A battery status assessment system

Abstract

The present invention provides a battery state evaluation system, including a characteristic acquisition unit, a scenario prediction unit, a state prediction unit, a health state prediction unit and a communication unit. The characteristic acquisition unit receives and processes passive and active data to generate shape variables and improved variables; the scenario prediction unit uses an artificial intelligence model to predict the battery usage mode; the state prediction unit combines various variables, dynamically outputs the internal variables of the secondary battery cell and updates the battery state model; the health state prediction unit evaluates the health and charging level of the battery; and the communication unit transmits the results to the user or system in real time. The system realizes accurate monitoring and efficient management of the battery state.

Description

Technical Field

[0001] The present invention relates to the field of battery evaluation, and in particular to a battery status evaluation system. Background Art

[0002] With the continuous growth of global energy demand and the enhancement of environmental awareness, the importance of battery technology in modern energy storage and electrification has become increasingly prominent. The performance and life of batteries not only directly affect the operating efficiency of various applications such as electric vehicles, renewable energy storage systems and portable electronic devices, but also have a profound impact on user experience and energy efficiency. Therefore, how to accurately evaluate and predict the state and health of batteries has become a core issue in the research of battery management systems (BMS).

[0003] Traditional battery management systems (BMS) rely on monitoring external battery parameters, such as voltage, temperature, and current, and use these parameters in combination with empirical models to predict battery status. However, this approach has significant limitations in practical applications, mainly in the following aspects:

[0004] First, traditional methods rely too much on the external parameters of the battery, making it difficult to directly observe and reflect the physical and chemical processes inside the battery. During the operation of the battery, the internal electrochemical reactions are complex and changeable, especially under high-power discharge, fast charging or drastic temperature changes. The chemical reaction rate, electrolyte conductivity, internal resistance, etc. inside the battery will change significantly. However, external measurement data such as voltage and temperature can only reflect part of the results of these processes, but cannot directly reveal the internal state changes. This lack of information leads to greater uncertainty in the state prediction of traditional empirical models.

[0005] Secondly, in complex usage environments, traditional prediction methods based on external measurement data are difficult to maintain stable prediction accuracy. For example, in the case of sudden temperature changes or high-power operations, the chemical processes inside the battery may change dramatically, affecting the overall performance of the battery. Traditional empirical models are trained based on data under specific conditions. When the usage environment deviates from the range covered by the training data, the prediction results of the model may have large deviations, or even error accumulation, which will directly affect the effectiveness of the battery management system and fail to provide users with reliable decision support.

[0006] In recent years, researchers have proposed a battery state estimation method based on a circuit equivalent model. This method provides more detailed and accurate battery state information by simulating the physical and chemical processes inside the battery. The circuit equivalent model can characterize the dynamic behavior of the battery through a series of parameters related to the physical properties of the battery, including but not limited to electrolyte resistance, double layer capacitance, diffusion impedance, etc.

[0007] The circuit equivalent model simulates the dynamic behavior of the battery by simplifying the complex physical and chemical processes inside the battery into several equivalent circuit elements related to physical quantities such as current, voltage and impedance. For example, the classic RC model simplifies the battery into a circuit consisting of resistors and capacitors in series and parallel, which can reflect the charge and discharge characteristics of the battery to a certain extent. The extended electrochemical impedance model (EIS) obtains the impedance information of each component inside the battery by measuring the battery's response to an AC signal within a specific frequency range, thereby providing a more accurate state assessment.

[0008] Although the battery state estimation method based on the circuit equivalent model has high accuracy and applicability in theory, its practical application still faces a series of challenges. First, the parameters of the circuit equivalent model often need to be calibrated in a complex manner for different types of batteries and their different working conditions. Each battery has its own unique physical and chemical properties, and the model parameters may also change under different working conditions. Therefore, in practical applications, the precise parameterization and calibration of the circuit equivalent model is complex and time-consuming, which increases the difficulty and cost of the model application.

[0009] Secondly, the physical and chemical processes inside the battery are extremely complex. As the battery is used for a long time, the aging effect and the degradation of the battery materials will cause its internal state to change significantly. Although the circuit equivalent model can capture these changes to a certain extent, the applicability and accuracy of the model are still limited under long-term use and extreme conditions. In addition, in actual operation, the battery system is often affected by external environmental factors such as temperature, humidity and mechanical vibration, causing the model's prediction results to deviate from the actual situation. Summary of the invention

[0010] In order to solve the above problems in the prior art, the present invention proposes a battery state evaluation system, including a characteristic acquisition unit, a scene prediction unit, a state prediction unit, a health state prediction unit and a communication unit;

[0011] The characteristic acquisition unit receives input data, the input data including passive data collected from passive measurements of the battery and active data collected from active measurements of the battery in response to a stimulus signal applied to the battery;

[0012] The characteristic acquisition unit performs variable identification based on shape characteristics based on input data related to the circuit equivalent model to acquire shape variables; performs variable identification based on improvement based on input data related to the circuit equivalent model to acquire improved variables;

[0013] The scenario prediction unit predicts the potential usage pattern of the battery using an artificial intelligence model based on the battery usage scenario and historical data, and generates a usage pattern variable;

[0014] The state prediction unit combines the shape variable and the improvement variable through a determination algorithm, and dynamically adjusts the weight and determination strategy of the determination algorithm according to the usage mode variable to output the secondary battery cell internal variable, wherein the secondary battery cell internal variable includes at least a fixed phase element coefficient, an electrolyte impedance, and a current transfer impedance;

[0015] The state prediction unit updates the internal state model of the battery according to the internal variables of the secondary battery cell;

[0016] The health state prediction unit is used to parse the internal state model to determine the current health state of the battery, including the health and charge level of the battery;

[0017] The communication unit is used to send one or more health prediction values ​​and charge level prediction values ​​to a user, a related computing system, or a combination of the two.

[0018] The passive data includes voltage, current and temperature data generated by the battery during normal operation, and the active data is obtained by applying a stimulation signal to the battery and measuring the battery response, wherein the stimulation signal is a current or voltage signal with a certain amplitude and frequency.

[0019] The shape variables include the slope and peak amplitude of the voltage response curve; the improvement variables include the internal electrolyte impedance and double-layer capacitance of the battery.

