JP7415654B2 - battery monitoring system - Google Patents

Description

The present invention relates to a battery monitoring system.

Conventionally, various systems for early detection of abnormalities in various devices are known. For example, Patent Document 1 describes a normal model that models the normal state of the device using multiple types of sensor data acquired from a gas engine device and mutual correlation coefficients calculated from the multiple types of sensor data. A configuration is disclosed in which a failure is diagnosed when observed data deviates from a normal model by a predetermined degree after the model is created. In such a configuration, by modeling a normal state using a correlation coefficient based on sensor data and diagnosing it by comparing it with observed data, it is easier to diagnose a failure than by simply comparing observed data with a predetermined threshold. The aim is to improve the diagnostic accuracy of

JP 2017-10263 Publication

The inventors of the present application have discovered that the normal state of secondary batteries may fluctuate depending on how the secondary battery is used, and sensor data acquired from the secondary battery may fluctuate in short cycles such as seasonal fluctuations. I discovered that. On the other hand, in the configuration disclosed in Patent Document 1, if the normal state of the device to be diagnosed cannot be uniquely determined, the accuracy of failure diagnosis decreases unless the normal model is constantly updated. If the normal state is uniquely defined, observed data will tend to deviate from the normal model, increasing the risk of erroneously diagnosing the normal state as a failure. Therefore, when the configuration disclosed in Patent Document 1 is applied to a secondary battery, there is a further concern that the accuracy of failure diagnosis will decrease.

The present invention has been made in view of the above problems, and it is an object of the present invention to provide a battery monitoring system that improves the accuracy of failure diagnosis of secondary batteries.

One aspect of the present invention is a battery monitoring system (1) that monitors the state of a secondary battery (2), comprising:
a data acquisition unit (10) that acquires multiple types of monitoring data for monitoring the secondary battery;
a calculation unit (20) that calculates a feature amount using the monitoring data;
an extraction unit (40) that extracts the trend component from a time-series variation including a trend component and a seasonal variation component or an irregular variation component as a systematic variation of the feature amount based on a change in the feature amount over a predetermined period; ,
a failure determination unit (50) that determines that the secondary battery has failed when the systematic fluctuation deviates from a predetermined reference range (R) ;
The calculation unit is in a battery monitoring system that calculates a partial correlation coefficient from the components of an inverse covariance matrix in which the monitoring data is used as a variable as the feature quantity .

The inventors of the present application have determined that there is a gap between the origin of systematic fluctuations extracted from changes over a predetermined period of feature values calculated using multiple types of monitoring data obtained from secondary batteries and the initial failure point of the secondary battery. We have newly discovered that there is a correlation. Based on this knowledge, the battery monitoring system extracts systematic fluctuations based on changes over a predetermined period of feature values calculated using multiple types of monitoring data acquired from secondary batteries, and If the absolute value of the fluctuation deviates from a predetermined reference range, it is determined that the secondary battery has failed. In this way, by monitoring the state of the secondary battery based on the systematic fluctuations, the influence of how the secondary battery is used and short-term fluctuations can be suppressed, and the accuracy of failure determination of the secondary battery can be improved. can be improved.

As described above, according to the present invention, it is possible to provide a battery monitoring system that improves the accuracy of fault diagnosis of secondary batteries.

Note that the numerals in parentheses described in the claims and means for solving the problem indicate correspondence with specific means described in the embodiments described later, and do not limit the technical scope of the present invention. It's not a thing.

