Battery degradation monitoring assessment and management

Abstract

The vehicle includes a battery with a set of battery sensors. A controller communicates with this set of battery sensors. The controller stores instructions for performing the following operations: receiving a dataset from the set of battery sensors, comprising multiple data sources and multiple data collection periods; classifying the data sources within the dataset based on the frequency of change of the data from the data sources; processing the data by determining the average value of each data element across each data collection period for data originating from the data sources in the first category and by partitioning the data originating from the data sources into a variable space; applying the processed data to at least one risk assessment model and identifying indications of the likelihood of battery health degradation; and issuing a warning function for the battery in response to the likelihood of battery health degradation.

Description

Technical Field

[0001] This subject matter relates to battery system monitoring and management, and more specifically to systems and processes for monitoring battery degradation. Background Technology

[0002] Electric vehicles and hybrid electric vehicles utilize battery systems to provide some or all of the power to the vehicle while it is in operation. It should be understood that continuous use of batteries results in repeated charge and discharge cycles, and may ultimately lead to a reduction in the battery's charging capacity and / or other reduced efficiency and effectiveness.

[0003] Existing systems typically operate by monitoring battery capacity (e.g., the maximum charge a battery can hold) on a per-cell basis. While this monitoring effectively identifies the end-of-life and near-end-of-life states of individual cells, it cannot determine what specific health mechanism or combination of health mechanisms leads to the end of a cell's lifespan, nor can this form of monitoring proactively identify and respond to risks. Therefore, monitoring is purely reactive and can only identify when an individual cell needs to be replaced or will need to be replaced in the near future.

[0004] Therefore, it is desirable to provide a battery monitoring process that can detect battery degradation in its early stages and identify the potential root causes of battery degradation. Summary of the Invention

[0005] In one exemplary embodiment, the vehicle includes a battery having a set of battery sensors. A controller communicates with the set of battery sensors and has a non-transitory memory and a processor. The memory stores instructions for causing the controller to perform the following operations: receive a dataset from the set of battery sensors comprising multiple data sources and multiple data collection cycles for each data source; classify the data sources within the dataset into a first category and a second category based on the frequency of change of the data from the data sources; process the data in the dataset by determining the average value of each data element across each data collection cycle for the data originating from the data sources in the first category and by partitioning the data originating from the data sources in the second category into a variable space; apply the processed data in the dataset to at least one risk assessment model and identify an indication of the likelihood of battery health degradation; and automatically change at least one function of the battery in response to the indication of the identified likelihood of battery health degradation.

[0006] In addition to one or more features described in this paper, the first category is slow-moving dynamic data, and the second category is fast-moving dynamic data.

[0007] In addition to one or more features described in this paper, partitioning data from data sources in the second category into a variable space involves dividing the variable space into multiple grids and evaluating the applicability of the partitioned variable space by comparing the number of grids in the variable space with a predefined value.

[0008] In addition to one or more features described herein, the operation also includes responding to a number of grids exceeding a predetermined value by determining that the segmented variable space is suitable for transmission.

[0009] In addition to one or more features described herein, the operation also includes responding to a grid number being less than or equal to a predefined value by determining that the segmented variables are unsuitable.

[0010] In addition to one or more features described herein, the operation also includes responding to the inadequacy of the segmented variable space by identifying regions of interest comprising a grid that satisfies a predefined set of parameters, and optimizing the cutoff values ​​that constrain the segmented variable space based on the identified regions of interest.

[0011] In addition to one or more features described in this paper, the operation also includes resegmenting the variable space using optimized cutoff values.

[0012] In addition to one or more features described herein, the controller communicates with an external data analysis system, and wherein the external data analysis system performs the following actions: determining the average value of each data element across each data collection cycle for data from data sources in the first category; partitioning data from data sources in the second category into a variable space; applying the processed data in the dataset to at least one risk assessment model; and identifying indications of the likelihood of battery health degradation.

[0013] In addition to one or more features described herein, the operations also include losslessly encoding the dataset into a low-dimensional latent space, transferring the low-dimensional latent space from the vehicle to an external data analysis system, and losslessly decoding the low-dimensional latent space.