[0020] The scenario prediction unit predicts the battery usage pattern through an artificial intelligence model, and the artificial intelligence model is a long short-term memory network model. The training process of the long short-term memory network model is based on a historical data set and is trained by a supervised learning method, using mean square error as a loss function to optimize the model's prediction accuracy.

[0021] The usage pattern variables generated by the scenario prediction unit include expected temperature changes, load change patterns, and charge and discharge rates, and the usage pattern variables are output to the state prediction unit or other modules.

[0022] The decision algorithm initially assigns weights to shape variables, improvement variables, and usage mode variables based on the importance of the variables and historical data. The initially assigned weights can be expressed by the following formula:

[0023]

[0024] in, represents the weight of the i-th variable at time t, is the prediction accuracy of the variable at time t-1; : represents the sum of the prediction accuracy of all n variables at time t-1;

[0025] The determination algorithm dynamically adjusts the weight of each variable by analyzing the error between the actual prediction result and the measured data; the adjustment formula is:

[0026]

[0027] in, is the learning rate, L is the loss function, and the weights are adjusted by minimizing L ; is the updated weight of the i-th variable at time t+1; is the current weight of the loss function L on the i-th variable The partial derivative of

[0028] The determination algorithm generates secondary cell internal variables by determining and measuring shape variables, improvement variables and usage mode variables according to current weights; the determination strategy is expressed as:

[0029]

[0030] in, is the internal variable of the secondary cell after determination, is the ith input variable, is a nonlinear combination function; n represents the total number of input variables; m represents the number of nonlinear combination functions.

[0031] The basic form of the state equation of the internal state model of the updated battery is:

[0032]

[0033] Represents state variables, including the battery's state of charge (SOC) and state of health (SOH);

[0034] Represents state variables The derivative with respect to time t;

[0035] u(t) is the input control variable, including the external control input to the battery;

[0036] The internal variables of the secondary battery cell after determination include fixed phase element coefficient, electrolyte impedance, and current transfer impedance;

[0037] function It is a nonlinear function, reflecting the dynamic response of the battery;

[0038] function Then through the internal variables of the secondary battery Modify the model.

[0039] The health status includes a health level determined by analyzing the battery's degradation characteristics, internal resistance changes, and electrolyte degradation, and a charge level estimated based on the battery's voltage, current, and capacity.

[0040] The communication unit supports wireless communication and wired communication protocols and is integrated with a battery management system or a user terminal.

[0041] The battery status evaluation system is applied to energy storage power stations, electric vehicles, and electronic equipment.

[0042] Beneficial effects:

[0043] The battery status assessment system of the present invention can accurately monitor and predict the health status and charge level of the battery, thereby greatly improving the effectiveness and accuracy of battery management. By collecting passive and active data through the characteristic acquisition unit, and combining the circuit equivalent model, the system can generate shape variables and improvement variables that reflect the internal state of the battery. Using the artificial intelligence model of the scenario prediction unit, the system can predict the potential usage mode of the battery, and dynamically adjust the judgment algorithm through the state prediction unit to generate key secondary battery cell internal variables. Based on these variables, the health status prediction unit can accurately assess the health and charge level of the battery, and finally transmit it to the user or related system in real time through the communication unit, thereby realizing real-time monitoring and efficient management of the battery status, extending the battery life and improving the reliability of system operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present application, but do not constitute an improper limitation of the present invention. In the drawings:

[0045] Figure 1 It is a schematic diagram of the overall structure of the battery status evaluation system of the present invention.

[0046] Figure 2 It is a schematic diagram of data interaction among modules of the battery status assessment system of the present invention. DETAILED DESCRIPTION

[0047] The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments, wherein the illustrative embodiments and descriptions are only used to explain the present invention but are not intended to limit the present invention.

[0048] Example 1: Structure of a battery status evaluation system

[0049] like Figure 1 As shown, this embodiment provides a battery status evaluation system 100, which includes a characteristic acquisition unit 101, a scenario prediction unit 102, a status prediction unit 103, a health status prediction unit 104 and a communication unit 105. These units work together to achieve real-time monitoring of the battery status and accurate evaluation of the health status.

[0050] Example 1.1 Characteristic acquisition unit

[0051] The characteristic acquisition unit 101 is used to receive input data, which includes passive measurement data and active measurement data. Passive measurement data is mainly obtained by monitoring parameters such as voltage, current, temperature, etc. generated by the battery during normal operation. Active measurement data is obtained by applying a specific stimulation signal to the battery and measuring the battery response. The characteristic acquisition unit performs variable identification based on shape characteristics based on input data related to the circuit equivalent model to obtain shape variables; at the same time, it also performs variable identification based on improvement to output improved variables. These variables can more accurately reflect the internal state of the battery and provide basic data for subsequent state evaluation. The specific implementation method is as follows:

[0052] Passive measurement data refers to data obtained by monitoring the battery's voltage, current, temperature and other parameters during normal operation of the battery. These data do not require external stimulation of the battery, so they are called passive measurement data. In order to ensure the accuracy and real-time nature of the data, the characteristic acquisition unit 101 is equipped with a high-precision sensor network, including but not limited to:

[0053] Voltage sensor: used to monitor the voltage of each battery cell in real time. The sampling rate of the sensor should be high enough to capture transient changes in the battery operation process. The voltage sensor can be directly connected to the positive and negative electrodes of the battery to achieve measurement, and the measurement error is required to be no more than ±0.1%.

[0054] Current sensor: used to monitor the battery's charge and discharge current. To accommodate different current ranges, the current sensor can be a Hall effect sensor or a shunt resistor sensor, depending on the battery's power level and application scenario. The sensor's measurement range should cover the battery's maximum charge and discharge current, and the accuracy requirement should be no more than ±0.5%.

[0055] Temperature sensor: used to monitor the operating temperature of the battery, installed on the battery surface or in key locations inside, such as near the battery cell. The temperature sensor can be a thermocouple, thermistor or integrated temperature sensor chip, and the measurement accuracy is required to be within ±0.5°C. These temperature data are crucial for analyzing the performance of the battery thermal management system.

[0056] These passive measurement data are collected through the real-time data acquisition module of the characteristic acquisition unit. After digital processing, the collected data are transmitted to the data processing module for further analysis and variable identification.