1 is a block diagram showing the configuration of a battery monitoring system in Embodiment 1. FIG. 1 is a conceptual diagram showing the configuration of a vehicle equipped with a battery monitoring system in Embodiment 1. FIG. 2 is a flow diagram showing a control mode of the battery monitoring system in Embodiment 1. FIG. FIG. 3 is a conceptual diagram showing a plurality of pieces of monitoring data acquired in the battery monitoring system in Embodiment 1. FIG. FIG. 3 is a conceptual diagram showing feature amounts calculated in the battery monitoring system in Embodiment 1. FIG. FIG. 2 is a conceptual diagram of (a) a trend component, (b) a seasonal variation component, and (c) an irregular variation component calculated in the battery monitoring system in Embodiment 1. FIG. FIG. 3 is a conceptual diagram showing a time change in battery performance of a secondary battery in Embodiment 1. FIG. FIG. 3 is a block diagram showing the configuration of a battery monitoring system in Embodiment 2. FIG. 7 is a flow diagram showing a control mode of the battery monitoring system in Embodiment 2. FIG. FIG. 7 is a conceptual diagram showing monitoring data acquired in the battery monitoring system in Embodiment 2. FIG. FIG. 7 is a conceptual diagram of trend components calculated in the battery monitoring system in Embodiment 2. FIG. 7 is a block diagram showing the configuration of a battery monitoring system in Embodiment 3. FIG. 7 is a flowchart showing a control mode of the battery monitoring system in Embodiment 3. FIG. 7 is a conceptual diagram showing a plurality of pieces of monitoring data acquired in the battery monitoring system in Embodiment 3. FIG. FIG. 7 is a conceptual diagram of trend components calculated in the battery monitoring system in Embodiment 3. FIG. 3 is a block diagram showing the configuration of a battery monitoring system in Embodiment 4. FIG. 7 is a flowchart showing a control mode of the battery monitoring system in Embodiment 4. FIG. FIG. 7 is a conceptual diagram showing the configuration of a battery monitoring system in Embodiment 5. FIG. 3 is a conceptual diagram showing the configuration of a battery monitoring system in modification 1. FIG. FIG. 7 is a conceptual diagram showing the configuration of a battery monitoring system in modification 2. FIG.

(Embodiment 1)
An embodiment of the battery monitoring system described above will be described using FIGS. 1 to 7.
A battery monitoring system 1 according to the present embodiment shown in FIG. 1 monitors the state of a secondary battery 2.
As shown in FIG. 1, it includes a data acquisition section 10, a calculation section 20, an extraction section 40, and a failure determination section 50.
The data acquisition unit 10 acquires multiple types of monitoring data for monitoring the secondary battery 2.
The calculation unit 20 calculates feature amounts using the monitoring data.
The extraction unit 40 extracts systematic fluctuations in the feature amount based on changes in the feature amount calculated by the calculation unit 20 over a predetermined period.
The failure determination unit 50 determines that the secondary battery 2 has failed when the systematic fluctuation deviates from a predetermined reference range.

Hereinafter, the battery monitoring system 1 of this embodiment will be explained in detail.
As shown in FIG. 2, the battery monitoring system 1 of this embodiment monitors a secondary battery 2 that constitutes a power source mounted on a vehicle 100 such as an electric vehicle or a hybrid vehicle. In this embodiment, as shown in FIG. 2, the secondary battery 2 is an assembled battery formed by combining a plurality of battery cells 21, 22, and 23. Note that the secondary battery 2 may be a single battery cell.

The data acquisition unit 10 shown in FIG. 1 acquires multiple types of monitoring data on the secondary battery 2. Examples of the monitoring data include closed circuit voltage, charging/discharging current, SOC (State of Charge), battery capacity, battery temperature, environmental temperature around the battery, cumulative charging time, cumulative discharging time, and cumulative current. Furthermore, examples of monitoring data for the assembled battery include the total closed circuit voltage of the assembled battery, the maximum and/or minimum closed circuit voltage of the secondary battery in the assembled battery, battery temperature, SOC, and the like. The data acquisition unit 10 can be a sensor that can detect the monitoring data to be acquired, such as a voltage sensor, a current sensor, a temperature sensor, etc., and the SOC is calculated based on these values. and can be obtained. In this embodiment, as shown in FIG. 2, a data acquisition unit 10 is configured by a voltage sensor 101, a current sensor 102, and a temperature sensor 103 provided in a BMU (battery management unit) mounted on a vehicle 100. .

The storage unit 11 shown in FIG. 1 is made of a rewritable nonvolatile memory, and stores multiple types of monitoring data acquired by the data acquisition unit 10. In this embodiment, the storage unit 11 is included in the BMU.

