An automatic power-off control method for a wireless charging system

Abstract

The application relates to the technical field of wireless charging and discloses an automatic power-off control method for a wireless charging system, which comprises the following steps: acquiring historical charging sample data; modeling capacity attenuation data by using a Weibull distribution to construct a Weibull distribution model; modeling overvoltage fault discrete data by using a beta distribution to construct a beta distribution model; fusing the Weibull distribution model and the beta distribution model to construct a hybrid loss function and calculate expected loss values under different off-voltages; selecting an off-voltage that makes the expected loss value minimum as an optimal off-voltage output; judging whether a real-time collected port voltage is less than the optimal off-voltage, if yes, the system normally charges and continuously monitors, otherwise, the system stops charging and shuts down; after charging is completed, characteristic data of this time of charging are recorded and updated; and the application solves the technical problem that a battery is overcharged or undercharged due to the adoption of a fixed voltage threshold.

Description

Technical Field

[0001] This invention relates to the field of wireless charging technology, and in particular to an automatic power-off control method for a wireless charging system. Background Technology

[0002] With the increasing maturity of wireless power transfer technology, medium- and high-power wireless charging systems have been widely used in various fields such as power tools, smart home devices, service robots, and new energy vehicles. In these applications, the receiver, as the core component for energy reception and conversion, directly determines charging efficiency and battery safety through its control strategy. Especially at the end of the charging process, precisely controlling the shutdown of the receiver circuit to prevent overcharging or undercharging is a crucial aspect of wireless charging technology.

[0003] In existing technological systems, power-off control at the charging end primarily relies on the battery management system or the receiver's own voltage monitoring mechanism. A common approach is direct control from the receiver without an independent battery management system. In these systems, the receiver chip directly determines when to turn off the synchronous rectifier by sampling the port voltage or battery voltage. Currently, most of these solutions employ a fixed threshold comparison method. This static control strategy based on a fixed threshold has significant drawbacks in practical applications.

[0004] First, it ignores the dynamic changes in battery characteristics. The full-charge voltage of a battery is not an absolutely constant physical quantity; it drifts with variations in ambient temperature, battery aging, and charge / discharge rate. A fixed threshold cannot detect these changes, leading to a high risk of false full charge and premature shutdown at low temperatures or when the battery's internal resistance increases due to aging, resulting in wasted battery capacity; or failure to shut down in time when battery characteristics drift, leading to overvoltage risks.

[0005] Secondly, existing technologies lack the ability to utilize historical data. Current receiver controllers are typically short-sighted, making decisions solely based on the current instantaneous voltage while discarding crucial information such as the voltage change rate at the end of previous charging cycles and whether overcurrent protection was triggered. This means that regardless of how long the user uses the device, the charging strategy remains stuck in its initial factory state, unable to adaptively optimize based on actual usage habits and battery degradation.

[0006] Finally, in low-cost systems lacking a battery management system, the receiver often faces more stringent response speed requirements. If the fixed threshold is not set conservatively enough, the receiver must complete the shutdown process in an extremely short time, placing extremely high demands on the response speed of the hardware circuitry. To mitigate this risk, existing technologies often tend to set the threshold lower, directly sacrificing usable battery capacity and resulting in a degraded user experience. Therefore, how to establish an adaptive shutdown mechanism that dynamically balances charging capacity and battery safety using the receiver's own historical data, without the involvement of an expensive battery management system or when battery characteristics are uncertain, is a pressing technical problem to be solved in the field of medium-to-high power wireless charging. Summary of the Invention

[0007] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art. The present invention proposes an automatic power-off control method for wireless charging systems, which aims to solve the technical problem in the prior art that the use of a fixed voltage threshold cannot adapt to the aging and temperature change characteristics of the battery, resulting in "overcharging" or "undercharging".

