JP7003844B2 - Battery system - Google Patents

Description

The present disclosure relates to a battery system comprising a nickel metal hydride battery.

Japanese Unexamined Patent Publication No. 2018-14210 (Patent Document 1) discloses a battery system including a nickel hydrogen battery. In this battery system, the internal resistance of the nickel-metal hydride battery is estimated using the current change amount and the voltage change amount in a predetermined period by using a current sensor and a voltage sensor (see Patent Document 1).

Japanese Unexamined Patent Publication No. 2018-14210

Generally, in a control system or the like, learning control may be used to improve the accuracy of calculated values. Also in the calculation of the internal resistance of the nickel-metal hydride battery, in order to improve the calculation accuracy, for example, as in Patent Document 1, it is conceivable to perform learning control using the estimated internal resistance of the nickel-metal hydride battery.

Here, in the learning control, the learning value converges to the true value (theoretical value) of the internal resistance by continuously performing the learning more than a certain number of times. In learning control, it is desirable to bring the learning value closer to the true value at an early stage.

However, in learning control, robustness and responsiveness are in a trade-off relationship. If the responsiveness is improved in order to bring the learning value closer to the true value at an early stage, the robustness will decrease. Therefore, when the internal resistance value with a large estimation error is used for learning, the learning value deviates from the true value. There is a possibility that it will be done. On the other hand, if the robustness is improved, the responsiveness is lowered, so that it may take time for the learning value to converge to the true value. Therefore, in learning control, it is desired not to decrease the robustness while improving the responsiveness.

The present disclosure has been made to solve the above problems, and an object thereof is to improve the calculation accuracy of internal resistance in a battery system including a nickel hydrogen battery.

In the battery system according to this disclosure, the internal resistance of the nickel hydrogen battery is estimated using the current change amount and the voltage change amount of the nickel hydrogen battery and the nickel hydrogen battery in a predetermined period, and learning is performed using the estimated internal resistance. It is equipped with a control device that calculates the internal resistance of the nickel-hydrogen battery. When the amount of change in current is large, the control device increases the ratio of reflecting the estimated internal resistance in learning as compared with the case where the amount of change in current is small.

For example, when the internal resistance is estimated from the current change amount and the voltage change amount of the nickel-metal hydride battery in a predetermined period by the IV plot method, the larger the current change amount, the better the estimation accuracy of the internal resistance. According to the above configuration, when the amount of change in current is large, the ratio of reflecting the estimated internal resistance in learning is larger than in the case where the amount of change in current is small. As a result, the higher the estimation accuracy (the larger the amount of current change), the larger the ratio reflected in the learning value. Therefore, since the learning value can be converged to the true value at an early stage, the calculation accuracy of the internal resistance can be improved.

According to the present disclosure, it is possible to improve the calculation accuracy of the internal resistance in a battery system including a nickel hydrogen battery.

It is a figure which shows schematic the whole structure of the vehicle which carried out the battery system which concerns on embodiment. It is a figure for demonstrating the method of estimating the internal resistance in the battery system which concerns on embodiment. It is a figure for demonstrating the 1st condition. It is a figure which shows the experimental result of the relationship between the current change amount and the estimated internal resistance. It is a figure which shows the relationship between the magnitude of a current change amount and a correction learning coefficient. FIG. 4 is an enlarged view of a portion where the amount of change in current in FIG. 4 is positive. It is a figure which showed schematic about the correction of a learning coefficient. It is a figure which compared the convergence speed of the learning value when the learning coefficient and the correction learning coefficient are applied. It is a flowchart which shows the procedure of the calculation process of the internal resistance of an assembled battery in a battery system.

Hereinafter, the present embodiment will be described in detail with reference to the drawings. The same or corresponding parts in the drawings are designated by the same reference numerals and the description thereof will not be repeated.

<Embodiment>

FIG. 1 is a diagram schematically showing an overall configuration of a vehicle 1 equipped with a battery system 5 according to the present embodiment. The vehicle 1 is an electric vehicle such as an electric vehicle, a hybrid vehicle, a plug-in hybrid vehicle, and a fuel cell vehicle. In the present embodiment, an example in which the vehicle 1 is an electric vehicle will be described.