[0057] Active measurement data is obtained by applying specific external stimulus signals to the battery and monitoring the battery's response to these signals. This measurement method can provide more in-depth information about the internal state of the battery, especially the analysis of the internal impedance characteristics of the battery.

[0058] Generation of stimulus signals: The stimulus signals are current or voltage signals of a certain amplitude and frequency, which are used to stimulate the dynamic response of the battery. The characteristic acquisition unit generates these stimulus signals through a built-in signal generator, which can output AC signals ranging from millivolts to several volts and frequencies from millihertz to kilohertz to meet the measurement requirements of different types of batteries.

[0059] Response signal measurement: After applying the stimulus signal, the characteristic acquisition unit monitors the battery response signal through a high-precision data acquisition system. The response signal includes the battery voltage change, current change and phase difference, which are the basis for analyzing battery impedance and other electrochemical characteristics.

[0060] Data processing and variable identification: The response signal received by the characteristic acquisition unit is filtered, amplified and digitized before entering the variable identification module. The variable identification module first identifies the shape characteristics and extracts shape variables related to the battery's electrochemical process, such as the slope of the voltage response curve and the peak amplitude. These shape variables reflect the transient response characteristics of the battery under the stimulus signal and can be used to evaluate the battery's health and performance.

[0061] In addition, the variable identification module also performs variable identification based on the improved algorithm, and calculates the improved variables through more complex mathematical models (such as extended electrochemical impedance spectroscopy analysis). These improved variables may include more in-depth physical and chemical parameters such as the internal electrolyte impedance and double-layer capacitance of the battery. This process involves mathematical tools such as Fourier transform and linear regression analysis. The specific algorithm can be described as:

[0062]

[0063] in, is the total impedance of the battery, is the ohmic resistance of the battery, is the charge transfer resistance, is the double layer capacitance, for impedance, is the angular frequency of the signal, and j is the imaginary unit.

[0064] By fitting the parameters of the above formula, the characteristic acquisition unit can extract more accurate improved variables, which directly reflect the electrochemical characteristics inside the battery.

[0065] After completing the identification of shape variables and improved variables, the characteristic acquisition unit makes preliminary judgments and sorts these variables to generate a basic data set. This data set includes both simple measurement data (such as voltage and temperature) and advanced variables (such as impedance spectrum parameters) after complex identification processing. Finally, the characteristic acquisition unit transmits these data to the state prediction unit or other modules to provide detailed basic data for battery state assessment and health prediction.

[0066] The characteristic acquisition unit can fully and accurately reflect the internal state and operating characteristics of the battery through a combination of passive and active measurements, providing strong data support for the battery management system. This design ensures high accuracy and reliability of the system and is suitable for various battery types and application scenarios.

[0067] Example 1.2 Scene prediction unit

[0068] The scenario prediction unit 102 uses an artificial intelligence model to predict the potential usage mode of the battery based on the battery usage scenario and historical data, and generates usage mode variables. The battery usage scenario may include various environmental parameters such as temperature, load, current, etc., and the historical data comes from the long-term operation history of the battery. By analyzing these data, the scenario prediction unit can identify the behavior characteristics of the battery under different usage conditions and predict its future usage mode.

[0069] The scenario prediction unit 102 is a key component for predicting the potential battery usage mode and generating usage mode variables using an artificial intelligence model based on the battery usage scenario and historical data. The following is a detailed description of the technical implementation of the scenario prediction unit.

[0070] The scene prediction unit 102 first collects various data related to battery use through the data acquisition module. The main data sources include:

[0071] Environmental parameter data: This data set includes parameters such as temperature, humidity, and pressure in the battery operating environment. These data are collected in real time through sensors installed around or inside the battery. Temperature sensors can measure temperature changes on the surface and inside of the battery, humidity sensors monitor the humidity level of the surrounding air, and pressure sensors can detect changes in the external pressure of the battery. These parameters are crucial for evaluating the performance of the battery in different environments.

[0072] Operation parameter data: including the battery’s current, charge and discharge rate, power output, etc. during use. These data are collected through current sensors, power meters and other equipment, and can directly reflect the load condition and working intensity of the battery in actual applications.

[0073] Historical data: Historical data is derived from the long-term operation records of the battery and is saved by the battery management system (BMS). These data cover the use of the battery in different time periods, including the number of past charge and discharge cycles, the maximum and minimum temperatures experienced, and the performance under different load conditions. The integrity and accuracy of historical data have an important impact on the training of the prediction model and the reliability of the prediction results.

[0074] After data collection is completed, the scene prediction unit preprocesses the data. The preprocessing steps include data cleaning, normalization and feature extraction:

[0075] Data cleaning: Eliminate noise and outliers in the data to ensure the accuracy of the input data. For example, remove extreme values ​​caused by sensor failure or fill in data lost due to data transmission interruption.

[0076] Data normalization: Due to the large differences in the magnitude of different data types, the data needs to be normalized to unify all data into a relative range to facilitate model training and prediction.

[0077] Feature extraction: Extract representative feature variables from the raw data, such as temperature change rate, volatility of charge and discharge rates, number of cycles in historical data, etc. These feature variables will be used as input to the artificial intelligence model to improve prediction accuracy.

[0078] The scenario prediction unit establishes a model for predicting battery usage patterns by using artificial intelligence (AI) technology. Artificial intelligence models include neural network models, support vector machines (SVMs), random forests, etc. In this embodiment, a deep learning model such as a long short-term memory network (LSTM) is preferably used because it can process time series data and has significant advantages in predicting long-term trends.

[0079] Model architecture: The LSTM model consists of multiple recursive layers and fully connected layers, which can effectively capture the time dependency in the data. The input layer accepts the preprocessed feature variables, and after calculations by several LSTM units, extracts the time series features of the data and finally outputs the prediction results. The output of the model is the usage pattern variables, which describe the possible usage of the battery in the future.

[0080] Training process: The training process of the model uses historical data as the training set and supervised learning methods for training. Specifically, the model inputs the feature variables in the historical data and outputs the known usage patterns corresponding to these features. By calculating the error between the predicted results and the actual usage patterns and adjusting the model parameters through back propagation, the prediction accuracy of the model is gradually improved. Loss functions such as mean square error (MSE) can be used to evaluate the performance of the model.