[0008] The technical solution of the present invention is: an automatic power-off control method for a wireless charging system, the method comprising the following steps: S1. Obtain historical charging sample data, which includes at least capacity decay data and discrete overvoltage fault data. S2. The capacity decay data is modeled using a Weibull distribution to construct a Weibull distribution model for quantifying undercharging risk; S3. The beta distribution is used to model the discrete data of the overvoltage fault, and a beta distribution model for quantifying the overvoltage risk is constructed. S4. Combine the Weibull distribution model and the Beta distribution model to construct a hybrid loss function and calculate the expected loss value under different turn-off voltages; S5. Select the shutdown voltage that minimizes the desired loss value as the optimal shutdown voltage output for shutdown control during the charging process. S6. Real-time acquisition of port voltage during charging, and determination of whether the port voltage is less than the optimal shutdown voltage. If so, the system maintains normal charging mode and continues to monitor; otherwise, the system stops charging and enters the shutdown process. S7. After charging is completed, record and update the characteristic data at the end of this charging cycle, which will be used to correct and optimize the risk model and shutdown voltage calculation for the next cycle.

[0009] The historical charging sample data includes the shutdown voltage, actual capacity, ambient temperature, and fault information for the previous N cycles.

[0010] S2 includes the following steps: S2.1 Define the probability density function of the Weibull distribution: , Where t is the battery's survival time at the end of charging when it reaches a specific voltage threshold or capacity state; k is a shape parameter used to determine the evolution trend of undercharge risk. This is a scale parameter used to reflect the overall characteristic lifetime or characteristic capacity of the battery pack. The larger the value of k, the lower the risk of undercharging; when k < 1, it means the failure rate decreases over time; when k = 1, the Weibull distribution degenerates into an exponential distribution, indicating that the risk of undercharging is random and unrelated to charging time; when k > 1, it means the failure rate increases over time. S2.2 Based on the probability density function, the probability of battery failure before reaching the target state t is calculated using the cumulative distribution function: ; S2.3. Based on the capacity decay data, the shape parameter k and scale parameter of the Weibull distribution in this charging cycle are solved using the maximum likelihood estimation method. The implicit equation for the shape parameter k is as follows: , The implicit equation for the shape parameter k is expressed as a recursive function: , in: , Based on k for each period, the scale parameter The explicit equation is: , The calculation process is then performed in the logarithmic domain, and finally the exponent is taken back to restore the original value. The transformed calculation formula is as follows: ; Where N is the number of samples, N samples .

[0011] S3 includes the following steps: S3.1 Constructing the mathematical expression for the beta distribution probability density function: , Where t is a random variable representing the battery's lifespan at the end of charging when it reaches a specific voltage threshold or capacity state, and the beta distribution is defined on the interval (0,1). It is the beta function, a normalization constant that ensures the integral of the total power is 1. and The beta distribution has two shape parameters; S3.2, Solve for the shape parameters of the beta distribution in this charging cycle using the discrete data of the overvoltage fault. and : , , Theoretical Expectation for: , Theoretical variance is : , By using algebraic transformations to make the theoretical expectation equal to the sample expectation, a system of equations is established: , Introducing intermediate variables After dimensionality reduction, the solution is obtained as follows: , in, , Let be the sample estimates of the two shape parameters of the beta distribution. The sample mean. For sample variance, .

[0012] S4 includes the following steps: S4.1 Constructing the under-filled loss component based on the Weibull distribution model When the expected capacitance value corresponding to the set shutdown voltage is lower than the target full capacitance value C target This will result in an underfilling penalty, the expression of which is: , Overpressure loss components are constructed based on the beta distribution model. : , The hybrid loss function is formed by weighting and combining the under-filling loss component and the over-voltage loss component. Its expression is: ; in, Parameters characterizing the state of the battery; For candidate turn-off voltages, the decision space A corresponds to the set of "turn-off voltages" that the charger can select. Each possible voltage setpoint represents a specific decision. ; and These are weighted coefficients, representing the degree of preference for "expected capacity due to undercharging" and "fault penalty due to overcharging," respectively. S4.2. Based on historical charging sample data, perform Bayesian analysis on battery charging risk and construct the posterior expected loss function. For the posterior expected loss function The integral expectation over the parameter space Θ is expressed by the following formula: , Where X represents historical observation data, and E represents the conditional expectation based on the observation data X. This is a posterior probability distribution based on historical data; S4.3 During real-time charging, regarding the candidate turn-off voltage Based on the offline pre-computed beta distribution cumulative distribution function lookup table, the posterior expected loss function is calculated using linear interpolation. The integral result; S4.4, Traverse all candidate turn-off voltages Calculate the corresponding expected loss values ​​for each, and select the posterior expected loss function that satisfies the given condition. Candidate turn-off voltage that achieves minimum value This serves as the optimal turn-off voltage for this cycle.

