Intelligent control method and system for rapid charging of lithium ion battery

Abstract

The invention discloses an intelligent control method and system for rapid charging of a lithium ion battery, and relates to the technical field of battery charging, and the system comprises a state monitoring module, an evaluation module, a parameter adjustment module, a prediction module, a charging control module, a safety protection module and a central processing unit. According to the method, the voltage, current, temperature and internal resistance data of the battery are collected in real time through a multi-source sensor, the health state of the battery and the charging state evaluation result are fused, the charging parameters are dynamically adjusted through fuzzy logic and PID control, then the optimal charging curve is predicted, and a multi-stage self-adaptive charging protocol is implemented; and finally, full-process safety control is realized through deep learning fault diagnosis and a multi-stage protection mechanism, intelligent perception, dynamic decision and closed-loop control are realized, and the charging speed, the battery life and the safety are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of battery charging technology, and in particular to an intelligent control method and system for fast charging of lithium-ion batteries. Background Technology

[0002] A lithium-ion battery is a rechargeable battery that uses lithium ions as charge carriers and can be repeatedly charged and discharged. Its core consists of a positive electrode, a negative electrode, an electrolyte, and a separator. It stores and releases electrical energy by inserting / deintercalating lithium ions between the positive and negative electrodes. Its core function is to efficiently store and supply electrical energy, replacing traditional fossil fuels and disposable batteries, while taking into account environmental protection, high efficiency, and portability.

[0003] Currently, under complex operating conditions, the battery status signals collected by sensors are easily affected by electromagnetic interference and circuit noise, leading to data distortion; traditional status assessment methods are difficult to accurately track the nonlinear characteristic changes caused by battery aging, resulting in limited assessment accuracy; charging strategies are mostly fixed modes and cannot be dynamically optimized according to the real-time battery status, making it difficult to balance charging speed and battery life.

[0004] Therefore, an intelligent control method and system for fast charging of lithium-ion batteries is proposed to solve the above problems. Summary of the Invention

[0005] The main objective of this invention is to provide an intelligent control method and system for fast charging of lithium-ion batteries, so as to solve the problems mentioned in the background above.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is: an intelligent control method and system for fast charging of lithium-ion batteries, the method comprising the following steps:

[0007] S1. Real-time monitoring of multiple state parameters of lithium-ion batteries, including voltage, current, temperature and internal resistance;

[0008] S2. Based on the aforementioned state parameters, assess the battery's health and charging status;

[0009] S3. Based on the assessed health status and charging status, dynamically adjust the parameters of charging current and charging voltage;

[0010] S4. Utilize machine learning algorithms to predict the optimal charging curve for the battery, in order to minimize charging time and extend battery life.

[0011] S5. Based on the predicted optimal charging curve, implement a multi-stage charging protocol, including a constant current charging stage, a constant voltage charging stage, and a pulse charging stage.

[0012] S6. During the entire charging process, real-time safety protection and fault handling are performed, including overvoltage, overcurrent, and overheat protection.

[0013] Preferably, the real-time monitoring of battery status parameters in step S1 includes the following steps:

[0014] S11. Collect battery voltage, current and temperature data through high-precision sensors;

[0015] S12. Measure the internal resistance of the battery using an electrical signal decoupling algorithm based on a dual-pulse sequence;

[0016] S13. Sample and filter the data at a preset sampling frequency, and use wavelet transform to remove noise to ensure the accuracy and real-time performance of the data.

[0017] Preferably, the assessment of the battery's health and charging status in S2 includes the following steps:

[0018] S21. The state of charge is estimated by fusing the open-circuit voltage method and the coulomb counting method, and the parameters of the estimation model are optimized by introducing the particle swarm optimization algorithm to improve the accuracy.

[0019] S22. The health status of the battery is assessed by analyzing its capacity decay and internal resistance changes, and the lifespan is predicted by combining the support vector machine model.

[0020] S23. Use an extended Kalman filter to smooth and correct the battery health and charge state estimation results to handle nonlinear battery characteristics.

[0021] Preferably, the dynamic adjustment of charging parameters in S3 includes the following steps:

[0022] S31. Calculate the maximum allowable charging current in real time based on the battery's health status and charging status, and dynamically adjust the charging voltage threshold based on the battery temperature.

[0023] S32. A fuzzy logic controller is used to adaptively adjust the charging parameters, wherein the fuzzy rules are updated in real time based on the battery aging model and environmental conditions.

[0024] S33. Introduce a feedback control loop to use a PID controller to compensate for errors in parameter adjustment and ensure charging stability.

[0025] Preferably, the step S32, which uses a fuzzy logic controller to adaptively adjust the charging parameters, includes the following steps:

[0026] S321. Define the input variables as the battery's health state, charging state, and temperature, and the output variables as the charging current and voltage, and use the Gaussian membership function to quantize the fuzzy set.