[0081] After being fully trained, the scenario prediction unit can generate battery usage pattern variables based on the real-time collected environmental parameters and operating parameters. These usage pattern variables include the following:

[0082] Estimated temperature variation: Based on current temperature and historical trends, predict the range of temperature variations that the battery may experience during future use.

[0083] Load change pattern: predicts the future load conditions of the battery, including the switching frequency between high load and low load, the average load level, etc. These variables can reflect the working intensity of the battery in actual applications.

[0084] Charge and discharge rate: Predict the charge and discharge rate change trend of the battery in future cycles to help evaluate the battery's charging needs and discharge capacity.

[0085] These usage pattern variables are output to the state prediction unit 103 or other modules, providing important reference for battery health state prediction and life assessment.

[0086] The scenario prediction unit has the function of real-time update and dynamic prediction. As new data is continuously input, the model can automatically adjust its prediction results to ensure that the output usage pattern variables always reflect the latest status and usage trends of the battery. By regularly retraining the model (online learning), the system can adapt to the gradual decline of battery performance and changes in environmental conditions, thereby improving the long-term effectiveness and reliability of the prediction.

[0087] In short, the scenario prediction unit can efficiently and accurately predict the future usage pattern of the battery through sophisticated data processing and advanced artificial intelligence models, providing key predictive variables for the battery management system. This technical design ensures the efficiency and adaptability of the system and can be widely used in different types of batteries and complex usage scenarios.

[0088] Example 1.3 State prediction unit

[0089] The state prediction unit 103 combines the shape variables, the improvement variables and the usage mode variables through a determination algorithm. The determination algorithm dynamically adjusts the weights and determination strategies according to the actual situation, thereby outputting the internal variables of the secondary battery cell. The internal variables of the secondary battery cell include at least the fixed phase element coefficient, the electrolyte impedance and the current transfer impedance, which can accurately reflect the internal state of the battery. Based on these internal variables of the secondary battery cell, the state prediction unit further updates the internal state model of the battery to provide an accurate prediction of the battery state.

[0090] The state prediction unit 103 plays a key role in the battery state evaluation system. It processes variables from different sources through a determination algorithm, thereby outputting secondary cell internal variables that reflect the internal state of the battery. The specific implementation scheme of the state prediction unit will be described in detail below.

[0091] The determination algorithm used by the state prediction unit 103 is a multi-source data determination technology that can comprehensively analyze the shape variables, improvement variables from the characteristic acquisition unit, and the usage mode variables generated by the scenario prediction unit. The core of the determination algorithm is to dynamically adjust the weight of each variable according to the actual working conditions, and determine different variables through adaptive strategies to output the secondary battery cell internal variables that best reflect the current internal state of the battery.

[0092] Shape variables: variables extracted by the characteristic acquisition unit based on the response curve characteristics of the battery, such as voltage response slope, peak current, etc. These variables reflect the transient behavior of the battery under different operating conditions.

[0093] Improved variables: Variables based on more in-depth physical and chemical models, such as electrolyte impedance, double-layer capacitance, etc. These variables provide more detailed information about the electrochemical processes inside the battery.

[0094] Usage pattern variables: variables predicted by the scenario prediction unit based on battery usage scenarios and historical data, such as expected temperature changes, load patterns, etc. These variables provide the decision algorithm with information about the possible future operating conditions of the battery.

[0095] The core of the decision algorithm is a weighted decision strategy based on multiple variables. The algorithm uses the following steps to process:

[0096] 1. Weight initialization: The weights initially assigned to shape variables, improvement variables, and usage pattern variables are calculated based on the importance of the variables and historical data. These weights can be expressed by the following formula:

[0097]

[0098] in, : represents the weight of the i-th variable at time t. This weight is used to measure the importance or influence of the variable at the current moment.

[0099] : represents the prediction accuracy of the i-th variable at time t-1. The prediction accuracy reflects the prediction effect or performance of the variable at the previous moment. The variable with higher accuracy may be assigned a larger weight at the next moment.

[0100] : represents the sum of the prediction accuracy of all n variables at time t-1. This term is used to normalize the prediction accuracy of each variable in order to calculate the weight of each variable at the current moment.

[0101] Through this formula, the weight at the current moment The prediction accuracy of the previous moment Determined, it reflects the influence of each variable relative to other variables at the current moment.

[0102] 2. Adaptive adjustment: As new data is input, the decision algorithm dynamically adjusts the weights of each variable by analyzing the error between the actual prediction results and the measured data. The adjustment formula can be implemented through methods such as gradient descent or Bayesian updating:

[0103]

[0104] in, : represents the updated weight of the i-th variable at time t+1. This weight is obtained by adjusting based on the new data input at time t.

[0105] : represents the current weight of the i-th variable at time t. This weight is used before time t+1 and is dynamically adjusted with the input of new data and analysis of errors.

[0106] : represents the learning rate, which controls the step size of each weight adjustment. The learning rate determines the magnitude of the update.

[0107] : represents the current weight of the loss function L on the i-th variable The partial derivative of . It measures the impact of the current weight on the loss function. Through this partial derivative, we can determine the contribution of this variable to the overall error.

[0108] L: represents the loss function, usually the mean square error (MSE) or other functions related to the prediction error. The goal of the loss function L is to measure the error between the predicted results and the actual measured data, and to optimize the model performance by minimizing L.

[0109] 3. Execution of the decision strategy: The decision algorithm generates the internal variables of the secondary battery cell by weighted summation or nonlinear combination of shape variables, improvement variables and usage mode variables according to the current weights. The specific decision strategy is expressed as:

[0110]

[0111] in, : Represents the internal variable of the secondary cell after judgment. This variable is the final result calculated by the judgment algorithm, based on the input variables and their weighted sum or nonlinear combination.

[0112] : represents the i-th input variable The weight reflects the importance of the variable in the decision process, and the value of the weight depends on the adaptive adjustment process in the previous step.

[0113] : represents the i-th input variable. These input variables can be shape variables, improvement variables or usage mode variables, representing different feature data, and are input into the judgment algorithm.

