DESC:SYSTEM FOR ASSESSING STATE OF HEALTH VALUES OF BATTERY CELLS
CROSS REFERENCE TO RELATED APPLICTIONS
The present application claims priority from Indian Provisional Patent Application No. 202421014546 filed on 28-02-2024, the entirety of which is incorporated herein by a reference.
TECHNICAL FIELD
The present disclosure generally relates to battery management systems. Further, the present disclosure particularly relates to systems for assessing state of health values of battery cells.

BACKGROUND
Electric vehicles rely on power packs comprising multiple battery cells to store and deliver energy for propulsion. Lithium-ion battery cells are commonly used due to their high energy density, long life cycle, fast charging capability and low self-discharge rate. However, the performance of battery cells deteriorates over time due to factors such as cycle ageing, calendar ageing, lithium plating and mechanical stress.
The state of health (SOH) of a battery cell represents a remaining capacity and reliability thereof. Accurate assessment of the SOH of the battery is necessary to schedule timely maintenance or replacement to prevent system failures. Conventional methods for state of health assessment include statistical models and machine learning techniques. Statistical models rely on historical data to identify degradation trends but fail to capture complex interdependencies among degradation factors. Machine learning techniques improve predictive accuracy but often suffer from inadequate training methods, limited access to high-quality data and reliance on voltage-based parameters. Voltage-based parameters are influenced by charging profiles, leading to reduced reliability in predictions.
Internal resistance is a reliable parameter for assessing battery health due to strong correlation thereof with capacity degradation over usage cycles. Internal resistance is unaffected by variations in charging and discharging profiles, making the parameter a robust predictor of the SOH of the battery cell. However, conventional methods do not effectively utilise internal resistance for accurate predictions.
In light of the above discussion, there exists an urgent need for a system that utilizes internal resistance as the primary parameter to accurately assess the SOH of battery cells. Such a solution must address the limitations of conventional techniques and provide reliable predictions to support better battery health management.

SUMMARY
The present disclosure generally relates to systems for assessing the state of health (SOH) of battery cells.
In an aspect, the present disclosure provides a system for assessing the SOH of a battery cell. The system comprises a memory unit to store a machine learning model. The system further comprises a sensing unit to sense an internal resistance of the battery cell, with the sensed internal resistance indexed with a state of charge (SOC) level of the battery cell at the time of sensing the internal resistance. The system also comprises a data processing unit operatively connected to the memory unit and the sensing unit. The data processing unit inputs the sensed internal resistance into the machine learning model and assesses the SOH of the battery cell based on the inputted internal resistance and predefined relationships modelled by the machine learning model. The system further comprises an output module operatively connected to the data processing unit to display or transmit the computed SOH value to a user interface or external computing arrangement.
In an embodiment, the machine learning model is a gradient-boosting-based machine learning model selected from XGBoost, LightGBM, CatBoost, HistGradientBoostingClassifier, GradientBoostingClassifier, NGBoost or AdaBoost.
In another embodiment, the memory unit comprises a data repository to store historical data associated with the battery cell for iteratively updating the machine learning model.
In yet another embodiment, the sensing unit senses a temperature of the battery cell during sensing of the internal resistance. The sensed temperature is used by the data processing unit to recalibrate the assessed SOH value.
In an additional embodiment, the data processing unit computes an estimated remaining useful life (RUL) of the battery cell based on the assessed SOH value and historical degradation trends.
In another embodiment, the data processing unit recommends an action for battery maintenance or replacement based on the assessed SOH value.
In yet another embodiment, the data processing unit refines the assessed SOH value based on one or more of a specific usage pattern, a charging current, a capacity of the battery cell, a charging mode and charging and discharging cycles.
In an embodiment, the machine learning model is trained using data from multiple battery cells associated with different battery chemistries.
In another embodiment, the data processing unit reassesses the SOH value based on a charging cycle count data associated with the battery cell.
In yet another embodiment, the sensing unit senses the internal resistance of the battery cell as a DC resistance by applying a constant current and measuring a resulting voltage drop.
In an additional embodiment, the DC resistance is sensed by the sensing unit using a pulse current method, wherein a short-duration current pulse is applied to the battery cell and the voltage response is used to calculate the resistance.
In an embodiment, the sensing unit dynamically measures DC resistance during the operation of the battery cell without interrupting ongoing charging or discharging cycles.
