JP5595361B2 - Secondary battery charge state estimation device - Google Patents

Description

  The present invention relates to a state-of-charge estimation device for a secondary battery, and more particularly to estimation of a state of charge using an extended Kalman filter.

  A technology for estimating a state of charge (SOC) of a secondary battery such as a nickel metal hydride secondary battery or a lithium ion secondary battery and controlling the charge / discharge of the secondary battery based on the estimated SOC is known. Yes.

  The following patent documents include a sensing unit that measures charge / discharge current flowing through a battery, battery temperature and terminal voltage, a prediction unit that accumulates charge / discharge current to estimate battery SOC, battery temperature, charge / discharge current. A data rejection unit that generates information corresponding to an error generated by the measurement model according to at least one of the time change rate of SOC, SOC, and charge / discharge current, and the measurement model and information corresponding to the error are used A battery management system including a measurement unit that corrects the estimated SOC of the battery is disclosed. Specifically, the data rejection unit sets a modeled battery equivalent circuit as a measurement model, and generates a gain corresponding to the variance of errors generated by the measurement model. Then, a Kalman gain is generated using the dispersion gain of the measurement model error, and the estimated SOC is corrected using the generated Kalman gain.

  Specifically, the prediction unit predicts the voltage Vdiff applied to the battery SOC and the diffusion impedance, which are the state variables (x), using the state equation. The prediction unit generates a covariance for the predicted state variable and the estimation error of the state variable and supplies the generated covariance to the measurement unit. The measurement unit predicts a value that can be measured using the SOC predicted by the prediction unit and the voltage Vdiff, that is, the terminal voltage of the battery. The measurement unit solves the non-linearity between SOC and OCV and uses differential equations to use the Kalman filter. Then, the measurement unit generates a Kalman gain for correcting the predicted SOC and voltage Vdiff. The Kalman gain is set to a value that minimizes the covariance.

JP 2008-10420 A

  In the above prior art, a reliable period and an unreliable period are set for the temperature, SOC, time change rate of current, and current magnitude, respectively, and the error variance is adjusted according to the section. That is, the error variance is fixed to a predetermined value in a reliable period, but is changed by the data rejection unit in an unreliable period. For example, in an interval where the magnitude of the current exceeds a predetermined reference value and is not reliable, the error variance value increases, the Kalman gain decreases, and the SOC predicted by the prediction unit is corrected by the measurement model in the measurement unit. Decrease.

  However, in the above prior art, the Kalman filter is used to correct the predicted SOC and the weight for correction is set in advance. However, when the battery is in an abnormal state such as overcharge, the charging efficiency is increased. Even though the accuracy of the measurement model is reduced due to deterioration, the SOC is corrected with a fixed weight set in advance, and there is a problem that the abnormal state of the battery cannot be detected quickly.

  The object of the present invention is to estimate the state of charge of the secondary battery and correct the estimated state of charge using the measurement model of the secondary battery in the abnormal state of the secondary battery. Is to provide a state-of-charge estimation device for a secondary battery capable of appropriately detecting the state of charge.

  The present invention is a secondary battery charge state estimation device, a voltage detection means for detecting a terminal voltage of the secondary battery, a charge state estimation means for estimating a charge state of the secondary battery, and a predetermined measurement Estimated based on a terminal voltage estimation means for estimating the terminal voltage of the secondary battery by a model, a weight calculated using a weight indicating the uncertainty of the measurement model and a weight indicating the uncertainty of the voltage detection means. Correction value calculating means for calculating a correction value for correcting the state of charge, and charge state correcting means for correcting the charge state estimated using the calculated correction value, the correction value calculating means Changing the gain by changing at least one of a weight indicating the uncertainty of the measurement model and a weight indicating the uncertainty of the voltage detection unit according to the state of the secondary battery. To.

  In one embodiment of the present invention, the gain is characterized in that the gain increases as the weight indicating the uncertainty of the measurement model increases, and increases as the weight indicating the uncertainty of the voltage detection means decreases. The weight indicating the uncertainty has a larger characteristic as the uncertainty of the measurement model is larger, and the weight indicating the uncertainty of the voltage detecting unit has a larger characteristic as the uncertainty of the voltage detecting unit is larger, and the correction value The calculating means increases the gain by relatively reducing the weight indicating the uncertainty of the voltage detecting means in accordance with the state of the secondary battery.