[0013] If a serious hardware fault is detected during charging, the fault interruption mechanism will be triggered immediately to forcibly stop charging in order to protect the circuit. Attached Figure Description

[0014] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a schematic diagram of the software flow of the present invention; Figure 3 This is a schematic diagram of a wireless charging system applicable to this method. Detailed Implementation

[0015] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0016] like Figures 1 to 3 As shown, in one aspect, the present invention discloses an automatic power-off control method for a wireless charging system, the method comprising the following steps: S1. Obtain historical charging sample data, which includes at least capacity decay data and discrete overvoltage fault data. S2. The capacity decay data is modeled using a Weibull distribution to construct a Weibull distribution model for quantifying undercharging risk; S3. The beta distribution is used to model the discrete data of the overvoltage fault, and a beta distribution model for quantifying the overvoltage risk is constructed. S4. Combine the Weibull distribution model and the Beta distribution model to construct a hybrid loss function and calculate the expected loss value under different turn-off voltages; S5. Select the shutdown voltage that minimizes the desired loss value as the optimal shutdown voltage output for shutdown control during the charging process. S6. Real-time acquisition of port voltage during charging, and determination of whether the port voltage is less than the optimal shutdown voltage. If so, the system maintains normal charging mode and continues to monitor; otherwise, the system stops charging and enters the shutdown process. S7. After charging is completed, record and update the characteristic data at the end of this charging cycle, which will be used to correct and optimize the risk model and shutdown voltage calculation for the next cycle.

[0017] In this embodiment, the historical charging sample data is stored in a non-volatile memory. The historical charging sample data includes the shutdown voltage, actual capacity and ambient temperature in the previous N cycles, as well as fault information. The fault information is used to determine whether the cycle is an "output overvoltage" fault or another fault shutdown.

[0018] set up Let X be a sample vector extracted from the historical charging sample data. Sample X takes values ​​in the sample space χ. We assume that the samples follow a certain parametric statistical model, i.e., there exists a family of sample distributions. Its probability density function is .

[0019] Here, θ represents the implicit state parameter affecting the battery charging characteristics, and Θ represents the parameter space. The goal of statistical inference is to infer the unknown parameter θ or its function based on the observed sample x.

[0020] For a specific statistical problem, we need to take some action or make a decision. The set of all possible decisions is called the decision space, denoted by A. Since the battery power-off process applied in this invention can also be regarded as a classic binary hypothesis testing experiment, in this type of estimation problem, the decision is the estimate of the parameter θ, where A = Θ.

[0021] In interval estimation, a decision can be described as an interval, i.e.:

[0022] In hypothesis testing, decisions are typically binary, for example... , ,but .

[0023] In the application scenario of this invention, the decision space A corresponds to the set of selectable "shutdown voltages" for the charger. Each possible voltage setting represents a specific decision. .

[0024] The core idea of ​​statistical decision-making is that every decision produces corresponding consequences. To quantify these consequences, we introduce a loss function. The loss function is used to represent the decision to be made when the true state is θ. The resulting "losses" or "costs" are considered. If the decision is correct or close to reality, the losses are small; if the decision deviates significantly from reality, the losses are large.

[0025] Based on the above theory, this invention constructs a specific statistical decision-making model. "Undercharging risk" and "overvoltage risk" are respectively modeled using Weibull and Beta distributions, and a hybrid loss function is constructed. This function not only quantifies the potential harm of a single decision (setting a certain cut-off voltage) under a specific battery state (parameter θ), but also calculates the posterior expected loss (i.e., Bayesian risk) by combining prior information with a Bayesian method. This invention aims to find an optimal decision, i.e., an optimal cut-off voltage, that minimizes the posterior expected loss, thereby achieving a globally optimal balance between charging safety and efficiency.