[0114] : indicates input variable The partial results are generated after weighted summation. The weighted summation is the value of each input variable. Multiply by its weight The sum of the variables is used to determine their linear contribution to the final decision variable.

[0115] : indicates input variable The nonlinear combination part of . It is a nonlinear combination function, which can be a polynomial, a neural network, etc. They are used to capture the complex and nonlinear relationships between input variables.

[0116] : represents the jth nonlinear combination function. These functions process the input variables nonlinearly and can reveal more complex interactions between variables. Functions may represent different nonlinear processing methods, such as polynomial combinations, neural networks, or other nonlinear mapping methods.

[0117] The two parts of this formula - the linear weighted summation and the nonlinear combination - together determine the final secondary cell internal variables , thereby more comprehensively reflecting the changes in the internal state of the battery.

[0118] n: represents the total number of input variables, usually including shape variables, improvement variables, and usage mode variables. This means that the first part of the formula Perform a weighted sum of n input variables.

[0119] m: represents the number of nonlinear combination functions. This means that the second part of the formula In the above example, m nonlinear combination functions are summed, each function Perform nonlinear processing on n input variables.

[0120] In short, n refers to the number of input variables involved in the linear weighted summation part, while m refers to the number of nonlinear combination functions used to generate the internal variables of the secondary battery cell.

[0121] Through the judgment algorithm, the state prediction unit can output a set of key secondary battery internal variables. These variables include but are not limited to:

[0122] Fixed phase element coefficient: A constant used to characterize the internal polarization process of the battery. This coefficient has an important influence on the current response characteristics of the battery.

[0123] Electrolyte impedance: reflects the resistance characteristics of the electrolyte inside the battery. This parameter can reveal the changes in the conductivity of the battery under different temperature and current conditions.

[0124] Current transfer impedance: describes the charge transfer impedance between the electrode surface and the electrolyte interface. This variable is crucial to evaluating the dynamic response characteristics of the battery.

[0125] These secondary cell internal variables are output in real time through the state prediction unit and used to further update the internal state model of the battery.

[0126] The state prediction unit uses the internal variables of the secondary battery cells generated above to dynamically update the internal state model of the battery. This model is the core part of the battery management system for predicting the overall health status and operating performance of the battery. The specific model update process is as follows:

[0127] In the battery management system, the internal state model uses nonlinear differential equations or state space models to describe the physical and chemical processes inside the battery. These models can reflect the dynamic behavior and state changes of the battery under different operating conditions. In order to ensure that the model can accurately reflect the actual situation of the battery, the internal variables of the secondary battery cell are introduced into the state equation, and the model is corrected and updated through these variables.

[0128] The basic form of the state equation can be expressed as:

[0129]

[0130] Represents state variables, including the battery's state of charge (SOC), state of health (SOH), etc., which describe the internal state of the battery at a specific time t.

[0131] Represents state variables The derivative with respect to time t, that is The rate of change over time. It is often used to describe the dynamic behavior or rate of change of a system.

[0132] Specifically in the battery management system, Represents the rate of change of the internal state of the battery (such as the state of charge SOC or the state of health SOH) over time t, reflecting the dynamic response of the battery. For example:

[0133] x(t) can represent the current state of charge (SOC) or state of health (SOH) of the battery.

[0134] It indicates how these states change over time, such as the rate of increase or decrease of the state of charge, the rate of degradation of the health state, etc.

[0135] u(t) is the input control variable, which represents the external control input of the battery, such as charge and discharge current, ambient temperature, etc. These variables directly affect the dynamic behavior of the battery.

[0136] To determine the internal variables of the secondary battery cell, key parameters such as fixed phase element coefficient, electrolyte impedance, current transfer impedance, etc. are included. These parameters directly reflect the physical and chemical properties inside the battery and change under different operating conditions.

[0137] In the state equation, the function Describes the state variables With input variables The relationship between is a nonlinear function that reflects the dynamic response of the battery. Function Then through the internal variables of the secondary battery The model is modified so that it can more accurately reflect the actual state of the battery. For example, when the electrolyte impedance increases, function B will reflect the effect of increased internal resistance of the battery, thus affecting the overall performance prediction of the battery.

[0138] By modifying the state equation, the system can update the internal state model of the battery in real time to keep it consistent with the actual operating conditions, ensuring that the model can accurately predict the state changes of the battery under different operating conditions.

[0139] In the battery management system, state prediction is a continuous process. The system needs to predict the next state of the battery at each time step. In order to improve the accuracy of state prediction, methods such as Kalman filtering or particle filtering are used to correct the prediction results.

[0140] The corrected state prediction can be expressed as:

[0141]

[0142] is the revised state prediction, which represents the system's prediction result of the battery state at time t+1.

[0143] Based on the current state prediction and input variables The calculated state update function reflects the dynamic changes of the battery.

[0144] It is the internal variable of the secondary battery The correction term for the state combines various internal parameters of the battery through the function Fine-tune the model to make the predictions more consistent with actual conditions.

[0145] is the gain matrix, which is used to adjust the difference between the model prediction and the actual observation. The gain matrix is ​​dynamically calculated through algorithms such as Kalman filtering or particle filtering to ensure that the correction process can effectively reduce the prediction error.

[0146] is the actual observed value at time t, measured directly by the sensor.

[0147] is the predicted observation value at time t, based on the current state prediction and models for calculation.

[0148] In this correction process, the system first calculates the predicted state and predicted observation value based on the current state and input variables, and then compares the predicted observation value with the actual observation value. Correct the predicted state. This correction process can significantly improve the accuracy of the state prediction, especially in the face of environmental changes or nonlinear dynamic behavior. The corrected prediction can better reflect the actual state of the battery.

[0149] Through the above steps, the state prediction unit can continuously and accurately track the health of the battery and adjust the management strategy according to the latest prediction results, thereby effectively extending the battery life and improving its performance.

[0150] The state prediction unit 103 integrates shape variables, improvement variables and usage mode variables through the judgment algorithm, dynamically generates secondary cell internal variables that reflect the internal state of the battery, and updates the internal state model of the battery accordingly, laying a solid foundation for the accurate prediction of the battery state. This technical solution ensures that the battery management system can provide high-precision state evaluation based on real-time data and prediction models, effectively extending the battery's service life and improving its operating efficiency.

