Battery soc estimation method and system for electric aircraft

Abstract

This invention discloses a method and system for estimating the State of Charge (SOC) of an electric aircraft battery. The method first constructs a second-order RC equivalent circuit model and sets up a two-dimensional lookup table module. By inputting the real-time operating temperature and the estimated SOC value into the lookup table module, the corresponding open-circuit voltage value is obtained and assigned to the equivalent circuit model. Then, based on this model, an extended Kalman filter algorithm is applied to estimate the SOC value in real time. During algorithm execution, the real-time current and temperature of the battery are monitored, and the process noise covariance matrix and observation noise covariance matrix are dynamically adjusted according to the current change rate and temperature change rate to optimize the SOC estimation result. This invention effectively overcomes the technical defects of existing second-order RC models, such as significant polarization effects and insufficient dynamic response under extreme temperatures, through a temperature-compensated two-dimensional lookup table module. Combined with an adaptively adjusted extended Kalman filter algorithm, it significantly improves the estimation accuracy and reliability of the electric aircraft battery management system under complex operating conditions.

Description

Technical Field

[0001] This invention relates to the field of battery management technology, and in particular to a method and system for estimating the state of charge (SOC) of a battery in an electric aircraft. Background Technology

[0002] With the advancement of green aviation concepts, electric aircraft have become an important development direction in the aviation field. As their core power source, the performance and safety of lithium-ion batteries directly affect flight safety. The Battery Management System (BMS) is crucial for ensuring the safe and efficient operation of batteries, and the accurate estimation of the State of Charge (SOC) is a core challenge. The SOC estimation results directly provide a basis for pilots' range judgment and energy management, making its accuracy paramount.

[0003] Currently, in the field of BMS (Battery Management System), state estimation algorithms based on equivalent circuit models are widely used. Among them, the second-order RC model is often adopted due to its balance between model complexity and accuracy. However, existing technical solutions are facing serious challenges in the practical application of electric aircraft. Practice shows that when electric aircraft experience a wide temperature environment from high-altitude cruise to ground charging, or when instantaneous high power output is required during takeoff and climb, the BMS's estimation of SOC (State of Charge) will show significant deviations, and may even suddenly become inaccurate. Such deviations and inaccuracies can lead to serious distortions in the aircraft's range prediction, affecting mission planning; in severe cases, they may cause misjudgments during critical flight phases, creating potential safety hazards.

[0004] The applicant believes that the root cause of the above problems lies in the insufficient ability of existing estimation schemes to cope with complex, dynamic, and extreme operating conditions. However, how to fundamentally improve the accuracy and reliability of SOC estimation in real flight environments, especially to maintain stability under the coupling effect of wide temperature range and dynamic power loads, is a long-standing and intractable problem that has not yet been effectively solved. Summary of the Invention

[0005] The purpose of this invention is to overcome the above-mentioned shortcomings of the prior art and provide a method and system for estimating the state of charge (SOC) of batteries for electric aircraft. This method aims to effectively improve the accuracy, stability and reliability of SOC estimation under wide temperature range and dynamic power conditions, so as to ensure the flight safety of electric aircraft.

[0006] To achieve the above objectives, a first aspect of the present invention provides a method for estimating the state of charge (SOC) of a battery in an electric aircraft, comprising the following steps: Construct a second-order RC equivalent circuit model and set up a two-dimensional lookup table module; The real-time operating temperature and real-time SOC estimate of the target battery are input into the two-dimensional lookup table module to query and output the corresponding open-circuit voltage value of the target battery, and the open-circuit voltage value is assigned to the second-order RC equivalent circuit model. Based on the second-order RC equivalent circuit model with the given open-circuit voltage value, the extended Kalman filter algorithm is applied to estimate the SOC value of the target battery in real time. During the execution of the extended Kalman filter algorithm, the real-time current and temperature of the target battery are monitored, and the process noise covariance matrix Q and observation noise covariance matrix R in the algorithm are dynamically adjusted based on the calculated current change rate and / or temperature change rate to optimize the estimation result of the SOC value.

[0007] A second aspect of the present invention provides a battery SOC estimation system for an electric aircraft, comprising: The model building module is used to build and maintain a second-order RC equivalent circuit model. A two-dimensional lookup table module, connected to the model building module, is used to receive the real-time operating temperature and real-time SOC estimate of the target battery, and query and output the corresponding open-circuit voltage value of the target battery to the model building module so as to assign values ​​to the second-order RC equivalent circuit model. An extended Kalman filter estimation module, connected to the model building module, is used to estimate the SOC value of the target battery in real time based on the second-order RC equivalent circuit model with the given open-circuit voltage value. The dynamic adjustment module, connected to the extended Kalman filter estimation module, is used to monitor the real-time current and temperature of the target battery, calculate the rate of change of current and / or the rate of change of temperature, and dynamically adjust the process noise covariance matrix Q and the observation noise covariance matrix R in the extended Kalman filter estimation module based on the rate of change to optimize the estimation result of the SOC value.

[0008] Compared with existing technologies, the target battery SOC estimation method and system for electric aircraft provided by this invention have the following technical advantages: Under various typical operating conditions, especially in extreme temperature and dynamic power scenarios, the system can stably control the SOC estimation error to a low level, significantly outperforming existing technologies and providing more reliable range prediction for electric aircraft. The system maintains stable and accurate estimation results across a wide temperature range and under high-rate dynamic charge and discharge conditions, significantly improving the adaptability and safety of electric aircraft in complex environments. The system can quickly track sudden changes in battery state, effectively reducing estimation delay and improving real-time response and control efficiency in transient power scenarios. When facing complex and variable real-world flight environments, the solution demonstrates stronger anti-interference capabilities and reliability, ensuring the continuous and stable functioning of the BMS system as an intelligent central hub. Attached Figure Description

[0009] Figure 1 A flowchart illustrating a method for estimating the state of charge (SOC) of a battery in an electric aircraft, provided in an embodiment of the present invention. Figure 2 A Simulink simulation diagram of a second-order RC equivalent circuit model of a lithium battery provided in an embodiment of the present invention; Figure 3 A simulation model diagram of a second-order RC lithium battery with temperature adaptive adjustment provided in an embodiment of the present invention; Figure 4 A structural block diagram of an electric aircraft battery management system provided in an embodiment of the present invention; Figure 5 This is a Simulink simulation diagram of a battery remaining capacity calculation method in real time using the ampere-hour integration method provided in an embodiment of the present invention. Figure 6 This is a schematic diagram of a battery SOC estimation system for an electric aircraft provided in an embodiment of the present invention; Figure label: Model building module - 100, two-dimensional table lookup module - 200, extended Kalman filter estimation module - 300, dynamic adjustment module - 400. Detailed Implementation

[0010] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0011] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0012] Example 1 To clearly demonstrate how the method of this invention solves the technical challenge of improving the accuracy and reliability of SOC estimation for electric aircraft under extreme temperature and dynamic power conditions through a systematic process, particularly how it overcomes the limitations of traditional equivalent circuit models with fixed parameters that cannot adapt to complex operating conditions, the following will provide a detailed explanation of each step of the method. See [link to relevant documentation]. Figure 1 This embodiment provides a flowchart illustrating a method for estimating the state of charge (SOC) of a battery in an electric aircraft. The method includes the following steps: S100. Construct a second-order RC equivalent circuit model and initialize the parameters.

[0013] This step aims to establish a mathematical model foundation capable of simulating the dynamic characteristics of a battery. The second-order RC equivalent circuit model is a circuit model used to simulate the dynamic characteristics of lithium-ion batteries. It consists of a voltage source (simulating open-circuit voltage), an ohmic internal resistance, and two parallel resistor-capacitor (RC) networks, which can describe the electrochemical polarization (fast dynamics) and concentration polarization (slow dynamics) processes of the battery, respectively. This model is simple in structure and computationally efficient, and is a commonly used model for state estimation in battery management systems (BMS).