In another embodiment, the machine learning model is trained using historical DC resistance data collected from multiple battery cells associated with the same chemistry, capacity and usage pattern.

BRIEF DESCRIPTION OF DRAWINGS
The summary above, as well as the following detailed description of illustrative embodiments, is better understood when read in conjunction with the appended drawings. For the purpose of illustrating the present disclosure, exemplary constructions of the disclosure are shown in the drawings. However, the present disclosure is not limited to specific methods and instrumentalities disclosed herein. Moreover, those in the art will understand that the drawings are not to scale. Wherever possible, like elements have been indicated by identical numbers.
Embodiments of the present disclosure will now be described, by way of example only, with reference to the following diagrams wherein:
FIG. 1 illustrates a block diagram of a system for assessing a state of health (SOH) value of a battery cell, in accordance with an embodiment of the present disclosure;
FIG. 2 illustrates a graph illustrating a relationship between the sensed internal resistance versus SOH for a battery cell (such as the battery cell of FIG. 1), in accordance with an embodiment of the present disclosure; and
FIG. 3 illustrates a a scatter plot illustrating a relationship between actual and predicted values of SOH of a battery cell (such as the battery cell of FIG. 1), in accordance with an embodiment of the present disclosure.
In the accompanying drawings, an underlined number is employed to represent an item over which the underlined number is positioned or an item to which the underlined number is adjacent. A non-underlined number relates to an item identified by a line linking the non-underlined number to the item. When a number is non-underlined and accompanied by an associated arrow, the non-underlined number is used to identify a general item at which the arrow is pointing.
DETAILED DESCRIPTION
The following detailed description illustrates embodiments of the present disclosure and ways in which they can be implemented. Although some modes of carrying out the present disclosure have been disclosed, those skilled in the art would recognise that other embodiments for carrying out or practising the present disclosure are also possible.
The description set forth below in connection with the appended drawings is intended as a description of certain embodiments of a system for assessing the SOH of a battery cell and is not intended to represent the only forms that may be developed or utilised. The description sets forth the various structures and/or functions in connection with the illustrated embodiments; however, it is to be understood that the disclosed embodiments are merely exemplary of the disclosure that may be embodied in various and alternative forms. The figures are not necessarily to scale; some features may be exaggerated or minimised to show details of particular components. Therefore, specific structural and functional details disclosed herein are not to be interpreted as limiting, but merely as a representative basis for teaching one skilled in the art to variously employ the present invention.
While the disclosure is susceptible to various modifications and alternative forms, specific embodiment thereof has been shown by way of example in the drawings and will be described in detail below. It should be understood, however, that it is not intended to limit the disclosure to the particular forms disclosed, but on the contrary, the disclosure is to cover all modifications, equivalents, and alternatives falling within the scope of the disclosure.
The terms “comprise”, “comprises”, “comprising”, “include(s)”, or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a setup, system that includes a list of components or steps does not comprise only those components or steps but may include other components or steps not expressly listed or inherent to such setup or system. In other words, one or more elements in a system or apparatus preceded by “comprises... a” does not, without more constraints, preclude the existence of other elements or additional elements in the system or apparatus.
In the following detailed description of the embodiments of the disclosure, reference is made to the accompanying drawings, and which are shown by way of illustration specific embodiments in which the disclosure may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the disclosure, and it is to be understood that other embodiments may be utilized and that changes may be made without departing from the scope of the present disclosure. The following description is, therefore, not to be taken in a limiting sense.
The present disclosure will be described herein below with reference to the accompanying drawings. In the following description, well known functions or constructions are not described in detail since they would obscure the description with unnecessary detail.
Referring to FIG. 1, there is shown a block diagram of a system 100 for assessing a state of health (SOH) value of a battery cell 102, in accordance with an embodiment of the present disclosure. The term “system” as used throughout the present disclosure relates to an arrangement configured to assess a state of health (SOH) value of a battery cell. The system 100 comprises various interconnected components working together to compute the SOH value using specific input parameters. The system 100 improves diagnostic accuracy and enables proactive maintenance of batteries by leveraging precise measurements and advanced computational techniques.