  In another embodiment of the present invention, the correction value calculation means changes the gain when the secondary battery is in an overcharged state.

  In another embodiment of the present invention, the overcharged state of the secondary battery is a threshold value of at least one of a terminal voltage of the secondary battery, a temperature increase rate of the secondary battery, and an internal pressure of the secondary battery. It is detected based on whether or not

  In another embodiment of the present invention, the charging state estimation unit estimates a current offset included in the current detection unit, and corrects a charge / discharge current detected by the current detection unit based on the current offset. Offset correction value calculation that estimates the charge state of the secondary battery and calculates the correction value of the estimated current offset using an error between the estimated terminal voltage and the terminal voltage detected by the voltage detection means And offset correction means for correcting the estimated current offset using the correction value calculated by the offset correction value calculation means.

  According to the present invention, in the abnormal state of the secondary battery, the weight is changed by changing the gain by changing at least one of the weight indicating the uncertainty of the measurement model and the weight indicating the uncertainty of the voltage detection means. Compared to the case where it is constant, the charged state of the secondary battery can be estimated appropriately even in the abnormal state of the secondary battery. Thereby, for example, even in an overcharged state, the state of charge can be appropriately estimated, and the overcharged state of the secondary battery can be suppressed based on the estimated state of charge.

It is a system configuration figure of an embodiment. It is a lineblock diagram of the rechargeable battery and battery ECU of an embodiment. It is a functional block diagram of a control part of an embodiment. It is an equivalent circuit diagram of a polarization voltage. It is a graph which shows the relationship between an electromotive force and charge condition SOC. It is a processing flowchart of an embodiment. It is a graph which shows the current offset simulation result using two models. It is a graph which shows the relationship between the weight R and a terminal voltage.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings, taking a hybrid electric vehicle as an example. In addition, although this embodiment illustrates the hybrid electric vehicle which is one of the electric vehicles, it can be applied to other electric vehicles including a motor as a drive source.

1. Basic Configuration FIG. 1 shows a schematic configuration of a hybrid electric vehicle. The vehicle ECU 10 controls the inverter 50 and the engine ECU 40. The engine ECU 40 controls the engine 60. The battery ECU 20 functions as a charge state estimation device, receives information such as the battery voltage V, the charge / discharge current I, and the battery temperature T from the secondary battery 30 and estimates the state of charge (SOC) of the secondary battery 30. In addition, the battery ECU 20 transmits battery information such as the SOC and battery temperature of the secondary battery 30 to the vehicle ECU 10. The vehicle ECU 10 controls charging / discharging of the secondary battery 30 by controlling the engine ECU 40, the inverter 50, and the like based on various battery information.

  The secondary battery 30 supplies power to the motor 52. The inverter 50 converts DC power supplied from the secondary battery 30 into AC power and supplies it to the motor 52 when the secondary battery 30 is discharged.

  The engine 60 transmits power to the wheels via the power split mechanism 42, the speed reducer 44, and the drive shaft 46. The motor 52 transmits power to the wheels via the speed reducer 44 and the dry shaft 46. When the secondary battery 30 needs to be charged, a part of the power of the engine 60 is supplied to the generator 54 via the power split mechanism 42 and used for charging.

  The vehicle ECU 10 receives information on the operating state of the engine 60 from the engine ECU 40, the operation amount of the accelerator pedal, the operation amount of the brake pedal, the operation information such as the shift range set by the shift lever, and the battery information such as the SOC from the battery ECU 20. Is output to the engine ECU 40 or the inverter 50 to drive the engine 60 or the motor 52.

  FIG. 2 shows configurations of the secondary battery 30 and the battery ECU 20. The secondary battery 30 is configured, for example, by connecting battery blocks B1 to B20 in series. Battery blocks B <b> 1 to B <b> 20 are accommodated in battery case 32. The battery blocks B1 to B20 are each configured by electrically connecting a plurality of battery modules in series, and each battery module is configured by electrically connecting a plurality of single cells (cells) in series. A plurality of temperature sensors 34 are provided in the battery case 32.