[0027] S322. Design a fuzzy rule base, set rules based on expert knowledge, and dynamically optimize rule weights using a genetic algorithm;

[0028] S323. Use the Mamdani-type fuzzy inference system to calculate the optimal charging parameters in real time, and combine it with the defuzzification method to output precise control values.

[0029] Preferably, the step S4, which uses a machine learning algorithm to predict the optimal charging curve, includes the following steps:

[0030] S41. Collect historical charging data, including charging curves, battery status and performance degradation data, and construct a multi-source data fusion feature set;

[0031] S42. Train a neural network model to predict the optimal charging curve that minimizes charging time and capacity loss under a given battery condition.

[0032] S43. Use reinforcement learning algorithms to optimize the charging curve online, where the reward function is dynamically adjusted based on charging efficiency and battery life indicators.

[0033] Preferably, training the neural network model in S42 includes the following steps:

[0034] S421. Construct a deep belief network as a prediction model. Its input layer includes battery voltage, current, temperature, health status and charging status features.

[0035] S422. Use the backpropagation algorithm and adaptive moment estimator to optimize the model parameters, and introduce dropout technology to prevent overfitting;

[0036] S423. Evaluate model performance through cross-validation and update the model regularly using an online learning mechanism to adapt to battery aging characteristics.

[0037] Preferably, implementing the multi-stage charging protocol in S5 includes the following steps:

[0038] S51. During the constant current charging phase, charge the battery with a constant current until the battery voltage reaches the first threshold, and dynamically adjust the threshold according to the battery's health status.

[0039] S52. During the constant voltage charging phase, charge at a constant voltage until the charging current drops to the second threshold, and introduce current decay rate monitoring to optimize the switching timing.

[0040] S53. During the pulse charging phase, based on the real-time battery charging status, temperature, and polarization voltage feedback, a pulse current with matching variable frequency, duty cycle, and amplitude is applied to eliminate the polarization effect, improve charging efficiency, and adjust the pulse parameters according to the real-time temperature.

[0041] Preferably, the safety protection and fault handling in S6 includes the following steps:

[0042] S61. Set up a multi-level protection mechanism. When overvoltage, overcurrent or overheating is detected, immediately reduce the charging current or stop charging, and start the backup power supply to maintain system safety.

[0043] S62. Use a deep learning-based fault diagnosis algorithm to identify battery anomalies, such as internal short circuits or accelerated aging, and analyze voltage fluctuation patterns through a convolutional neural network.

[0044] S63. Record fault events and trigger alarms or maintenance notifications, while remotely monitoring and updating protection policies through a cloud platform.

[0045] An intelligent control system for fast charging of lithium-ion batteries, the system comprising:

[0046] Status monitoring module: Real-time monitoring of battery voltage, current, temperature and internal resistance, and integrates high-precision sensors and data acquisition units;

[0047] Evaluation module: Evaluates the battery's health and charging status, including the processor and memory storing algorithm code;

[0048] Parameter adjustment module: dynamically adjusts charging current and charging voltage, including fuzzy logic controller and PID compensation unit;

[0049] Prediction module: Utilizes machine learning algorithms to predict the optimal charging curve, including a neural network processor and a reinforcement learning engine;

[0050] Charging control module: Implements multi-stage charging protocols, including constant current, constant voltage, and pulse charging controllers;

[0051] Safety protection module: Provides real-time safety protection and fault handling, including multi-level protection circuits and fault diagnosis units;

[0052] Central processing unit: coordinates the work of each module and connects to external devices through a communication interface.

[0053] The present invention has the following beneficial effects:

[0054] 1. In this invention, during the rapid charging of lithium-ion batteries, voltage, current, temperature and internal resistance data are collected synchronously by multiple source sensors. The signal processing technology combining wavelet transform and electrical signal decoupling algorithm can effectively separate and suppress electromagnetic interference and circuit noise, ensuring the purity of battery status signals collected under complex charging conditions, reducing evaluation errors caused by noise interference, and thus improving the accuracy and reliability of battery status monitoring.

[0055] 2. In this invention, when dynamically optimizing the charging strategy, by integrating deep belief networks and reinforcement learning algorithms, and fusing battery health status and real-time operating parameters, the optimal charging curve can be intelligently predicted and dynamically adjusted to achieve precise control of the charging process, avoid battery overcharging or performance degradation caused by fixed charging modes, and ensure the best balance between fast charging and battery life protection.

