JP7705756B2 - BATTERY STATE ESTIMATION METHOD AND BATTERY STATE ESTIMATION DEVICE - Google Patents

Description

The present invention relates to a battery state estimation method and a battery state estimation device.

The state of charge (SOC) of a battery (cell) is usually estimated using the cell's charge and discharge current, terminal battery and surface temperature.

For example, since there is a close relationship between the cell terminal voltage (open circuit voltage (OCV)) when not charging or discharging and the SOC, one method is to measure the OCV and convert it to SOC. Another method is to use a cell equivalent circuit model that uses the OCV, resistance, and polarization (resistor and capacitor connected in parallel) to calculate the OCV from the cell terminal voltage (closed circuit voltage (CCV)) during charging and discharging, and then convert it to SOC. Another method is to measure or calculate the OCV, find the initial SOC, and then integrate the charge and discharge current to calculate the SOC.

In these SOC estimations, if the current and voltage sensing errors are large, the SOC error will be large.

Patent document 1 discloses an apparatus that detects the charge/discharge current and terminal voltage of a secondary battery, estimates a current offset, and corrects the detected charge/discharge current using the current offset to estimate the state of charge of the secondary battery, estimating the terminal voltage of the secondary battery using a predetermined measurement model, calculating a first correction value for the estimated current offset using the error between the estimated terminal voltage and the detected terminal voltage, and correcting the estimated current offset using the first correction value. Patent document 1 also discloses that the predetermined measurement model is based on an equivalent circuit model of the polarization voltage of the secondary battery, and that an extended Kalman filter is used.

Patent No. 5616464

The secondary battery charge state estimation device described in Patent Document 1 does not take into account cases where the battery has deteriorated.

Normally, when a battery deteriorates, its capacity decreases and its internal resistance increases, causing the resistance and polarization values of the cell equivalent circuit to fluctuate. This results in low accuracy in estimating the SOC with conventional methods. Therefore, there is room for improvement in terms of accuracy when estimating the SOC, taking into account battery deterioration during charging and discharging.

An object of the present invention is to estimate the SOC with high accuracy even when the current differs between when the battery is charged and when it is discharged .

The battery state estimation method of the present invention is a method for estimating the state of a battery using an apparatus having a Kalman filter calculation unit, a battery equivalent circuit model recording unit, and a system noise covariance calculation unit, and includes a step in which the Kalman filter calculation unit sequentially estimates the charging rate and capacity deterioration rate of the battery, the current offset of the current sensor, and the internal resistance of the battery set in the battery equivalent circuit model using measurement values of a voltage sensor, a current sensor, and a temperature sensor installed on the battery and a battery equivalent circuit model recorded in the battery equivalent circuit model recording unit, and a step in which the system noise covariance calculation unit changes the system noise covariance value, which is a setting value used in the Kalman filter calculation unit, between when used for estimation during charging of the battery and when used for estimation during discharging of the battery, and the system noise covariance value includes the system noise covariance value of the current offset and the internal resistance, and the system noise covariance value of the current offset and the internal resistance is made smaller during discharging than during charging.

According to the present invention, even when the current during charging and discharging of the battery is different , the SOC can be estimated with high accuracy.

1 is a configuration diagram showing a battery state estimating device according to a first embodiment; FIG. 2 is a configuration diagram showing a state estimation unit 552 of FIG. 1. FIG. 2 is a circuit diagram showing a battery equivalent circuit model used when calculating a CCV. FIG. 2 is a flow diagram showing a battery state estimation method according to the first embodiment. FIG. 11 is a configuration diagram showing a battery state estimating device according to a second embodiment.

Embodiments of the present disclosure will be described below with reference to the drawings.

(Embodiment 1)
FIG. 1 is a configuration diagram showing a battery state estimating device according to a first embodiment.

In this figure, the battery state estimation device 500 is composed of a Kalman filter calculation unit 510, a battery equivalent circuit model recording unit 520, a system noise covariance calculation unit 530, and a battery deterioration detection unit 540. The Kalman filter calculation unit 510 includes a state estimation unit 552 and an error variance calculation unit 554.

The Kalman filter calculation unit 510 receives as input the measured values from the voltage sensor 200, which measures the terminal voltage CCV (closed circuit voltage) of the battery 100 (cell), the current sensor 300, which measures the charge/discharge current I of the battery 100, and the temperature sensor 400, which measures the temperature T of the battery 100. From these inputs, the state estimation unit 552 sequentially estimates the SOC (charge rate) and SOH (capacity degradation rate) of the battery 100, Io, which is the current error offset value of the current sensor 300, and the gain of the resistor R0 in the battery equivalent circuit model. In addition, the error variance calculation unit 554 estimates the error variance when estimating the gain. The battery equivalent circuit model is also called a "cell equivalent circuit model."

The Kalman filter calculation unit 510 uses the above estimated values to calculate and output the SOC and SOH of the battery 100.

Figure 2 shows the configuration of the state estimation unit 552 of the Kalman filter calculation unit 510 in Figure 1.