[0014] In practice, the model structure is first defined in an embedded system or simulation environment. The second-order RC equivalent circuit model consists of the following components: an ideal voltage source (representing the open-circuit voltage OCV), an ohmic internal resistance (R0), and two parallel resistor-capacitor networks (R1-C1 and R2-C2). The R1-C1 branch is used to simulate the electrochemical polarization process of the battery (second-level dynamics), and the R2-C2 branch is used to simulate the concentration polarization process (minute-level dynamics).

[0015] In one possible implementation, the model parameters are obtained through offline identification using a combination of laboratory pulse charge-discharge tests and nonlinear least squares method. Specifically, the target battery is placed in a constant temperature chamber at 25°C, and a 1C discharge pulse is applied for 10 seconds, followed by a 40-second rest period. This process is repeated until the battery voltage reaches the cutoff voltage. A high-precision data acquisition card (such as NI PXIe-4309) is used to record the battery terminal voltage and current data at a sampling rate of 1kHz. Based on the acquired data, the initial values ​​of the model parameters are fitted in MATLAB using the Levenberg-Marquardt optimization algorithm: R0 = 0.025Ω, R1 = 0.015Ω, C1 = 1200F, R2 = 0.008Ω, C2 = 48000F. The convergence tolerance for parameter identification is set to 1×10⁻⁶. -6 The maximum number of iterations is 200.

[0016] As an alternative, frequency domain impedance spectroscopy combined with equivalent circuit fitting can be used. Specifically, AC impedance testing is performed in the frequency range of 10 mHz to 10 kHz using an electrochemical workstation (such as the Gamry Interface 5000), and parameters are obtained through CNLS fitting using ZView software. This method can more accurately reflect the electrochemical characteristics of the battery, but the equipment cost is high and the testing cycle is long. In contrast, time-domain pulse testing is more suitable for rapid parameter acquisition in engineering settings.

[0017] The model established in this step provides the mathematical foundation for subsequent state estimation. Its key innovation lies in designing the model parameters, which are traditionally considered fixed values, as variables that can be updated online through subsequent steps, thus laying the foundation for handling complex operating conditions. After model initialization, all parameters are stored in the BMS's non-volatile memory, with an operating temperature range covering -40°C to 85°C.

[0018] S200, configure the two-dimensional lookup table module and load the temperature-SOC-OCV mapping data.

[0019] The core of this step lies in establishing a precise mapping relationship between battery open-circuit voltage, temperature, and SOC, addressing the insufficient applicability of traditional one-dimensional OCV-SOC curves across a wide temperature range. The two-dimensional lookup table module is a software or hardware function module that queries and outputs corresponding data based on two input variables. In this embodiment, the module uses battery temperature and SOC as row and column indices to store and output the corresponding open-circuit voltage (OCV) values. This module enables a precise and rapid mapping of the nonlinear relationship between battery open-circuit voltage and temperature and SOC. The SOC value (State of Charge) refers to the battery's state of charge. It represents the percentage of remaining usable charge in the current state compared to its nominal total capacity when fully charged, and is a key parameter for measuring remaining battery energy. Accurate SOC estimation is crucial for predicting the range of electric aircraft and optimizing energy management.

[0020] One possible implementation involves using a two-dimensional lookup table method based on experimental data. Specifically, at five characteristic temperature points (-20°C, -10°C, 0°C, 25°C, 35°C), the steady-state open-circuit voltage of the battery at different SOC points (0, 0.1, 0.2, ..., 1.0) is collected using a charge-discharge tester (such as an Arbin BT-5HC). During testing, the battery is first charged and discharged at a constant current rate of 0.3C to the target SOC point, then left to stand for 2 hours to allow the polarization voltage to fully relax. Finally, the stable voltage value is recorded as the OCV value for that temperature-SOC combination.

[0021] The obtained experimental data are stored in the BMS's FLASH memory in the form of a two-dimensional matrix. The matrix row index corresponds to the temperature value, the column index corresponds to the SOC value, and the matrix elements are the OCV values. In the program implementation, a bilinear interpolation algorithm is used to calculate the OCV value for any temperature-SOC combination in real time. The interpolation algorithm specifically includes: first, determining which two temperature points T lie between the current temperature T. i With T i+1 Between which two SOC points does the current SOC lie? j With SOC +1 Between; then respectively at T i and T i+1 Linear interpolation of SOC at temperature yields OCV1 and OCV2; finally, linear interpolation of OCV1 and OCV2 at temperature yields the final OCV value.

[0022] As an alternative, a polynomial surface fitting method can be used to establish... A continuous functional relationship. For example, using a bivariate sixth-degree polynomial: This method saves storage space, but may overfit in the boundary regions. In contrast, the lookup table method, although requiring more storage space (typically 2KB), offers higher accuracy and better stability, making it particularly suitable for aerospace applications with extremely high accuracy requirements.

[0023] Key parameters in this step include: temperature lookup range -40°C to 60°C, resolution 1°C; SOC lookup range 0-1, resolution 0.001; OCV output accuracy ±1mV. This step achieves accurate compensation for the effects of temperature on battery characteristics, providing an accurate voltage reference for subsequent model calculations.

[0024] S300. Real-time acquisition of battery status data and updating of model output voltage: Input the real-time operating temperature and real-time SOC estimate of the target battery into the two-dimensional lookup table module, query and output the corresponding open-circuit voltage value of the target battery, and assign the open-circuit voltage value to the second-order RC equivalent circuit model.

[0025] The open circuit voltage (OCV) refers to the potential difference between the positive and negative electrodes of a battery after it has been left to rest for a sufficiently long time (reaching electrochemical equilibrium). OCV has a definite correspondence with the state of charge (SOC) of the battery and is affected by temperature, thus it is a core state-dependent parameter in battery models.

[0026] This step achieves the above functions through the following specific operations: The system performs the following operations in 100ms cycles: synchronously acquires the battery terminal voltage (range 0-5V, accuracy ±1mV) through a 16-bit precision ADC module (such as ADS8588S); acquires the charging and discharging current (range -200A to +200A) through a Hall current sensor (such as ACS758, accuracy ±0.5%); and acquires the battery surface temperature through a digital temperature sensor (such as DS18B20, accuracy ±0.5°C). All sensor data are processed by a 4th-order Butterworth low-pass filter (cutoff frequency 10Hz) and then stored in a circular buffer.

[0027] Based on the collected real-time data, the system performs a model update: First, the current temperature T(k) and the previous SOC estimate SOC(k-1) are input into the two-dimensional lookup table module configured in S200 to obtain the current OCV value; then, this OCV value is assigned to the second-order RC equivalent circuit model established in S100. The state-space representation of the model is updated as follows: ; , where the state vector x = [U1, U2] T U1 and U2 are the polarization voltages of the two RC branches, respectively; system matrix Input matrix B = [1 / C1, 1 / C2] T Output matrix C = [1, 1]; I is the load current (positive for discharge).

[0028] As an alternative, an online parameter identification method based on electrochemical impedance spectroscopy can be used, which estimates model parameters in real time by injecting a small-amplitude AC signal and analyzing the response. This method is highly accurate but complex to implement and may interfere with normal battery operation. In contrast, the lookup table method combined with a fixed-parameter model used in this embodiment achieves an optimal balance between computational complexity and reliability while ensuring accuracy.

[0029] The innovation of this step lies in the deep integration of the traditionally independent temperature compensation stage with the equivalent circuit model. By dynamically updating the model's core parameter OCV through real-time table lookup, the model gains the ability to adapt to temperature changes. The data acquisition cycle can be configured between 50-200ms, with 100ms being the preferred option to balance real-time performance and computational load.