[0151] Example 1.4 Health status prediction unit

[0152] The health state prediction unit 104 is used to analyze the internal state model to determine the current health state of the battery, including the health and charge level of the battery. The health is determined by analyzing the battery's decay characteristics, internal resistance changes, electrolyte degradation, and other aspects, while the charge level is estimated based on the battery's current charge state and capacity. With this information, the health state prediction unit can provide a comprehensive battery health assessment to help users take necessary maintenance measures before battery performance deteriorates.

[0153] The health state prediction unit 104 is a core component in the battery state evaluation system for parsing the internal state model and determining the current health state of the battery. The technical implementation scheme of the health state prediction unit will be described in detail below.

[0154] The health state prediction unit 104 first obtains key internal parameters of the battery by analyzing the internal state model output by the state prediction unit 103. These parameters include but are not limited to battery internal resistance, electrolyte impedance, fixed phase element coefficient, etc.

[0155] By analyzing these parameters, the health status prediction unit can extract features that are closely related to battery health and charging level, providing a basis for subsequent status assessment.

[0156] State of Health (SOH) is a comprehensive evaluation indicator of the current health status of the battery, expressed as a percentage (for example, 100% represents the health status of a new battery). The evaluation of health involves multiple aspects, mainly including the battery's decay characteristics, internal resistance changes, and electrolyte degradation.

[0157] The battery degradation characteristics are determined by analyzing the battery capacity decay and power drop. The health status prediction unit calculates the degradation characteristic parameters by monitoring the deviation between the actual capacity and the nominal capacity of the battery, as well as the output power change of the battery under different load conditions. The calculation formula is as follows:

[0158]

[0159] in Indicates the state of health (SOH) of the battery capacity, reflecting the ratio of the actual capacity of the current battery to the rated capacity. It is used to evaluate the capacity attenuation of the battery. The closer the value is to 100%, the closer the battery capacity is to its initial state. is the actual capacity of the current battery, is the rated capacity of the battery.

[0160]

[0161] in Indicates the state of health (SOH) of the battery power, reflecting the ratio of the actual output power of the current battery to the rated output power. It is used to evaluate the power drop of the battery under different load conditions. is the actual output power of the current battery, is the rated output power of the battery.

[0162] Through the above calculations, the health status prediction unit can obtain health indicators of battery capacity attenuation and power reduction.

[0163] The change of battery internal resistance is another important indicator for evaluating battery health. The health status prediction unit monitors the change of battery internal resistance under different charging and discharging conditions to evaluate its impact on battery health. The increase of battery internal resistance means the increase of battery aging. The health status prediction unit dynamically adjusts the battery health evaluation result according to the magnitude of the internal resistance change.

[0164] Internal resistance change health It can be calculated by the following formula:

[0165]

[0166] in, is the initial internal resistance of a new battery, is the currently measured internal resistance of the battery.

[0167] Electrolyte degradation is characterized by changes in electrolyte impedance. The health status prediction unit obtains electrolyte impedance data through frequency domain analysis (such as electrochemical impedance spectroscopy, EIS), and determines the degree of electrolyte degradation based on its change trend over time. This impedance data can be compared with the historical impedance data of the battery to obtain the health index of electrolyte degradation.

[0168] Electrolyte Deterioration Health The calculation formula is:

[0169]

[0170] in, is the initial electrolyte impedance of a new cell, is the currently measured electrolyte impedance.

[0171] The health status prediction unit weights the above multiple health indicators to obtain the comprehensive health of the battery. The formula for weighted judgment is:

[0172]

[0173] in, It is a weight coefficient determined according to the battery type and application scenario, satisfying .

[0174] The state of charge (SOC) is a measure of the percentage of the battery's current stored charge relative to its total capacity, and is a key indicator in the battery management system. Accurate estimation of SOC is crucial to ensure the effective operation and life extension of the battery. The estimation of SOC mainly depends on parameters such as the battery's voltage, current, and rated capacity. Through comprehensive analysis of these data, the current state of charge of the battery can be obtained.

[0175] In the health status prediction unit, the SOC estimation process involves real-time monitoring of the battery's voltage and current data. The voltage data provides important information about the battery's instantaneous state, while the current data reflects the battery's charge and discharge rate. These data are combined with the battery's known charge and discharge curves and processed by algorithms to accurately estimate the current SOC value. The charge and discharge curves are predetermined based on the battery's chemical composition, structure, and historical performance data, and reflect the battery's voltage and current characteristics under different charge and discharge conditions.

[0176] To improve the accuracy of SOC estimation, the health state prediction unit not only relies on instantaneous voltage and current data, but also combines the battery's historical usage records and dynamic performance characteristics. For example, under high load or extreme temperature conditions, the battery's voltage curve may show obvious nonlinear changes, which must be taken into account in the SOC estimation. By combining these factors, the health state prediction unit can provide a more reliable SOC value, allowing the battery management system to perform corresponding control and optimization based on the precise charging level.

[0177] In addition, the SOC estimation process can also be optimized using a variety of algorithms, such as the combination of the coulomb counting method and the open circuit voltage method, which can effectively reduce the estimation error. This multi-dimensional data fusion and algorithm optimization makes the SOC estimation more accurate, which can better reflect the actual charging level of the battery and provide strong protection for the safe operation and long life of the battery.

[0178] After calculating the battery's health and charge level, the health prediction unit generates a comprehensive battery health assessment report. The report includes the battery's current SOH, SOC, and predicted remaining useful life (RUL). Based on the assessment results, the system can provide users with specific maintenance recommendations, such as:

[0179] When the health level falls below a certain threshold, it is recommended to replace the battery or perform deep cycle maintenance;

[0180] Optimize charging strategy or adjust load configuration based on SOC value and battery usage.

[0181] This information is transmitted to the user or the battery management system (BMS) through the system's communication module, helping the user to take necessary maintenance measures before the battery performance deteriorates, thereby extending the battery life and improving its usage efficiency.

[0182] Example 1.5 Communication unit

[0183] The communication unit 105 plays a key role in information transmission in the battery status evaluation system. The communication unit is responsible for timely transmitting the health prediction value and charge level prediction value generated by the health status prediction unit to the user or related computing system to ensure real-time acquisition and effective processing of battery status information.