The vehicle 1 includes an assembled battery 10, a system main relay 20, a power control unit (hereinafter, also referred to as “PCU (Power Control Unit)”) 40, a motor generator (MG) 50, and the like. It includes a drive wheel 60, an ECU (Electronic Control Unit) 100, and a monitoring unit 200. The battery system 5 includes an assembled battery 10, an ECU 100, and a monitoring unit 200.

The assembled battery 10 has a configuration in which a plurality of batteries are stacked. The battery is a nickel-metal hydride battery that can be charged and discharged. The monitoring unit 200 includes, for example, a voltage sensor 210, a current sensor 220, a temperature sensor 230, and the like. The voltage sensor 210 detects the voltage of the assembled battery 10 and outputs a signal VB indicating the detection result to the ECU 100. The current sensor 220 detects the current input / output to / from the assembled battery 10 and outputs a signal IB indicating the detection result to the ECU 100. When the battery current IB is a positive value, it indicates that the battery is discharged, and when it is a negative value, it indicates that the battery is charged. The temperature sensor 230 detects the temperature of the assembled battery 10 and outputs a signal TB indicating the detection result to the ECU 100. It should be noted that the voltage and temperature do not necessarily have to be monitored in units of assembled batteries, and for example, the voltage and temperature may be monitored in units of batteries. In this case, the voltage sensor 210 detects the voltage of each battery included in the assembled battery 10 and outputs a signal VB indicating the detection result to the ECU 100. The temperature sensor 230 detects the temperature of each battery included in the assembled battery 10 and outputs a signal TB indicating the detection result to the ECU 100.

One end of the system main relay 20 is electrically connected to the assembled battery 10, and the other end is electrically connected to the PCU 40. The open / closed state of the system main relay 20 is switched according to the control signal from the ECU 100. When the system main relay 20 is in the open state, the power supply from the assembled battery 10 to the PCU 40 is cut off. When the system main relay 20 is in the closed state, electric power can be supplied from the assembled battery 10 to the PCU 40.

The PCU 40 collectively shows a power conversion device for driving a motor generator 50 by receiving electric power from the assembled battery 10. For example, the PCU 40 includes an inverter for driving the motor generator 50, a converter that boosts the power output from the assembled battery 10 and supplies the power to the inverter.

The motor generator 50 is an AC rotary electric machine, for example, a permanent magnet type synchronous motor including a rotor in which a permanent magnet is embedded. The rotor of the motor generator 50 is mechanically connected to the drive wheels 60 via a power transmission gear (not shown). The motor generator 50 can generate electric power by the rotational force of the drive wheels 60 during the regenerative braking operation of the vehicle 1, and outputs the generated electric power to the PCU 40.

The ECU 100 includes a CPU (Central Processing Unit) 100a, a memory (more specifically, a ROM (Read Only Memory) and a RAM (Random Access Memory)) 100b, and an input / output port for inputting / outputting various signals (shown in the figure). It is composed of including). The ECU 100 controls each device based on signals from each sensor and device, a program stored in the memory 100b, and the like. It should be noted that various controls are not limited to processing by software, but can also be processed by dedicated hardware (electronic circuit).

The ECU 100 further includes a timer circuit 100c. The timer circuit 100c is configured to be capable of measuring a predetermined period (described later) set in advance.

(Estimation of internal resistance)

The ECU 100 estimates the internal resistance of the assembled battery 10 using the battery information (battery voltage VB, battery current IB, battery temperature TB) acquired from the monitoring unit 200. For example, the so-called IV plot method is used for estimating the internal resistance. Specifically, the ECU 100 acquires a set (IB, VB) of the battery current IB and the battery voltage VB from the monitoring unit 200 every unit time, and takes the current IB on the horizontal axis and the voltage VB on the vertical axis. Plot to the two-dimensional coordinates. Then, the slope of a straight line that approximates the plurality of plotted points can be calculated, and the slope of the calculated straight line can be used as the internal resistance of the assembled battery 10. In general, in the present embodiment, the internal resistance of the assembled battery 10 is estimated from the slope of a straight line obtained from the battery current IB and the battery voltage VB at two times. The estimation of the internal resistance of the assembled battery 10 in the present embodiment will be specifically described with reference to FIG.