[0014] In addition to one or more features described herein, the indication for identifying the likelihood of battery health degradation also includes identifying at least one fleet-wide modification to battery operation based on the indication of the identified likelihood of battery health degradation.

[0015] In another exemplary embodiment, a method for monitoring a vehicle battery includes receiving a dataset from a set of battery sensors at a vehicle controller. The dataset includes multiple data sources and multiple data collection periods for each data source. The method categorizes the data sources in the dataset into a first category and a second category based on the frequency of change of data from the data sources. The method processes the data in the dataset by determining the average value of each data element across each data collection period for data originating from data sources in the first category and by partitioning data originating from data sources in the second category into a variable space. The method applies the processed data in the dataset to at least one risk assessment model and identifies indications of the likelihood of battery health degradation. The method automatically modifies at least one function of the battery in response to the identified indications of the likelihood of battery health degradation.

[0016] In addition to one or more features described in this paper, the first category is slow-moving dynamic data, and the second category is fast-moving dynamic data.

[0017] In addition to one or more features described in this paper, partitioning data from data sources in the second category into a variable space involves dividing the variable space into multiple grids and evaluating the applicability of the partitioned variable space by comparing the number of grids in the variable space with a predefined value.

[0018] In addition to one or more features described herein, the method also includes responding to a number of grids exceeding a predetermined value by determining that the segmented variable space is suitable for transmission.

[0019] In addition to one or more features described herein, the method also includes responding to a number of grids being less than or equal to a predefined value by determining that the segmented variables are unsuitable.

[0020] In addition to one or more features described herein, the method also includes responding to the inappropriateness of the segmented variable space by identifying regions of interest comprising a grid that satisfies a predefined set of parameters, and optimizing the cutoff values ​​that constrain the segmented variable space based on the identified regions of interest.

[0021] In addition to one or more features described in this paper, the method also includes resegmenting the variable space using optimized cutoff values.

[0022] In addition to one or more features described herein, the vehicle controller communicates with an external data analysis system, wherein the external data analysis system processes the data in the dataset by determining the average value of each data element across each data collection period for data from data sources in the first category, by partitioning data from data sources in the second category into a variable space, by applying the processed data in the dataset to at least one risk assessment model, and by identifying an indication of the likelihood of battery health degradation of the battery.

[0023] In addition to one or more features described herein, the method also includes losslessly encoding the dataset into a low-dimensional latent space, transferring the low-dimensional latent space from the vehicle to an external data analysis system, and losslessly decoding the low-dimensional latent space.

[0024] In addition to one or more features described herein, the indication for identifying the likelihood of battery health degradation also includes identifying at least one fleet-wide modification to battery operation based on the indication of the identified likelihood of battery health degradation.

[0025] The above-described features and advantages, as well as other features and advantages, of this disclosure will become apparent when taken in conjunction with the accompanying drawings and the following detailed description. Attached Figure Description

[0026] Other features, advantages, and details appear by way of example only in the following detailed description, which is described in detail with reference to the accompanying drawings, wherein:

[0027] Figure 1 This is a top-down view of the vehicle communicating with the battery monitoring system;

[0028] Figure 2 yes Figure 1 Vehicle and battery monitoring systems monitor and respond to the process of battery degradation over time;

[0029] Figure 3 It is a process used to divide fast-moving data;

[0030] Figure 4 It is the process of encoding and decoding data for transmission; and

[0031] Figure 5 This is an example table that associates risk patterns with recommendation responses. Detailed Implementation

[0032] The following description is exemplary in nature only and is not intended to limit this disclosure, its application, or use. It should be understood that throughout the drawings, corresponding reference numerals denote the same or corresponding parts and features. As used herein, the term "module" refers to processing circuitry that may include application-specific integrated circuits (ASICs), electronic circuitry, processor (shared, dedicated, or group) and memory executing one or more software or firmware programs, combinational logic circuitry, and / or other suitable components that provide the described functionality.

[0033] As used herein, the term controller refers to a system that includes at least a processor and memory, wherein the system is configured to perform or cause to perform at least one operation. The system may be a dedicated controller including a single-purpose processor and memory, a general-purpose controller including one or more modules for performing operations, a distributed system including multiple controllers communicating with each other and configured to control operations, or any similar system.