[0184] To achieve this function, the communication unit 105 supports multiple communication protocols, including wireless communication (such as Wi-Fi, Bluetooth, cellular network) and wired communication (such as Ethernet). This diverse communication mode enables the system to flexibly adapt to different application scenarios and environmental requirements. For example, in a remote or distributed battery monitoring system, wireless communication can achieve seamless data transmission, while in a fixed industrial environment, wired communication can provide higher transmission stability and data security.

[0185] The communication unit 105 is tightly integrated with the battery management system (BMS) or the user terminal, and can transmit key evaluation results such as the battery health and charging level to the user or the monitoring system in real time. This process includes data encryption, packaging, and transmission through appropriate communication protocols to ensure data integrity and security. During the transmission process, the system can also cache and resend data to cope with network interruptions or data loss that may occur during the communication process.

[0186] In addition, the communication unit 105 can also be connected to a cloud computing platform or a big data analysis system, so that the battery status data can be shared and analyzed on a larger scale. This not only helps to monitor the status of a single battery in real time, but also provides a basis for predicting the health status of the entire battery group and optimizing maintenance strategies through data aggregation and analysis.

[0187] The design of the communication unit 105 not only focuses on the timeliness and reliability of data transmission, but also takes into account the flexibility and scalability of the system, so as to meet the needs of different users and application scenarios and ensure the accurate communication and efficient processing of battery status information.

[0188] Example 2: Data interaction process of battery status assessment system

[0189] As attached Figure 2 As shown, this embodiment describes in detail a workflow of a battery state evaluation system, which includes a characteristic acquisition unit 101, a scenario prediction unit 102, a state prediction unit 103, a health state prediction unit 104 and a communication unit 105. The interaction process between the units and their specific functions will be described below in the form of data interaction.

[0190] 1. User data input and feature acquisition

[0191] First, the user provides the battery status assessment system with battery usage data. This data may include information such as the battery's current working environment, usage conditions, and historical operation records. The user's data input triggers the system to start evaluating the battery status.

[0192] After receiving the battery usage data provided by the user, the characteristic acquisition unit 101 immediately starts processing the data collected from the passive and active measurements of the battery. The passive data includes the voltage, current, temperature, etc. generated by the battery during normal operation, while the active data comes from the response measurement after applying a specific stimulus signal to the battery.

[0193] After processing these data, the characteristic acquisition unit 101 performs analysis based on the circuit equivalent model, obtains shape variables through shape characteristic identification, and outputs improved variables through improved variable identification. These variables accurately reflect the physical and chemical state inside the battery and are important basic data for subsequent evaluation.

[0194] 2. Usage pattern prediction

[0195] The scenario prediction unit 102 uses the battery's historical data and current usage scenarios to predict the battery's potential usage patterns based on an artificial intelligence model. The historical data includes the battery's performance records under different conditions, while the usage scenarios may include the current ambient temperature, load conditions, etc.

[0196] By analyzing these data, the scenario prediction unit 102 generates usage pattern variables, which predict the possible working mode of the battery in the future and provide a prediction basis for subsequent state evaluation.

[0197] 3. State assessment and model update

[0198] The state prediction unit 103 receives the shape variables and the improvement variables from the characteristic acquisition unit 101, and the usage mode variables from the scene prediction unit 102. Through the determination algorithm, the state prediction unit 103 comprehensively analyzes these variables, dynamically adjusts the weights and determination strategies of the algorithm according to the actual situation, and finally outputs the secondary cell internal variables reflecting the internal state of the battery.

[0199] These secondary cell internal variables include at least fixed phase element coefficients, electrolyte impedance and current transfer impedance, which can accurately describe the physical and chemical processes inside the battery. Based on these variables, the state prediction unit 103 updates the internal state model of the battery in real time to ensure that the model can always accurately reflect the current state of the battery.

[0200] 4. Health status analysis and prediction

[0201] The updated internal state model is passed to the health state prediction unit 104. The health state prediction unit 104 parses the model to determine the current health state of the battery, including the state of health (SOH) and the charge level (SOC). The health assessment involves analyzing multiple aspects such as the battery's decay characteristics, internal resistance changes, and electrolyte degradation, while the charge level is estimated based on the battery's current voltage, current, and capacity data. Through these comprehensive analyses, the health state prediction unit 104 can provide a comprehensive health status assessment of the battery, ensuring that users can take appropriate maintenance measures before the battery performance deteriorates.

[0202] 5. Information transmission and user feedback

[0203] After completing the health status assessment, the health status prediction unit 104 sends the assessment results, including the health status prediction value and the charge level prediction value, to the communication unit 105. The communication unit 105 transmits these assessment results to the user or the related computing system in real time through a predetermined communication protocol (such as wireless communication, Ethernet, etc.).

[0204] These evaluation results can be fed back to the user in a variety of ways, such as displaying the current status of the battery on the user terminal, or uploading the data to the central control system for centralized monitoring and management. In this way, the user can obtain the health information of the battery in a timely manner and perform appropriate operations or maintenance according to the system's recommendations.

[0205] Through the collaborative work of the above units, the battery status assessment system can comprehensively and accurately assess the health status and charge level of the battery. This system not only improves the safety and reliability of battery use, but also helps users optimize battery use and maintenance strategies, thereby extending the battery life. The interaction process between each unit ensures the timeliness and accuracy of data processing, allowing the entire system to effectively respond to different usage scenarios and environmental conditions.

[0206] Example 3: Application of battery status assessment system in energy storage power station

[0207] In energy storage power stations, battery systems, as core equipment for large-scale energy storage, are used to balance the difference between electricity supply and demand and ensure the stable operation of the power grid.

[0208] Example 3.1 Overview of energy storage power station

[0209] Energy storage power stations are composed of multiple battery modules, which are connected in parallel or in series to form battery packs that can store a large amount of electrical energy. The status of the battery pack directly affects the efficiency and reliability of the entire energy storage power station. Therefore, real-time monitoring of the health of the battery pack is crucial to ensure the normal operation of the energy storage power station.