[0030] S400. Perform SOC estimation using the extended Kalman filter algorithm: Based on the second-order RC equivalent circuit model with the given open-circuit voltage value, the extended Kalman filter algorithm is applied to estimate the SOC value of the target battery in real time.

[0031] The Extended Kalman Filter (EKF) algorithm is an application of the standard Kalman filter to nonlinear systems. Its core principle is to locally linearize the nonlinear system (usually using a first-order Taylor expansion) and then apply the prediction and update steps of the Kalman filter to achieve the optimal estimate of the system's internal state (such as the battery's state of charge).

[0032] The core of this step lies in using advanced estimation algorithms to handle model uncertainties and measurement noise, thereby achieving high-precision estimation of SOC.

[0033] In one possible implementation, the estimation process is accomplished through iterative calculations using the extended Kalman filter algorithm, specifically including the following sub-steps: S410 State Prediction: Based on the posterior state estimate at time k-1 k-1 + And error covariance P k-1 + and the current measurement value I at time k. k Predict the prior state estimate at time k. k - and prior error covariance P k - .

[0034] ;P k - =A k-1 ·P k-1 + ·A k-1 +Q k The state transition matrix Δt is the sampling interval of 0.1s; input matrix Qn is the rated capacity of the battery.

[0035] S420 Kalman Gain Calculation: Calculating Kalman Gain K based on prior error covariance and observation equation. k .

[0036] K k = P k - ·C k ·(C k ·P k - ·C k + R k ) - ¹, where the observation matrix C k = [ h / x, h / [SOC], where h is the observation equation, which needs to be obtained by linearizing the state equation.

[0037] S430 Status Update: Using Voltage Measurements Update the state estimate.

[0038] ; ,in To predict the output voltage for the model, The third dimension is the SOC estimate.

[0039] As alternatives, nonlinear filtering algorithms such as unscented Kalman filtering or particle filtering can be used. UKF avoids Jacobian matrix calculations through sigma point propagation, achieving higher accuracy but increasing computational cost by approximately 40%. PF is suitable for strongly nonlinear systems, but its computational complexity increases linearly with the number of particles. In contrast, EKF achieves the best balance between accuracy and computational cost, making it particularly suitable for BMS applications with high real-time requirements.

[0040] The key innovation of this step lies in incorporating SOC as a state variable into the estimation system, overcoming the cumulative error problem of the ampere-hour integration method through real-time correction. In the algorithm, the process noise covariance Q is set to diag[1×10⁻⁶]. -6 1×10 -6 1×10 -8 The observation noise covariance R is set to 1×10⁻⁶. -4 These parameters will be dynamically adjusted in the S500.

[0041] S500. Dynamically adjust the noise covariance matrix to optimize estimation performance: During the execution of the extended Kalman filter algorithm, monitor the real-time current and temperature of the target battery, and based on the calculated current change rate and / or temperature change rate, dynamically adjust the process noise covariance matrix Q and the observation noise covariance matrix R in the algorithm to optimize the estimation results of the SOC value.

[0042] The process noise covariance matrix Q characterizes the uncertainty or error of the system's state equations (model). It reflects the degree of deviation between the model's predicted values ​​and the actual state evolution, such as noise introduced by model simplification, inaccurate parameters, or unknown external disturbances. The observation noise covariance matrix R characterizes the uncertainty or error of sensor measurements. It reflects the degree of deviation between sensor (e.g., voltage, current sensors) measurements and the actual physical quantity, such as noise caused by sensor accuracy, resolution, and random disturbances.

[0043] This step achieves performance optimization through the following adaptive mechanism: The system calculates the rate of change of current in real time. dI / dt and rate of temperature change dT / dt As an indicator for identifying operating conditions: dI / dt = |I k - I k-1 | / Δt ; dT / dt = |T k - T k-1 | / ΔtBased on these indicators, the process noise covariance matrix Q and the observation noise covariance matrix R are dynamically adjusted: when dI / dt >5 A / s or dT / dt When the temperature exceeds 1 °C / s, it is considered a drastic change condition. In this case, the process noise covariance Q is increased to diag[1×10]. -4 1×10 -4 1×10 -6 [This also reduces the observation noise covariance R to 1×10] -5 This makes the filter more trusting of sensor measurements.

[0044] when dI / dt <0.5 A / s and dT / dt When the temperature is <0.1 °C / s, it is considered a steady operating condition. At this point, the process noise covariance Q is reduced to diag[1×10]. -8 1×10 -8 1×10 -10 At the same time, the observation noise covariance R is increased to 1×10. -3 This makes the filter more confident in the model's predictions.

[0045] Under moderately varying operating conditions, a linear interpolation method is used to smoothly transition between extreme values: Q = Q min + (Q max - Q min )·(dI / dt - 0.5) / (5 - 0.5) ; R = R max + (R min - R max )·(dI / dt - 0.5) / (5 - 0.5) .

[0046] As an alternative, fuzzy logic control or neural networks can be used to adaptively adjust the Q and R matrices. Fuzzy logic methods achieve smoother adjustments by defining membership functions for the rates of change of current and temperature and using a fuzzy rule base; neural network methods establish a mapping relationship between operating conditions and optimal parameters by training on historical data. These methods offer better adjustment results but are complex to implement and computationally intensive. In contrast, the threshold judgment method used in this implementation is simple, reliable, and computationally efficient, making it suitable for implementation in embedded systems with limited computing resources.

[0047] The innovation of this step lies in overcoming the technical bias of fixed parameters in traditional EKF algorithms. Through condition identification and adaptive parameter adjustment, the algorithm maintains optimal estimation performance in various flight phases of electric aircraft (takeoff, cruise, climb, and landing). Tests show that this dynamic adjustment mechanism can reduce the SOC estimation error under dynamic conditions by approximately 60%.

[0048] The embodiments of this invention, through the coordinated execution of the above steps, form a complete technical solution. In 3C high-rate discharge testing at -20°C, the traditional method's SOC estimation error reaches 8.2%, while the error of this method is controlled within 1.8%. Under conditions where the temperature drops sharply from 25°C to -10°C, the traditional method exhibits an estimation deviation of 3.5%, while the deviation of this method does not exceed 1.2%. The calculation cycle of the entire solution can be controlled within 80ms on the STM32F407 platform, meeting the real-time requirements of electric aircraft BMS.

[0049] Compared with existing technologies, this invention has at least the following technical advantages: First, it achieves adaptation to a wide temperature range environment through a "two-dimensional lookup table-model update" mechanism; second, it solves the estimation accuracy problem under dynamic power conditions by "dynamically adjusting EKF parameters"; and third, it constructs a complete technical closed loop of "data acquisition-model update-state estimation-parameter optimization". This invention provides a high-precision and high-reliability SOC estimation solution for electric aircraft, and has significant engineering application value.

[0050] Example 2 This embodiment, based on Embodiment 1, further provides several preferred implementation schemes to more completely demonstrate the diversity and feasibility of the technical solutions of the present invention. By combining the technical features of the following preferred embodiments, this embodiment will elaborate in detail the specific implementation methods, technical effects, and roles of each technical solution in the overall technical solution, focusing on the reasons, advantages, and technical effects of each preferred embodiment. Those skilled in the art will understand that the following specific embodiments are only used to explain the present invention and are not intended to limit the scope of protection of the present invention. To achieve sufficient disclosure, each preferred embodiment is fully described in terms of the specific implementation details, material composition, connection method, functional limitations, and technical effects of the technical solution, ensuring that those skilled in the art can realize the invention based on this description. This embodiment of the present invention provides a method for estimating the SOC of a battery in an electric aircraft, including the following steps: Construct a second-order RC equivalent circuit model and set up a two-dimensional lookup table module; The real-time operating temperature and real-time SOC estimate of the target battery are input into the two-dimensional lookup table module to query and output the corresponding open-circuit voltage value of the target battery, and the open-circuit voltage value is assigned to the second-order RC equivalent circuit model. Based on the second-order RC equivalent circuit model with the given open-circuit voltage value, the extended Kalman filter algorithm is applied to estimate the SOC value of the target battery in real time. During the execution of the extended Kalman filter algorithm, the real-time current and temperature of the target battery are monitored, and the process noise covariance matrix Q and observation noise covariance matrix R in the algorithm are dynamically adjusted based on the calculated current change rate and / or temperature change rate to optimize the estimation result of the SOC value.