The term “battery cell” as used throughout the present disclosure relates to an energy storage device capable of storing electrical energy and delivering the stored energy as needed. The battery cell 102 may be of various chemistries, such as lithium-ion, nickel-metal hydride or lead-acid, depending on the intended application. The performance degradation is detected early by monitoring the battery cell 102, which reduces risks of unexpected failures and enhances operational safety in applications such as electric vehicles. The ability of the system 100 to evaluate individual cells in a battery pack ensures even utilization, which extends the overall lifespan of the pack.
The system 100 assesses the SOH value of the battery cell 102 by utilising interconnected components. The battery cell 102 serves as the primary source of data for the system 100. The battery cell 102 is typically a lithium-ion battery due to widespread application thereof in electric vehicles and portable devices, but the system 100 can be adapted for other chemistries depending on the use case. In implementations involving electric vehicles, the battery cell 102 is part of a battery pack comprising multiple interconnected cells. The system prevents overloading of degraded cells by evaluating individual cells, which mitigates thermal runaway risks and improves vehicle safety.
The system 100 comprises a memory unit 104 to store a machine learning model 106. The term “memory unit” as used throughout the present disclosure relates to a storage device within the system that holds data, programs or instructions necessary for the operation of the system 100. The memory unit 106 stores the machine learning model 106 used for processing and analysis. The term “machine learning model” as used throughout the present disclosure relates to an algorithmic representation stored in the memory unit 106. The machine learning model 106 predicts outcomes based on input parameters such as the sensed internal resistance of the battery cell 102. The system achieves a high degree of accuracy in assessing SOH across different battery chemistries and usage conditions by utilizing a pre-trained model. Such adaptability enhances the utility of the system 100 in electric vehicles, where battery packs operate under variable load profiles and temperatures. The ability of the machine learning model 106 to capture complex nonlinear relationships between degradation parameters and SOH enables reliability in predictive outcomes. Storing the machine learning model 106 in the memory unit 104 allows the system 100 to adapt and update predictive capabilities thereof through iterative learning. Such an operation contributes to long-term precision in SOH assessments by incorporating new degradation patterns and environmental conditions into its analysis.
The memory unit 104 is located within the system and stores the machine learning model 106 along with other data such as historical degradation patterns or configurations for processing. The memory unit 104 may be implemented using flash memory, solid-state drives or other storage technologies, enabling robust and secure data retention. Storing degradation trends allows the system 100to provide insights into long-term battery health, enabling electric vehicle operators to plan replacements or upgrades based on predictable wear patterns.
The system 100 further comprises a sensing unit 108 to sense an internal resistance of the battery cell 102. The sensed internal resistance is indexed with a state of charge (SOC) level of the battery cell 102 at a time of the sensing of the internal resistance. The term “sensing unit” as used throughout the present disclosure relates to a component or set of components capable of measuring the internal resistance of the battery cell 102. The sensing unit 108 performs measurements indexed with the state of charge (SOC) level of the battery cell 102 at the time of sensing. The sensing unit 108 minimizes errors caused by fluctuating voltage and current profiles by providing precise measurements of internal resistance, thereby enhancing the reliability of SOH predictions. In electric vehicles, the sensing unit 108 allows consistent monitoring across all cells in a battery pack, reducing the likelihood of performance bottlenecks caused by a single degraded cell. For example, in a battery pack used in an electric two-wheeler, sequential sensing of internal resistance for each cell enables detection of potential imbalances, leading to timely corrective actions.
The sensing unit 108 interacts directly with the battery cell 102 to measure internal resistance thereof. The internal resistance is an important parameter as the internal resistance correlates with the SOH of the battery cell 102. The sensing unit 108 measures the internal resistance at a specific state of charge (SOC) level for consistency and accuracy of the data. The variations due to charge levels do not affect the reliability of SOH assessments by indexing the measurement to SOC. In electric vehicles, the sensing unit 108 can sequentially evaluate cells within a large battery pack, providing an integrated view of the SOH of the battery pack while identifying individual weak cells. Such an operation improves energy distribution and reduces uneven ageing of cells.
Moreover, the system 100 comprises a data processing unit 110 operatively connected to the memory unit 104 and the sensing unit 108. The data processing unit 110 inputs the sensed internal resistance into the machine learning model 106 and assesses the SOH value of the battery cell 102 based on the inputted internal resistance and predefined relationships modelled by the machine learning model 106. The term “data processing unit” as used throughout the present disclosure relates to a processing element in the system responsible for inputting data into the machine learning model 106 and performing analysis to assess the SOH value of the battery cell 102. The data processing unit 110 eliminates noise from raw data and only high-quality inputs are provided to the machine learning model 106. Such an operation of the data processing unit 110 improves computational efficiency and enhances predictive accuracy. In electric vehicles, the data processing unit 110 can aggregate SOH values from individual cells, enabling real-time assessment of overall battery health. Such a functionality prevents catastrophic failures by identifying critical cells and enabling focused maintenance.