[0026] S2 includes the following steps: S2.1 Define the probability density function of the Weibull distribution: , Where t is the battery's survival time at the end of charging when it reaches a specific voltage threshold or capacity state; k is a shape parameter used to determine the evolution trend of undercharge risk. This is a scale parameter used to reflect the overall characteristic lifetime or characteristic capacity of the battery pack. The larger the value of k, the longer the battery batch can be charged statistically or the higher its actual capacity, and the lower the risk of undercharging. When k < 1, the failure rate decreases over time, corresponding to the instability of the battery in the early stages of charging. When k = 1, the Weibull distribution degenerates into an exponential distribution, indicating that the risk of undercharging is random and unrelated to the charging time. When k > 1, the failure rate increases over time, meaning that as charging progresses, battery polarization increases, and the probability of reaching the overvoltage protection threshold and being forced to stop prematurely (undercharging) gradually increases. S2.2 Based on the probability density function, the probability of battery failure before reaching the target state t is calculated using the cumulative distribution function: ; S2.3. Based on the capacity decay data, the shape parameter k and scale parameter of the Weibull distribution in this charging cycle are solved using the maximum likelihood estimation method. The implicit equation for the shape parameter k is as follows: , Expressing the implicit equation for the shape parameter k as a recursive function, and using an iterative algorithm, moving the terms on the right-hand side of the equation to the left-hand side and setting them equal to 0, we obtain the following expression: , in: , Based on k for each period, the scale parameter The explicit equation is: , The calculation process is then performed in the logarithmic domain, and finally the exponent is taken back to restore the original value. The transformed calculation formula is as follows: ; Where N is the number of samples, N samples .

[0027] In this embodiment, the system uses historical data (valid samples filtered after fault information labeling) to fit the parameters of the current battery pack. The shape parameter k and scale parameter k of the Weibull distribution are calculated using the cumulative distribution function (CDF) to determine the under-sufficiency probability under a given decision. Maximum likelihood estimation (MLE) is then implemented on a computer or microcontroller (MCU) to solve for the shape parameter k and scale parameter k of the Weibull distribution. Essentially, this is a numerical optimization problem. Because the Weibull distribution's MLE equations have no analytical solution—meaning they cannot be calculated directly using formulas in one step—it is necessary to rely on iterative algorithms to approximate the optimal solution.

[0028] S3 includes the following steps: S3.1 Constructing the mathematical expression for the beta distribution probability density function: , Where t is a random variable representing the battery's lifespan at the end of charging when it reaches a specific voltage threshold or capacity state, and the beta distribution is defined on the interval (0,1). It is the beta function, a normalization constant that ensures the integral of the total power is 1; and The beta distribution has two shape parameters; S3.2, Solve for the shape parameters of the beta distribution in this charging cycle using the discrete data of the overvoltage fault. and : , , Theoretical Expectation for: , Theoretical variance is : , By using algebraic transformations to make the theoretical expectation equal to the sample expectation, a system of equations is established: , Introducing intermediate variables To perform dimensionality reduction, algebraic calculations are first performed to compute intermediate variables. : , Solving for the shape parameters yields: , in, , Let be the sample estimates of the two shape parameters of the beta distribution. The sample mean. This represents the sample variance.

[0029] In this embodiment, It is the beta function, a normalization constant that ensures the integral of the total power is 1. Its relationship is: , It is a gamma function, similar to a factorial. A higher-order generalization of . Extended to the non-integer field by the Euler integral form, its expression is: .parameter and The beta distribution consists of two shape parameters. In this invention, these two parameters correspond to the weighted counts of "healthy survival" and "failure" in the historical operating data of the battery pack, respectively. After reading the historical charging sample data, the two shape parameters of the beta distribution in the current charging cycle are calculated using these discrete data to determine the overcharge risk of the current charging cycle.