  Next, the configuration of the battery ECU 20 will be described.

  The voltage measuring unit 22 measures the terminal voltage of the secondary battery 30. The voltage measuring unit 22 measures the terminal voltages of the battery blocks B1 to B20 and outputs them to the control unit 26. The control unit 26 stores the voltage measurement value in the storage unit 28. The output of the voltage data to the control unit 26 is performed at a preset period, for example, 100 msec. The control unit 26 calculates the battery voltage V by summing the terminal voltages of the battery blocks B1 to B20.

  The current measuring unit 23 measures the charging / discharging current I during charging / discharging of the secondary battery 30 and outputs the measured charging / discharging current I to the control unit 26. The control unit 26 stores the current measurement value in the storage unit 28. The current measuring unit 23 generates a current measurement value, for example, when charging is negative and when discharging is positive.

The temperature measurement unit 24 measures the battery temperature of the secondary battery 30 and outputs it to the control unit 26. The control unit 26 stores the temperature data in the storage unit 28 .

  The control unit 26 includes a DCIR (internal resistance) unit 261, a polarization voltage calculation unit 262, an electromotive force calculation unit 263, a charge state estimation unit 264, an extended Kalman filter (EKF) 265, and a correction unit 266. The control unit 26 estimates the state of charge (SOC) of the secondary battery 30 based on the charge / discharge current I. Specifically, the SOC of the secondary battery 30 is estimated by integrating the charge / discharge current I, and information corresponding to the error generated by the measurement model is generated. Then, a Kalman gain is generated using the dispersion gain of the measurement model error, and the estimated SOC is corrected or updated using the generated Kalman gain.

  FIG. 3 shows a detailed functional block diagram of the control unit 26. Based on a predetermined measurement model, the control unit 26 uses an internal resistance voltage drop calculated by the DCIR unit 261, a long-term polarization voltage and a short-term polarization voltage calculated by the polarization voltage calculation unit 262, and an electromotive force calculation unit 263. The estimation unit 267 that estimates the terminal voltage of the secondary battery 30 using the calculated electromotive force, the error calculation unit 268 that calculates the error between the voltage measured by the voltage measurement unit 22 and the estimated terminal voltage, and the error An extended Kalman filter (EKF) 265 that generates a Kalman gain by using it and calculates various correction values, and a correction unit 266 that performs correction using the calculated correction values are provided. The electromotive force calculation unit 263 calculates the estimated SOC obtained by integrating the charging current in the charge state estimation unit 264 by converting it into an electromotive force using a predetermined electromotive force map. The extended Kalman filter (EKF) 265 calculates a current sensor offset correction value (first correction value) and an SOC correction value (second correction value) as correction values, and a long-term polarization voltage correction value (third correction value). A short-term polarization correction value (fourth correction value) is calculated. The correction unit 266 corrects the estimated current sensor offset value included in the current measurement unit 23 using the current sensor offset correction value, and corrects the estimated SOC calculated by the charging state estimation unit 264 using the SOC correction value. . Further, the correction unit 266 corrects the long-term polarization voltage and the short-term polarization voltage using the long-term polarization voltage correction value and the short-term polarization correction value, respectively.

  Hereinafter, the SOC estimation method in the control unit 26 will be described. As described above, the gain (Kalman gain) is a transfer function for calculating a correction value for correcting the estimated SOC. In this embodiment, the gain is calculated by using an estimated terminal voltage calculated by the measurement model and In addition to the error from the terminal voltage calculated by the voltage detection means, a weight indicating the uncertainty of the measurement model and a weight indicating the uncertainty of the voltage detection means are also considered. Specifically, the gain is generated in consideration of the usage ratio of the terminal voltage calculated by the measurement model and the terminal voltage calculated by the voltage detection means (which one is used more). That is, the gain is changed by changing the utilization ratio according to the battery state, and a suitable correction value is calculated. More specific description will be given below.