[0056] 3. In this invention, when performing charging safety protection and fault diagnosis, by constructing a multi-level protection mechanism and a deep learning fault diagnosis model, it is possible to intelligently identify abnormal battery states based on real-time voltage fluctuation characteristics, realize early warning and active protection against potential faults, improve the system's response capability and timeliness in dealing with safety hazards such as overvoltage, overcurrent, and overheating, and enhance the safety and reliability of the charging process. Attached Figure Description

[0057] Figure 1 This is a flowchart of an intelligent control method for fast charging of a lithium-ion battery according to the present invention;

[0058] Figure 2 This is a framework diagram of an intelligent control system for fast charging of lithium-ion batteries according to the present invention.

[0059] Figure 3 This is a framework diagram of the state monitoring module of an intelligent control system for fast charging of lithium-ion batteries according to the present invention.

[0060] Figure 4 This is a framework diagram of the parameter adjustment module of an intelligent control system for fast charging of lithium-ion batteries according to the present invention.

[0061] Figure 5 This is a framework diagram of the safety protection module of an intelligent control system for fast charging of lithium-ion batteries according to the present invention. Detailed Implementation

[0062] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] Please see Figure 1-5 This invention provides a technical solution: an intelligent control method for fast charging of lithium-ion batteries, the method comprising the following steps:

[0064] S1. Real-time monitoring of multiple state parameters of lithium-ion batteries, including voltage, current, temperature and internal resistance;

[0065] S2. Based on state parameters, assess the battery's health and charging status;

[0066] S3. Based on the assessed health status and charging status, dynamically adjust the parameters of charging current and charging voltage;

[0067] S4. Utilize machine learning algorithms to predict the optimal charging curve for the battery, in order to minimize charging time and extend battery life.

[0068] S5. Based on the predicted optimal charging curve, implement a multi-stage charging protocol, including a constant current charging stage, a constant voltage charging stage, and a pulse charging stage.

[0069] S6. During the entire charging process, real-time safety protection and fault handling are performed, including overvoltage, overcurrent, and overheat protection.

[0070] Real-time monitoring of battery status parameters in S1 includes the following steps:

[0071] S11. Collect battery voltage, current and temperature data through high-precision sensors;

[0072] S12. Measure the internal resistance of the battery using an electrical signal decoupling algorithm based on a dual-pulse sequence, including the following steps:

[0073] S121. Set the dual-pulse sequence parameters: first pulse current amplitude (in (Battery rated capacity), pulse width Its main purpose is to excite the battery to measure the pure ohmic response, and due to its extremely short timescale, it can effectively avoid the significant establishment of slow polarization processes such as concentration polarization; then, after a short resting interval (For example, after 50ms) a second pulse is applied, with a current amplitude of Pulse width The width of the second pulse is designed to be sufficient to excite the electrode polarization and concentration polarization processes of the battery, while maintaining a high sampling rate. Synchronous acquisition of voltage response data and current data Record the voltage changes before, during, and after the pulse is applied;

[0074] S122. Using an electrical signal decoupling algorithm, the ohmic internal resistance and polarization internal resistance are calculated based on the double-pulse response data. First, the ohmic internal resistance is calculated. The calculation utilizes the voltage surge instantaneous upon application of the first pulse, as the ohmic resistance response completes within microseconds and is unaffected by subsequent polarization processes. The calculation formula is:

[0075] ;

[0076] in It is the instant the pulse is applied (time point) The voltage change, and These are the instantaneous voltage values ​​before and after the pulse application, respectively. It is the current amplitude of the first pulse;

[0077] Then, the polarization resistance is extracted from the voltage response curve. The voltage relaxation (recovery) process after the second pulse ends is analyzed. The relaxation voltage curve reflects the dissipation of polarization charge. Fitting this process can accurately quantify the polarization resistance. An exponential model containing two time constants is used for fitting to distinguish between electrochemical polarization and concentration polarization.

[0078] ;

[0079] in It is the steady-state voltage after relaxation ends. It is the amplitude. It is a time constant, and the parameters are fitted using the least squares method. Polarization internal resistance The calculation is as follows: Total internal resistance The sum of ohmic resistance and polarization resistance: ;

[0080] S123. By averaging measurements at multiple state points, the accuracy of internal resistance measurement is improved. Temperature compensation is applied, and multiple pulses are applied at different SOC points of the battery to calculate the average internal resistance value.

[0081] ;

[0082] in The number of pulses, and also based on battery temperature. To compensate for internal resistance, the Arrhenius equation is used:

[0083] ;

[0084] in It is the activation energy of battery materials. It is the gas constant. This is the reference temperature, and the compensated internal resistance. For subsequent control purposes;

[0085] S13. Sample and filter the data at a preset sampling frequency, and use wavelet transform to remove noise to ensure data accuracy and real-time performance, including the following steps:

[0086] S131, Preset sampling frequency Voltage, current, and temperature data are sampled and filtered using a first-order Butterworth low-pass filter with a cutoff frequency of [missing information]. To remove high-frequency noise;

[0087] S132. Select the Daubechies wavelet basis function to perform multi-resolution decomposition on the filtered data, setting the decomposition level to 5 levels. The wavelet decomposition formula is as follows:

[0088] ;

[0089] in Represents wavelet coefficients, For scale parameters ( ), For translation parameters, The filtered signal These are the Daubechies wavelet basis functions;

[0090] A soft thresholding denoising method is used for high-frequency detail coefficients, and a threshold is set. The soft thresholding formula is:

[0091] ;

[0092] in For high-frequency detail coefficients, These are the denoised coefficients.