[0030] S4 includes the following steps: S4.1 Constructing the under-filled loss component based on the Weibull distribution model When the expected capacitance value corresponding to the set shutdown voltage is lower than the target full capacitance value C target This will result in an underfilling penalty, the expression of which is: , Overpressure loss components are constructed based on the beta distribution model. : , The hybrid loss function is formed by weighting and combining the under-filling loss component and the over-voltage loss component. Its expression is: ; in, Parameters characterizing the state of the battery; For candidate turn-off voltages, the decision space A corresponds to the set of "turn-off voltages" that the charger can select. Each possible voltage setpoint represents a specific decision. ; and These are weighting coefficients, representing the degree of preference for "expected capacity due to undercharging" and "fault penalty due to overcharging," respectively. These coefficients can be adjusted during actual commissioning. S4.2. Based on historical charging sample data, perform Bayesian analysis on battery charging risk and construct the posterior expected loss function. For the posterior expected loss function The integral expectation over the parameter space Θ is expressed by the following formula: , Where X represents historical observation data, and E represents the conditional expectation based on the observation data X. This is a posterior probability distribution based on historical data; S4.3 During real-time charging, regarding the candidate turn-off voltage Based on the offline pre-computed beta distribution cumulative distribution function lookup table, the posterior expected loss function is calculated using linear interpolation. The integral result; S4.4, Traverse all candidate turn-off voltages Calculate the corresponding expected loss values ​​for each, and select the posterior expected loss function that satisfies the given condition. Candidate turn-off voltage that achieves minimum value This serves as the optimal turn-off voltage for this cycle.

[0031] Due to the actual state of the battery during this cycle Since the loss is unknown, this invention cannot rely on prior calculations to determine the loss in the current period; instead, it requires calculating the posterior expected loss function. After determining the posterior expected loss function Then, the solution for the optimal turn-off voltage can be simplified into a minimum optimization problem of a single-variable function. In this embodiment, we choose to use an offline pre-computed beta cumulative distribution function (CDF) lookup table combined with linear interpolation to quickly approximate the integral result, thereby transforming the complex integral operation into a simple table lookup and arithmetic operation. Based on the historical data of the previous N cycles, we perform posterior Bayesian analysis to obtain the optimal turn-off voltage for the current cycle.

[0032] If a serious hardware fault is detected during charging, the fault interruption mechanism will be triggered immediately to forcibly stop charging in order to protect the circuit.

[0033] This embodiment provides a Bayesian decision-based adaptive shutdown control method for wireless charging, applied to the control chip of a wireless charging receiver. This invention determines the shutdown timing during system operation based on port voltage and historical operating data.

[0034] This invention also discloses a wireless charging system employing an automatic power-off control method for wireless charging systems. This invention is applied to the receiver end of a wireless charging system and has no special requirements regarding the resonant cavity topology of the transmitter converter. The structure of the wireless charging system to which this invention is applicable is as follows: Figure 3 As shown. A wireless charging system consists of the following components: Transmitter-side converter assembly: The front end typically includes a PFC (Power Factor Control), and depending on the specific needs, a three-phase Vienna or totem-pole topology may be selected. Common transmitter-side converters are phase-shifted full-bridge inverters, and some systems use a buck converter plus an inverter. Since these structures themselves have no impact on the shutdown timing, only a simplified illustration is provided here. This invention does not have special requirements regarding the topology type of the transmitter-side converter.

[0035] Transmitter-side resonant cavity assembly: Common transmitter-side resonant cavity networks include LCC, LCL, and SS, etc. The choice of these cavity networks will affect performance characteristics such as transmission characteristics. However, they have no impact on the power-off timing. Therefore, only a simple illustration is provided here. This invention does not have special requirements for the resonant cavity topology type of the transmitter converter.

[0036] Transmitting coil and receiving coil: important components of a wireless charging system, serving as the core physical medium for wireless power transmission between the transmitting and receiving coils.

[0037] Receiver-side resonant cavity assembly: Common receiver-side resonant cavities can be selected based on the type of resonant cavity on the transmitter side, such as LCC and SS. The choice of these cavity networks will affect performance characteristics such as transmission features. However, it has no impact on the power-off timing.

[0038] The receiver-side rectifier bridge assembly in this invention uses a rectifier bridge type, which can be either synchronous or passive rectification. This is because the power-off process in this invention is achieved by short-circuiting the lower diode and then issuing a power-off command. Some wireless charging systems add a buck converter to the receiver-side rectifier bridge for impedance matching. This invention does not use this approach; instead, it directly connects the rectifier bridge to the load.

[0039] Battery: As the final load of the wireless charging system, the characteristics of the battery directly determine the setting of the charging cutoff voltage and current.