The calculation unit 20 shown in FIG. 1 includes a processor capable of executing a predetermined program for calculating feature amounts. In this embodiment, the calculation unit 20 is provided outside the vehicle 100. Note that the calculation unit 20 may be configured by causing a processor installed in a BMU provided in the vehicle 100 to execute the program. The calculation unit 20 calculates the feature amount of the secondary battery 2 using multiple types of monitoring data stored in the storage unit 11. The feature amount can be, for example, a correlation coefficient between variables of the monitoring data, and in this embodiment, a partial correlation coefficient is calculated from the components of the inverse covariance matrix of the monitoring data. The inverse covariance matrix can be calculated by the calculation unit 20 performing sparse regularization using multiple types of monitoring data as variables.

As a method for calculating feature amounts by the calculation unit 20 in this embodiment, a case will be described in which the data acquisition unit 10 acquires six types of monitoring data X1 to X6 as shown in FIG. 4. Note that the number of types of monitoring data to be acquired may be two or more, and is not limited to six types. As shown in FIG. 4, each of the monitoring data X1 to X6 changes over time. The calculation unit 20 first performs sparse regularization using the monitoring data X1 to X6 acquired in the first period P1 as variables, and calculates a partial correlation coefficient matrix Λ as a covariance inverse matrix. The partial correlation coefficient matrix Λ can be expressed, for example, as follows.

In the above equation, the component λnm (n, m=1 to 6) of the partial correlation coefficient matrix Λ indicates the partial correlation coefficient between the monitoring data Xn and the monitoring data Xm. That is, λ12 means a partial correlation coefficient between the monitoring data X1 and the monitoring data X2. Furthermore, since the partial correlation coefficient matrix Λ is a symmetric matrix, some of the partial correlation coefficients λ are omitted from the above equation. Furthermore, since all the components on the main diagonal of the partial correlation coefficient matrix Λ are 1, their description is omitted. Then, the calculation unit 20 calculates the component λnm (n, m=1 to 6) of the partial correlation coefficient matrix Λ as a feature quantity. The partial correlation coefficient λnm, which is a feature amount, approaches 1 or -1 when the correlation between two types of monitoring data is high. Moreover, when the correlation is low, it approaches 0. As shown in FIG. 5, the partial correlation coefficient λnm, which is a feature amount, changes over time.

In this embodiment, the feature amount calculated by the calculation unit 20 is stored in the feature amount storage unit 30 shown in FIG. The feature amount storage unit 30 is composed of a rewritable nonvolatile memory. In this embodiment, the feature storage unit 30 is provided outside the vehicle 100. Note that the BMU provided in the vehicle 100 may include the feature storage unit 30.

The extraction unit 40 shown in FIG. 1 includes a processor capable of executing a predetermined program for extracting systematic variations in feature amounts. In this embodiment, the extraction unit 40 is provided outside the vehicle 100. Note that the extraction unit 40 may be configured by causing a processor installed in a BMU provided in the vehicle 100 to execute the program. In this embodiment, the extraction unit 40 extracts a trend component as a systematic variation of the feature amount by smoothing processing of factor decomposition in time series analysis. Factor analysis in time series analysis can be performed using a known method. Furthermore, as the smoothing process, a moving average that calculates the average of the most recent feature amounts, or Kalman smoothing that extracts past systematic fluctuations based on the feature amounts up to the present may be adopted. Note that the period of the feature used to extract systematic variations is not particularly limited.

In this embodiment, for example, as a feature quantity, a partial correlation coefficient λnm, which is a specific component in the partial correlation coefficient matrix Λ, can be shown as shown in FIG. Then, the extraction unit 40 performs factor decomposition in time series analysis on the partial correlation coefficient λnm, thereby generating the trend component, seasonal variation component, and irregular variation component shown in FIGS. 6(a) to 6(c). The trend components are extracted as systematic fluctuations. The trend component shows long-period fluctuations, the seasonal variation component shows short-period fluctuations, and the irregular fluctuation component shows random fluctuations.