2. SOC Estimation Method The control unit 26 considers that the current measurement value from the current measurement unit 23 includes a current sensor offset, sets the current sensor offset estimation value δI, and determines the current measurement value I and the current sensor offset estimation. Difference calculation with the value δI is performed. That is, I−δI is calculated. The difference value is supplied to a DCIR (internal resistance) unit 261, a polarization voltage calculation unit 262, and a charge state estimation unit 264.

  The DCIR (internal resistance) unit 261 plots a set of current measurement values and voltage measurement values in advance, calculates the DCIR (internal resistance) of the secondary battery 30 from the slope of the primary approximate line, and uses this DCIR voltage Calculate the drop (voltage drop). That is, DCIR · (I−δI) is calculated. The DCIR unit outputs the calculation result to the terminal voltage estimation unit 267.

  The polarization voltage calculation unit 262 calculates the polarization voltage based on an equivalent circuit model of the polarization voltage. The polarization voltage is calculated based on the current measurement value. The polarization voltage includes a short-term polarization voltage and a long-term polarization voltage, and the polarization voltage calculation unit 262 calculates a short-term polarization estimated value and a long-term polarization estimated value. The polarization voltage calculation unit 262 outputs each calculation result to the terminal voltage estimation unit 267.

  The charge state estimation unit 264 calculates an SOC estimated value by integrating the amount (I−δI) obtained by subtracting the current sensor offset from the measured current value. The charge state estimation unit 264 outputs the estimated SOC value obtained by integrating the current to the electromotive force calculation unit 263.

  The electromotive force calculation unit 263 stores an electromotive force map that preliminarily defines the relationship between the SOC and the electromotive force in the memory, and obtains an electromotive force corresponding to the estimated SOC value with reference to the electromotive force map. The electromotive force calculation unit outputs the obtained electromotive force to the terminal voltage estimation unit 267.

であり、このモデルから電圧は

となる。 The terminal voltage estimation unit 267 estimates the voltage of the secondary battery 30 based on a predetermined measurement model. The measurement model is different for discharging and charging.

And the voltage from this model is

It becomes.

であり、このモデルから電圧は

となる。
但し、
ｘ：状態推定値
Ｖｐｌ：長期分極
Ｖｐｓ：短期分極
δＩ：電流センサオフセット推定値
ｕ＝Ｉ：計測電流値
ｙ＝Ｖｂ：計測電圧値
Ｃｐｌ：分極電圧の等価回路モデルにおける長期分極キャパシタ
Ｃｐｓ：分極電圧の等価回路モデルにおける短期分極キャパシタ
Ｒｐｌ：分極電圧の等価回路モデルにおける長期分極抵抗
Ｒｐｓ：分極回路の等価回路モデルにおける短期分極抵抗
ＤＣＩＲ：内部抵抗
ηｃ：充電効率
τδ_I：電流センサオフセットの相関時間
Ｖｅｍｆ（ＳＯＣ）：ＳＯＣ推定値から算出された起電力
である。また、本実施形態では、電流センサオフセットを、時間変化量が現在の値によって決まる、いわゆるＥＣＲＰ（Exponentially correlated Random Process）でモデル化して、δＩ（・）＝−δＩ／τδ_I＋σ（２／τδ_I）^0.5ωとしている。ここで、（・）は時間微分を表し、ωはランダムノイズを表す。 On the other hand, in the case of charging,

And the voltage from this model is

It becomes.
However,
x: state estimated value Vpl: long-term polarization Vps: short-term polarization δI: current sensor offset estimated value u = I: measured current value y = Vb: measured voltage value Cpl: long-term polarization capacitor Cps: polarization voltage in an equivalent circuit model of polarization voltage Short-term polarization capacitor Rpl in the equivalent circuit model of the long-term polarization resistance Rps in the equivalent circuit model of the polarization voltage RIR: short-term polarization resistance in the equivalent circuit model of the polarization circuit DCIR: internal resistance ηc: charging efficiency τδ _I : correlation time Vemf of the current sensor offset (SOC): An electromotive force calculated from the estimated SOC value. Further, in the present embodiment, the current sensor offset, determined by the time variation of the current value, and modeled by the so-called ECRP (Exponentially correlated Random Process), δI (·) = - δI / τδ I + σ (2 / τδ _I ) ^0.5 Ω. Here, (•) represents time differentiation, and ω represents random noise.