[0034] According to an exemplary embodiment, Figure 1 Vehicle 10 is shown. Exemplary vehicle 10 is an electric vehicle that includes a battery system (battery 20). Battery 20 is connected to charging port 22, which is configured to receive power from an external power source and charge battery 20. Controller 30 communicates with battery 20 and is configured to control the operation of battery 20 and monitor battery parameters (e.g., charging rate, capacity, discharge rate, etc.) using battery sensor 24.

[0035] In addition, sensor kit 40 provides controller 30 with usage statistics of vehicle 10 and any other relevant sensor information. Sensor kit 40 may include any number of sensor types and configurations. Controller 30 includes data collection module 32 and data transmission module 34. Data collection module 32 collects and verifies data provided from battery sensor 24 and sensor kit 40.

[0036] In one example, controller 30 also includes a wireless or wired communication 36 capable of connecting to and communicating with an external data analysis system 50. The external data analysis system 50 includes a data extraction module 52 and a data analysis module 54. Although described herein as four software modules, the data collection module 32, data transmission module 34, data extraction module 52, and data analysis module 54 can be implemented in some practical embodiments as any number of different interconnected software modules distributed throughout controller 30, the external data analysis system 50, and / or any additional computer or controller system communicating with controller 30.

[0037] In yet another practical implementation, software modules 32, 34, 52, and 54 may be implemented entirely within controller 30 and / or other vehicle-based controllers. In such an example, the portion described herein relating to the transfer of collected data from vehicle 10 to external analysis system 50 may be omitted.

[0038] The combined process of software modules 32, 34, 52, and 54 provides a system and functionality capable of detecting and isolating the degradation of battery 20 over time based on its usage history. The combined process can detect early degradation by monitoring battery 20 usage and assessing the risk of failure mode development. The combined process provides insights that can help identify the root causes of battery 20 failures and identify potential failure modes, and can recommend responses and modifications to battery 20 operation to mitigate battery 20 degradation.

[0039] In the implementation using the external data analysis system 50, the process described herein provides data transmission that uses significantly reduced bandwidth compared to existing data transmission processes to transfer relevant collected data from vehicle 10 to the external data analysis system 50.

[0040] In a system where vehicle 10 is part of a fleet that includes substantially similar battery 20 monitoring characteristics, an external data analytics system 50 can participate in fleet-wide analysis and recommendations. Risk identification is not limited to a specific vehicle 10. Furthermore, recommendations for risk mitigation are not limited to individual vehicles. For example, fleet risk mitigation may include actions such as assigning vehicles to less frequent services or moving vehicle 10 to warmer or colder environments.

[0041] In another example, the process described herein can be updated to an existing vehicle as a software update, thereby allowing the existing vehicle to also implement the process.

[0042] Continue to refer to Figure 1 , Figure 2 A process 200 is shown for collecting vehicle usage data and recommending changes to the operation of battery 20 in order to improve battery life cycle.

[0043] Initially, process 200 collects sensor readings from battery sensor 24 and sensor suite 40 in data collection step 210. Sensed data includes charging capacity, charging rate, discharging rate, battery voltage, battery current, battery temperature, state of charge (SOC), estimated battery capacity, estimated battery resistance, and similar battery health indicators. In some embodiments, data collection may be optimized based on a risk model used to analyze the risk of failure modes within battery 20. In other embodiments, the data used is selected based on relevance to the risk being analyzed, feature importance scores, or any other method used to identify the most relevant data from all available data.

[0044] Then, in the data source classification step 220, the collected data is classified into two types based on how much each specific data element changes across a given data collection period (e.g., two minutes). Each data element is assigned to one of the two categories: fast dynamic data sources and slow dynamic data sources. Fast dynamic data sources are those that include elements that change rapidly or frequently within a given data collection period. In contrast, slow dynamic data sources are those that include elements that change slowly (or not at all) within a given data collection period.

[0045] In a practical example, the data collection period can be based on each driving trip, and a fast dynamic data source is any data source that changes during the data collection period. A slow dynamic data source is a data source that changes by less than a minimum amount (e.g., 1%) during the data collection period.

[0046] Once categorized, the data from each data source is processed in data processing step 230. Data processing step 230 is performed locally on controller 30 before the data is transferred to external data analysis system 50.