As shown in FIG. 2, the state estimation unit has an SOC estimation unit 511, an SOH estimation unit 512, an Io estimation unit 513, an RoG estimation unit 514, a Q estimation unit 515, a CCV estimation unit 516, a Q Kalman gain estimation unit 521, an SOC Kalman gain estimation unit 522, an Io Kalman gain estimation unit 523, and an RoG Kalman gain estimation unit 524. Here, Q is the full charge capacity of the cell. RoG is the internal resistance in the equivalent circuit model of the cell.

The state estimation unit receives I, T, and V from each sensor. It then outputs SOC, SOH, Io, and RoG estimated using I, T, and V. A predetermined Kalman gain is estimated and used when making these estimates.

SOC, SOH, Io, and RoG are calculated sequentially using the following formulas (1) to (4).

Here, Qini is the capacity of the new cell. The unit is Ah. Gsoc is the Kalman gain of SOC, Gq is the Kalman gain of Q, Gio is the Kalman gain of Io, and Grog is the Kalman gain of RoG. These Kalman gains are calculated sequentially by the Kalman filter calculation unit 510 (Figure 1). t is the sensing period of voltage, current, and temperature, or the calculation period of SOC, SOH, Io, and RoG. For example, if every second, then t=1.

Figure 3 shows the battery equivalent circuit model used to calculate CCV.

In the battery equivalent circuit model recording unit 520 in FIG. 1, a map of the OCV, the resistance values R0, R1, and R2 constituting the battery equivalent circuit, and the capacitor time constants τ1 and τ2 are recorded in advance as parameters of the battery equivalent circuit for each SOC and T. Then, the battery parameters are estimated from the SOC, T, and RoG input to the CCV estimation unit 516, and the CCV is calculated from the input current (I) and current offset (Io).

The difference ΔCCV between the calculated CCV and the battery terminal voltage (V) measured by the voltage sensor is calculated, and the difference is multiplied by the Kalman gains (Gsoc, Gq, Gio, Grog) using the above formulas (1) to (4), and the SOC, Q, Io, and RoG are updated by adding the result to SOC, Q, Io, and RoG, respectively.

The calculation formula for the Kalman gain G calculated by the Kalman filter calculation unit 510 (Figure 1) is as shown in the following formulas (5) and (6).

In the formula, P^(k) is the prior error covariance, P(k) is the posterior error covariance, A and b are matrices indicating the state, C is a matrix indicating the observation value, σ _w ² is the observation noise covariance, and σ _v ² is the system noise covariance. Note that prior and posterior mean before and after using the "latest data", respectively.

The Kalman gain for each state is calculated sequentially together with the error covariance. In addition, the system noise covariance is calculated for each state estimate of SOC, Q, Io, and RoG by the system noise covariance calculation unit 530 in FIG. 1. The observation noise covariance is the covariance value when the battery terminal voltage is measured by the voltage sensor 200 (FIG. 1), and is the setting value used in the Kalman filter calculation unit 510.

In this way, the Kalman filter calculation unit 510 sequentially estimates SOC, SOH, Io, and RoG.

However, as the battery deteriorates, the parameters in the battery equivalent circuit model in Figure 3 fluctuate, decreasing the accuracy of estimating SOC and SOH.

To prevent this, the parameters of the calculation formula are set so that battery degradation is detected and the system noise covariance value of the system noise covariance calculation unit 530 in FIG. 1 is changed. In other words, the system noise covariance calculation unit 530 changes the system noise covariance value, which is the setting value used in the Kalman filter calculation unit 510, in response to battery degradation.

From the above equations (5) and (6), we can see that if the system noise covariance value changes, the Kalman gain G changes.

Specifically, the battery degradation detection unit 540 uses the SOH estimated by the Kalman filter calculation unit 510. The SOH is normally 100% for a new battery, and as the battery deteriorates, the value drops to 90% and 80%. For this reason, the system noise covariance value is changed when a predetermined SOH is reached. With regard to the system noise covariance value, the Kalman gain is increased by increasing the system noise covariance values of Io and RoG, which affect the battery equivalent circuit model in particular. The Kalman gain obtained in this way is used to correct (change) the parameters of the battery equivalent circuit model. This makes it possible to estimate the SOC and SOH with high accuracy even when the battery is deteriorated.

In summary, the deterioration of the battery 100 is represented by a value corresponding to a decrease in the capacity (or SOH) of the battery 100, an increase in the actual internal resistance of the battery 100, or the actual usage time of the battery 100. When any of these values changes by a predetermined amount with respect to its initial value, the system noise covariance calculation unit 530 increases the system noise covariance value.

The frequency of changing the system noise covariance value may be only once within a predetermined period, but from the viewpoint of improving the accuracy of the estimated value, the system covariance value may be changed stepwise multiple times as the SOH decreases. In addition, in this embodiment, an example is shown in which battery deterioration is detected using the SOH estimated by the Kalman filter calculation unit 510, but this is not limited to this, and battery deterioration may be detected by other methods. For example, the fluctuations in the charging and discharging current and the fluctuations in the voltage may be measured, the internal resistance of the battery may be calculated, and battery deterioration may be detected by an increase in the internal resistance. In addition, the actual usage time of the battery may be counted, and battery deterioration may be detected based on the usage time. Here, the usage time means the discharge time or charge time of the battery, or the sum of the discharge time and the charge time.