[0051] In one possible implementation, the second-order RC equivalent circuit model is built using the Simulink environment. Its core structure includes an ohmic internal resistance R0, two parallel RC branches (R1-C1 and R2-C2), and a controlled voltage source (representing the open-circuit voltage OCV). Specifically, R0 is connected in series in the main circuit path, R1-C1 and R2-C2 are connected in parallel and then in series with R0, and the controlled voltage source is connected in parallel with R0. The model parameters are calibrated based on battery characteristics: typical values ​​for R0 are 0.072Ω, R1 is 0.03Ω, C1 is 1000F, R2 is 0.05Ω, and C2 is 2000F (these parameters can be obtained experimentally). In terms of hardware connectivity, the model connects to the individual battery cells via a data acquisition module. This module includes a voltage sensor (using a high-precision differential amplifier such as INA188, accuracy ±1mV), a current sensor (using a closed-loop Hall sensor such as ACS758, accuracy ±0.5%), and a temperature sensor (using a PT1000 platinum resistance thermometer, accuracy ±0.5°C). The sensor signals are converted to digital signals via an analog-to-digital converter (ADC, such as ADS131M04, 24-bit resolution) and transmitted to the main control unit (MCU, such as TI's TMS320F28379D, 200MHz clock frequency) via an SPI bus (1MHz clock frequency). The SPI bus connection includes SCLK, MOSI, MISO, and CS signal lines. The signal line impedance is controlled at 50Ω ±5%, and the wiring length does not exceed 50mm to minimize signal integrity loss.

[0052] Figure 2 This is a Simulink simulation structure diagram of a second-order RC equivalent circuit model of a lithium battery provided in an embodiment of the present invention, as shown below. Figure 2As shown, the model includes core circuit modules: a current input module, an ohmic internal resistance R0, a first RC branch (resistor R1 and capacitor C1 in parallel), a second RC branch (resistor R2 and capacitor C2 in parallel), an integrator, a multiplier, and a summing module. The output of the current input module is electrically connected to one end of the ohmic internal resistance R0, one end of the first RC branch, and one end of the second RC branch, respectively; the other ends of the ohmic internal resistance R0, the first RC branch, and the second RC branch are all electrically connected to the input of the summing module; the input of the integrator is electrically connected to the output of the current input module, the output of the integrator is electrically connected to the input of the multiplier, and the output of the multiplier is electrically connected to the input of the summing module. Finally, the battery terminal voltage Vmdl is output through the summing module. The ohmic internal resistance R0 is used to simulate the instantaneous voltage drop of the battery; the first RC branch is used to simulate the electrochemical polarization effect, with a time constant of R1×C1, corresponding to a fast dynamic response on the order of seconds; the second RC branch is used to simulate the concentration polarization effect, with a time constant of R2×C2, corresponding to a slow dynamic response on the order of minutes. The integrator is used to solve the polarization voltage differential equation: and Where U1 and U2 are the polarization voltages of the first and second RC branches, respectively, and I is the real-time current. A multiplier and summator module are used to superimpose the ohmic internal resistance voltage drop, polarization voltage, and open-circuit voltage, outputting a terminal voltage consistent with the actual battery dynamic characteristics. This model, through independent design of dual RC branches, accurately fits the instantaneous voltage fluctuations and gradual changes during battery charging and discharging. Under 3C rate discharge conditions, the voltage prediction deviation is ≤50mV, and the dynamic voltage response accuracy is ≥40% higher than that of the first-order RC model.

[0053] The two-dimensional lookup table module is implemented in Simulink as a 2-D Lookup Table module. Its row index is the SOC value (0 to 1, step size 0.1), and its column index is the temperature value (-40°C to 60°C, step size 5°C). The table data is based on the experimentally measured OCV-T-SOC relationship, as shown in Table 1 below. This lookup table module is connected to the MCU's Flash memory via memory mapping, with storage addresses from 0x0800_0000 to 0x0800_0FFF, occupying approximately 2KB of space to ensure fast lookup. The real-time SOC estimate and temperature value are sampled by the ADC, calculated by the MCU, and input into the lookup table module. The OCV value is output and assigned to the controlled voltage source of the second-order RC model. During the lookup, a bilinear interpolation algorithm is used to handle non-integer points. The interpolation formula is: ;in , and , For the nearest temperature and SOC point, to This represents the OCV value of the corresponding point.

[0054] Table 1: Experimental Data of SOC-T-OCV

[0055] The extended Kalman filter (EKF) algorithm was implemented in Matlab and deployed to run on an MCU. The core formulas include the state equation and the observation equation: State equations (discretized): Where x = [U1, U2, SOC] T Where u is the current and w is the process noise. Matrices A and B are obtained by discretizing the model parameters: ， ， in The sampling interval is 1s, and Qn is the battery capacity (2Ah).

[0056] Observation equation: , where v is the observation noise.

[0057] The strategy for dynamically adjusting the Q and R matrices is based on the rate of change of current. di / dt and rate of temperature change dT / dt When | di / dt |>1A / s or| dT / dt At |>1°C / s, increase Q and R to quickly track mutations (Q adjusted to R is adjusted to 0.1); when the change is gradual (| di / dt |<0.1A / s and| dT / dt |<0.1°C / s), reduce Q and R to improve steady-state accuracy (Q is (R is 0.01). The adjustment coefficient scales linearly according to the rate of change and is calculated in real time by the MCU's arithmetic logic unit (ALU).

[0058] Figure 3 This is a simulation model diagram of a second-order RC lithium battery with temperature adaptive adjustment provided in an embodiment of the present invention, as shown below. Figure 3As shown, the model includes core functional modules: a current signal generation module (Current Signal Builder), a temperature signal input module (Signal), a SOC calculation submodule (soc calculate), an RC parameter calculation submodule (RCParameter Calculate), a 2-DT(u) lookup table module (SOC-T-OCV two-dimensional lookup table), a data output module (ToWorkspace1), and a waveform observation module (Scope). The output of the current signal generation module is electrically connected to the input of the SOC calculation submodule and the first input of the RC parameter calculation submodule; the output of the temperature signal input module is electrically connected to the second input of the RC parameter calculation submodule and the first input of the 2-DT(u) lookup table module; the output of the SOC calculation submodule is electrically connected to the second input of the 2-DT(u) lookup table module; the outputs of the RC parameter calculation submodule and the 2-DT(u) lookup table module are both electrically connected to the input of the waveform observation module; and the output of the waveform observation module is electrically connected to the data output module. The temperature signal input module can output temperature signals from -40℃ to 60℃, covering the entire operating temperature range of the electric aircraft. The 2-DT(u) lookup module incorporates SOC-temperature-OCV experimental data (as shown in Table 3-2). By receiving the current SOC value and temperature signal, it outputs the corresponding open-circuit voltage value, replacing the fixed open-circuit voltage source and eliminating the influence of temperature on the OCV-SOC mapping relationship. The RC parameter calculation submodule dynamically adjusts the ohmic internal resistance R0 and the parameters of the dual RC branches (R1, C1, R2, C2) based on the temperature signal. Specifically, at low temperatures (-40℃), the values ​​of R0, R1, and R2 are increased by 20%-30% compared to normal temperatures (25℃), while the values ​​of C1 and C2 are decreased by 15%-20%. At high temperatures (60℃), the parameter values ​​are adjusted in the opposite direction. Within the temperature range of -40℃ to 60℃, the model achieves a terminal voltage prediction deviation of ≤60mV and an SOC estimation error of ≤2%, with an SOC estimation accuracy improvement of ≥50% under extreme temperatures.