である。 FIG. 4 shows an equivalent model of the polarization voltage. The change in polarization voltage over time is

It is.

  FIG. 5 shows an example of an electromotive force map, that is, a map that defines the relationship between the estimated SOC value and the electromotive force.

  The terminal voltage estimator 267 performs the long-term polarization voltage Vpl, the short-term polarization voltage Vps, the electromotive force Vemf (SOC) calculated from the estimated SOC value, and the internal resistance drop DCIR in accordance with the above formula (2) or (4). The measurement voltage Vb is calculated using (I−δI).

  The error calculation unit 268 calculates a difference between the voltage measurement value actually measured by the voltage measurement unit 22 and the measurement voltage Vb calculated by the terminal voltage estimation unit 267, that is, an error caused by the measurement model. The calculated error is supplied to an extended Kalman filter (EKF) 265.

  The extended Kalman filter (EKF) 265 generates a Kalman gain by using the dispersion gain of the measurement model error, calculates an SOC correction value by using the generated Kalman gain, and also calculates a current sensor offset correction value, a long-term polarization. A correction value and a short-term polarization correction value are calculated. These correction values are supplied to the correction unit 266.

  The SOC correction value is supplied to a correction unit 266 that corrects the SOC. Correction unit 266 corrects the SOC estimated value obtained by charge state estimating unit 264 using the SOC correction value. The corrected SOC is transmitted to the vehicle ECU 10.

  The current sensor offset correction value is supplied to a correction unit 266 that corrects the current sensor offset. The correction unit 266 corrects the current sensor offset estimated value using the current sensor offset correction value.

  The long-term polarization correction value and the short-term polarization correction value are supplied to the respective correction units 266 and used for correcting the long-term polarization estimation value and the short-term polarization estimation value.

であり、ｘは状態ベクトル、ｕは入力ベクトル、関数ｆはｘとｕについて非線形な関数である。これを、ｘの微分がｘに依存しないゲインとなるように近似する。すなわち、連続時間では、

であり、離散時間では、

である。ここで、Ａは状態行列、Ｂは入力行列である。Ａは、具体的には、（１）式、（３）式を微分して得られ、放電の場合には、

であり、充電の場合には、

である。 FIG. 6 shows a processing flowchart in the present embodiment. First, the measurement model is linearized (S101). Nonlinear systems are

Where x is a state vector, u is an input vector, and function f is a non-linear function with respect to x and u. This is approximated so that the derivative of x becomes a gain independent of x. That is, in continuous time,

And in discrete time,

It is. Here, A is a state matrix and B is an input matrix. Specifically, A is obtained by differentiating equations (1) and (3). In the case of discharge,

In the case of charging,

It is.

で与えられる。ここで、Ｅは、

である。ΔＴは、離散時間における最小の時間単位（サンプリング時間）である。ソフトウェアによる処理では、動作周期（例えば１００ｍｓｅｃ）が最小の時間単位として用いられる。 The discretized (k-1) component Adk-1 of A is

Given in. Where E is

It is. ΔT is the minimum time unit (sampling time) in discrete time. In processing by software, an operation cycle (for example, 100 msec) is used as a minimum time unit.

である。なお、ハット（ ^ ）は推定値を表す。また、

は時間更新状態推定値であり、

は観測更新状態推定値である。 After linearization, next, time update processing is performed (S102). The time update process

It is. Hat (^) represents an estimated value. Also,

Is the time update state estimate,

Is the observed update state estimate.