[0093] S133. Reconstruct the denoised signal using the inverse wavelet reconstruction formula:

[0094] ;

[0095] in These are the denoised wavelet coefficients. It is the reconstructed signal;

[0096] The denoising effect was verified by calculating the signal-to-noise ratio (SNR) and root mean square error (RMSE):

[0097] ;

[0098] in and The power of the signal and noise are respectively, and real-time performance is ensured by guaranteeing processing time. This ensures that the requirements of the control system are met.

[0099] The S2 assessment of battery health and charge status includes the following steps:

[0100] S21. The state of charge is estimated based on the fusion of the open-circuit voltage method and the coulomb counting method, and the parameters of the estimation model are optimized by introducing the particle swarm optimization algorithm to improve the accuracy. This includes the following steps:

[0101] S211. First, after the battery has been left to stand for a sufficient period of time, measure its open-circuit voltage (OCV) and then use the pre-stored OCV-SOC lookup table function. Obtain the SOC estimate using the open-circuit voltage method;

[0102] Meanwhile, the coulomb counting method is used for integral calculation:

[0103] ;

[0104] in For the previous moment value, For the battery's rated capacity, This represents the real-time current (positive for charging, negative for discharging). Coulomb efficiency related to current;

[0105] S212. Construct a fusion estimation model and apply the particle swarm optimization algorithm to dynamically optimize the model weight parameters. The fusion estimation model is as follows:

[0106] ;

[0107] in The estimated state of charge after fusion; This is the estimated state of charge obtained based on the open-circuit voltage lookup table method; The state of charge estimate is obtained based on the Coulomb integral method; For adaptive weights, and satisfying ;

[0108] Dynamically optimize weights using the particle swarm optimization algorithm. Define the particle position vector as The velocity vector is fitness function Defined as The negative of the root mean square error (RMSE) of the difference between the estimated value and the true reference value (e.g., measured by high-precision equipment):

[0109] ;

[0110] in The fitness value of particle i The total number of reference data points used for evaluation, The true reference value of the charging state obtained by measuring the j-th data point using a high-precision device. To use the weight of particle i The estimated state of charge obtained by fusing the j-th data point This is a calculation for the root mean square error (RMSE).

[0111] The update formula for PSO is:

[0112] ;

[0113] ;

[0114] in For inertial weights, As a learning factor, A random number within the interval [0,1]. and Let be the velocity and position of particle i in the t-th iteration, respectively. This represents the position of particle i with the highest fitness value in its own history (up to generation t). The optimal weight is obtained by iteratively optimizing the position with the highest fitness value in the entire history of the particle swarm (up to generation t). It is used for real-time SOC fusion estimation.

[0115] S22. Assess battery health by analyzing capacity decay and internal resistance changes, and predict battery life using a support vector machine model.

[0116] S23. Smooth and correct the battery health and state of charge estimation results using an extended Kalman filter to handle nonlinear battery characteristics, including the following steps:

[0117] S231. Establish the Kalman state equation and the observation equation. The state equation is:

[0118] ;

[0119] The observation equation is:

[0120] ;

[0121] in This includes process noise and observation noise. The battery charging state at time k; The voltage across the RC link is used to simulate the electrochemical polarization process, which reflects the kinetic polarization during charge transfer at the electrode / electrolyte interface. The voltage across the RC link is used to simulate the concentration polarization process. This voltage reflects the concentration gradient caused by the limited diffusion of lithium ions in the bulk phase of the electrode active material or the electrolyte.

[0122] S232. Execute the extended Kalman filter recursive algorithm to jointly estimate the SOC and model parameters, and output the smoothed result. The prediction steps are as follows:

[0123] ;

[0124] ;

[0125] in It is a state transition function exist Jacobian matrix at the location;

[0126] The update steps are as follows:

[0127] ;

[0128] ;

[0129] ;

[0130] in It is the observation function exist The Jacobian matrix at that location.

[0131] The S3's dynamic adjustment of charging parameters includes the following steps:

[0132] S31. Calculate the maximum allowable charging current in real time based on the battery's health status and charging status, and dynamically adjust the charging voltage threshold based on the battery temperature.

[0133] S32. A fuzzy logic controller is used to adaptively adjust the charging parameters, wherein the fuzzy rules are updated in real time based on the battery aging model and environmental conditions.