[0059] This embodiment significantly improves the accuracy of State of Charge (SOC) estimation under extreme temperatures (-40°C to 60°C) and dynamic operating conditions (such as instantaneous power demand > 3C during takeoff) by introducing a temperature-compensated two-dimensional lookup table module and a dynamic noise-adjusted EKF algorithm. Simulation results show that the SOC estimation error can be controlled within 2%, and the terminal voltage prediction deviation is less than 50mV, which is better than the traditional fixed-parameter model. Experimental data show that at a low temperature of -20°C, the SOC estimation error is reduced from more than 5% to 1.5%, and the voltage prediction deviation is only 30mV at high-rate discharge (3C).

[0060] In a preferred embodiment, the two-dimensional lookup table module includes pre-stored experimental data on the open-circuit voltage of the target battery at different temperatures and different SOCs.

[0061] In one possible implementation, experimental data is obtained through constant current pulse discharge testing, with test conditions covering a temperature range of -40°C to 60°C (controlled by a temperature chamber) and a State of Charge (SOC) range of 0 to 1 (set via charge-discharge cycles). Data is stored as a two-dimensional array in non-volatile memory (e.g., EEPROM AT24C1024, 1Mb capacity), in row-major format (SOC as the row, temperature as the column). The memory is connected to the MCU via an I2C bus at a communication rate of 400kHz, with bus pins including SCL and SDA and pull-up resistors of 4.7kΩ. To reduce storage space, data is in 16-bit fixed-point format (0.1mV precision), with a total storage capacity of 11 (temperature points) × 10 (SOC points) × 2 bytes = 220 bytes. During querying, the MCU calculates the OCV values ​​of non-integer SOC and temperature points using bilinear interpolation. The interpolation algorithm is implemented in the MCU firmware, using a floating-point unit (FPU) to accelerate the calculation.

[0062] This embodiment utilizes a lookup table module constructed from high-precision experimental data to provide accurate OCV input for the equivalent circuit model, effectively overcoming the errors caused by the nonlinearity of the OCV-SOC curve in traditional methods. Actual measurements show that the SOC estimation error decreased from over 5% to less than 2% at -20°C, and remained at 1.8% at 60°C. Furthermore, the lookup table module's query response time is <1ms, meeting real-time requirements.

[0063] In a preferred embodiment, the method further includes identifying parameters in the second-order RC equivalent circuit model online using a recursive least squares method that runs in parallel with the extended Kalman filter algorithm.

[0064] In one possible implementation, Recursive Least Squares (RLS) is executed in parallel with EKF as a task in the MCU, with a sampling period of 1ms. The RLS algorithm model is as follows: + ,in For the terminal voltage error, θ = [R0, R1, R2, C1, C2] T The parameter vector to be identified, The regression vector contains historical values ​​of current and voltage. The RLS update formula is: ; ; ,in The forgetting factor is set to 0.995 initially, and the P matrix is ​​initially set to... The parameter identification results interact with the EKF algorithm via shared memory (dual-port RAM), and EKF uses the updated parameters to correct the state equation in real time. The shared memory address mapping is from 0x2000_0000 to 0x2000_0014, and a mutex lock is used to avoid data races.

[0065] This embodiment uses online parameter identification to dynamically track parameter changes caused by battery aging, avoiding SOC estimation drift caused by battery degradation in fixed-parameter models. In actual testing, after 500 battery cycles, the SOC estimation error remained within 3%, and the parameter identification convergence time was <10s. On the hardware side, the RLS algorithm has a computational load of approximately 50kFLOPs per cycle, efficiently processed by the MCU's DSP core.

[0066] In a preferred embodiment, the parameters include the ohmic internal resistance R0 of the target battery, used to assess the state of health (SOH) of the target battery in real time.

[0067] In one possible implementation, the ohmic internal resistance R0 is identified in real time using the RLS algorithm and compared with an initial value (e.g., 0.072Ω) to calculate the state of harmonics (SOH). ,in The battery's internal resistance value at the factory is stored in the MCU's Flash memory (address 0x0800_1000). The SOH (State of Health) assessment results are sent to the aircraft's main controller via the CAN bus for early warning and maintenance decisions. In terms of hardware, internal resistance measurement is achieved using high-precision current and voltage sensors with a sampling rate of 1kHz to ensure accurate identification. The current sensor uses a shunt (100μΩ) in conjunction with a differential amplifier, while the voltage sensor uses a resistor divider network (100:1 ratio) and an ADC.

[0068] This embodiment utilizes the strong correlation between internal resistance and aging degree to achieve real-time monitoring of SOH (State of Health). Experimental data shows that when SOH drops to 80%, internal resistance increases by approximately 30%, which coincides with capacity decay, providing a reliable basis for battery life prediction. The SOH estimation error is <5%, meeting the requirements for aerospace applications.

[0069] In a preferred embodiment, the recursive least squares method includes an algorithm with a forgetting factor, and the forgetting factor is dynamically adjusted based on the fluctuation of the target battery current. When the rate of change of current is lower than a first preset threshold, a smaller forgetting factor is used to enhance the role of historical data, and when the rate of change of current is higher than a second preset threshold, a larger forgetting factor is used to quickly track parameter changes.

[0070] In one possible implementation, the forgetting factor The dynamic adjustment strategy is as follows: calculate the rate of change of current. .when When <0.1A / s (first threshold), set =0.999 to enhance the weight of historical data and improve parameter stability; when When >1A / s (second threshold), set =0.98 to quickly track parameter mutations. The adjustment logic is implemented through conditional instructions from the MCU to ensure the response speed of parameter identification under high-dynamic conditions such as electric aircraft takeoff. Threshold data is stored in Flash and can be updated via a configuration interface.

[0071] This embodiment balances the steady-state accuracy and dynamic response of parameter identification through an adaptive forgetting factor. In actual measurements at 3C rate discharge, the parameter tracking delay is reduced from the 10ms level to less than 5ms, significantly improving the accuracy of SOC estimation under transient conditions. The RLS algorithm exhibits a parameter error of <2% under dynamic conditions.

[0072] In a preferred embodiment, the method further includes dynamically updating the OCV-T-SOC data in the two-dimensional lookup table module based on the target battery's cycle aging count or cumulative capacity decay data using a predefined aging model.

[0073] In one possible implementation, the aging model is based on an electrochemical empirical formula: ,in , The aging factor (calibrated through accelerated aging tests, typical value) =0.001, =0.01), cycle is the number of cycles, and ΔQ is the capacity decay. The update mechanism is executed by the MCU after each charge-discharge cycle: it reads the cycle counter (stored in EEPROM) and the current capacity (estimated through a full charge-discharge test), calculates the new OCV-T-SOC table, and writes it to the update area of ​​Flash (address 0x0800_2000). To reduce computational burden, the update is only performed during calibration cycles (e.g., every 100 cycles). Capacity testing is achieved by constant current discharge to the cutoff voltage (2.8V), with a current of 1C and an accuracy of ±1%.

[0074] This embodiment compensates for the impact of battery aging on the OCV-SOC relationship by dynamically updating the lookup table data. Experiments show that after 1000 battery cycles, the SOC estimation error can still be maintained within 2.5%, extending the effective lifespan of the BMS. The update process takes less than 10 seconds and does not affect the real-time operation of the system.