[0040] The wireless charging system is equipped with a controller, which executes the following control logic: it collects battery charging end condition data, fits Weibull distribution model and Beta distribution model respectively, constructs a hybrid loss function, traverses a preset voltage lookup table to calculate the expected loss corresponding to each candidate turn-off voltage, and selects the voltage corresponding to the minimum expected loss as the optimal turn-off voltage; when the battery charging voltage reaches the optimal turn-off voltage, it controls the lower tube of the rectifier bridge on the receiving side to short-circuit and depressurize, and issues a shutdown command to cut off the power output on the transmitting side.

[0041] An automatic power-off control method for a wireless charging system, based on Bayesian decision-making, is applied to the control chip of the wireless charging receiver, such as... Figure 2 As shown, the present invention includes the following steps: System Initialization and Data Loading: After the wireless charging receiver is powered on, the software initialization process is first executed to configure the ADC sampling interface, communication interface, and storage unit. Subsequently, the system reads historical charging sample data from non-volatile memory (such as Flash or EEPROM). This historical data includes records of the shutdown voltage of previous charging cycles, actual charging capacity, and whether overvoltage protection was triggered, which are used for subsequent risk model construction. Constructing a Weibull distribution model for undercharging risk: The system utilizes historical charging sample data, particularly data on battery capacity decay and end-charging polarization characteristics, to calculate and establish a Weibull distribution model for undercharging risk. This model describes the probability distribution of the battery not being fully charged if charging stops prematurely in the current battery state, thus quantifying the risk of "not being fully charged." Constructing an overvoltage risk beta distribution model: The system utilizes discrete statistical data on overvoltage fault (OVP) occurrences from historical samples to calculate and establish an overvoltage risk beta distribution model. This model describes the probability distribution of continued charging under current charging conditions, leading to voltage exceeding the safety threshold and thus damaging the device, thereby quantifying the risk of "overcharging." Calculate the hybrid loss function and solve for the optimal value: The system combines the two risk models mentioned above to construct a hybrid loss function that includes under-filling loss weights and over-voltage penalty weights. Using a numerical search algorithm (e.g., traversing a pre-defined voltage lookup table), the expected loss value under different candidate turn-off voltages is calculated, and the point that minimizes the expected loss is found. Determine the optimal shutdown voltage for the current cycle: The system determines the voltage value corresponding to the minimum expected loss as the optimal shutdown voltage for the current charging cycle. This optimal shutdown voltage is used as a threshold, which is dynamically generated to adapt to the current aging level of the battery and environmental conditions. Real-time monitoring and cyclic judgment: After determining the threshold, the system enters the normal charging mode. In this mode, the control chip collects the battery's port voltage and charging current in real time. The system compares the real-time collected port voltage with the optimal shutdown voltage: If the real-time port voltage is less than the optimal shutdown voltage, it means the battery has not yet reached the optimal stop point, and the system maintains the normal charging mode and continues to collect data in real time; if the real-time port voltage is greater than or equal to the optimal shutdown voltage, it means the preset safe and efficient stop point has been reached, and the system executes the stop charging operation and enters the shutdown process. Anomaly Handling and Data Update: During the charging process, if the system detects a serious hardware-level fault, such as a sudden output overvoltage interruption, it will immediately trigger the fault interruption mechanism to forcibly stop charging to protect the circuit.

[0042] Whether the charging stops normally due to reaching the optimal shutdown voltage or ends for other reasons, the system will record the characteristic data at the end of this charging process, such as the final shutdown voltage and the amount of electricity charged, and update this new data in the memory as the basis for optimizing the model during the next charging, thereby realizing the system's self-learning and iteration.

[0043] This invention addresses the challenges of small space and poor heat dissipation in wireless charging receivers. It utilizes Weibull and Beta distributions to quantify the risks of "undercharging" and "overvoltage," respectively. By calculating the minimum expected loss, it dynamically determines the optimal shutdown voltage, replacing the traditional fixed threshold and improving charging efficiency and safety. Employing an engineering architecture of "offline modeling + online table lookup," it transforms complex probabilistic statistical calculations into simple table lookup operations, enabling the control algorithm to run smoothly on low-cost, low-computing-power receiver MCUs, significantly reducing hardware BOM costs. Furthermore, this invention establishes a data closed loop within the receiver, automatically updating model parameters using historical charging samples. This allows the control strategy to automatically adjust as the battery ages, solving the consistency problem caused by individual differences and aging in miniaturized devices, eliminating the need for manual calibration.