である。Ｐは、真の状態と推定値の誤差の共分散値であり、状態数×状態数の行列で本実施形態では４×４である。ｘｋを真の状態とすると、誤差は

であり、誤差共分散推定値は、

である。対角成分が状態推定値の誤差共分散値である。具体的には、

である。 The error covariance estimate is

It is. P is a covariance value of the error between the true state and the estimated value, and is a matrix of the number of states × the number of states, which is 4 × 4 in this embodiment. If xk is true, the error is

And the error covariance estimate is

It is. The diagonal component is the error covariance value of the state estimation value. In particular,

It is.

で定義される。例えば、分極モデルによる長期分極の推定誤差が０．１Ｖと推定できるならば、Ｑ１＝０．１となる。 Further, Q in the equation (16) is a weight matrix of the state estimation error,

Defined by For example, if the estimation error of long-term polarization based on the polarization model can be estimated to be 0.1 V, Q1 = 0.1.

である。 After performing the time update process, linearization is performed (S103). That is,

It is.

である。ｘｋにおける（−）は補正値で補正される前の値であり、（＋）は補正値で補正された後の更新値である。ここで、（２７）式の右辺第２項が補正値であり、カルマンゲイン、すなわち観測更新ゲインを

として算出される。具体的に示すと、長期分極電圧推定値、短期分極電圧推定値、ＳＯＣ推定値、電流センサオフセット推定値は、それぞれ

によって更新処理される。 Finally, observation update processing is performed (S104). The observation update process

It is. (−) in xk is a value before being corrected with the correction value, and (+) is an updated value after being corrected with the correction value. Here, the second term on the right side of equation (27) is the correction value, and the Kalman gain, that is, the observation update gain is

Is calculated as Specifically, the long-term polarization voltage estimated value, the short-term polarization voltage estimated value, the SOC estimated value, and the current sensor offset estimated value are respectively

Is updated.

として定義される。式（３３）から分かるように、電圧計測の不確かさ、すなわち電圧計測の誤差が大きくなると重み行列Ｒの値は大きくなる。言い換えれば、電圧計測の誤差が小さく電圧計測の信頼性が高い場合には、重み行列Ｒの値は小さくなる。重み行列Ｑについても同様であり、式（２３）を再び参照すれば分かるように、測定モデルの不確かさ、すなわち推定誤差が大きくなると重み行列Ｑの値は大きくなる。言い換えれば、測定モデルの推定誤差が小さく測定モデルの信頼性が高い場合には、重み行列Ｑの値は小さくなる。 Note that R in equation (25) is a measurement error weight matrix,

Is defined as As can be seen from the equation (33), the value of the weight matrix R increases as the uncertainty of voltage measurement, that is, the error in voltage measurement increases. In other words, when the voltage measurement error is small and the voltage measurement reliability is high, the value of the weight matrix R is small. The same applies to the weight matrix Q. As can be understood by referring to the equation (23) again, the value of the weight matrix Q increases as the uncertainty of the measurement model, that is, the estimation error increases. In other words, when the estimation error of the measurement model is small and the reliability of the measurement model is high, the value of the weight matrix Q is small.

  Therefore, by adjusting the balance between the weight matrix Q of the state estimation error and the weight matrix R of the measurement error, it is possible to set which of the estimation based on the measurement model and the observation value is used more. For example, when the estimation accuracy of the measurement model is high and the reliability is high, the SOC estimation by the measurement model is emphasized by setting Q <R. On the other hand, when it is assumed that the actual deviation from the measurement model is large, it is possible to perform SOC estimation based on the measured voltage by setting Q> R. Qualitatively, when it is evaluated that the reliability of the measurement model is low, Q is relatively increased or R is relatively decreased. That is, when it is evaluated that Q is Q0 and R is R0 in the normal state and the secondary battery 30 is in an abnormal state and the reliability of the measurement model is lowered, the Q at that time Is made relatively larger than Q0, or R at that time is made relatively smaller than R0. The abnormal state of the secondary battery 30 is, for example, an overcharged state, and can be detected based on whether the measured voltage of the secondary battery 30 exceeds a predetermined upper limit threshold (overcharged state). Since the measured voltage at this time is considered to accurately reflect the abnormal state of the secondary battery 30, R is relatively decreased as the measured voltage is increased, and SOC estimation based on the measured voltage is performed.