Regarding the period of each fluctuation, for example, when monitoring seasonal changes in one year, the fluctuation component with a period of one year is considered a short-period fluctuation, and the fluctuation component with a period longer than one year is considered a long-period fluctuation. , and a fluctuation component with a period shorter than one year can be defined as a random fluctuation. In addition, when monitoring changes in usage over a week, the fluctuation components with a period of one week are considered short-period fluctuations, and the fluctuation components with a period longer than one week are considered long-period fluctuations. , a fluctuation component with a period shorter than one week can be defined as a random fluctuation. As described above, the period of each fluctuation can be defined based on the target of change.

Abnormalities in the secondary battery 2 appear as long-period fluctuations in the partial correlation coefficient, and conditions that have a low correlation with the abnormality in the secondary battery 2, such as temperature changes and usage of the secondary battery 2, This appears in seasonal fluctuation components and irregular fluctuation components in relational numbers. Therefore, by extracting a trend component from the partial correlation coefficient as a systematic variation, it is possible to make a long-period variation that reflects the abnormality of the secondary battery 2 visible.

The failure determination unit 50 shown in FIG. 1 includes a processor capable of executing a predetermined program for determining whether or not the secondary battery 2 has failed. In this embodiment, the failure determination unit 50 is provided outside the vehicle 100. Note that the failure determination unit 50 may be configured by causing a processor installed in a BMU provided in the vehicle 100 to execute the program. The failure determination unit 50 compares the systematic variation with a predetermined reference range, and when the comparison result indicates that the systematic variation deviates from the predetermined reference range R, the failure determination unit 50 determines whether the secondary battery 2 It is determined that there is a failure.

Next, the control flow of the battery monitoring system 1 will be explained using FIG. 3.
First, in step S1 shown in FIG. 3, the data acquisition unit 10 acquires multiple types of monitoring data X1 to X6 shown in FIG. 4 from the secondary battery 2. The acquired monitoring data X1 to X6 are stored in the storage unit 11 shown in FIGS. 1 and 2.

After step S1, in step S2 shown in FIG. 3, the calculation unit 20 calculates a feature amount using the monitoring data. In this embodiment, the calculation unit 20 performs sparse regularization using multiple types of monitoring data as variables to calculate an inverse covariance matrix, and calculates a partial correlation coefficient as a feature quantity from the components of the inverse covariance matrix. . In this embodiment, feature amounts are calculated over the first period P1 shown in FIG. 4. Then, in step S3, the feature amount for the certain period is stored in the feature amount storage section 30 shown in FIGS. 1 and 2.

After step S3, in step S4 shown in FIG. 3, the extracting unit 40 extracts systematic fluctuations from the feature quantities stored in the feature quantity storage unit 30. In this embodiment, the extraction unit 40 performs a smoothing process using factor decomposition in time series analysis on the partial correlation coefficient as a feature stored in the feature storage 30. ) to (c) are separated into trend components, seasonal fluctuation components, and irregular fluctuation components. Thereby, the extraction unit 40 extracts a trend component as a systematic variation shown in FIG. 6(a) from changes in the feature amount over a predetermined period.

Next, in step S5 shown in FIG. 3, the failure determination unit 50 determines whether the systematic fluctuation has deviated from a predetermined reference range R. In this embodiment, as shown in FIG. 6(a), the reference range R is set in advance when designing the secondary battery 2 based on the range of systematic variation in bench evaluation. If the failure determination unit 50 determines that the systematic fluctuation deviates from the reference range R, the process proceeds to step S5: Yes, and in step S6, the failure determination unit 50 determines that the secondary battery 2 is malfunctioning. It is determined that this is the case, and this control flow is ended. For example, as shown in FIG. 6A, at time T1, a trend component that is a systematic fluctuation deviates from the reference range R, and it is determined that the secondary battery 2 is malfunctioning. As shown in FIG. 7, the maintenance rate of the full charge capacity of the secondary battery 2 when the initial state is 100% gradually decreases with use as shown as an assumed value. In this case, the maintenance rate of full charge capacity is lower than the expected value. According to the battery monitoring system 1 of the first embodiment, at the initial failure Fi in FIG. 7 corresponding to time T1 in FIG. A failure of the secondary battery 2 can be determined. On the other hand, if the failure determination unit 50 determines in step S5 that the systematic fluctuation does not deviate from the reference range R, the process advances to No in step S5 and returns to step S1 again.