[0047] For data from a slow dynamic data source, the data follows a first processing path 232. On this first processing path 232, in the slow dynamic data averaging step 234, each data element is averaged over the data collection period, and the average value is provided as an output from processing step 230. Due to the low rate of change of the average data within each data collection period, a strong approximation of the data element trajectory is provided. This approximation can be used for multiple risk models.

[0048] When data comes from a fast, dynamic data source, the data follows the second processing path 236 and is initially partitioned in the fast, dynamic data partitioning step 238. (Continue to refer to...) Figure 1-2 , Figure 3 The example of a quick partitioning path 236 is shown. Initially, based on... (For k=0, i=1,2,…,n), the dataset is cut off in step 302, thus providing the start and end times of the data, where min(vi) is the minimum value of the data element (variable vi), max(vi) is the maximum value of the data element, k is the iteration number, k starts from iteration 0, i is each reading of the data element across the data collection cycle, and n is the dimension of the variable space. It is the endpoint along variable i after the k-th iteration.

[0049] Once partitioned, the variable space of data elements is limited based on... (Where i = 1, 2, ..., n) and is divided into N grids along the variable vi. A grid is defined as... , where k is the iteration index and p is the grid index. For each grid, determine the count, mean, and variance of the degradation risk values ​​within that grid.

[0050] After partitioning the variable space and determining the count, mean, and variance values ​​for each grid cell, the second processing path 236 evaluates the partition to determine whether the partition is satisfactory at check 306. This is achieved when the number of grid cells N is at least equal to the predefined number of grid cells (N...). max A partition is considered satisfactory if the grid size is the same as the grid size, or if there is no grid in the region of interest. Otherwise, the partition is not considered satisfactory.

[0051] When the partitioning is satisfactory, the second processing path 236 ends at the stopping element 308, and considers the data to have been satisfactorily partitioned.

[0052] When the partitioning is unsatisfactory, the second processing path 236 identifies meshes from which the partitioning can be improved, and in the region of interest identification step 310, a set of mesh parameters is used to assign all such meshes to the region of interest (R). In one example, the region of interest includes any mesh with a variance greater than 0.001, a count greater than 1000, and an average value with a magnitude less than 0.2. In other examples, specific values ​​of the mesh parameters can be changed based on the expected available transmission bandwidth.

[0053] After identifying the region of interest (R), the second processing path 236 determines whether at least one mesh exists in the region of interest during the region of interest check 312. If no mesh is found in the region of interest, the second processing path 236 determines that the partition is the best possible partition and proceeds to the stopping element 308.

[0054] When at least one grid is identified in the region of interest, the second processing path 236 identifies, in grid selection step 314, which grid has the highest variance with the risk of degradation. In one example, the grid (P*) with the highest variance with the risk of degradation is determined according to:

[0055]

[0056]

[0057] Where argmaxVar() is the function used to identify the maximum variance, and p is... The index, V is In the data point, risk(V) is the degradation risk associated with the data point, and R is the region of interest.

[0058] After identifying the grid with the highest risk of degradation, a new cutoff value is added to the set of cutoff values ​​in step 316. The new cutoff value is added at the median position of all data points in the grid. In one example, a new cutoff value is added based on the following:

[0059] ,in

[0060] After adding the new cutoff value, the second processing path 236 re-divides the variable space using the updated cutoff value and identifies the axis that minimizes the sum of the squared deviations of the means across all grids based on the following:

[0061] Where N i It is the cutoff value med added along the i-axis. i The number of grids after that, It is the p-th grid after the cutoff value is added along the i-axis. It is a function to obtain the average value.

[0062] In the resegmentation step 318, the combined function of steps 316 and 318 is to optimize the mesh along an axis, thereby minimizing the sum of squared deviations from the mean across all meshes.

[0063] Once optimized, the second processing path updates the cutoff by optimizing the boundary along the selected axis in step 320. In one example, the boundary is optimized by optimizing the cutoff along the i-axis:

[0064]

[0065] { }as well as

[0066] Once optimized, the second processing path provides the optimized cutoff value to the segmentation step 304, and the second processing path 236 loops until a satisfactory segmentation is achieved in step 308.