  As can be seen from the equations (16) and (25), since R exists in the denominator of the gain Kk and Q exists in the numerator, the gain Kk increases as Q increases or decreases. Have. For simplicity, it can be said that the gain Kk is substantially proportional to Q / R. Referring to Equation (31), SOC value (after correction) = SOC value (before correction) + Kk (measured voltage value−model estimated voltage value), and the corrected SOC estimated value increases as the gain Kk increases. When R is relatively small, the gain Kk increases, and as a result, the estimated value of SOC also increases. Therefore, by reducing R in the overcharged state, the estimated SOC value increases, and an appropriate estimated SOC value according to the overcharged state of the secondary battery 30 can be output. That is, if the value of R is fixed even if the measured voltage exceeds the upper limit threshold, an SOC value lower than the original SOC value of the secondary battery 30 is estimated due to a decrease in the reliability of the measurement model. In this embodiment, the SOC value can be increased and corrected to be close to the original SOC value of the secondary battery 30 by changing R relatively small.

  FIG. 8 shows how the R changes according to the measured voltage. When the measurement voltage is equal to or less than the threshold Tr1, R is a constant value, for example, 10. This constant value is determined by the uncertainty of the voltage measurement set in advance by the equation (33).

  On the other hand, when the measured voltage exceeds the threshold value Tr1, R is corrected to be small and set to 5, for example. Further, when the measured voltage exceeds the threshold value Tr2 (Tr1 <Tr2), R is corrected to be further reduced and set to 1, for example. In other words, it can be said that such a change in R decreases the uncertainty of voltage measurement (the reliability of voltage measurement increases) as the measurement voltage increases.

  Of course, FIG. 8 is an example, and it is possible to define an arbitrary relationship such that R has a negative phase with respect to the measurement voltage, and not only when R changes stepwise with respect to the measurement voltage, You may set so that R may change continuously with respect to a measurement voltage.

  Moreover, in order to detect the overcharge state of the secondary battery 30, the temperature increase rate of the secondary battery 30 can also be used instead of the measurement voltage. Two parameters of measurement voltage and temperature rise rate may be used. For example, R is decreased when the measured voltage exceeds the threshold and the rate of temperature increase exceeds the threshold. It may be possible to reduce R when at least one of the measured voltage and the temperature rise rate exceeds a threshold value.

  Further, in order to detect the overcharged state of the secondary battery 30, the internal pressure of the secondary battery 30 may be used. For example, R is decreased when the internal pressure of the secondary battery 30 exceeds a threshold value.

  As described above, correction values for correcting the long-term polarization voltage estimated value, the short-term polarization voltage estimated value, the current sensor offset estimated value, and the SOC estimated value can be calculated and corrected by the extended Kalman filter (EKF). Therefore, the accuracy of SOC estimation can be improved. In particular, in the present embodiment, it is possible to output an appropriate SOC value corresponding to the overcharged state of the secondary battery 30, thereby taking measures such as immediately stopping charging of the secondary battery 30.

In the present embodiment, the current sensor offset, time variation is determined by the current value, and modeled by the so-called ECRP, δI (·) = - δI / τδ I + σ (2 / τδ I) is set to ^0.5 omega However, the present invention includes a case where the current sensor offset is modeled so as to change with time by ω (random noise) so that δI (·) = ω. In this case, the number (1) and (3), (9), in (10), -1 / τδ _I is "0". However, the model using ECRP is preferable because it reaches the true error earlier and has less variation. FIG. 7 shows simulation results of a model using ECRP and a model that changes with time at ω. In the figure, the horizontal axis represents time, and the vertical axis represents the current sensor offset δI. Reference numeral 100 denotes a model using ECRP, and reference numeral 102 denotes a model that changes over time with ω. In the model that changes over time with ω, a phenomenon occurs in which the estimated value varies near the true value. On the other hand, in the model using ECRP, the estimated value quickly approaches the true value and no variation occurs. From this simulation result, it is understood that a model using ECRP is more desirable.