FIG. 2 is a diagram for explaining a method of estimating the internal resistance in the battery system 5 according to the present embodiment. The horizontal axis of FIG. 2 shows the time t, and the vertical axis shows the battery current IB. FIG. 2 shows the time change of the battery current IB acquired for each unit time in a predetermined period.

The predetermined period is set to, for example, several hundred milliseconds or several seconds. The predetermined period may be set according to, for example, the battery temperature TB. In the present embodiment, an example in which the predetermined period is set to 1 second will be described as an example. The time from time t1 to time t11 shown in FIG. 2 corresponds to 1 second. In the present embodiment, the battery current acquired at time t1 is defined as the first current. Then, the battery current acquired at the time t2 after a unit time (for example, 0.1 second) from the time t1 is the second current i2, and the battery current acquired at the time t3 after the unit time from the time t2 is the third current. i3, the fourth current i4 to the eleventh current i11 are defined in the same manner. The first current i1 is the battery current at the start time (time t1) of the predetermined period, and the eleventh current i11 is the battery current at the end time (time t11) of the predetermined period. The voltage corresponding to the first current i1 is defined as "first voltage v1", and similarly, the voltage corresponding to the second current i2 to the eleventh current i11 is defined as the second voltage v2 to the eleventh voltage v11, respectively.

As shown in FIG. 2, battery information of 11 points (first current i1, first voltage v1) to (11th current i11, 11th voltage v11) is acquired in a predetermined period. In order to improve the estimation accuracy of the internal resistance, the ECU 100 is a battery acquired at time t1 when the battery information of the above 11 points satisfies the two conditions (first condition and second condition) described below. Using the information (i1, v1) and the battery information (i11, v11) acquired at time t11, the internal resistance of the assembled battery 10 is estimated from the amount of current change and the amount of voltage change between time t1 and time t11. do.

The first condition is a condition set to guarantee that there is a current change of a certain value or more when estimating the internal resistance of the assembled battery 10 by using the IV plot method. Specifically, the first condition is that the current change amount (hereinafter, also referred to as "first change amount") di1 between the first current i1 and the second current i2 is the first predetermined amount TH1 or more. The first predetermined amount TH1 is a value appropriately set according to the characteristics of the battery, the usage environment, and the like, and is set to, for example, about several tens of amperes. The first condition in the present embodiment is that the current change amount di1 between the first current i1 and the second current i2 is the first predetermined amount TH1 or more, but the present invention is not limited to this, and is predetermined. It suffices if there is a current change from the first current i1 to the first predetermined amount TH1 or more during the period. For example, when the amount of change in current between the first current i1 and the third current i3 is the first predetermined amount TH1 or more, the first condition may be satisfied.

FIG. 3 is a diagram for explaining the first condition. FIG. 3-1 is a diagram schematically showing the estimation of the internal resistance when the amount of change in current (di) is small. For example, as shown in the lower figure of FIG. 3-1 when the current change amount di between the current i1 and the current i11 is small, an approximate straight line is drawn at two adjacent points. In this case, for example, even if the error caused by the detection accuracy of the current sensor and / or the voltage sensor is small, the influence of the error on the slope of the approximate straight line becomes large. Therefore, there is a concern that the estimation accuracy of the internal resistance of the assembled battery 10 estimated from the slope of the approximate straight line will decrease.

FIG. 3-2 is a diagram schematically showing the estimation of the internal resistance when the amount of change in current (di) is large. For example, as shown in the lower figure of FIG. 3-2, when the current change amount di between the current i1 and the current i11 is large, the distance between the two points becomes large, so that the above error is the slope of the approximate straight line. The effect on is smaller than when the distance between two points is small. Therefore, by setting the distance between the two points to a certain value or more, it is possible to secure the estimation accuracy of the internal resistance of the assembled battery 10 estimated from the slope of the approximate straight line by the IV plot method.