[0134] S33. Introduce a feedback control loop to compensate for errors in parameter adjustment using a PID controller, ensuring charging stability. This includes the following steps:

[0135] S331. Design a PID controller that uses current / voltage tracking error as input, and define the current tracking error. Voltage tracking error The discrete form of the PID controller is:

[0136] ;

[0137] in For controlling the output (used to fine-tune the output of the charging power supply). These are the proportional, integral, and differential coefficients.

[0138] S332. Tune the PID parameters and integrate them into the charging control loop. Use the Ziegler-Nichols method for parameter tuning, integrate the PID controller into the main control loop, and adjust the output of the fuzzy logic controller. and Closed-loop tracking compensation is implemented to ensure the stability of the charging process;

[0139] The S32 uses a fuzzy logic controller to adaptively adjust the charging parameters, which includes the following steps:

[0140] S321. Define the input variables as the battery's health state, charging state, and temperature, and the output variables as the charging current and voltage, and use the Gaussian membership function to quantize the fuzzy set.

[0141] S322. Design a fuzzy rule base, set rules based on expert knowledge, and dynamically optimize rule weights using a genetic algorithm;

[0142] S323. Calculate the optimal charging parameters in real time using a Mamdani-type fuzzy inference system, and output precise control values ​​using a defuzzification method, including the following steps:

[0143] S3231. Employing Mamdani-type fuzzy inference, calculate the rule trigger strength for each input combination. ;

[0144] S3232. Use the area center method for deblurring and calculate the precise output value:

[0145] , ;

[0146] in The charging current adjustment factor is for precise output. The charging voltage adjustment factor is for precise output. The total number of activated fuzzy rules. and The output membership function center value corresponding to the i-th rule is given, and the final charging parameters are:

[0147] , ;

[0148] in This is the maximum allowable charging current calculated based on the battery state. This is the adjusted charging voltage threshold based on battery status and temperature. and This refers to the precise charging parameter setting value output by the fuzzy logic controller, which will be sent to the PID controller for tracking.

[0149] Predicting the optimal charging curve using machine learning algorithms in S4 includes the following steps:

[0150] S41. Collect historical charging data, including charging curves, battery status and performance degradation data, and construct a multi-source data fusion feature set;

[0151] S42. Train a neural network model to predict the optimal charging curve that minimizes charging time and capacity loss under a given battery condition.

[0152] S43. Optimize the charging curve online using a reinforcement learning algorithm, where the reward function is dynamically adjusted based on charging efficiency and battery life indicators, including the following steps:

[0153] S431. Construct a reinforcement learning environment, defining the state space, action space, and reward function. The state space S includes: current battery charging state, health state, voltage, and temperature; the action space A is the charging current adjustment amount. The reward function is designed as follows:

[0154] ;

[0155] in These are the weighting coefficients, and It is dynamically adjusted according to the charging stage; These are: charging speed bonus, temperature rise penalty, electrochemical stress penalty, and aging penalty;

[0156] S432. Online optimization is performed using the Deep Q-Network algorithm, and the network parameters are updated periodically. The action value function update of DQN adopts the Bellman equation:

[0157] ;

[0158] in The expected cumulative discounted return for taking action a in state s. The immediate reward obtained after taking action a For discount factor, In the new state Below, all possible actions The highest action value;

[0159] The target network parameters are updated every C steps: ,in These are parameters for soft updates.

[0160] Training a neural network model in S42 includes the following steps:

[0161] S421. Construct a deep belief network as a prediction model. Its input layer includes battery voltage, current, temperature, health status and charging status features.

[0162] S422. Optimize model parameters using the backpropagation algorithm and adaptive moment estimator, and introduce dropout technology to prevent overfitting, including the following steps:

[0163] S4221. Apply Dropout technology during the fine-tuning phase, using probability during forward propagation. Randomly discard hidden layer neurons and correct the output as follows:

[0164] ;

[0165] in The output value of the original hidden layer neurons; This represents the probability that the neuron is preserved. Let be a random variable following a Bernoulli distribution, with probability . The value is 1, with probability The value is 0; The output value of the neuron after applying Dropout;

[0166] S4222. Use the Adam optimizer to adaptively adjust the learning rate. The parameter update rule for the Adam optimizer is as follows:

[0167] ;

[0168] ;

[0169] , ;

[0170] ;

[0171] in This represents the current iteration number. For the loss function with respect to the parameters In the current iteration gradient, This is an estimate of the exponential moving average of the first moment (mean) of the gradient. This is an estimate of the exponential moving average of the second moment of the gradient (the uncentered variance). To Estimation after bias correction To Estimation after bias correction For the exponential decay rates of the first and second moments, For a very small constant , These are the model parameters at the t-th iteration. These are the updated model parameters;

[0172] S423. Evaluate model performance through cross-validation and update the model periodically using an online learning mechanism to adapt to battery aging characteristics, including the following steps:

[0173] S4231. Use k-fold cross-validation to evaluate the model's generalization ability. Divide the dataset into k=5 parts, alternating between using 4 parts as the training set and 1 part as the test set. The model performance metrics are:

[0174] ;

[0175] in The mean absolute percentage error, Let be the total number of samples in the test set, and be the true value of the i-th sample. Let be the model's predicted value for the i-th sample;

[0176] Require The verification can be completed;

[0177] S4232. Establish an online learning mechanism to periodically update model parameters. After every N=50 charging cycles, update the model with new data increments, and decay the learning rate to 0.1 times the initial value.