In summary, battery degradation is represented by a value corresponding to a decrease in the battery capacity, an increase in the actual internal resistance of the battery, or the actual usage time of the battery. When any of these values changes by a predetermined amount from its initial value, the system noise covariance calculation unit 530 increases the system noise covariance value.

Figure 4 is a flow diagram showing the battery state estimation method of this embodiment.

In this diagram, the Kalman filter calculation unit receives the measurement values of the voltage sensor, current sensor, and temperature sensor installed in the battery (step S11). Then, using these measurement values and the battery equivalent circuit model recorded in the battery equivalent circuit model recording unit, the Kalman filter calculation unit sequentially estimates the battery's SOC and SOH, the Io of the current sensor, and the battery's RoG set in the battery equivalent circuit model (step S12).

The system noise covariance calculation unit changes the system noise covariance value, which is the setting value used in the Kalman filter calculation unit, in response to battery deterioration (step S13).

(Embodiment 2)
In the first embodiment, the condition for changing the system noise covariance value is set according to the deterioration of the battery. In the second embodiment, however, the system noise covariance value is switched based on charging and discharging of the battery.

Figure 5 is a configuration diagram showing the battery state estimation device of this embodiment.

In this figure, the battery state estimation device 500 does not have the battery degradation detection unit 540 of FIG. 1. Instead, the system noise covariance calculation unit 530 is configured to change the system noise covariance value when used for estimation during charging of the battery 100 and when used for estimation during discharging. In other words, the system noise covariance calculation unit 530 is configured to set the system noise covariance value to a different value when used for estimation during charging of the battery 100 and when used for estimation during discharging.

In battery-powered home appliances, electric vehicles, and the like, the discharge current is usually greater than the charge current. Therefore, according to the above formula (1), the effect of the current offset Io during discharge is smaller. In addition, making the Kalman gain smaller also reduces the SOC error.

Specifically, the system noise covariance values of Io and RoG are made smaller during discharging compared to during charging. This makes it possible to estimate SOC and SOH with high accuracy.

100: Battery, 200: Voltage sensor, 300: Current sensor, 400: Temperature sensor, 500: Battery state estimation device, 510: Kalman filter calculation unit, 511: SOC estimation unit, 512: SOH estimation unit, 513: Io estimation unit, 514: RoG estimation unit, 515: Q estimation unit, 516: CCV estimation unit, 520: Battery equivalent circuit model recording unit, 521: Q Kalman gain estimation unit, 522: SOC Kalman gain estimation unit, 523: Io Kalman gain estimation unit, 524: RoG Kalman gain estimation unit, 530: System noise covariance calculation unit, 540: Battery deterioration detection unit, 552: State estimation unit, 554: Error variance calculation unit.

Claims (2)

A method for estimating a state of a battery using a device having a Kalman filter calculation unit, a battery equivalent circuit model recording unit, and a system noise covariance calculation unit, comprising:
a step of the Kalman filter calculation unit sequentially estimating the charging rate and capacity deterioration rate of the battery, the current offset of the current sensor, and the internal resistance of the battery set in the battery equivalent circuit model, using measurement values of a voltage sensor, a current sensor, and a temperature sensor installed in the battery, and a battery equivalent circuit model recorded in the battery equivalent circuit model recording unit;
The system noise covariance calculation unit changes a system noise covariance value, which is a setting value used in the Kalman filter calculation unit, between a case where the system noise covariance value is used for estimation during charging of the battery and a case where the system noise covariance value is used for estimation during discharging of the battery,
the system noise covariance value includes a system noise covariance value of the current offset and the internal resistance;
The method for estimating a battery state, wherein the system noise covariance value of the current offset and the internal resistance is set to be smaller during discharging than during charging. An apparatus for estimating a state of a battery, comprising a Kalman filter calculation unit, a battery equivalent circuit model recording unit, and a system noise covariance calculation unit,
the Kalman filter calculation unit sequentially estimates the charging rate and capacity deterioration rate of the battery, the current offset of the current sensor, and the internal resistance of the battery set in the battery equivalent circuit model, using measurement values of a voltage sensor, a current sensor, and a temperature sensor installed in the battery, and a battery equivalent circuit model recorded in the battery equivalent circuit model recording unit;
the system noise covariance calculation unit changes a system noise covariance value, which is a setting value used in the Kalman filter calculation unit, between a case where the system noise covariance value is used for estimation during charging of the battery and a case where the system noise covariance value is used for estimation during discharging of the battery;
the system noise covariance value includes a system noise covariance value of the current offset and the internal resistance;
The battery state estimation device, wherein the system noise covariance value of the current offset and the internal resistance is made smaller during discharging than during charging.

Images