Returning to FIG. 2, the second condition is a condition set to guarantee that the battery information acquired at time t11 is a value acquired in a stable state of the current input / output to / from the assembled battery 10. Is. Batteries are known to be affected by battery charge / discharge history (hysteresis). When the battery is charged and discharged, polarization occurs in the battery, but hysteresis occurs because the speed at which the polarization is generated and the speed at which the polarization is eliminated are different. When hysteresis occurs, it may not be possible to accurately acquire the voltage corresponding to the current acquired at a certain time. Therefore, in order to accurately estimate the internal resistance of the assembled battery 10, it is desirable to use the battery information acquired in a state where the influence of hysteresis is mitigated.

Therefore, the ECU 100 determines whether or not the influence of the hysteresis is mitigated by determining whether or not the second condition is satisfied. The second condition is that the fluctuation range di2 of the battery current after the current change from the first current i1 to the first change amount di1 or more is equal to or less than the second predetermined amount TH2. In the present embodiment, the fluctuation range of the battery current from the second current i2 to the eleventh current i11 corresponds to the fluctuation range di2. If the fluctuation range di2 of the battery current from time t2 to time t11 is within the second predetermined amount TH2, the current fluctuation range is small, so that the influence of hysteresis has little influence on the estimation accuracy of the internal resistance. It can be said that it is (relaxed). That is, when the second condition is satisfied, the battery information acquired at the time t11 (end time) used for estimating the internal resistance of the assembled battery 10 is the information acquired in a stable state of the assembled battery 10. Is guaranteed. The second predetermined amount TH2 is set to a value smaller than the first predetermined amount TH1. The second predetermined amount TH2 is set to, for example, about several amperes.

In the present embodiment, it is guaranteed that the current change amount di between the first current i1 and the eleventh current i11 becomes a certain value or more by satisfying the first condition and the second condition.

(Learning of internal resistance)

As described above, the estimation accuracy of the internal resistance is improved. However, the internal resistance estimated by one estimation process may cause an estimation error due to, for example, the signal line of each sensor being affected by noise. Therefore, it is conceivable to calculate the current internal resistance by learning using the estimated internal resistance instead of using the internal resistance estimated by one estimation process as the current internal resistance. This makes it possible to improve the accuracy of calculating the internal resistance. For learning, for example, the following equation (1) using the learning coefficient τ is used.

This time learning value = last learning value × ((τ-1) / τ) + this time estimated value × (1 / τ)… (1)

The ECU 100 reads the learning value (previous learning value) learned up to the previous time from the memory 100b. The ECU 100 multiplies the estimated internal resistance (estimated value this time) by a predetermined fixed ratio (1 / τ) and adds it to the previously learned value. The ECU 100 uses the current learning value calculated by the equation (1) as the current internal resistance. Then, the ECU 100 stores the calculated learning value this time in the memory 100b. The learning coefficient τ is a coefficient for expressing the ratio of reflecting the estimated internal resistance in the learning value this time, and is set to an appropriate value for the system to which the learning control is applied. The smaller the learning coefficient τ, the higher the ratio of reflecting the estimated internal resistance in the learning value this time. In the following, "1 / τ" multiplied by the estimated value this time is also referred to as "gain".

In this way, the internal resistance estimated by one estimation process is not used as the internal resistance of the battery at that time, but the internal resistance is calculated by learning the estimated internal resistance. As a result, even if an error in estimating the internal resistance occurs, the value of the internal resistance used for battery control does not fluctuate significantly, and so-called robustness is improved.

Here, in learning control, it is a well-known fact that the improvement of robustness and the improvement of responsiveness have a contradictory relationship. Robustness is improved by reducing the gain (1 / τ) in the equation (1), that is, increasing the learning coefficient τ. On the other hand, the responsiveness is improved by increasing the gain (1 / τ), that is, decreasing the learning coefficient τ. In learning control, it is desirable to appropriately adjust the balance between the improvement of robustness and the improvement of responsiveness.