[0075] In a preferred embodiment, the method is implemented in a distributed BMS architecture, where the SOC estimation task assigned to the battery module controller for local execution employs a computationally reduced lightweight extended Kalman filter algorithm.

[0076] In one possible implementation, the distributed BMS architecture includes one master control unit (BCU) and multiple slave control units (BMUs), with each BMU responsible for one battery module (e.g., 12 cells). The BMUs employ low-power MCUs (e.g., STM32G474, 170MHz) and run a lightweight EKF algorithm: the state vector is simplified to [SOC] (single state), and the observation equation is simplified to... The process noise covariance Q and observation noise covariance R are fixed at empirical values ​​(Q=1e-6, R=0.01). The algorithm's computational cost is reduced by approximately 70% compared to the full-order EKF, requiring only 10kFLOPs per cycle, making it suitable for resource-constrained BMUs. The BMU uploads its local SOC estimate to the BCU for fusion via a CAN bus (e.g., CAN 2.0B). The CAN bus connection includes CAN_H and CAN_L, with a 120Ω terminating resistor and a transmission rate of 500kbps.

[0077] Figure 4 This is a structural block diagram of an electric aircraft battery management system provided by an embodiment of the present invention, such as... Figure 4As shown, this system embodies a control architecture of "centralized management + distributed data acquisition." The system includes a control unit component, a signal acquisition module, a function execution module, and an external interaction interface. The control unit component includes a master control unit (Battery Management Unit) and at least one slave control unit (Battery Control Unit). The master control unit and slave control units achieve bidirectional data interaction via a CAN bus or LIN bus. The slave control units are directly connected to the battery module and are used to acquire voltage and temperature signals from individual battery cells. The master control unit receives the acquired data uploaded by the slave control units and performs estimations of SOC (State of Charge), SOH (State of Health), SOP (State of Power), and SOE (State of Energy), as well as scheduling of global balancing and thermal management strategies. The signal acquisition module includes a voltage sensor, a current sensor, a temperature sensor, and a pressure sensor. The signal output terminals of each sensor are electrically connected to the signal input terminals of the slave control units. The voltage sensor has an acquisition accuracy of ±1mV, the current sensor has an acquisition accuracy of ±0.5%, and the temperature sensor has an acquisition accuracy of ±0.5℃. The functional execution module includes a balancing management module, a thermal management module, and a fault diagnosis module. The balancing management module is electrically connected to the output of the main control unit, with an active balancing efficiency of ≥85%, controlling the voltage difference between individual cells within the battery pack to within 30mV. The thermal management module is used to control the battery pack temperature within a safe range of 20℃-40℃. The fault diagnosis module communicates bidirectionally with the main control unit, monitoring overvoltage, undervoltage, overcurrent, overtemperature, and short-circuit anomalies, with an anomaly protection response time of <50ms. The external interaction interface allows the main control unit to connect to external systems via an Ethernet interface, enabling the uploading of battery status data and the reception of external control commands. This architecture reduces the data processing pressure on the main control unit through distributed collaboration, improves system redundancy and fault tolerance, and adapts to the safe flight requirements of electric aircraft through high-precision acquisition and rapid protection design.

[0078] This embodiment distributes the computational load through edge computing, improving system redundancy and response speed. In actual testing, the error in module-level SOC estimation is less than 3%, and the main control unit load is reduced by 50%, making it suitable for high-cell-count battery packs (such as electric aircraft with ≥200 cells). The system supports hot-swapping, and fault isolation time is <100ms.

[0079] In a preferred embodiment, the method further includes real-time monitoring of the voltage, current and temperature of the target battery, and triggering protection measures within 50 milliseconds when overvoltage, undervoltage, overcurrent or overtemperature conditions exceeding a preset safety threshold occur.

[0080] In one possible implementation, the safety protection module is implemented in collaboration with a hardware comparator (such as the TI TLV3201, with a response time of 1μs) and an MCU: voltage thresholds (4.2V overvoltage, 2.8V undervoltage), current thresholds (100A overcurrent), and temperature thresholds (60°C overtemperature) are pre-configured at the comparator's reference voltage terminal (set via a DAC). When the sensor signal exceeds the threshold, the comparator triggers an interrupt in microseconds. Upon responding to the interrupt, the MCU executes protection actions within 50ms, including disconnecting the main relay (via a MOSFET driver circuit such as the IR2110), activating the fuse (such as a surface-mount fuse), and recording the fault code to an EEPROM. For communication, protection events are sent to the aircraft's main controller via CAN frames, with a frame ID of 0x100 and a data field containing the fault type and timestamp.

[0081] This embodiment employs a hardware-accelerated protection mechanism to ensure rapid response in the event of a fault. Experimental verification shows a response time of <50ms, effectively preventing safety accidents such as thermal runaway. The protection circuit power consumption is <10mW, meeting aviation low-power requirements.

[0082] In a preferred embodiment, the method further includes estimating the state of health (SOH) of the target battery based on parameter changes or capacity decay of the second-order RC equivalent circuit model.

[0083] In one possible implementation, SOH estimates the parameters (R0, R1, C1) and capacity Qn based on RLS identification: The weight , Typical values ​​were determined through regression analysis. =0.6, =0.4). The capacity Qn is calibrated using the ampere-hour integration method over a full charge-discharge cycle. The SOH result is updated every 24 hours, stored in Flash, and uploaded via the communication module. In terms of hardware, parameter identification and capacity calibration share ADC resources and employ time-division multiplexing.

[0084] This embodiment integrates parameters and capacity information, improving the robustness of SOH estimation. The error between measured and laboratory capacity test results is <5%, and the consistency of SOH tracking is good throughout the cycle life. The algorithm has low computational requirements, only 5kFLOPs per update.

[0085] In a preferred embodiment, the method further includes target battery equalization management based on the estimated SOC value to reduce the charge difference between individual cells within the target battery pack.

[0086] In one possible implementation, equalization management employs active equalization technology. Each BMU integrates equalization circuitry (such as a bidirectional DC-DC converter), controlling energy transfer via MOSFET switches (such as the IRF7416 with an internal resistance of 10mΩ). The equalization strategy is based on SOC difference: when the SOC difference between individual cells exceeds 3%, equalization is initiated, transferring energy from the high-SOC cell to the low-SOC cell. The equalization current is set to 1A and controlled by a PWM signal (frequency 100kHz). The equalization circuitry includes an inductor (100μH) and a capacitor (470μF), with efficiency improved through synchronous rectification technology.

[0087] This embodiment effectively reduces battery pack inconsistencies through SOC-guided equalization. Experiments show that the voltage difference between individual cells after equalization is <30mV, extending the overall battery pack lifespan. The equalization response time is <1s, and the energy transfer efficiency is >90%.

[0088] In a preferred embodiment, the target battery equalization management employs active equalization technology with an equalization efficiency of not less than 85%, and controls the voltage difference between individual cells of the target battery pack to within 30mV.

[0089] In one possible implementation, the active balancing circuit employs a flying capacitor structure with a capacitance of 100μF (ceramic capacitor, X7R material) and a switching frequency of 100kHz. Efficiency is measured using a power analyzer (such as a Yokogawa WT1800) to ensure it is above 85%. Voltage difference is monitored using a high-precision ADC (such as an ADS8588S, 16-bit resolution), and the control algorithm uses PID control to adjust the PWM duty cycle, ensuring the voltage difference is <30mV. PID parameters: Kp=0.1, Ki=0.01, Kd=0.001.

[0090] This embodiment improves energy utilization efficiency through efficient active balancing. The measured balancing efficiency is ≥85%, and the voltage difference is ≤30mV, which is superior to passive balancing schemes. The circuit is small in size, making it suitable for the compact spaces of aviation applications.