[0178] ;

[0179] in For the updated learning rate, The initial learning rate, This represents the total number of charging cycles that have been completed so far. The number of charging cycle intervals that trigger model updates. To Round down;

[0180] By using a sliding window to retain the most recent M=1000 sets of data, the model is ensured to always adapt to the current aging state of the battery.

[0181] Implementing a multi-stage charging protocol in the S5 includes the following steps:

[0182] S51. During the constant current charging phase, charge the battery with a constant current until the battery voltage reaches a first threshold, and dynamically adjust this threshold according to the battery's health status, including the following steps:

[0183] S511. Dynamically calculate the voltage threshold for the constant current charging stage based on the battery state of health (SOH). The adjustment is made dynamically based on the battery's health status, and the calculation formula is as follows:

[0184] ;

[0185] in The standard charging cutoff voltage for brand new batteries, SOH compensation coefficient, Battery health percentage;

[0186] S512. Implement constant current charging control, monitor voltage changes in real time, and use a PID controller to maintain a constant charging current. The control law is:

[0187] ;

[0188] in The adjusted control signal, Let k be the current tracking error at time k. The set constant charging current value, The actual measured charging current value at time k. These are the proportional, integral, and derivative coefficients of the PID controller. The sampling time interval;

[0189] When the battery voltage is detected Immediately switch to the constant voltage charging stage;

[0190] S52. During the constant voltage charging phase, charge at a constant voltage until the charging current drops to the second threshold, and introduce current decay rate monitoring to optimize the switching timing, including the following steps:

[0191] S521. Set a second current threshold based on battery health status and temperature, and a cutoff current threshold for the constant voltage charging phase. The calculation is as follows:

[0192] ;

[0193] in For the reference cutoff current ratio, For the rated capacity of the battery, This is the temperature compensation function;

[0194] S522: Dynamically monitor the charging status and determine the optimal switching time based on the cumulative charging amount and current decay trend. In specific implementation, the following two key parameters are calculated in real time:

[0195] Accumulated charging amount during constant voltage phase :

[0196] ;

[0197] in This represents the total amount of electricity charged from the start of the constant voltage phase up to time k. Let j be the actual charging current at time j. This marks the sampling point at the start of the constant pressure phase. Sampling time interval;

[0198] Moving average of current decay rate :

[0199] ;

[0200] in To calculate the time window, For smoothing coefficients;

[0201] The conditions for switching to the pulse charging phase have been revised to allow switching to occur when any of the following conditions are met:

[0202] Condition A (Basic battery capacity is met): ,in This indicates that 95% of the target power has been replenished, ensuring that the main power has been charged.

[0203] Condition B (Inefficient): and ,in This is a very small threshold, indicating that the current has dropped to a safe limit and hardly decays anymore, and continuing constant voltage charging will result in extremely low efficiency.

[0204] Condition C (Abnormal Protection): This condition serves as a safety backup, ensuring that the current will not drop indefinitely under any circumstances;

[0205] S53. During the pulse charging phase, based on real-time battery charging status, temperature, and polarization voltage feedback, a pulse current with matched variable frequency, duty cycle, and amplitude is applied to eliminate polarization effects and improve charging efficiency. The pulse parameters are adjusted according to the real-time temperature, including the following steps:

[0206] S531. Calculate the optimal pulse parameters and pulse frequency in real time based on battery status. Adjust according to battery health status:

[0207] ;

[0208] in As the reference frequency, Let SOC be the influence function; This is the temperature compensation function. As temperature increases, ion diffusion accelerates, so the frequency can be appropriately increased. For health status compensation function, A decrease in SOH indicates battery aging and increased polarization, and the frequency should be reduced accordingly.

[0209] Pulse duty cycle According to polarization voltage Adjustment:

[0210] ;

[0211] in As the baseline duty cycle, For real-time polarization voltage, For adjustment coefficients, This refers to the battery's nominal voltage.