Specifically, when the robustness is improved, the ratio of the estimated internal resistance reflected in the learning value becomes small. Therefore, for example, even if an internal resistance estimation error occurs, it is given to the learning value. The impact can be reduced. When the responsiveness is improved, the ratio of the estimated internal resistance reflected in the learning value increases. Therefore, for example, when there is a change in the internal resistance, the internal resistance after the learning value changes early (true). Value) converges. In learning control, it is desired not to reduce robustness while improving responsiveness.

(About correction of learning coefficient)

Therefore, the ECU 100 according to the present embodiment makes the learning coefficient different depending on the estimated accuracy of the estimated internal resistance. Specifically, the learning coefficient is corrected so that the learning coefficient τ becomes smaller as the magnitude of the estimated current change amount di of the internal resistance becomes larger. As a result, the higher the estimation accuracy of the internal resistance, the greater the reflection in the learning value. In the following, the corrected learning coefficient τa is also referred to as “corrected learning coefficient τa”. As described in FIG. 3, the larger the magnitude of the current change amount di, the higher the estimation accuracy of the internal resistance.

FIG. 4 is a diagram showing the experimental results of the relationship between the current change amount di and the estimated internal resistance. In FIG. 4, the horizontal axis shows the current change amount di, and the vertical axis shows the internal resistance. In FIG. 4, the internal resistance estimated when the current change amount di is variously changed is plotted for a certain battery. More specifically, the internal resistance when the current change amount di is variously changed is estimated by the IV plot method, and the estimated internal resistance is plotted.

As shown in FIG. 4, it can be seen that the larger the magnitude (absolute value) of the current change amount di, the smaller the variation in the estimated internal resistance, and the better the estimation accuracy of the internal resistance. Specifically, as the magnitude of the current change amount di becomes larger, the estimated internal resistance shows a value near the resistance Rt.

The ECU 100 according to the present embodiment corrects the gain (1 / τ) from the default value in the range of 1 to 2 times according to the estimation accuracy of the internal resistance (the magnitude of the current change amount di). That is, the ECU 100 corrects the learning coefficient τ from the default value in the range of 1/2 to 1 times (τ / 2 ≦ τa) according to the estimation accuracy of the internal resistance (the magnitude of the current change amount di). ≤τ). The degree of correction is not limited to the above, and can be appropriately set according to the system to which the present disclosure is applied.

FIG. 5 is a diagram showing the relationship between the magnitude of the current change amount di and the correction learning coefficient τa. In FIG. 5, the horizontal axis represents the magnitude of the current change amount di | di |, and the vertical axis represents the correction learning coefficient τa. The magnitude | di | of the current change amount di changes within the range of the minimum value Adimin and the maximum value Adimax of the predetermined current change amount di magnitude. Details of the minimum value Adimin and the maximum value Adimax will be described later. The correction learning coefficient τa changes in the range of τ / 2 to τ according to the magnitude | di | of the current change amount di.

As shown in FIG. 5, the correction learning coefficient τa according to the present embodiment is set so as to change linearly in proportion to the magnitude | di | of the current change amount di. As the magnitude | di | of the current change amount di increases, the correction learning coefficient τa decreases, and the ratio of the estimated internal resistance reflected in the learning increases.

FIG. 6 is an enlarged view of a portion where the current change amount di in FIG. 4 is positive. Adimin and Adimax shown in FIG. 6 indicate the minimum value and the maximum value of the predetermined magnitude (absolute value) of the current change amount di, respectively. Adimin can be the same as the magnitude of the first predetermined amount TH1 under the first condition. As a result, the current change amount di used for estimating the internal resistance becomes equal to or higher than a certain level, and the estimation accuracy of the internal resistance is ensured. Adimax is appropriately set according to the characteristics of the battery, the usage environment, and the like. By setting Adimax, the value when the current change amount di suddenly increases beyond the expected standard, for example, when a clear false detection of the current sensor occurs is internal. Excluded from resistance estimation. The relationship between the current change amount di and the minimum value Adimin and the maximum value Adimax of the current change amount is expressed by the following equation (2).

Adimin ≤ | di | ≤ Adimax ... (2)

Equation (3) can be obtained by modifying equation (2).