[0044] The present invention has the following beneficial effects: 1. This invention addresses the problem of balancing charging capacity and safety caused by using a fixed voltage threshold for charging shutdown control in existing technologies by constructing a decision model based on posterior expected loss. Existing technologies typically set a uniform shutdown voltage based on experience, which cannot adapt to individual battery differences and characteristic drift caused by aging. This invention quantifies undercharge loss and overvoltage penalty loss, and uses a lookup table method combined with a numerical optimization algorithm to dynamically solve for the optimal shutdown voltage. This maximizes the actual charging capacity of the battery while ensuring the safety of the receiving circuit, achieving an adaptive balance between charging efficiency and safety.

[0045] 2. This invention employs a dual-distribution model to perform Bayesian analysis on the risks at the charging end, improving the accuracy of state assessment. Specifically, it uses a Weibull distribution to fit the battery's capacity decay characteristics at the charging end, and a Beta distribution to probabilistically model historical discrete data of overvoltage faults. This statistical modeling approach can more accurately reflect the battery's true physical characteristics under its current state of health (SOH), significantly reducing the misjudgment rate caused by battery polarization or internal resistance changes compared to traditional threshold judgment methods, thus improving the system's robustness.

[0046] 3. This invention possesses strong engineering practicality and is suitable for resource-constrained wireless charging receiver microcontrollers (MCUs). Although this invention is based on Bayesian statistical decision theory, at the computational level, it avoids floating-point underflow by introducing logarithmic domain transformation and employs an offline pre-computed cumulative distribution function (CDF) lookup table combined with linear interpolation to transform complex integral operations into simple table lookups and arithmetic operations. Combined with a low-complexity single-variable optimization strategy, this invention can run in real-time on embedded chips with limited computing power, without relying on cloud computing, thus reducing system latency and communication costs.

[0047] 4. This invention possesses adaptive update capabilities throughout its entire lifecycle. The system can utilize end-of-charge data (such as shutdown voltage, actual charging capacity, and whether overvoltage protection is triggered) to update the characteristic parameters of the Weibull and Beta distributions in real time through maximum likelihood estimation or moment estimation. This means that as the battery ages, the control strategy will automatically adjust the decision boundary of the optimal shutdown voltage without requiring manual parameter recalibration, thus extending the effective lifespan of the battery module.

[0048] 5. This invention introduces a hardware safety boundary clamping mechanism at the algorithm output. After calculating the theoretically optimal turn-off voltage, it is forcibly limited to a physically safe voltage range allowed by the receiving circuit. This "algorithm optimization + physical fallback" architecture effectively prevents the risk of voltage exceeding limits due to sensor noise or algorithm anomalies, further improving the safety of the wireless charging system.

[0049] Finally, it should be emphasized that the above description is not intended to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An automatic power-off control method for a wireless charging system, the method comprising: The method includes the following steps: S1. Obtain historical charging sample data, which includes at least capacity decay data and discrete overvoltage fault data. S2. The capacity decay data is modeled using a Weibull distribution to construct a Weibull distribution model for quantifying undercharging risk; S3. The beta distribution is used to model the discrete data of the overvoltage fault, and a beta distribution model for quantifying the overvoltage risk is constructed. S4. Combine the Weibull distribution model and the Beta distribution model to construct a hybrid loss function and calculate the expected loss value under different turn-off voltages; S5. Select the shutdown voltage that minimizes the desired loss value as the optimal shutdown voltage output for shutdown control during the charging process. S6. Real-time acquisition of port voltage during charging, and determination of whether the port voltage is less than the optimal shutdown voltage. If so, the system maintains normal charging mode and continues to monitor; otherwise, the system stops charging and enters the shutdown process. S7. After charging is completed, record and update the characteristic data at the end of this charging cycle, which will be used to correct and optimize the risk model and shutdown voltage calculation for the next cycle.

2. The automatic power-off control method for a wireless charging system according to claim 1, characterized in that, The historical charging sample data includes the shutdown voltage, actual capacity, ambient temperature, and fault information for the previous N cycles.