The data processing unit 110 receives the sensed internal resistance from the sensing unit 108 and inputs the data into the machine learning model 106 stored in the memory unit 104. By utilizing the data processing unit 110, the sensed internal resistance is processed to remove noise or inconsistencies before providing the pre-processed internal resistance to the machine learning model 106. Such a pre-processing step improves fidelity of the input data, which is critical for accurate predictions. In electric vehicles, the data processing unit 110 may also generate alerts for early signs of degradation, allowing users to take preventative measures such as load balancing or targeted cell replacement.
Moreover, the system 100 comprises an output module 112 operatively connected to the data processing unit 110. The output module 112 displays or transmits the computed SOH value to a user interface or external computing arrangement (not shown). The term “output module” as used throughout the present disclosure relates to a component that communicates the computed SOH value to external systems or a user interface. The output module 112 facilitates informed decision-making by providing actionable information to users or external systems. For example, in electric vehicle applications, the output module 112 can relay critical SOH values to fleet management systems, enabling predictive scheduling of battery pack replacements, which reduces downtime and optimizes operational efficiency.
The output module 112 is connected to the data processing unit 110 and acts as an interface for communicating the computed SOH value. The output module 112 can be configured to display the SOH value on a user interface, such as a screen or transmit the data to an external computing arrangement for further analysis. The output module 112 improves operational efficiency by enabling real-time access to SOH values. In electric vehicle applications, the output module 112 integrates with onboard displays to provide drivers with actionable feedback, such as estimated range or remaining battery lifespan, enhancing the user experience.
In a first working example, the system 110 is employed in an electric car. The sensing unit 108 measures the internal resistance of each battery cell 102 within the battery pack during regular operation at a 50% SOC level. The system 100 prevents uneven ageing by detecting deviations in internal resistance among cells, thus prolonging the overall life of the battery pack. The data processing unit 110 processes the sensed internal resistance values and provides the data to the machine learning model 106, which computes the SOH value for each cell. The output module 112 then displays the aggregate SOH information on the vehicle dashboard, enabling the user to monitor the health of the entire battery pack in real time.
In a second working example, the system 100 is used in an electric bus fleet. The sensing unit periodically measures the internal resistance of battery cells in multiple buses at a 30% SOC level. The system 100 identifies patterns that predict potential failures by comparing SOH trends across the fleet. The data processing unit 110 processes the collected data and assesses the SOH values using the machine learning model 106. The output module 112 transmits the SOH values to a central fleet management server, which schedules maintenance or replacements based on the reported SOH values, leading to reliability across the fleet.
In an embodiment, the machine learning model 106 is a gradient-boosting-based machine learning model selected from XGBoost, LightGBM, CatBoost, HistGradientBoostingClassifier, GradientBoostingClassifier, NGBoost or AdaBoost. The term “gradient-boosting-based machine learning model” as used throughout the present disclosure relates to an ensemble learning technique that builds multiple predictive models and combines outputs thereof to improve accuracy and robustness. The selection of the specific model is based on factors such as computational efficiency, dataset characteristics and the operational requirements of the system 100. For example, XGBoost is suitable for scenarios requiring fast training and scalability while CatBoost effectively handles categorical data without extensive preprocessing. Such a selection enables the system 100 to adapt to a wide range of applications including monitoring of battery packs in electric vehicles where variations in environmental and usage conditions can affect the accuracy of SOH assessments.
The system 100 achieves higher predictive reliability across diverse battery chemistries by integrating advanced learning algorithms, which enhances operational efficiency. For example, NGBoost is particularly effective for generating probabilistic predictions that provide confidence intervals around the SOH value, aiding decision-making processes in fleet management.
In another embodiment, the memory unit 104 comprises a data repository to store historical data associated with the battery cell 102 for iteratively updating the machine learning model 106. The term “data repository” as used throughout the present disclosure relates to a structured storage system within the memory unit 104 that retains information such as previously measured internal resistance values, corresponding SOH values and environmental conditions. The system 100 refines the machine learning model 106 over time by storing historical data, which enhances predictive accuracy thereof. For example, by incorporating data from varying operational cycles and environmental conditions, the machine learning model 106 adjusts parameters thereof to better reflect real-world degradation patterns.