0 ≤ (| di | -Adimin) / (Adimax-Adimin) ≤ 1 ... (3)

Then, the equation (4) can be obtained by adding the variable k to each side of the equation (3).

k≤((| di | -Adimin) / (Adimax-Adimin)) + k≤k + 1 ... (4)

Then, using the equation (4), the equation (5) showing the relationship between the learning coefficient τ and the correction learning coefficient τa is expressed.

τa = τ / ((| di | -Adimin) / (Adimax-Adimin) + k) ... (5)

Since the correction learning coefficient τa according to the present embodiment takes the range of τ / 2 ≦ τa ≦ τ, the variable k is set to 1 (k = 1). When the equation (5) is modified with k = 1, it is expressed as the equation (6).

τa = τ / ((| di | -Adimin) / (Adimax-Adimin) +1) ... (6)

The value of the equation (6) is determined by the current change amount di. The correction learning coefficient τa is determined in proportion to the current change amount di, and changes linearly in the range of τ / 2 ≦ τa ≦ τ. Specifically, when the magnitude of the current change amount di is the maximum value Adimax (| di | = Adimax), the value of the denominator is 2, so τa = τ / 2. When the magnitude of the current change amount di is the minimum value Adimin (| di | = Adimin), the value of the denominator is 1, so τa = τ.

The internal resistance is calculated by learning using the estimated internal resistance by the equation (7) using the correction learning coefficient τa calculated in this way.

This time learning value = last learning value × ((τa-1) / τa) + this time estimated value × (1 / τa)… (7)

The above can be summarized as FIG. 7. FIG. 7 is a diagram schematically showing the correction of the learning coefficient τ. The horizontal axis of FIG. 7 shows the magnitude of the current change amount di | di |, and the vertical axis shows τb (τb = 1 / τa), which is the reciprocal of the correction learning coefficient τa, that is, the gain. There is. By transforming equation (7) using the reciprocal τb of the correction learning coefficient τa, it can be expressed by equation (8).

This time learning value = last learning value × (1-τb) + this time estimated value × τb ... (8)

First, when the magnitude | di | of the current change amount di is less than the minimum value Adimin, the learning coefficient is not substantially corrected in the present embodiment (τa = τ). That is, the learning coefficient τ will be used. In reality, the first condition is not satisfied, so the internal resistance is not estimated.

When the magnitude | di | of the current change amount di is not less than the minimum value Adimin and not more than the maximum value Adimax, the reciprocal τb of the correction learning coefficient τa is determined according to the magnitude of the current change amount di. To. That is, in this case, the ratio of the estimated internal resistance reflected in the learning increases in proportion to the magnitude | di | of the current change amount di.

FIG. 8 is a diagram comparing the convergence speeds of the learning values when the learning coefficient τ and the correction learning coefficient τa are applied. The horizontal axis of FIG. 8 shows the number of learnings, and the vertical axis shows the internal resistance. The dotted line L1 shows the learning value when the learning coefficient τ is applied, and the solid line L2 shows the learning value when the correction learning coefficient τa is applied.

In both the dotted line L1 and the solid line L2, the learning value converges to the resistance Rt (true value) as the number of learning times increases. Specifically, when the number of learnings is N3, the learning values of both the dotted line L1 and the solid line L2 converge to the resistance Rt. On the other hand, when the number of learnings is N1, N2, or N1 or less, the solid line L2 shows a value closer to the resistance Rt in any case. In this way, when the correction learning coefficient τa is applied, the learning value can be converged to the resistance Rt earlier than when the learning coefficient τ is applied. Further, even after the learning value converges to the resistance Rt, the ratio of the internal resistance having high estimation accuracy reflected in the learning value increases, so that the responsiveness of the learning value when the internal resistance changes can be improved. can. As described above, by correcting the learning coefficient τ, it is possible to improve the responsiveness in the learning control but not to reduce the robustness. Thereby, when the internal resistance is calculated by learning using the estimated internal resistance, the calculation accuracy of the internal resistance can be improved.