[0210] Example 3.2 System Integration

[0211] In this embodiment, the battery status evaluation system 100 is integrated into the monitoring system of the energy storage power station to evaluate the status of the battery pack in real time. The characteristic acquisition unit 101 is connected to the battery pack through a sensor network to continuously collect passive and active data of the battery pack. These data include the voltage, current, temperature, etc. of each battery module, as well as the response data under a specific stimulus signal.

[0212] The scenario prediction unit 102 analyzes the operation data of the energy storage power station to predict the behavior mode of the battery pack under different operating conditions. For example, during the peak load period of the power grid, the battery pack needs to discharge quickly to supplement the power grid load, and during the low load period, the battery pack enters the charging mode. By predicting these scenarios, the scenario prediction unit can provide important usage mode variables for the state prediction unit.

[0213] The state prediction unit 103 combines these variables, calculates the internal variables of the secondary battery cell through a determination algorithm, and updates the internal state model of the battery pack. Based on these models, the system can evaluate the current state of the battery pack in real time and predict its future performance change trend.

[0214] The health status prediction unit 104 further analyzes the internal status model and generates a report on the health and charge level of the battery pack. These reports can serve as an important reference for power station operation and maintenance, helping operators to promptly discover and deal with potential battery problems and extend the service life of the battery pack.

[0215] Example 3.3 Communication and Control

[0216] The battery status evaluation system 100 sends the evaluation results to the central control system or remote monitoring center of the energy storage power station through the communication unit 105. The operator can view the health status of the battery pack in real time through the monitoring system and make corresponding adjustments and maintenance operations based on the system's prediction results. In addition, the system can also share data with the power grid management system to optimize power dispatch and improve the overall operation efficiency of the power grid.

[0217] The above description is only a preferred embodiment of the present invention, so all equivalent changes or modifications made according to the structure, characteristics and principles described in the scope of the patent application of the present invention are included in the scope of the patent application of the present invention.

Claims

1. A battery status assessment system, Features: The battery state assessment system (100) comprises a characteristic acquisition unit (101), a scene prediction unit (102), a state prediction unit (103), a health state prediction unit (104) and a communication unit (105); The characteristic acquisition unit (101) receives input data, the input data comprising passive data collected from passive measurements of the battery and active data collected from active measurements of the battery in response to a stimulus signal applied to the battery; The characteristic acquisition unit (101) performs variable identification based on shape characteristics based on input data related to the circuit equivalent model to obtain shape variables; performs variable identification based on improvement based on input data related to the circuit equivalent model to obtain improved variables; The scenario prediction unit (102) predicts the potential usage pattern of the battery using an artificial intelligence model based on the battery usage scenario and historical data, and generates a usage pattern variable; The state prediction unit (103) combines the shape variable and the improvement variable through a determination algorithm, and dynamically adjusts the weight and determination strategy of the determination algorithm according to the usage mode variable to output the secondary battery cell internal variable, wherein the secondary battery cell internal variable at least includes a fixed phase element coefficient, an electrolyte impedance and a current transfer impedance; The state prediction unit (103) updates the internal state model of the battery according to the internal variables of the secondary battery cell; The health state prediction unit (104) is used to analyze the internal state model to determine the current health state of the battery, including the health and charge level of the battery; The communication unit (105) is used to send one or more health prediction values ​​and charge level prediction values ​​to a user, a related computing system, or a combination of the two; The decision algorithm initially assigns weights to shape variables, improvement variables, and usage mode variables based on the importance of the variables and historical data. The initially assigned weights can be expressed by the following formula: in, Indicates The variables in The weight of the moment, is The prediction accuracy of the variable at the moment; : represents the sum of the prediction accuracy of all n variables at time t-1; The determination algorithm dynamically adjusts the weight of each variable by analyzing the error between the actual prediction result and the measured data; the adjustment formula is: in, is the learning rate, is the loss function, by minimizing To adjust the weight ; is the updated weight of the i-th variable at time t+1; is the current weight of the loss function L on the i-th variable The partial derivative of The determination algorithm generates secondary cell internal variables by determining and measuring shape variables, improvement variables and usage mode variables according to current weights; the determination strategy is expressed as: in, is the internal variable of the secondary cell after determination, It is input variables, is a nonlinear combination function; n represents the total number of input variables; m represents the number of nonlinear combination functions; is the i-th input variable The weight of; The basic form of the state equation of the internal state model of the updated battery is: Represents state variables, including the battery's state of charge and health status; Represents state variables The derivative with respect to time t; u(t) is the input control variable, including the external control input to the battery; The internal variables of the secondary battery cell after determination include fixed phase element coefficient, electrolyte impedance, and current transfer impedance; function It is a nonlinear function, reflecting the dynamic response of the battery; function Then through the internal variables of the secondary battery Modify the model.

2. A battery status evaluation system as claimed in claim 1, Features: The passive data includes voltage, current and temperature data generated by the battery during normal operation, and the active data is obtained by applying a stimulation signal to the battery and measuring the battery response, wherein the stimulation signal is a current or voltage signal with a certain amplitude and frequency.

3. A battery status evaluation system as claimed in claim 2, Features: The shape variables include the slope and peak amplitude of the voltage response curve; the improvement variables include the internal electrolyte impedance and double-layer capacitance of the battery.

4. A battery status evaluation system as claimed in claim 1, Features: The scenario prediction unit predicts the battery usage pattern through an artificial intelligence model, and the artificial intelligence model is a long short-term memory network model. The training process of the long short-term memory network model is based on a historical data set and is trained by a supervised learning method, using mean square error as a loss function to optimize the model's prediction accuracy.

5. A battery status evaluation system as claimed in claim 4, Features: The usage pattern variables generated by the scenario prediction unit include expected temperature changes, load change patterns, and charge and discharge rates, and the usage pattern variables are output to the state prediction unit or other modules.

6. A battery status evaluation system as claimed in claim 1, Features: The health status includes a health level determined by analyzing the battery's degradation characteristics, internal resistance changes, and electrolyte degradation, and a charge level estimated based on the battery's voltage, current, and capacity.

7. A battery status evaluation system as claimed in claim 1, Features: The communication unit supports wireless communication and wired communication protocols and is integrated with a battery management system or a user terminal.

8. A battery status assessment system according to any one of claims 1 to 7, Features: The battery status evaluation system is applied to energy storage power stations, electric vehicles, and electronic equipment.

Images