portion y3 is reduced, and the parameters (R ₂ , C ₃ , p _3, R _3, executes the sigma ₃₎ only the process of adjusting to the weighted error of the fourth curved portion y4 is reduced (step S21).

  As described above, steps S18 to S21 correspond to the functional blocks (inductance part (i), positive electrode part (ii), negative electrode part (iii), diffusion part (iv)) included in the equivalent circuit 1E of the battery 1. The parameter groups I to IV thus grouped are adjusted independently so that the individual errors of the curve portions y1 to y4 corresponding to the parameter groups I to IV are reduced.

On the other hand, when it is determined as YES in step S17 and it is confirmed that the amount of change of all parameters is equal to or less than the threshold value (fitting accuracy is almost equal), the calculation unit 11c is included in the impedance function Z (jω). Among all the parameters, only the parameters belonging to the parameter group I (L ₀ , q ₀ , R ₀ ,..., Σ _s ) are adjusted so that the entire weighted error of the impedance curve Y is reduced ( Step S22).

Similarly, the calculation unit 11c adjusts only the parameters (C ₁ , p ₁ , R ₁ ) belonging to the parameter group II so that the weighted error of the entire curve is reduced (step S23), and the parameter group III A process of adjusting only the parameters (C ₂ , p ₂ ) belonging to the entire curve so as to reduce the weighted error of the entire curve (step S 24), and parameters (R ₂ , C ₃ , p ₃ , R ₃ , A process of adjusting only σ ₃ ) so as to reduce the weighted error of the entire curve (step S25) is executed.

  As described above, the parameter groups I to IV are adjusted independently in the steps S22 to S25 as well, but the adjustment is different from the steps S18 to 21 so that the error of the entire impedance curve Y is reduced. To be done.

After the parameter is adjusted by the process of steps S18 to S21 or the process of steps S22 to S25, the calculation unit 11c executes a process of storing the adjusted parameter in the storage unit 11a (step S26). It executes processes for updating the weighting coefficients w ₁ to w ₄ based on the error after the curve (step S27).

  Thereafter, the computing means 11c returns to step S11 and repeats the subsequent processing. When the loop count (parameter adjustment count) reaches a predetermined count, the subroutine (curve fitting) in FIG. 5 is terminated.

  When the curve fitting is completed by the processing as described above, the estimation processing unit 12 of the control unit 10 proceeds to step S4 in FIG. 4 and adjusts the parameters so as to follow the fitted impedance curve Y, that is, the actually measured value. After that, a process for extracting a predetermined geometric feature amount from the impedance curve Y defined by the impedance function Z (jω) is executed.

  The geometric feature amount to be extracted from the impedance curve Y is not limited as long as it has a correlation with the state amount of the battery 1 to be estimated (here, the charged state SOC and the deteriorated state SOH). Values as shown in FIG. 6 are considered as candidates. That is, one or more of values such as the length of the sections (1) to (16) in the impedance curve Y, the angle of the inclined portions (17) and (18), and the frequency of the points (19) to (23). It is. Further, a predetermined function (for example, n multiplier, trigonometric function, sum or difference) using any one or more of these values (1) to (23) as a variable may be extracted as a geometric feature. Good. Among the various feature quantities exemplified in FIG. 6, which value has a close correlation with the state of charge SOC and the deterioration state SOH of the battery has been examined in advance by experiments or the like. In step S4, Only specific features that are known to be correlated are extracted.

  When a predetermined geometric feature amount in the impedance curve Y is extracted as described above, the estimation processing unit 12 charges the battery 1 based on a predetermined prediction formula using the extracted geometric feature amount. A process for estimating the state SOC and the deterioration state SOH is executed (step S5). Note that the prediction equation here is a so-called multiple regression equation specified by a statistical method in order to calculate the state of charge SOC and the deterioration state SOH (objective variable) using the geometric feature amount (explanatory variable). is there. Such a prediction equation (multiple regression equation) is examined in advance by experiments or the like and stored in the control unit 10. In step S5, the charged state SOC and the deteriorated state SOH of the battery 1 are obtained by substituting the geometric feature into the stored prediction formula. As a matter of course, the prediction formula for obtaining the SOC and the prediction formula for obtaining the SOH are different, and the feature amounts used in the respective prediction formulas may be different from each other.