(Procedure for calculating internal resistance)

FIG. 9 is a flowchart showing a procedure for calculating the internal resistance of the assembled battery 10 in the battery system 5. Each step shown in this flowchart is repeatedly executed by the ECU 100 when the vehicle 1 is in operation. Each step of the flowchart shown in FIG. 9 describes a case where it is realized by software processing by the ECU 100, but a part or all of the steps may be realized by hardware (electric circuit) manufactured in the ECU 100.

The ECU 100 is initialized by substituting 1 for the variable n which is an index (step 100, hereinafter the step is abbreviated as "S"). n is a natural number, and takes a value of 1 to 11 when battery information is acquired at each of the times t1 to t11 in a predetermined period, for example. The ECU 100 determines the value assigned to the variable n (S105).

When the variable n is 1 (n = 1 in S105), the ECU 100 acquires the battery current and the battery voltage at the start time t1 of the predetermined period from the monitoring unit 200 as the first current i1 and the first voltage v1, respectively, and stores the memory. Store in 100b (S110).

The ECU 100 determines whether or not the variable n is N (S130). In the example of the present embodiment shown in FIG. 2, the value of N is 11. Since the value of the variable n is 1 (n ≠ N), the ECU 100 selects NO in S130 and advances the process to S135. In S135, the ECU 100 sets the variable n to 2 and returns the process to S105.

When n is 2 (n = 2 in S105), the ECU 100 acquires the battery current and the battery voltage at time t2 from the monitoring unit 200 as the second current i2 and the second voltage v2, respectively, and stores them in the memory 100b ( S115).

The ECU 100 determines whether or not the first condition is satisfied in S120. Specifically, the ECU 100 determines whether or not the first change amount di1 which is the magnitude of the difference between the first current i1 and the second current i2 is equal to or more than the first predetermined amount TH1. When the first change amount di1 is smaller than the first predetermined amount TH1 (di1 <TH1) (NO in S120), the ECU 100 can secure a certain level of accuracy or higher in estimating the internal resistance of the assembled battery 10 by the IV plot method. It is determined that the current has not changed to a certain extent, and the process is terminated without estimating the internal resistance of the assembled battery 10.

On the other hand, when the first change amount di1 is the first predetermined amount TH1 or more (di1 ≧ TH1) (YES in S120), the ECU 100 advances the process to S130. Then, since the value of the variable n is 2 (n ≠ N), the ECU 100 advances the process to S135, sets the variable n to 3, and returns the process to S105.

When the variable n is 3 or more (n ≧ 3 in S105), the ECU 100 acquires the battery current and the battery voltage at the time tun from the monitoring unit 200 as the nth current in and the nth voltage vn, respectively, and stores them in the memory 100b. (S125). The ECU 100 repeatedly executes this process until the variable n becomes N.

When the variable n reaches N in S130 (YES in S130), the ECU 100 determines whether or not the second condition is satisfied (S140). Specifically, the ECU 100 determines whether or not the fluctuation range di2 of the battery current from the second current i2 to the eleventh current i11 is equal to or less than the second predetermined amount TH2 (di2 ≦ TH2).

In the ECU 100, when the fluctuation width di2 is larger than the second predetermined amount TH2 (di2> TH2) (NO in S140), the battery information (11th current i11, 11th voltage v11) acquired at time t11 is the assembled battery 10. It is determined that the current input / output to / from is not acquired in a stable state, and the process is terminated without estimating the internal resistance of the assembled battery 10.

On the other hand, when the fluctuation width di2 is the second predetermined amount TH2 or less (di2≤TH2) (YES in S140), the ECU 100 calculates the magnitude | di | of the current change amount in the predetermined period and (| di | = | I1-i11 |), the internal resistance of the assembled battery 10 is estimated (S145). The ECU 100 uses the battery information (first current i1, first voltage v1) acquired at time t1 and the battery information (11th current i11, 11th voltage v11) acquired at time t11 to time t1. The internal resistance of the assembled battery 10 is estimated from the amount of change in current and the amount of change in voltage from to time t11.