[0212] S532. Adjust the pulse current amplitude in real time according to the temperature. The temperature compensation formula is:

[0213] ;

[0214] in , This is the temperature compensation coefficient. The optimal temperature;

[0215] S533. Implement pulse charging control and monitor the elimination of polarization effect in real time by monitoring the voltage recovery rate during relaxation. To evaluate the polarization elimination effect:

[0216] ;

[0217] in Relaxation period Battery voltage at any given moment;

[0218] when (in When the polarization effect is fully eliminated, charging is considered complete.

[0219] S6's safety protection and fault handling include the following steps:

[0220] S61. Set up a multi-level protection mechanism. When overvoltage, overcurrent or overheating is detected, immediately reduce the charging current or stop charging, and start the backup power supply to maintain system safety.

[0221] S62. Use a deep learning-based fault diagnosis algorithm to identify battery anomalies, such as internal short circuits or accelerated aging, and analyze voltage fluctuation patterns using a convolutional neural network, including the following steps:

[0222] S621. Construct a CNN network for voltage sequence feature extraction, with the input layer receiving a voltage sequence of length N=100. The first convolutional layer is calculated as follows:

[0223] ;

[0224] The kernel size is 3 and the stride is 1. For the i-th convolution output value, To represent a segment from the i-th element to the (i+2)-th element in a voltage sequence, For convolution kernel weights, For bias, For ReLU activation function: ;

[0225] The pooling layer uses max pooling:

[0226] ;

[0227] in This is the i-th output value after pooling. and This represents two adjacent values ​​in the output of the convolutional layer.

[0228] S622. Train a multi-class fault diagnosis model, and use the Softmax function in the output layer to calculate the probability of each type of fault:

[0229] ;

[0230] in This represents the probability of fault type j under input X. For the value of the j-th output node, Corresponding fault types: Normal, Internal short circuit, Accelerated aging, Abnormal connection;

[0231] The loss function uses cross-entropy:

[0232] ;

[0233] in This represents the cross-entropy loss value. Indicates the number of samples. This represents the indicator value of the j-th class in the true label of the i-th sample. This represents the probability that the model predicts the i-th sample belongs to the j-th class;

[0234] Parameters are updated using the Adam optimizer, and the learning rate is... ;

[0235] S63. Record fault events and trigger alarms or maintenance notifications, while remotely monitoring and updating protection policies through a cloud platform.

[0236] An intelligent control system for fast charging of lithium-ion batteries, the system comprising:

[0237] Status monitoring module: Real-time monitoring of battery voltage, current, temperature and internal resistance, and integrates high-precision sensors and data acquisition units;

[0238] Evaluation module: Evaluates the battery's health and charging status, including the processor and memory storing algorithm code;

[0239] Parameter adjustment module: dynamically adjusts charging current and charging voltage, including fuzzy logic controller and PID compensation unit;

[0240] Prediction module: Utilizes machine learning algorithms to predict the optimal charging curve, including a neural network processor and a reinforcement learning engine;

[0241] Charging control module: Implements multi-stage charging protocols, including constant current, constant voltage, and pulse charging controllers;

[0242] Safety protection module: Provides real-time safety protection and fault handling, including multi-level protection circuits and fault diagnosis units;

[0243] Central processing unit: coordinates the work of each module and connects to external devices through a communication interface.

[0244] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus.

[0245] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart control method for fast charging of lithium-ion batteries, characterized in that, The method includes the following steps: S1. Real-time monitoring of multiple state parameters of lithium-ion batteries, including voltage, current, temperature and internal resistance; S2. Based on the aforementioned state parameters, assess the battery's health and charging status; S3. Based on the assessed health status and charging status, dynamically adjust the parameters of charging current and charging voltage; S4. Utilize machine learning algorithms to predict the optimal charging curve for the battery, in order to minimize charging time and extend battery life. S5. Based on the predicted optimal charging curve, implement a multi-stage charging protocol, including a constant current charging stage, a constant voltage charging stage, and a pulse charging stage. S6. During the entire charging process, real-time safety protection and fault handling are performed, including overvoltage, overcurrent, and overheat protection.

2. The intelligent control method for fast charging of a lithium-ion battery according to claim 1, characterized in that: The real-time monitoring of battery status parameters in S1 includes the following steps: S11. Collect battery voltage, current and temperature data through high-precision sensors; S12. Measure the internal resistance of the battery using an electrical signal decoupling algorithm based on a dual-pulse sequence; S13. Sample and filter the data at a preset sampling frequency, and use wavelet transform to remove noise to ensure the accuracy and real-time performance of the data.

3. The intelligent control method for fast charging of a lithium-ion battery according to claim 1, characterized in that: The assessment of the battery's health and charge status in step S2 includes the following steps: S21. The state of charge is estimated by fusing the open-circuit voltage method and the coulomb counting method, and the parameters of the estimation model are optimized by introducing the particle swarm optimization algorithm to improve the accuracy. S22. The health status of the battery is assessed by analyzing its capacity decay and internal resistance changes, and the lifespan is predicted by combining the support vector machine model. S23. Use an extended Kalman filter to smooth and correct the battery health and charge state estimation results to handle nonlinear battery characteristics.