In the control flow of the battery monitoring system 1 of the first embodiment, feature quantities are calculated using multiple types of monitoring data acquired from the secondary battery 2, and systematic fluctuations are calculated based on changes in the feature quantities over a predetermined period. Extract. Then, when the systematic fluctuation deviates from a predetermined reference range R, it is determined that the secondary battery 2 has failed. Thereby, failure of the secondary battery 2 can be detected early.

As shown in the conceptual diagram shown in FIG. 7, when the normal state of the secondary battery 2 continues, the battery performance of the secondary battery 2 gradually decreases from the solid line L0 to the solid line L1. On the other hand, when a failure occurs in the secondary battery 2, the battery performance of the secondary battery 2 shifts from the solid line L0 to the solid line L2, as shown in FIG. do.

In this embodiment, as described above, failure determination is performed using systematic fluctuations based on changes in the feature amount acquired from the secondary battery 2 over a predetermined period. Since the systematic fluctuation can be detected even at the stage of initial failure Fi, a failure determination can be made at the stage of early failure Fi, which is before the battery performance of the secondary battery 2 has significantly decreased. I can do it. In other words, the initial failure Fi of the secondary battery 2 can be detected.

Next, the effects of the battery monitoring system 1 of the first embodiment will be described in detail.
In the battery monitoring system 1 of the first embodiment, the origin of systematic fluctuations extracted from changes over a predetermined period of feature values calculated using multiple types of monitoring data acquired from the secondary battery 2 and the Based on the new knowledge that there is a correlation with the initial failure point, systematic fluctuations are calculated based on changes over a predetermined period of feature quantities calculated using multiple types of monitoring data acquired from the secondary battery 2. If the absolute value of the systematic variation deviates from a predetermined reference range, it is determined that the secondary battery has failed. In this way, by monitoring the state of the secondary battery 2 based on the systematic fluctuations, the influence of how the secondary battery 2 is used and short-term fluctuations can be suppressed, and failures of the secondary battery 2 can be prevented. The accuracy of determination can be improved.

Further, in the present embodiment, the data acquisition unit 10 acquires at least the closed circuit voltage, charging/discharging current, charging state, and battery temperature of the secondary battery 2 as the plurality of types of monitoring data X1 to X6. Thereby, by acquiring these as the monitoring data X1 to X6, the abnormality of the secondary battery 2 is more likely to be reflected in the feature amount, so that the accuracy of failure determination can be improved.

Furthermore, in this embodiment, the calculation unit 20 calculates correlation coefficients using the monitoring data X1 to X6 as variables as feature quantities. Thereby, it is possible to calculate a feature value that reflects the abnormality of the secondary battery 2 from the plurality of types of monitoring data X1 to X6, and it is possible to improve the accuracy of failure determination of the secondary battery 2.

Further, in the present embodiment, the calculation unit 20 calculates a partial correlation coefficient from the components of the inverse covariance matrix as the correlation coefficient. As a result, the feature amount further reflects the abnormality of the secondary battery 2, so that the accuracy of failure determination of the secondary battery 2 can be further improved.

Further, in this embodiment, the calculation unit 20 calculates the above covariance inverse matrix by performing sparse regularization. As a result, the feature amount more reflects the abnormality of the secondary battery 2, so that the accuracy of failure determination of the secondary battery 2 can be further improved.

In addition, the present embodiment further includes a feature amount storage unit 30 that stores the feature amount calculated by the calculation unit 20, and the extraction unit 40 systematically performs smoothing processing on the feature amount stored in the feature amount storage unit 30. Extract fluctuations. As a result, it is possible to extract trend components that reflect long-term fluctuations as systematic fluctuations, and effectively suppress the generation of noise based on seasonal fluctuation components and irregular fluctuation components. The accuracy of failure determination can be further improved.

Furthermore, in this embodiment, as shown in FIG. The failure determination section 50 is provided in an external device 8 disposed outside the vehicle 100. When diagnosing the failure of the secondary battery 2, the external device 8 is connected to the vehicle 100, and the monitoring data X1 to X6 acquired by the data acquisition section 10 is transmitted to the external device 8. In this way, by providing part of the configuration of the battery monitoring system 1 in the external device 8, the number of devices mounted on the vehicle 100 can be reduced.