(3) Operation, etc. As described above, the vehicle of this embodiment is equipped with a state estimation device that estimates the state of the battery 1. This state estimation device includes a signal generation unit 13 that can output a plurality of AC signals having different frequencies to the battery, a frequency characteristic calculation unit 11 that calculates a frequency characteristic of the AC impedance of the battery 1, and a frequency of the calculated AC impedance. And an estimation processing unit 12 that estimates the state of charge SOC and the deterioration state SOH of the battery 1 based on the characteristics. The frequency characteristic calculation unit 11 includes a storage unit 11a that stores an impedance function Z (jω) that defines the frequency characteristic of the AC impedance, and a response signal (voltage sensor) output from the battery 1 in response to a signal from the signal generation unit 13 And the impedance function Z (jω) is adjusted so as to reduce the error with respect to the measured actual value data, and after the adjustment. And a calculation means 11c for performing a fitting process for storing the above function in the storage means 11a. The storage unit 11a includes, as an impedance function Z (jω), a plurality of functional blocks included in the equivalent circuit 1E of the battery 1 (inductance unit (i), positive electrode unit (ii), negative electrode unit (iii), diffusion unit in FIG. 2). (Iv)) stores a polynomial function including a plurality of terms (inductance term (I), positive term (II), negative term (III), diffusion term (IV)) in equation (1)) . The impedance function Z (jω) includes various parameters (parameter groups I to IV) classified into a plurality of groups corresponding to each functional block of the equivalent circuit 1E. In the process, it has a function of adjusting the parameters independently for each group (see steps S18 to S21 or S22 to S25 in FIG. 5). According to such a configuration, there is an advantage that the state of the battery 1 can be estimated accurately in a short time.

  That is, in the above embodiment, the parameters included in the impedance function Z (jω) of the battery 1 are classified into a plurality of groups corresponding to the functional blocks of the battery 1, and the parameters are adjusted according to the measured value data. In the fitting process (curve fitting), the parameters are adjusted independently for each of the above groups. For example, even when the fitting accuracy of the impedance curve Y before adjustment varies greatly depending on the location, The adjustment amount (change amount) can be set relatively freely for each group, and the optimum parameter can be obtained more efficiently. For this reason, unlike the case where all parameters can always be adjusted only simultaneously, high fitting accuracy can be obtained in a short time.

  For example, if all parameters are to be adjusted at the same time even though the fitting accuracy of the impedance curve Y before adjustment varies greatly depending on the location, the fitting accuracy is affected by the error of the poor fitting accuracy. Even the parameters corresponding to good parts may be changed greatly, and even if the number of adjustments is repeated, the accuracy of fitting is not stable, and it may take a long time to obtain good results. is there.

  On the other hand, when the parameters can be adjusted independently for each group corresponding to each functional block of the equivalent circuit 1E as in the above embodiment, the above situation is avoided. A high fitting accuracy can be obtained even with a relatively small number of adjustments.

  FIG. 7 shows the results of an experiment conducted by the inventor of the present application in order to confirm the effects as described above. In this figure, curve fitting was performed under a plurality of conditions for a specific type of lithium ion battery, and the results were compared. Specifically, (a) when a new battery is SOC = 0%, (b) when a new battery is SOC = 100%, (c) a battery after 200 cycles in a 25 ° C. environment is SOC = When 0%, (d) When the battery after 200 cycles in a 25 ° C. environment is SOC = 100%, (e) When the battery after 200 cycles in a 60 ° C. environment is SOC = 0%, f) Curve fitting was performed under the respective conditions when the battery after use of 200 cycles in a 60 ° C. environment had SOC = 100%. As a method of curve fitting, when the method described in the above embodiment is used, that is, when parameters are grouped according to the functional blocks of the equivalent circuit 1E (the parameters are adjusted independently for each of the parameter groups I to IV). 2 types) and a case where parameter grouping was not performed. In any of these fitting methods, the number of parameter adjustments is the same.

  As is clear from FIG. 7, when the parameter grouping is not performed (shown in the left column in the figure), there is a significant difference between the actually measured data indicated by the broken line and the calculated impedance curve indicated by the solid line. It can be seen that an error has occurred and sufficient fitting accuracy has not been obtained. On the other hand, according to the method of the above embodiment for adjusting the parameters for each group (shown in the right column in the figure), the calculated impedance curve is in good agreement with the actual measurement data, so the number of adjustments of the same parameter It can be seen that good fitting accuracy is obtained despite this.

  In the above embodiment, as a process of independently adjusting the parameters of the impedance function Z (jω) for each group, the parameters (parameter groups I to IV) of each group are set so that the overall error of the impedance curve Y is reduced. ) Sequentially (steps S22 to S25) and the error of each divided portion (curve portions y1 to y4) when the impedance curve Y is divided according to each functional block of the equivalent circuit 1E is individually reduced. Includes the process of sequentially adjusting the parameters of each group (steps S18 to S21) according to the variation in the amount of change during the previous parameter adjustment (more specifically, before and after the adjustment of all parameters). Depending on whether the amount of change falls within a certain range), one of the above two processes is selectively executed. . According to such a configuration, the parameters (parameter groups I to IV) of each group can be adjusted in consideration of the error of the entire impedance curve Y, and the related curve portion (any one of y1 to y4). It is also possible to make adjustments taking into account only the error. Then, by selecting the adjustment according to any of these procedures according to the variation in the previous parameter adjustment amount, it is possible to perform an appropriate parameter adjustment according to the situation, and further improve the efficiency of the fitting process. .