The adaptability of the system 100 is improved in electric vehicle applications by using the data repository. For example, an electric car operating in a region with high temperature variability may experience degradation patterns distinct from those in milder climates. The predictions remain accurate and context-sensitive by continuously updating the machine learning model 106 with localized data stored in the data repository. Moreover, the need for frequent retraining is reduced by iteratively updating the machine learning model 106, which conserves computational resources while maintaining precision in SOH assessments.
In yet another embodiment, the sensing unit 108 senses a temperature of the battery cell 102 during sensing of the internal resistance. The term “temperature” as used throughout the present disclosure relates to the thermal condition of the battery cell 102 at the time of measurement. The sensed temperature is used by the data processing unit 110 to recalibrate the assessed SOH value. The accuracy of the SOH assessment is maintained by accounting for temperature variations that affect internal resistance. For example, higher temperatures typically reduce internal resistance while lower temperatures increase the internal resistance, which can otherwise lead to inaccurate SOH predictions if uncorrected.
The sensing unit 108 improves the reliability of SOH assessments by dynamically monitoring the temperature of the battery cell 102 along with internal resistance thereof. The sensed temperature is indexed with the state of charge (SOC) level at the time of measurement for consistent recalibration by the data processing unit 110. In applications involving electric vehicles, the sensing unit 108 enables real-time temperature monitoring across multiple cells within a battery pack. For example, in an electric car operating in extreme climates, the sensing unit 108 allows the system 100 to detect temperature-induced imbalances in individual cells and recalibrate SOH values accordingly. Such functionality prevents premature degradation for optimal energy distribution within the battery pack.
In an additional embodiment, the data processing unit 110 computes an estimated remaining useful life (RUL) of the battery cell 102 based on the assessed SOH value and historical degradation trends. The term “remaining useful life” or “RUL” as used throughout the present disclosure relates to the projected duration or number of cycles for which the battery cell 102 is expected to operate effectively before requiring replacement. The system 100 improves battery management by providing actionable insights into the operational lifespan of the battery cell 102.
The data processing unit 110 enhances predictive maintenance strategies by utilizing the assessed SOH value and historical degradation trends to calculate the RUL of the battery cell 102. For example, degradation trends such as capacity fade and internal resistance growth are correlated with the SOH value to determine the remaining operational capability of the battery cell 102. In an electric vehicle fleet, the data processing unit 110 enables centralized planning by predicting RUL for all battery cells within a fleet. For example, in an electric bus, the system 100 can project RUL values for each cell in the battery pack, allowing the operator to schedule replacements in advance and avoid unexpected downtimes.
In another embodiment, the data processing unit 110 recommends an action for battery maintenance or replacement based on the assessed SOH value. The term “action” as used throughout the present disclosure relates to a specific instruction or operation aimed at addressing the health status of the battery cell 102. The system 100 improves operational reliability by enabling timely maintenance or replacement actions through precise SOH assessments.
The data processing unit 110 generates actionable recommendations based on predefined thresholds of the assessed SOH value. For example, if the SOH value of the battery cell 102 falls below a critical level, the system 100 recommends immediate replacement to prevent potential failures. In another case, the system 100 may recommend routine maintenance if the SOH value indicates early signs of degradation. In applications involving electric vehicles, the data processing unit 110 allows recommendations account for the operational demands of the vehicle. For example, in a fleet of electric delivery vans, the system 100 may prioritize replacement actions for battery cells 102 in vehicles scheduled for long-distance routes for uninterrupted service.
In yet another embodiment, the data processing unit 110 refines the assessed SOH value based on one or more of a specific usage pattern, a charging current, a capacity of the battery cell 102, a charging mode and charging and discharging cycles. The term “refines” as used throughout the present disclosure relates to the process of enhancing the accuracy of the assessed SOH value by incorporating additional operational parameters. The SOH assessments are contextualized to reflect real-world usage conditions, thereby improving the precision of battery health evaluations.