Further, in this embodiment, the secondary battery 2 constitutes a battery pack made by combining a plurality of battery cells 21 to 23. Thereby, when a plurality of battery cells are provided, it is not necessary to provide the battery monitoring system 1 for each battery cell. In addition, when the secondary battery 2 consists of a single battery cell, it is good also as providing the said battery monitoring system 1 for every battery cell.

As described above, according to the present embodiment, it is possible to provide the battery monitoring system 1 in which the accuracy of failure diagnosis is improved.

(Embodiment 2)
As shown in FIG. 8, the battery monitoring system 1 of the second embodiment further includes a fluctuation storage unit 45 that stores systematic fluctuations, and a reference range setting unit 55 that sets the reference range based on the systematic fluctuations. Be prepared. Note that the other configurations are the same as those in the first embodiment, and are given the same reference numerals as those in the first embodiment, and the description thereof will be omitted.

In the control flow by the battery monitoring system 1 of the second embodiment shown in FIG. 9, steps S1 to S3 are performed similarly to the first embodiment shown in FIG. Thereafter, in step S4 shown in FIG. 9, the extraction unit 40 sets the partial correlation coefficient calculated from the components of the inverse covariance matrix of the monitoring data X1 to X6 in the period Pt shown in FIG. On the other hand, trend components are extracted as systematic fluctuations by performing smoothing processing using factor decomposition in time series analysis.

Thereafter, as shown in FIG. 9, step S7 and step S50 are performed in parallel. First, in step S7, the fluctuation storage section 45 stores systematic fluctuations over a predetermined period. For example, as shown in FIG. 10, the variation storage unit 45 stores all systematic variations acquired in the period Pt from the start to the present. Then, in step S8, the reference range setting section 55 sets the reference range R based on the systematic variation stored in the variation storage section 45. Note that if a reference range has been previously set, this is updated and set as the latest reference range R.

On the other hand, in step S50 shown in FIG. 9, the failure determination unit 50 determines whether the systematic fluctuation extracted in step S4 deviates from the latest reference range R. If the failure determination unit 50 determines that the latest systematic fluctuation deviates from the latest reference range R, the process advances to Yes in step S50, and as in the case of the first embodiment, in step S6, the failure is determined. The determination unit 50 determines that the secondary battery 2 is out of order, and ends this control flow. In the battery monitoring system 1 of the second embodiment, for example, at time T2 shown in FIGS. 10 and 11, the trend component, which is a systematic fluctuation, deviates from the reference range R, and the secondary battery 2 is out of order. It will be judged. On the other hand, if the failure determination unit 50 determines in step S50 that the systematic fluctuation does not deviate from the latest reference range R, the process advances to No in step S50 and returns to step S1 again.

Next, the effects of the battery monitoring system 1 of the second embodiment will be described in detail.
According to the second embodiment, by timely setting the reference range R for failure determination based on systematic fluctuations, failure diagnosis can be performed while updating the reference range R indicating the normal state of the secondary battery 2 in a timely manner. I can do it. Thereby, the accuracy of failure determination can be further improved. Note that the second embodiment also provides the same effects as the first embodiment.

(Embodiment 3)
The battery monitoring system 1 of the third embodiment includes a data acquisition section 10, a calculation section 20, an extraction section 40, and a failure determination section 50, as shown in FIG. 12, and a feature amount storage section in the first embodiment shown in FIG. 30 is not provided. In the battery monitoring system 1 of the first embodiment shown in FIG. 1, the extraction unit 40 extracts trend components as systematic fluctuations by smoothing the feature values for a predetermined period stored in the feature storage unit 30. . Instead, in the battery monitoring system 1 of the third embodiment shown in FIG. 12, the extraction unit 40 continues to perform filter processing on the feature amount calculated by the calculation unit 20 for a predetermined period, thereby detecting a trend as a systematic variation. Extract the ingredients. In the third embodiment, filter processing using a Kalman filter is adopted as the filter processing. Note that instead of the filter processing using the Kalman filter, filter processing using a low-pass filter may be employed in which frequency components lower than a predetermined cut-off frequency are hardly attenuated, while frequency components higher than the cut-off frequency are gradually attenuated. Note that in the third embodiment, the other configurations are the same as those in the first embodiment, are given the same reference numerals as in the first embodiment, and the explanation thereof will be omitted.