3. The automatic power-off control method for a wireless charging system according to claim 1, characterized in that, S2 includes the following steps: S2.1 Define the probability density function of the Weibull distribution: ， Where t is the battery's survival time at the end of charging when it reaches a specific voltage threshold or capacity state, and k is a shape parameter used to determine the evolution trend of undercharging risk. This is a scale parameter used to reflect the overall characteristic lifetime or characteristic capacity of the battery pack. The larger the value of k, the lower the risk of undercharging; when k < 1, it means the failure rate decreases over time; when k = 1, the Weibull distribution degenerates into an exponential distribution, indicating that the risk of undercharging is random and unrelated to charging time; when k > 1, it means the failure rate increases over time. S2.2 Based on the probability density function, the probability of battery failure before reaching the target state t is calculated using the cumulative distribution function: ； S2.

3. Based on the capacity decay data, the shape parameter k and scale parameter of the Weibull distribution in this charging cycle are solved using the maximum likelihood estimation method. The implicit equation for the shape parameter k is as follows: ， The implicit equation for the shape parameter k is expressed as a recursive function: ， in: ， Based on k for each period, the scale parameter The explicit equation is: ， The calculation process is then performed in the logarithmic domain, and finally the exponent is taken back to restore the original value. The transformed calculation formula is as follows: ； Where N is the number of samples, N samples .

4. The automatic power-off control method for a wireless charging system according to claim 1, characterized in that, S3 includes the following steps: S3.1 Constructing the mathematical expression for the beta distribution probability density function: ， Where t is a random variable representing the battery's lifespan at the end of charging when it reaches a specific voltage threshold or capacity state, and the beta distribution is defined on the interval (0,1). It is the beta function, a normalization constant that ensures the integral of the total power is 1. and The beta distribution has two shape parameters; S3.2, Solve for the shape parameters of the beta distribution in this charging cycle using the discrete data of the overvoltage fault. and : ， ， Theoretical Expectation for: ， Theoretical variance is : ， By using algebraic transformations to make the theoretical expectation equal to the sample expectation, a system of equations is established: ， Introducing intermediate variables After dimensionality reduction, the solution is obtained as follows: ， in, , Let be the sample estimates of the two shape parameters of the beta distribution. The sample mean. For sample variance, .

5. The automatic power-off control method for a wireless charging system according to claim 1, characterized in that, S4 includes the following steps: S4.1 Constructing the under-filled loss component based on the Weibull distribution model When the expected capacitance value corresponding to the set shutdown voltage is lower than the target full capacitance value C target This will result in an underfilling penalty, the expression of which is: ， Overpressure loss components are constructed based on the beta distribution model. : ， The hybrid loss function is formed by weighting and combining the under-filling loss component and the over-voltage loss component. Its expression is: ； in, Parameters characterizing the state of the battery; For candidate turn-off voltages, the decision space A corresponds to the set of "turn-off voltages" that the charger can select. Each possible voltage setpoint represents a specific decision. ; and These are weighted coefficients, representing the degree of preference for "expected capacity due to undercharging" and "fault penalty due to overcharging," respectively. S4.

2. Based on historical charging sample data, perform Bayesian analysis on battery charging risk and construct the posterior expected loss function. For the posterior expected loss function The integral expectation over the parameter space Θ is expressed by the following formula: ， Where X represents historical observation data, and E represents the conditional expectation based on the observation data X. This is a posterior probability distribution based on historical data; S4.3 During real-time charging, regarding the candidate turn-off voltage Based on the offline pre-computed beta distribution cumulative distribution function lookup table, the posterior expected loss function is calculated using linear interpolation. The integral result; S4.4, Traverse all candidate turn-off voltages Calculate the corresponding expected loss values ​​for each, and select the posterior expected loss function that satisfies the given condition. Candidate turn-off voltage that achieves minimum value This serves as the optimal turn-off voltage for this cycle.

6. The automatic power-off control method for a wireless charging system according to claim 1, characterized in that, If a serious hardware fault is detected during charging, the fault interruption mechanism will be triggered immediately to forcibly stop charging in order to protect the circuit.

Images