The data processing unit 110 enhances the reliability of SOH assessments by accounting for factors that influence battery degradation. For example, specific usage patterns such as frequent high-discharge cycles may accelerate wear, which the system 100 incorporates into the refinement process. Similarly, the system 100 adjusts SOH values based on variations in charging current, such as fast-charging scenarios that induce higher stress on the battery cell 102. In applications involving electric vehicles, the SOH values reflect the unique operational profile of the vehicle. For example, in an electric taxi, the system 100 may refine SOH values by incorporating frequent stop-and-go driving patterns and high-capacity utilization, ensuring precise battery management tailored to the operational environment.
In an embodiment, the machine learning model 106 is trained using data from multiple battery cells associated with different battery chemistries. The term “different battery chemistries” as used throughout the present disclosure relates to variations in the composition and electrochemical properties of battery cells, such as lithium-ion, nickel-metal hydride and lead-acid. Training the machine learning model 106 with data from diverse chemistries enhances the adaptability of the system 100 by enabling accurate SOH predictions across a wide range of battery types.
The machine learning model 106 utilizes a training dataset comprising operational and degradation data from multiple battery cells associated with varying chemistries. For example, the training dataset includes parameters such as internal resistance trends, temperature variations and cycle count for each chemistry type. The machine learning model 106 captures unique degradation patterns inherent to each chemistry, which improves reliability thereof in assessing SOH for different applications. In electric vehicles, such adaptability allows the system 100 to monitor battery packs containing cells of distinct chemistries, such as lithium-ion cells for propulsion and nickel-metal hydride cells for auxiliary systems.
In another embodiment, the data processing unit 110 reassesses the SOH value based on charging cycle count data associated with the battery cell 102. The term “charging cycle count” as used throughout the present disclosure relates to the cumulative number of complete charge and discharge cycles experienced by the battery cell 102. Reassessing the SOH value accounts for cumulative usage effects, which improves the accuracy of battery health evaluations.
The data processing unit 110 incorporates charging cycle count data into the analysis to refine the SOH value of the battery cell 102. For example, if the battery cell 102 has undergone a higher-than-average number of charging cycles, the system 100 adjusts the SOH value to reflect the additional wear and tear. In electric vehicles, this functionality is particularly useful for applications such as fleet management, where vehicles experience varying duty cycles. For example, the system 100 may reassess the SOH of battery cells in an electric delivery van that completes multiple short-range trips daily for health assessments reflect real-world operational demands.
In yet another embodiment, the sensing unit 108 senses the internal resistance of the battery cell 102 as a DC resistance by applying a constant current and measuring a resulting voltage drop. The term “DC resistance” as used throughout the present disclosure relates to the resistance exhibited by the battery cell 102 during the application of a direct current. The precise measurement of the internal resistance can be performed by eliminating the influence of alternating current (AC) components, which improves the accuracy of SOH assessments.
The sensing unit 108 applies a constant current to the battery cell 102 and measures the resulting voltage drop across the terminals of the battery cell 102. The internal resistance is calculated using Ohm’s law, where the voltage drop is divided by the applied current. For example, in an electric vehicle, the sensing unit 108 can measure the DC resistance of individual cells in a battery pack during periodic maintenance cycles for accurate detection of early-stage degradation. The system 100 eliminates noise from external AC signals by focusing on DC resistance, providing more reliable input data for the data processing unit 110 and the machine learning model 106.
In an additional embodiment, the DC resistance is sensed by the sensing unit 108 using a pulse current method, wherein a short-duration current pulse is applied to the battery cell 102 and the voltage response is used to calculate the resistance. The term “pulse current method” as used throughout the present disclosure relates to a sensing technique that evaluates the transient response of the battery cell 102 to a brief current input. The system 100 improves the efficiency of resistance measurements by minimizing the impact on normal battery operation, as the pulse method requires only a short measurement window.
The sensing unit 108 applies a current pulse of known magnitude and duration to the battery cell 102 and records the instantaneous voltage change. The internal resistance is calculated based on the ratio of the voltage change to the applied current. For example, in an electric two-wheeler, the sensing unit 108 can perform resistance measurements during brief idle periods without interrupting the ongoing charging or discharging cycles for integration into real-time monitoring. Such a method is particularly advantageous for high-capacity battery packs, as the method reduces the time required for sequential resistance measurements across multiple cells.