In the control flow by the battery monitoring system 1 of the third embodiment shown in FIG. 13, steps S1 and S2 are performed similarly to the first embodiment shown in FIG. Thereafter, in step S30 shown in FIG. 13, the extraction unit 40 extracts systematic variations from the feature amount calculated by the calculation unit 20. In the third embodiment, the extraction unit 40 performs filter processing using a Kalman filter on the partial correlation coefficient as the feature amount calculated by the calculation unit 20. Thereby, the extraction unit 40 extracts a trend component as a systematic variation from changes in the feature amount over a predetermined period. In the third embodiment, as shown in FIG. 14, the extraction unit 40 performs filter processing using a Kalman filter on the feature amount calculated based on the monitoring data acquired during the predetermined period Pn up to the current time Tn. , as shown in FIG. 15, a trend component as a systematic variation at the current time Tn is extracted. Thereafter, in the third embodiment, steps S5 and S6 shown in FIG. 13 are performed similarly to steps S5 and S6 of the first embodiment shown in FIG.

In the third embodiment, the feature amounts calculated by the calculation unit 20 are filtered to extract systematic variations. As a result, by performing filter processing in the extraction unit 40 every time a feature quantity is calculated by the calculation unit 20, systematic fluctuations can be extracted at a timing close to real time, thereby improving the real-time performance of failure determination. I can do it. Note that in the third embodiment as well, the same effects as in the first embodiment can be obtained except for the effects due to the smoothing process in the first embodiment.

(Embodiment 4)
As shown in FIG. 16, the battery monitoring system 1 of the fourth embodiment includes a data acquisition section 10, a calculation section 20, an extraction section 40, a fluctuation storage section 45, a failure determination section 50, and a reference range setting section 55. The feature amount storage unit 30 in Embodiment 2 shown in Embodiment 9 is not provided. Then, as in the case of the third embodiment, the extraction unit 40 in the fourth embodiment extracts a trend component as a systematic variation by continuing to filter the feature amount calculated by the calculation unit 20 for a predetermined period. do. In addition, in this Embodiment 4, other structures are equivalent to Embodiment 2, and the same code|symbol as in Embodiment 2 is attached|subjected and the description is abbreviate|omitted.

In the control flow by the battery monitoring system 1 of the fourth embodiment shown in FIG. 17, steps S1, S2, and S30 are performed similarly to the third embodiment shown in FIG. Thereafter, step S7 and step S50 and subsequent steps are performed similarly to the second embodiment shown in FIG. In Embodiment 4, the same effects as in Embodiments 1 and 2 can be obtained, except for the effects of the smoothing process in Embodiments 1 and 2, and the same effects as in Embodiment 3 can also be obtained. can get.

(Embodiment 5)
In the first embodiment described above, the external device 8 is directly connected to the external connection section 105 of the vehicle, but instead, a communication means is provided in the vehicle and the external device 8 is connected to the outside of the vehicle 100 via a network using a cloud system. It may also be connected to an external device 8. For example, in the third embodiment shown in FIG. 18, the vehicle 100 includes a transceiver 110 that can communicate with the external device 8. In the third embodiment, the secondary battery 2, the data acquisition section 10, the calculation section 20, the extraction section 40, and the failure determination section 50 are arranged in the vehicle 100, and the feature amount storage section 30 and the storage section 11 are arranged in the external device. It is set at 8. The external device 8 can be configured by a server, for example, and can be installed at a dealer of the vehicle 100 or the like. Then, using the transmitter/receiver 110, the monitoring data X1 to X6 and the partial correlation coefficient matrix Λ as the feature quantities are transmitted to the feature quantity storage unit 30 or the storage unit 11, or from the feature quantity storage unit 30 or the storage unit 11. can be received.