[0091] In a preferred embodiment, the method further includes fusing the SOC value estimated by the extended Kalman filter algorithm with the SOC reference value obtained by the ampere-hour integration method to correct the accumulated error.

[0092] In one possible implementation, the fusion algorithm uses a weighted average: The weight k is dynamically adjusted based on the SOC error covariance. When the EKF covariance is small, k increases (maximum 0.8); conversely, when the EKF covariance is large, the ampere-hour integration weight increases (minimum 0.2). The ampere-hour integration method calculates the weight through current integration. ; The initial SOC0 is obtained by looking up the OCV-SOC table, as shown in Table 2 below. Integration is triggered by the MCU's timer, with a sampling interval of 1 second.

[0093] Table 2: SOC-T-OCV

[0094] Figure 5 This is a Simulink simulation diagram of a battery remaining capacity calculation in real time using the ampere-hour integration method provided in an embodiment of the present invention, as shown below. Figure 5 As shown, the model includes a core calculation module: a current input interface, an integrator, a subtractor, a parameter configuration module (nominal capacity Cn, initial SOC value SOC0), and a SOC output interface. The output of the current input interface is electrically connected to the input of the integrator; the output of the integrator is electrically connected to the first input of the subtractor; the output of the parameter configuration module is electrically connected to the second input of the subtractor and the parameter configuration terminal of the integrator; the output of the subtractor is electrically connected to the SOC output interface to output the real-time SOC value. The current input interface receives real-time charging and discharging current signals from the battery. The charging current is positive, and the discharging current is negative, and it has a built-in charging and discharging efficiency coefficient η (range 0.95-0.98). In the parameter configuration module, the nominal capacity Cn is configured as 1.35 × 3600C (i.e., 2Ah), and the initial SOC value SOC0 is configured as 0.9. The integrator is used to perform the integration of current over time, with the integration formula as follows: The subtractor is used to calculate the real-time SOC based on the core formula of the ampere-hour integration method: The model has a pre-defined interface for integration with the Extended Kalman Filter (EKF) algorithm, which can transmit the output SOC baseline value to the EKF module and correct the accumulated error using voltage observations. In the uncorrected state, the SOC calculation error is ≤3%, and after EKF correction, the error can be reduced to within 2%.

[0095] This embodiment overcomes the cumulative error of ampere-hour integration and the model dependence of EKF by fusing multi-source data. Real-world testing shows that the SOC error remains stable within 2% during long-duration flight, the fusion algorithm's computational cost is <1kFLOPs, and it exhibits good real-time performance.

[0096] In a preferred embodiment, the method further includes interacting with the aircraft main controller via a communication module using condition monitoring data, condition estimation results, and fault diagnosis information.

[0097] In one possible implementation, the communication module uses a CAN bus (such as CAN 2.0B), with a data frame format including an ID (11 bits) and a data field (8 bytes). Status data (voltage, current, temperature, SOC, SOH) is transmitted every 100ms (ID0x200), while fault information is transmitted immediately (ID 0x100). The communication chip (such as MCP2562) connects to the MCU via SPI, with a bus termination resistor of 120Ω. Diagnostic information, including fault codes, timestamps, and recovery suggestions, is stored in a circular buffer (1KB in size).

[0098] This embodiment achieves seamless integration of the BMS with the aircraft system, supports real-time monitoring and diagnostics, and improves flight safety. The communication error rate is <1e-6, and the latency is <10ms, meeting aviation standards.

[0099] In a preferred embodiment, the electric aircraft BMS serves as the "intelligent central hub" of the aviation power battery, integrating functions such as state monitoring, state estimation, safety protection, equalization management, thermal management, communication, and diagnostics. In one possible implementation, the system is modeled using Simulink / Simscape, and an EKF algorithm module is written in Matlab. Experimental data, including OCV-SOC relationships (Table 3-3) and SOC-T-OCV (Table 3-1), are used to verify the model's accuracy.

[0100] Experimental content: In the constant current pulse discharge test, the initial SOC=0.9, current 1.35A, discharge 10s, rest 40s, the simulated terminal voltage compared with the actual voltage showed an error of <50mV. SOC estimation was achieved through the EKF algorithm, with an error of <2%. In addition, the distributed BMS architecture adopts a lightweight EKF, reducing the computational load by 70%, and the module-level SOC / SOH estimation error is <3%. The experimental setup includes a battery pack (10 cells in series), a temperature chamber (-40°C to 60°C), a load cell (0-100A), and a data acquisition system (sampling rate 1kHz).

[0101] Experimental results: At -20°C, the SOC estimation error is 1.5%, and the terminal voltage error is 40mV; during 3C discharge, the SOC error is 1.8%, and the voltage difference after equalization is 25mV. The safety protection response time is 45ms, and the communication delay is 8ms. The system power consumption is <5W, meeting aviation requirements.

[0102] This embodiment, through full-function integration and experimental verification, demonstrates the high precision, high reliability, and safety of the BMS system in electric aircraft applications, providing key technical support for the development of green aviation.

[0103] This invention, through the combination and refinement of the aforementioned preferred technical solutions, successfully constructs a high-precision, high-reliability SOC estimation and management system for electric aircraft batteries. Experimental data fully verifies the superior performance of this invention: under extreme temperatures (-40°C to 60°C) and high dynamic operating conditions (3C rate), the SOC estimation error is ≤2%, the terminal voltage prediction deviation is ≤50mV, the equalization efficiency is ≥85%, the voltage difference is ≤30mV, and the safety protection response time is <50ms. By introducing temperature compensation lookup tables, dynamic noise adjustment (EKF), online parameter identification, distributed architecture, and lightweight algorithms, the system's adaptability and robustness in real aviation environments are greatly improved. Furthermore, relying on communication and diagnostic integration, seamless interaction with the aircraft's main controller is achieved, ultimately achieving a comprehensive technical effect of high-precision estimation, efficient equalization, and low-power operation while ensuring flight safety.

[0104] Example 3 Based on the aforementioned method embodiments, this embodiment provides a battery SOC estimation system for electric aircraft. This system adopts a modular architecture, integrated into the battery management unit, and achieves high-precision estimation of the state of charge (SOC) of the electric aircraft's power battery through the coordinated operation of its various functional modules. See also... Figure 6 This is a schematic diagram of a battery SOC estimation system for an electric aircraft provided in an embodiment of the present invention. The system includes: a model building module 100, a two-dimensional lookup table module 200, an extended Kalman filter estimation module 300, and a dynamic adjustment module 400.

[0105] The model building module 100 is responsible for establishing and maintaining the second-order RC equivalent circuit model. Implemented through the digital signal processing kernel of a microprocessor, the core parameters of the model building module 100 include the ohmic internal resistance, and the resistance and capacitance values ​​of the two parallel RC branches. These parameters are stored in non-volatile memory based on experimental calibration results. The typical ohmic internal resistance is 0.072 ohms, the first RC branch has a resistance of 0.03 ohms and a capacitance of 1000 farads, and the second RC branch has a resistance of 0.05 ohms and a capacitance of 2000 farads. In terms of software implementation, this module is built using the Simulink environment and includes a current input interface, an RC branch calculation unit, and a voltage output interface, enabling accurate simulation of the battery's ohmic internal resistance voltage drop, electrochemical polarization, and concentration polarization effects. The module is connected to a current sensor and a temperature sensor via an analog-to-digital converter, with a sampling frequency of 1 kHz to ensure real-time data acquisition.