[0067] The partitioned data is represented as a variable space, where each grid in the variable space has a value. The total size of the variable space is generally smaller than the size of the complete dataset.

[0068] Then, in the data collection and transmission step 240, the processed dataset is collected and transmitted to the external data analysis system 50. When process 200 is performed entirely locally in the vehicle 10 where the data is collected (e.g., when process 200 is limited to a single battery 10 and no fleet-wide analysis is applied), the data collection and transmission step 240 can be omitted, and the controller 30 executes all steps of process 200.

[0069] Continue to refer to Figure 1-3 , Figure 4 An example system 400 is shown for performing the data collection and transmission step 240. The processed data 402 is configured as a grid, including slow-moving dynamic data, and the controller 30 encodes the data into a low-dimensional latent space using a lossless encoder 404 according to any known lossless compression process. The processed data and the lossless encoder 404 are functionalities of the controller 30.

[0070] The output of the lossless encoder 404 is transmitted to an external data analysis system 50 via a wireless or wired connection. Once at the external data analysis system 50, the data is decoded into a complete dataset 408 that matches the encoded dataset 402 using a lossless decoder 406. The lossless decoder 406 can be any lossless decoding process suitable for the dataset being sent and received.

[0071] Once data is received at the external data analysis system 50, or at the controller 30 during local execution, at least one risk model is used in step 250 of assessing risk and failure modes to analyze the data to identify potential risks of failure modes. When analyzing multiple potential failure modes, the data can be assessed multiple times via different assessment models.

[0072] For example, process 200 can assess data to identify electroplating risks, solid electrolyte (SEI) growth risks, loss of active material (LAM) risks, and electrolyte degradation risks. In this case, each risk is modeled separately, and the data is analyzed four times. The provision of multiple concurrent data types from battery sensor 24 and sensor suite 40 supports the use of multiple risk models by ensuring that all relevant data for each risk model is provided for analysis.

[0073] The assessed risks are output as a chart indicating the combinations of risks present based on the data. This chart is applied to a lookup table, database, or other reference stored in memory, which associates the identified risk set with one or more recommended actions to eliminate or minimize the risk in recommended action step 260.

[0074] Continue to refer to Figure 1-4 , Figure 5An exemplary lookup table 500 is shown, which includes a first part 502 identifying risk patterns and a second part 504 identifying recommendations for minimizing the set of risk patterns. For example, if process 200 identifies a high plating risk but does not indicate other potential risks (row 1 of table 500), the recommended action is to reduce the charging current supplied to battery 20, and no other actions are recommended.

[0075] When a high SEI growth risk is indicated and no other risks are indicated (Row 2 of Table 500), the recommended action is to reduce the maximum state of charge, increase the minimum state of charge, and reduce the battery temperature.

[0076] When a high LAM risk and a high electrolyte degradation risk are indicated, and no other risks are indicated (Row 3 of Table 500), the recommended actions are to reduce the maximum state of charge, increase the minimum state of charge, and reduce the battery temperature.

[0077] When a high electroplating risk and a high SEI growth risk are indicated, and no other risks are indicated (row 4 of Table 500), the recommended action is to reduce the maximum state of charge and reduce the charging current.

[0078] While the exemplary table 500 includes entries for four risk combinations, it should be understood that, in the case of analyzing four risk models, up to sixteen possible rows can be used to encompass all possible outcomes. Furthermore, when one or more risk models provide more than a binary risk indication (e.g., no risk, low risk, high risk), additional rows can be added to accommodate possible output combinations. Additionally, in some cases, risks can be mutually exclusive; in such cases, fewer rows can be used because the two risks cannot coexist. Therefore, it should be understood that those skilled in the art can determine the appropriate number of rows for such a lookup table 500 based on the risks being analyzed. When a recommended action is determined by the external analysis system 50, the action is transmitted to the controller 30.

[0079] Once at controller 30, the recommended action is implemented in action step 270. In some cases, where the recommended action can be adjusted via controller 30, the controller 30 implements the action automatically. One such case is an example where the recommendation includes adjusting the maximum state of charge of battery 20. The recommended implementation can be invisible, without notification to the vehicle owner or operator, or it can be visible, wherein notification of the change and / or any indirect effects of the change can be communicated to the vehicle owner or operator. Indirect effects of the change include reduced range, increased expected charging time, etc.