The ECU 100 determines whether or not the magnitude | di | of the amount of change in current is at least the minimum value Adimin and at least the maximum value Adimax (S147). When the magnitude | di | of the amount of change in the current is smaller than the minimum value Admin or larger than the maximum value Adimax (NO in S147), the ECU 100 advances the process to S155 without correcting the learning coefficient τ. When the magnitude | di | of the amount of change in the current is smaller than the minimum value Admin or larger than the maximum value Adimax, the ECU 100 ends the process without reflecting the estimated internal resistance in the learning. You may.

When the magnitude | di | of the current change amount is equal to or greater than the minimum value Adimin and is equal to or less than the maximum value Adimax (YES in S147), the ECU 100 corrects according to the magnitude | di | of the current change amount. The learning coefficient τa is calculated (S150). Specifically, the ECU 100 calculates the correction learning coefficient τa by using the magnitude | di | of the amount of current change calculated in S145 and the equation (6).

The ECU 100 calculates the internal resistance by learning using the correction learning coefficient τa calculated in S150 and the internal resistance estimated in S145 (S155). Specifically, the ECU 100 calculates the current learning value by using the internal resistance estimated in S145 and the equation (7), and uses the calculated current learning value as the current internal resistance.

Then, the ECU 100 stores the current learning value calculated in S155 in the memory 100b of the ECU 100 (S160). The current learning value stored in the memory 100b will be used as the previous learning value when the flowchart is executed next time.

As described above, the battery system 5 according to the present embodiment calculates the internal resistance by learning using the internal resistance estimated from the current change amount di and the voltage change amount in a predetermined period. The internal resistance estimated when the current change amount di is large is reflected in the learning value at a larger ratio than the internal resistance estimated when the current change amount di is small. As a result, the higher the estimation accuracy (the larger the current change amount di), the larger the ratio reflected in the learning value. Therefore, the learning value can be converged to the true value at an early stage. In addition, even after the learning value has converged to the true value, the ratio of the internal resistance with high estimation accuracy reflected in the learning value increases, so that the responsiveness of the learning value when the internal resistance changes can be improved. can. By correcting the learning coefficient τ, it is possible to improve the responsiveness in the learning control but not to reduce the robustness. Thereby, when the internal resistance is calculated by learning using the estimated internal resistance, the calculation accuracy of the internal resistance can be improved.

<Modification example>

In the embodiment, the correction learning coefficient τa was calculated by using the equation (6). However, the method for calculating the correction learning coefficient τa is not limited to using the equation (6). For example, the correction coefficient Y for calculating the correction learning coefficient τa may be predetermined according to the magnitude | di | of the amount of change in the current. Specifically, the correction learning coefficient τa is calculated by the following equation (9).

τa = Y × τ… (9)

The correction coefficient Y is calculated using a correction coefficient map that defines the relationship between the magnitude | di | of the amount of change in current and the correction coefficient Y. The correction coefficient map is obtained in advance by an experiment or the like and stored in the memory 100b of the ECU 100.

Even if a correction coefficient map is used instead of the equation (6), the same effect as that of the embodiment can be obtained.

The embodiments disclosed this time should be considered to be exemplary and not restrictive in all respects. The scope of the present disclosure is set forth by the claims rather than the description of the embodiments described above, and is intended to include all modifications within the meaning and scope of the claims.

1 vehicle, 5 battery system, 10 sets of batteries, 20 system main relay, 50 motor generator, 60 drive wheels, 100 ECU, 100a CPU, 100b memory, 100c timer circuit, 200 monitoring unit, 210 voltage sensor, 220 current sensor, 230 Temperature sensor, VB battery voltage, di1 first change amount, di2 fluctuation range, IB battery current, TB battery temperature.

Claims (1)

Nickel-metal hydride batteries and
The internal resistance of the nickel-metal hydride battery is estimated using the current change amount and the voltage change amount of the nickel-metal hydride battery in a predetermined period, and the internal resistance of the nickel-metal hydride battery is calculated by learning using the estimated internal resistance. Equipped with a control device to
The control device is a battery system in which when the amount of change in current is large, the ratio of reflecting the estimated internal resistance in learning is larger than in the case where the amount of change in current is small.

Images