  In the above embodiment, a predetermined geometric feature amount (see, for example, FIG. 6) is extracted from the impedance curve Y defined by the impedance function Z (jω) after parameter adjustment, and the geometric feature is extracted. Based on a predetermined prediction formula using the quantity, the state of charge SOC and the deterioration state SOH of the battery 1 are estimated. According to such a configuration, the state of charge SOC and the deterioration state SOH of the battery 1 are accurately obtained based on the geometric feature amount of the impedance curve Y adjusted with high accuracy by the fitting process (curve fitting) as described above. Can be estimated. In addition, since it is only necessary to calculate a prediction expression using the geometric feature amount of the impedance curve Y, the SOC and SOH can be obtained easily and in a short time.

  In the above embodiment, a plurality of AC signals (sine waves) having different frequencies are input from the signal generator 13 to the battery 1 and the AC impedance of the battery 1 is measured based on the output voltage of the battery 1 with respect to the input signal. However, the input signal generated from the signal generator 13 for impedance measurement may be a signal including a plurality of frequency components, and is not limited to an AC signal (sine wave).

  For example, an impulse wave approximated to a Dirac delta function may be used as an input signal from the signal generator 13. Since such an impulse wave theoretically includes all frequency components, if an output signal (voltage) from the battery 1 that has received the impulse wave is measured over a predetermined time, the battery for a plurality of frequencies is used. 1 AC impedance can be specified. As described above, the method of inputting the impulse wave to the battery 1 (hereinafter referred to as the FFT method) is compared with the method of sequentially inputting a plurality of sine waves having different frequencies as in the above embodiment (hereinafter referred to as the AC method). Since signals of a plurality of frequency components can be input to the battery 1 in a very short time, it is advantageous in shortening the measurement time. However, in the case of the FFT method, the computer (the control unit 10 in the above embodiment) needs to have a function of Fourier transforming input / output signal waveforms.

  On the other hand, in the case of the AC method as shown in the above embodiment, since it is necessary to sequentially input a plurality of sine waves having different frequencies to the battery 1, it takes a longer time to measure than the FFT method, but the Fourier transform. Therefore, the configuration of the computer can be simplified.

  In the above-described embodiment, the rechargeable ion battery mounted on the vehicle including the hybrid vehicle or the electric vehicle is the target of estimation. However, the target of estimation can be limited to the rechargeable ion battery for vehicles. However, the present invention can be applied to various industrial batteries.

DESCRIPTION OF SYMBOLS 1 Battery 1E Equivalent circuit 11 Frequency characteristic calculation part 11a Memory | storage means 11b Measuring means 11c Calculation means 12 Estimation processing part 13 Signal generation part

Claims (4)

A signal generator capable of outputting a signal including a plurality of frequency components to the battery, a frequency characteristic calculator for calculating the frequency characteristics of the AC impedance of the battery, and the battery based on the frequency characteristics of the AC impedance calculated by the frequency characteristics calculator A state estimation device for a battery comprising an estimation processing unit for estimating a predetermined amount of state regarding
The frequency characteristic calculation unit includes a storage unit that stores an impedance function that defines the frequency characteristic of the AC impedance, and an actual measurement value of the AC impedance based on a response signal output from a battery with respect to a signal from the signal generation unit. Measuring means for obtaining data, and arithmetic means for performing fitting processing for adjusting the impedance function so as to reduce an error with respect to the actual measurement data measured by the measuring means, and storing the adjusted impedance function in the storage means. And
The storage means stores, as the impedance function, a polynomial function including a plurality of terms respectively corresponding to a plurality of functional blocks included in an equivalent circuit of the battery,
The impedance function includes various parameters classified into a plurality of groups corresponding to the functional blocks of the equivalent circuit.
The battery state estimation device, wherein the calculation means has a function of adjusting the parameter independently for each of the groups in the fitting process. The battery state estimation device according to claim 1,
As a process of independently adjusting the parameters of the impedance function for each group, a first process of sequentially adjusting the parameters of each group so as to reduce the overall error of the impedance curve based on the impedance function, and the impedance A second process for sequentially adjusting the parameters of each group so that the error of each divided portion when the curve is divided according to each functional block of the equivalent circuit is individually reduced;
The battery state estimation device according to claim 1, wherein the arithmetic means selectively executes one of the first and second processes in accordance with variation in the amount of change during parameter adjustment performed last time. The battery state estimation device according to claim 1 or 2,
The estimation processing unit extracts a predetermined geometric feature amount from an impedance curve based on an impedance function after the fitting process is completed, and based on a predetermined prediction formula using the geometric feature amount. Thus, at least one of a charged state and a deteriorated state of the battery is estimated as the state quantity. In the battery state estimation device according to any one of claims 1 to 3,
In the equivalent circuit of the battery, as the functional block, an inductance part derived from the battery cell structure, a positive electrode part derived from a reaction at the positive electrode, a negative electrode part derived from a reaction at the negative electrode, and movement of ions of the positive electrode and the negative electrode Includes a diffusion part derived from
The battery state estimation device according to claim 1, wherein the impedance function includes an inductance term, a positive term, a negative term, and a diffusion term corresponding to each functional block.

Images