[0080] Particularly beneficial is the data segmentation and optimization process, which significantly reduces the size of the data being transmitted, thereby saving bandwidth and reducing the time spent transmitting data.

[0081] The terms “a” and “an” do not indicate a limitation of quantity, but rather that at least one of the referenced items is present. Unless the context clearly indicates otherwise, the term “or” means “and / or”. Throughout the specification, the reference to “aspect” means that a particular element described in connection with that aspect (e.g., a feature, structure, step, or characteristic) is included in at least one aspect described herein and may or may not be present in other aspects. Furthermore, it should be understood that the described elements may be combined in any suitable manner in the aspects.

[0082] When a component, such as a layer, film, region, or substrate, is referred to as being “on” another component, it can be directly on the other component, or there may be intermediate components. Conversely, when a component is referred to as being “directly” on another component, there are no intermediate components.

[0083] Unless otherwise stated herein, all test standards are the most recent standards in force up to the date of filing of this application, or, if priority is claimed, the date of filing of the earliest priority application in which a test standard appears.

[0084] Unless otherwise defined, the technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains.

[0085] While the foregoing disclosure has been described with reference to exemplary embodiments, those skilled in the art will understand that various changes can be made and elements can be substituted with equivalents without departing from its scope. Furthermore, many modifications can be made to adapt particular situations or materials to the teachings of this disclosure without departing from the basic scope of this disclosure. Therefore, it is intended that this disclosure be limited to the specific embodiments disclosed, but will include all embodiments falling within its scope.

Claims

1. A vehicle comprising: A battery, the battery including a set of battery sensors; A controller, which communicates with the set of battery sensors, has a non-transitory memory and a processor, wherein the non-transitory memory stores instructions for causing the controller to perform the following operations: A dataset is received from the set of battery sensors, the dataset comprising multiple data sources and multiple data collection cycles for each data source; Based on the frequency of change of data from the data source, the data sources within the dataset are classified into a first category and a second category; The data in the dataset is processed by determining the average value of each data element across each data collection period for data from data sources in the first category, and by partitioning data from data sources in the second category into a variable space. The processed data in the dataset is applied to at least one risk assessment model to identify indications of the likelihood of battery health degradation. and In response to an indication of a potential degradation in battery health, at least one function of the battery is automatically changed.

2. The vehicle according to claim 1, wherein, The first category is slow-moving dynamic data, and the second category is fast-moving dynamic data.

3. The vehicle according to claim 1, wherein, Dividing data from data sources in the second category into a variable space includes: dividing the variable space into multiple grids, and evaluating the applicability of the divided variable space by comparing the number of grids in the variable space with a predefined value.

4. The vehicle of claim 3, further comprising responding to a number of grids exceeding the preset value by determining that the segmented variable space is suitable for transmission.

5. The vehicle of claim 3, further comprising responding to a number of grids being less than or equal to the preset value by determining that the segmented variable is unsuitable.

6. The vehicle of claim 5, further comprising responding to the segmentation of the variable space being unsuitable by identifying regions of interest comprising a grid satisfying a predefined set of parameters, and optimizing a cutoff value defining the segmented variable space based on the identified regions of interest.

7. The vehicle of claim 6 further includes resegmenting the variable space using an optimized cutoff value.

8. The vehicle of claim 1, wherein the controller communicates with an external data analysis system, and wherein the processing of data in the dataset is performed at the external data analysis system by: determining the average value of each data element across each data collection period for data originating from a data source in the first category, and partitioning data originating from a data source in the second category into a variable space, and applying the processed data in the dataset to at least one risk assessment model, and identifying an indication of the likelihood of battery health degradation of the battery.

9. The vehicle of claim 8, further comprising losslessly encoding the dataset into a low-dimensional latent space, transmitting the low-dimensional latent space from the vehicle to the external data analysis system, and losslessly decoding the low-dimensional latent space.

10. The vehicle of claim 8, wherein the indication of the likelihood of battery health degradation of the battery further includes identifying at least one fleet-wide modification to battery operation based on the indication of the likelihood of battery health degradation.

Images