[0106] The two-dimensional lookup table module 200 is directly connected to the model building module 100 and is responsible for querying the corresponding open-circuit voltage value based on the real-time operating temperature and SOC estimate. This module is implemented using the microprocessor's lookup table unit, and the storage medium is electrically erasable programmable read-only memory with a storage capacity of 1 megabit. The lookup data is based on extensive experimental measurements, covering an SOC range from zero to one and a temperature range from -40°C to 60°C. The data format uses 16-bit fixed-point numbers to maintain an accuracy of 0.1 millivolts. For the lookup algorithm, bilinear interpolation is used to handle numerical calculations at non-integer points, ensuring accurate open-circuit voltage values ​​are obtained at any operating point and temperature. This module communicates with the temperature sensor and the SOC calculation submodule via a serial peripheral interface and transmits the queried open-circuit voltage value to the controlled voltage source input terminal of the model building module 100.

[0107] The extended Kalman filter estimation module 300 is tightly coupled with the model building module 100 and is responsible for performing real-time SOC estimation. This module runs on the microprocessor's floating-point unit, and the algorithm code is stored in flash memory. During module initialization, the process noise covariance matrix is ​​set to 10^-8 multiplied by a 3rd-order identity matrix, and the initial value of the observation noise covariance matrix is ​​0.01. The algorithm execution process includes two main stages: state prediction and measurement update, continuously revising the state estimate through recursive calculation. In the state prediction stage, the state vector and error covariance matrix for the next time step are predicted based on the system state equations; in the measurement update stage, the state estimate is corrected using the difference between the actual measured value and the predicted value, combined with the Kalman gain. This module establishes a connection with the model building module 100 and the dynamic adjustment module 400 via an interrupt signal line, and transmits the finally calculated SOC estimate to the main control unit via the controller area network bus.

[0108] The dynamic adjustment module 400, serving as the system's optimization unit, continuously monitors the real-time current and temperature changes of the battery. Integrated into the analog front-end of the microprocessor, this module includes sophisticated current and temperature sampling circuits. Current sampling employs a shunt combined with a differential amplifier, while temperature sampling utilizes a platinum resistance temperature sensor paired with an analog-to-digital converter. The module's core function is to dynamically adjust the noise covariance matrix in the extended Kalman filter algorithm by calculating the rates of change of current and temperature. When the detected rate of change of current exceeds 1 ampere per second or the rate of change of temperature exceeds 1 degree Celsius per second, the module automatically increases the values ​​of the process noise covariance matrix and the observation noise covariance matrix, enabling the algorithm to quickly track abrupt changes in state. When the rate of change is below 0.1 amperes per second and 0.1 degrees Celsius per second, the value of the noise covariance matrix is ​​appropriately decreased to improve the estimation accuracy under steady-state conditions. These adjustment parameters are directly written to the configuration register of the extended Kalman filter estimation module 300 via memory-mapped input / output.

[0109] The functional modules are interconnected via a system bus, forming a complete working loop. The model building module 100 provides a mathematical model of the battery's dynamic characteristics, the two-dimensional lookup table module 200 provides a temperature-compensated open-circuit voltage reference, the extended Kalman filter estimation module 300 executes the core estimation algorithm, and the dynamic adjustment module 400 optimizes the algorithm performance based on changes in operating conditions. The system employs a compact hardware design, with overall power consumption controlled below 5 watts, and is fully compatible with a distributed battery management system architecture. Experimental verification shows that the system's SOC estimation error does not exceed 2% within a temperature range of -40°C to 60°C; under 3x discharge conditions, the terminal voltage prediction deviation is less than 50 millivolts; and under the dynamic operating conditions unique to electric aircraft, the adaptive adjustment mechanism improves SOC estimation stability by 40%, fully meeting the stringent accuracy and reliability requirements of aviation applications.

[0110] This system effectively solves the challenge of battery state estimation in electric aircraft under complex operating conditions through innovative modular design and advanced adaptive algorithms. The collaborative work between the system's modules ensures estimation accuracy and real-time performance; the hardware design balances performance and power consumption; and the software algorithms provide ample flexibility and reliability. The implementation of this system provides technical support for electric aircraft battery management systems and has significant application value.

[0111] The above description is merely a preferred embodiment of this application and does not limit the patent scope of this invention. Any equivalent structural or procedural transformations made based on the description and drawings of this invention, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this invention.

Claims

1. A method for estimating the state of charge (SOC) of a battery in an electric aircraft, characterized in that, Includes the following steps: Construct a second-order RC equivalent circuit model and set up a two-dimensional lookup table module; The real-time operating temperature and real-time SOC estimate of the target battery are input into the two-dimensional lookup table module to query and output the corresponding open-circuit voltage value of the target battery, and the open-circuit voltage value is assigned to the second-order RC equivalent circuit model. Based on the second-order RC equivalent circuit model with the given open-circuit voltage value, the extended Kalman filter algorithm is applied to estimate the SOC value of the target battery in real time. During the execution of the extended Kalman filter algorithm, the real-time current and temperature of the target battery are monitored, and the process noise covariance matrix Q and observation noise covariance matrix R in the algorithm are dynamically adjusted based on the calculated current change rate and / or temperature change rate to optimize the estimation result of the SOC value.

2. The method according to claim 1, characterized in that, The two-dimensional lookup table module includes pre-stored experimental data on the open-circuit voltage of the target battery at different temperatures and SOCs.

3. The method according to claim 1, characterized in that, The method further includes the following steps: The parameters in the second-order RC equivalent circuit model are identified online using a recursive least squares method that runs in parallel with the extended Kalman filter algorithm.

4. The method according to claim 3, characterized in that, The parameters include the ohmic internal resistance R0 of the target battery, which is used to evaluate the state of health (SOH) of the target battery in real time.

5. The method according to claim 3, characterized in that, The recursive least squares method includes: An algorithm with a forgetting factor is adopted, and the forgetting factor is dynamically adjusted based on the fluctuation of the target battery current. When the current change rate is lower than the first preset threshold, a smaller forgetting factor is used to enhance the role of historical data. When the current change rate is higher than the second preset threshold, a larger forgetting factor is used to quickly track parameter changes.

6. The method according to claim 1, characterized in that, The method further includes the following steps: Based on the target battery's cycle aging count or cumulative capacity decay data, the OCV-T-SOC data in the two-dimensional lookup table module is dynamically updated through a predefined aging model.

7. The method according to claim 1, characterized in that, The method is implemented in a distributed BMS architecture, and the SOC estimation task assigned to the battery module controller for local execution adopts a lightweight extended Kalman filter algorithm with reduced computational load.

8. The method according to claim 1, characterized in that, The method further includes the following steps: It monitors the voltage, current and temperature of the target battery in real time, and triggers protection measures within 50 milliseconds when overvoltage, undervoltage, overcurrent or overtemperature conditions exceed the preset safety threshold.

9. The method according to claim 1, characterized in that, The method further includes the following steps: The target battery's state of health (SOH) is estimated based on parameter changes or capacity decay in the second-order RC equivalent circuit model.

10. A battery SOC estimation system for an electric aircraft, characterized in that, include: The model building module is used to build and maintain a second-order RC equivalent circuit model. A two-dimensional lookup table module, connected to the model building module, is used to receive the real-time operating temperature and real-time SOC estimate of the target battery, and query and output the corresponding open-circuit voltage value of the target battery to the model building module so as to assign values ​​to the second-order RC equivalent circuit model. An extended Kalman filter estimation module, connected to the model building module, is used to estimate the SOC value of the target battery in real time based on the second-order RC equivalent circuit model with the given open-circuit voltage value. The dynamic adjustment module, connected to the extended Kalman filter estimation module, is used to monitor the real-time current and temperature of the target battery, calculate the rate of change of current and / or the rate of change of temperature, and dynamically adjust the process noise covariance matrix Q and the observation noise covariance matrix R in the extended Kalman filter estimation module based on the rate of change to optimize the estimation result of the SOC value.

Images