4. The intelligent control method for fast charging of a lithium-ion battery according to claim 1, characterized in that: The S3 mid-level dynamic adjustment of charging parameters includes the following steps: S31. Calculate the maximum allowable charging current in real time based on the battery's health status and charging status, and dynamically adjust the charging voltage threshold based on the battery temperature. S32. A fuzzy logic controller is used to adaptively adjust the charging parameters, wherein the fuzzy rules are updated in real time based on the battery aging model and environmental conditions. S33. Introduce a feedback control loop to use a PID controller to compensate for errors in parameter adjustment and ensure charging stability.

5. The intelligent control method for fast charging of a lithium-ion battery according to claim 4, characterized in that: The S32 step of adaptively adjusting the charging parameters using a fuzzy logic controller includes the following steps: S321. Define the input variables as the battery's health state, charging state, and temperature, and the output variables as the charging current and voltage, and use the Gaussian membership function to quantize the fuzzy set. S322. Design a fuzzy rule base, set rules based on expert knowledge, and dynamically optimize rule weights using a genetic algorithm; S323. Use the Mamdani-type fuzzy inference system to calculate the optimal charging parameters in real time, and combine it with the defuzzification method to output precise control values.

6. The intelligent control method for fast charging of a lithium-ion battery according to claim 1, characterized in that: The step of predicting the optimal charging curve using a machine learning algorithm in S4 includes the following steps: S41. Collect historical charging data, including charging curves, battery status and performance degradation data, and construct a multi-source data fusion feature set; S42. Train a neural network model to predict the optimal charging curve that minimizes charging time and capacity loss under a given battery condition. S43. Use reinforcement learning algorithms to optimize the charging curve online, where the reward function is dynamically adjusted based on charging efficiency and battery life indicators.

7. The intelligent control method for fast charging of a lithium-ion battery according to claim 6, characterized in that: Training the neural network model in S42 includes the following steps: S421. Construct a deep belief network as a prediction model. Its input layer includes battery voltage, current, temperature, health status and charging status features. S422. Use the backpropagation algorithm and adaptive moment estimator to optimize the model parameters, and introduce dropout technology to prevent overfitting; S423. Evaluate model performance through cross-validation and update the model regularly using an online learning mechanism to adapt to battery aging characteristics.

8. The intelligent control method for fast charging of a lithium-ion battery according to claim 1, characterized in that: The implementation of the multi-stage charging protocol in S5 includes the following steps: S51. During the constant current charging phase, charge the battery with a constant current until the battery voltage reaches the first threshold, and dynamically adjust the threshold according to the battery's health status. S52. During the constant voltage charging phase, charge at a constant voltage until the charging current drops to the second threshold, and introduce current decay rate monitoring to optimize the switching timing. S53. During the pulse charging phase, based on the real-time battery charging status, temperature, and polarization voltage feedback, a pulse current with matching variable frequency, duty cycle, and amplitude is applied to eliminate the polarization effect, improve charging efficiency, and adjust the pulse parameters according to the real-time temperature.

9. The intelligent control method for fast charging of a lithium-ion battery according to claim 1, characterized in that: The safety protection and fault handling in S6 includes the following steps: S61. Set up a multi-level protection mechanism. When overvoltage, overcurrent or overheating is detected, immediately reduce the charging current or stop charging, and start the backup power supply to maintain system safety. S62. Use a deep learning-based fault diagnosis algorithm to identify battery anomalies, such as internal short circuits or accelerated aging, and analyze voltage fluctuation patterns through a convolutional neural network. S63. Record fault events and trigger alarms or maintenance notifications, while remotely monitoring and updating protection policies through a cloud platform.

10. An intelligent control system for fast charging of a lithium-ion battery, comprising the intelligent control method for fast charging of a lithium-ion battery according to any one of claims 1-9, characterized in that, The system includes: Status monitoring module: Real-time monitoring of battery voltage, current, temperature and internal resistance, and integrates high-precision sensors and data acquisition units; Evaluation module: Evaluates the battery's health and charging status, including the processor and memory storing algorithm code; Parameter adjustment module: dynamically adjusts charging current and charging voltage, including fuzzy logic controller and PID compensation unit; Prediction module: Utilizes machine learning algorithms to predict the optimal charging curve, including a neural network processor and a reinforcement learning engine; Charging control module: Implements multi-stage charging protocols, including constant current, constant voltage, and pulse charging controllers; Safety protection module: Provides real-time safety protection and fault handling, including multi-level protection circuits and fault diagnosis units; Central processing unit: coordinates the work of each module and connects to external devices through a communication interface.

Images