In an embodiment, the sensing unit 108 dynamically measures DC resistance during the operation of the battery cell 102 without interrupting ongoing charging or discharging cycles. The term “dynamically measures” as used throughout the present disclosure relates to the continuous or periodic measurement of DC resistance in real-time while the battery cell 102 is in use. The uninterrupted operation by integrating resistance measurement into normal charging or discharging processes, enhances monitoring efficiency and avoids disruption to the battery cell's functionality.
The sensing unit 108 measures DC resistance by monitoring current and voltage variations during operational phases of the battery cell 102. For example, during charging, the sensing unit 108 applies a minor current adjustment to observe the corresponding voltage response and calculates the DC resistance based on the observed parameters. Similarly, during discharging, the sensing unit 108 uses transient responses to determine resistance without affecting the energy delivery process. In electric vehicles, this functionality provides real-time monitoring of battery health across all cells in a battery pack, allowing timely detection of anomalies without requiring operational pauses. For example, in an electric taxi fleet, the system 100 dynamically measures resistance while vehicles are actively in use, enables uninterrupted service while maintaining accurate SOH assessments.
In another embodiment, the machine learning model 106 is trained using historical DC resistance data collected from multiple battery cells associated with the same chemistry, capacity and usage pattern. The term “historical DC resistance data” as used throughout the present disclosure relates to previously recorded resistance measurements and associated parameters that reflect the operational history and degradation trends of battery cells. Training the machine learning model 106 with such data customizes predictions to specific battery configurations and operational scenarios, improving the accuracy of SOH assessments.
The machine learning model 106 utilizes a dataset comprising historical resistance measurements, SOH values, and usage-specific parameters such as charging rates and discharge depths. For example, the system 100 trains the model using data from lithium-ion battery cells of a specific capacity used in electric scooters, where frequent stop-and-go patterns influence degradation. The machine learning model 106 captures nuances unique to specific configurations by focusing on a consistent chemistry, capacity and usage pattern, resulting in more reliable predictions. Such a functionality is particularly beneficial for electric vehicle manufacturers managing battery packs with standardized specifications, as the functionality allows for precise monitoring and maintenance strategies across fleets.
Referring to FIG. 2, there is shown a graph 200 illustrating a relationship between the sensed internal resistance versus SOH for a battery cell (such as the battery cell 102 of FIG. 1), in accordance with an embodiment of the present disclosure. The graph 200 illustrates two plotted curves: a decreasing curve representing the SOH and an increasing curve representing the sensed internal resistance of the battery cell. The x-axis represents the cycle number, while the y-axis on the left corresponds to the SOH values and the y-axis on the right corresponds to the internal resistance values.
The graph 200 highlights that as the cycle number increases, the SOH of the battery cell gradually decreases due to progressive degradation of the battery cell. Conversely, the internal resistance of the battery cell increases with the number of cycles, indicating the corresponding degradation effects such as electrode ageing and electrolyte decomposition. Such an inverse correlation between SOH and internal resistance forms the basis for using internal resistance as a reliable predictor for SOH assessments.
The system 100 leverages such a relationship to assess the SOH value of the battery cell with improved accuracy by utilizing the sensed internal resistance as a primary input parameter. For example, during an operational scenario such as long-term usage of an electric vehicle, the graph 200 provides insights into the progressive decline in battery health, enabling the system 100 to generate predictive alerts or maintenance recommendations based on observed trends.
Referring now to FIG. 3, there is shown a scatter plot 300 illustrating a relationship between actual and predicted values of SOH of a battery cell (such as the battery cell 102 of FIG. 1), in accordance with an embodiment of the present disclosure. The x-axis of the scatter plot 300 represents the actual SOH values, while the y-axis represents the predicted SOH values as calculated by the machine learning model 106.
The scatter plot 300 includes a series of data points distributed along a reference line depicted as a dashed red line. The reference line represents the ideal case where the predicted values match the actual values perfectly. The proximity of the data points to the reference line indicates the accuracy of the machine learning model 106 in predicting the SOH values of the battery cell 102.
The scatter plot 300 demonstrates that the predicted SOH values align closely with the actual SOH values, indicating that the machine learning model 106 provides reliable and precise predictions. Such accuracy aids the system 100 to make assessments of battery health and operational status. For example, in an electric vehicle, this level of precision allows the system 100 to detect early signs of degradation and recommend timely maintenance or replacement of the battery cell 102, thus optimal performance and reliability.