  Further, in the present embodiment, the correction is calculated using the extended Kalman filter from the error between the estimated terminal voltage and the terminal voltage detected by the voltage detection unit. Other adaptive filters can also be used. In this embodiment, the current sensor offset estimation value is corrected in order to improve the accuracy of current sensor offset estimation and thus improve the accuracy of SOC estimation. However, the correction need not necessarily be performed.

  In the present embodiment, the overcharged state is described as an example of the abnormal state of the secondary battery 30, but R can be similarly changed in other abnormal states in which the charging efficiency is deteriorated. In short, an abnormal state that decreases the reliability of the measurement model is detected, and when such an abnormal state is detected, Q and R are not maintained at constant values, but Q is relatively increased. Alternatively, by changing R so as to be relatively small, the weight indicating the uncertainty of the measurement model and the weight indicating the uncertainty of the voltage detection means may be changed.

  DESCRIPTION OF SYMBOLS 10 Vehicle ECU, 20 Battery ECU, 22 Voltage measurement part, 23 Current measurement part, 26 Control part.

Claims (7)

A rechargeable battery state of charge estimation device comprising:
Voltage detecting means for detecting a terminal voltage of the secondary battery;
Charging state estimation means for estimating a charging state of the secondary battery;
Terminal voltage estimating means for estimating the terminal voltage of the secondary battery according to a predetermined measurement model;
Correction value calculation means for calculating a correction value for correcting the estimated state of charge based on a gain calculated using a weight indicating the uncertainty of the measurement model and a weight indicating the uncertainty of the voltage detection means When,
Charge state correction means for correcting the state of charge estimated using the calculated correction value;
With
The correction value calculation means changes the gain by changing at least one of a weight indicating the uncertainty of the measurement model and a weight indicating the uncertainty of the voltage detection means according to the state of the secondary battery. ,
The gain is larger as the weight indicating the uncertainty of the measurement model is larger, and the gain is larger as the weight indicating the uncertainty of the voltage detection unit is smaller.
The weight indicating the uncertainty of the measurement model has a larger characteristic as the uncertainty of the measurement model is larger,
The weight indicating the uncertainty of the voltage detection means has a characteristic that the greater the uncertainty of the voltage detection means, the greater the characteristic,
The correction value calculation means increases the gain by relatively reducing the weight indicating the uncertainty of the voltage detection means according to the state of the secondary battery, and increasing the gain. apparatus. The rechargeable battery state of charge estimation device according to claim 1,
The secondary battery charge state estimation apparatus, wherein the correction value calculation means changes the gain when the secondary battery is in an overcharge state. In the rechargeable battery state of charge estimation device according to claim 2 ,
The overcharge state of the secondary battery is detected based on whether or not at least one of a terminal voltage of the secondary battery, a temperature increase rate of the secondary battery, and an internal pressure of the secondary battery exceeds a threshold value. A charging state estimation device for a secondary battery, which is characterized. In the secondary battery charge state estimation device according to any one of claims 1 to 3 ,
The apparatus for estimating a state of charge of a secondary battery, wherein the predetermined measurement model is based on an equivalent circuit model of a polarization voltage of the secondary battery. In the rechargeable battery state of charge estimation device according to claim 4 ,
The terminal voltage estimation means uses a terminal voltage using a long-term polarization voltage, a short-term polarization voltage, a voltage drop due to internal resistance in the equivalent circuit model, and an electromotive force corresponding to the charge state estimated by the charge state estimation means. A state of charge estimation device for a secondary battery, wherein In the secondary battery charge state estimation device according to any one of claims 1 to 5 ,
The correction value calculation means is an extended Kalman filter . In the secondary battery charge state estimation device according to any one of claims 1 to 6 ,
The charging state estimation unit estimates a current offset included in the current detection unit, corrects a charge / discharge current detected by the current detection unit based on the current offset, and estimates a charging state of the secondary battery,
An offset correction value calculating means for calculating a correction value of the estimated current offset using an error between the estimated terminal voltage and the terminal voltage detected by the voltage detecting means;
Offset correction means for correcting the estimated current offset using the correction value calculated by the offset correction value calculation means ;
A charging state estimating device for a secondary battery, comprising:

Images