According to the configuration of the fifth embodiment, the number of parts mounted on the vehicle 100 can be reduced, the weight of the vehicle 100 can be easily reduced, and the server constituting the external device 8 can collect monitoring data X1 to X6 and feature amounts. It becomes easier. Note that the fifth embodiment can also provide the same effects as those of the first to fourth embodiments.

In the fifth embodiment, the feature storage unit 30 and the storage unit 11 are arranged outside the vehicle 100, but instead of this, as in the first modification shown in FIG. 19, the secondary battery 2 and the data acquisition The unit 10, the storage unit 11, and the feature storage unit 30 may be arranged in the vehicle 100, and the calculation unit 20, the extraction unit 40, and the failure determination unit 50 may be provided in the external device 8. Further, as in a second modification shown in FIG. 20, the secondary battery 2 and the data acquisition section 10 are arranged in the vehicle 100, and the storage section 11, the calculation section 20, the feature amount storage section 30, the extraction section 40, and the failure determination section are arranged in the vehicle 100. 50 may be provided in the external device 8. In both the first modification and the second modification, data is transmitted and received between the external device 8 and the vehicle 100 via the transceiver 110 of the vehicle 100.

According to the configurations of the first modification and the second modification, it is possible to perform failure diagnosis in real time, manage data stored in the feature storage unit 30, update the program in the failure determination unit 50, and use AI technology. This makes it easier to run programs that have been written. Further, by configuring the calculation unit 20, the extraction unit 40, and the failure determination unit 50 using devices with high calculation speed, such as a server, failure determination can be performed in a shorter time. Note that the first modification and the second modification have the same effects as the fifth embodiment.

The present invention is not limited to the above-described embodiments and modifications, but can be applied to various embodiments without departing from the spirit thereof.

DESCRIPTION OF SYMBOLS 1... Battery monitoring system, 2... Secondary battery, 10... Data acquisition part, 11... Storage part, 20... Calculation part, 30... Feature amount storage part, 40... Extraction part, 45... Fluctuation storage part, 50... Failure determination Part, 55... Reference range setting part, 100... Vehicle

Claims (8)

A battery monitoring system (1) that monitors the state of a secondary battery (2),
a data acquisition unit (10) that acquires multiple types of monitoring data for monitoring the secondary battery;
a calculation unit (20) that calculates a feature amount using the monitoring data;
an extraction unit (40) that extracts the trend component from a time-series variation including a trend component and a seasonal variation component or an irregular variation component as a systematic variation of the feature amount based on a change in the feature amount over a predetermined period; ,
a failure determination unit (50) that determines that the secondary battery has failed when the systematic fluctuation deviates from a predetermined reference range (R) ;
The calculation unit calculates a partial correlation coefficient from a component of an inverse covariance matrix with the monitoring data as a variable as the feature quantity . The battery monitoring system according to claim 1, wherein the data acquisition unit acquires at least a closed circuit voltage, a charging/discharging current, a state of charge, and a battery temperature of the secondary battery as the plurality of types of monitoring data. The battery monitoring system according to claim 1 or 2 , wherein the calculation unit calculates the inverse covariance matrix by performing sparse regularization. The battery monitoring system according to any one of claims 1 to 3 , wherein the extraction unit extracts the systematic variation by filtering the feature quantity calculated by the calculation unit. further comprising a feature amount storage unit (30) that stores the feature amount calculated by the calculation unit,
4. The battery monitoring system according to claim 1 , wherein the extraction section extracts the systematic fluctuation by smoothing the feature amount stored in the feature storage section. a variation storage section (45) that stores the systematic variation extracted by the extraction section; and a reference range setting section (55) that sets the reference range based on the systematic variation stored in the variation storage section. The battery monitoring system according to any one of claims 1 to 5 , further comprising: . The secondary battery and the data acquisition unit are mounted on a vehicle (100),
The battery monitoring system according to any one of claims 1 to 6 , wherein at least one of the calculation section, the extraction section, and the failure determination section is arranged outside the vehicle. The battery monitoring system according to any one of claims 1 to 7 , wherein the secondary battery constitutes a battery cell or an assembled battery formed by combining a plurality of battery cells (21, 22, 23).

Images