In the description of the present invention, it is also to be noted that, unless otherwise explicitly specified or limited, the terms “disposed,” “mounted,” and “connected” are to be construed broadly, and may for example be fixedly connected, detachably connected, or integrally connected, either mechanically or electrically. They may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meaning of the above terms in the present invention can be understood in specific cases to those skilled in the art.
Modifications to embodiments and combination of different embodiments of the present disclosure described in the foregoing are possible without departing from the scope of the present disclosure as defined by the accompanying claims. Expressions such as “comprising”, “comprising”, “incorporating”, “have”, “is” used to describe and claim the present disclosure are intended to be construed in a non- exclusive manner, namely allowing for items, components or elements not explicitly described also to be present. Reference to the singular is also to be construed to relate to the plural where appropriate.
Although embodiments have been described with reference to a number of illustrative embodiments thereof, it should be understood that numerous other modifications and embodiments can be devised by those skilled in the art that will fall within the spirit and scope of the principles of this disclosure. More particularly, various variations and modifications are possible in the component parts and/or arrangements of the subject combination arrangement within the scope of the present disclosure, the drawings and the appended claims. In addition to variations and modifications in the component parts and/or arrangements, alternative uses will also be apparent to those skilled in the art.
,CLAIMS:WE CLAIM:
1. A system (100) for assessing a state of health (SOH) value of a battery cell (102), the system (100) comprising:
- a memory unit (104) to store a machine learning model (106);
- sensing unit (108) to sense an internal resistance of the battery cell (102), wherein the sensed internal resistance is indexed with a state of charge (SOC) level of the battery cell (102) at a time of the sensing of the internal resistance;
- a data processing unit (110) operatively connected to the memory unit (104) and the sensing unit (108), wherein the data processing unit (110):
- inputs the sensed internal resistance into the machine learning model (106); and
- assesses the SOH value of the battery cell (102) based on the inputted internal resistance and predefined relationships modelled by the machine learning model (106); and
- an output module (112) operatively connected to the data processing unit (110), wherein the output module (112) displays or transmits the computed SOH value to a user interface or external computing arrangement.
2. The system (100) as claimed in claim 1, wherein the machine learning model (106) is a gradient-boosting-based machine learning model (106) and wherein the gradient-boosting-based machine learning model (106) is selected from: XGBoost, LightGBM, CatBoost, HistGradientBoostingClassifier, GradientBoostingClassifier, NGBoost or AdaBoost.
3. The system (100) as claimed in claim 1, wherein the memory unit (104) comprises a data repository to store historical data associated with the battery cell (102) for iteratively updating the machine learning model (106).
4. The system (100) as claimed in claim 1, wherein the sensing unit (108) comprises a temperature sensor to sense a temperature of the battery cell (102) during sensing of the internal resistance and wherein the sensed temperature is used by the data processing unit (110) to recalibrate the assessed SOH value.
5. The system (100) of claim 1, wherein the data processing unit (110) computes an estimated remaining useful life (RUL) of the battery cell (102) based on the assessed SOH value and historical degradation trends.
6. The system (100) of claim 1, wherein the data processing unit (110) recommends an action for battery maintenance or replacement based on the assessed SOH value.
7. The system (100) of claim 1, wherein the data processing unit (110) refines the assessed SOH value based on one or more of: a specific usage pattern, a charging current, a capacity of the cell, a charging mode, charging and discharging cycles.
8. The system (100) as claimed in claim 1, wherein the machine learning model (106) is trained using data from multiple battery cell (102)s associated with different battery chemistries.
9. The system (100) as claimed in claim 1, wherein the data processing unit (110) reassesses the SOH value based on a charging cycle count data associated with the battery cell (102).
10. The system (100) of claim 1, wherein the sensing unit (108) senses the internal resistance of the battery cell (102) as a DC resistance by applying a constant current and measuring a resulting voltage drop.
11. The system (100) of claim 10, wherein the DC resistance is sensed by the sensing unit (108) using a pulse current method and wherein a short-duration current pulse is applied to the battery cell (102) and the voltage response is used to calculate the resistance.
12. The system (100) of claim 10, wherein the sensing unit (108) dynamically measures DC resistance during operation of the battery cell (102) without interrupting ongoing charging or discharging cycles of the battery cell (102).
13. The system (100) of claim 10, wherein the machine learning model (106) is trained using historical DC resistance data collected from multiple battery cell (102)s associated with same chemistry, capacity and usage pattern.