A PC battery endurance intelligent management method and system

Abstract

The application relates to the technical field of battery endurance management, in particular to a PC battery endurance intelligent management method and system, which comprises the following steps: collecting the terminal voltage and the discharge current of a battery according to a preset period, and obtaining the dynamic internal resistance of the battery; monitoring a memory scheduling stack; when loading signaling of a preset model is captured, extracting the peak current corresponding to the preset model, and calculating the user weight based on the input interruption frequency; calculating the predicted drop voltage; comparing the predicted drop voltage with a preset protection voltage; if the predicted drop voltage is greater than the preset protection voltage, obtaining the residual capacity of the battery, and issuing a joint scheduling strategy containing the upper limit of power consumption based on the residual capacity and the user weight; if the predicted drop voltage is not greater than the preset protection voltage, performing staggered peak control of reducing the display backlight and switching the radio frequency communication to the sleep state. The application can solve the problem that the dynamic internal resistance calculation lags behind the current jump, thereby causing the device to be powered off due to under-voltage.

Description

Technical Field

[0001] This application relates to the technical field of battery life management, and in particular to a PC battery life intelligent management method and system. Background Technology

[0002] When an AI-powered personal computer operates without an external power source and its battery ages, the battery cell exhibits abrupt changes in internal resistance at the physical level. Loading a local large language model triggers a transient step discharge current due to computational power. This high transient load current, passing through the high internal resistance of the aging battery cell, generates a large ohmic voltage divider, causing a sudden voltage drop on the device's power supply bus. When the bus transient voltage drops below the underlying hardware protection threshold, the device experiences a forced undervoltage shutdown, halting system operation.

[0003] Chinese invention patent application CN108918971A discloses a method and apparatus for calculating dynamic equivalent internal resistance. This method simultaneously samples the voltage and current of the object under test to obtain sampling data. The instantaneous equivalent internal resistance of the object is obtained by dividing the voltage sampling data by the current sampling data at the same time. The calculation module subtracts a preset constant internal resistance from the instantaneous equivalent internal resistance to calculate the dynamic equivalent internal resistance of the object at the corresponding time. The system constructs a data sequence from the dynamic equivalent internal resistances at multiple times and performs a filtering operation. Based on the output filtered sequence, it detects the details of the object's actions.

[0004] The above scheme relies on physical sampling data collected at the same time to perform calculations, and its control is based on post-event passive parameter measurement logic. When deployed in an AI personal computer environment, the control system can only obtain the basic data for calculating internal resistance after the power supply hardware has withstood a step current surge and generated a real voltage drop. Data acquisition and internal resistance calculation lag behind the physical discharge process of the current surge. When the upper layer of the system induces extreme computing power demands, the bus voltage has already reached the hardware protection threshold before the controller completes its calculation cycle. The lower-level power distribution unit cannot intervene to perform drop-down current limiting before the voltage penetrates the protection threshold. Summary of the Invention

[0005] To address the problem of device undervoltage shutdown caused by the lag of dynamic internal resistance calculation behind current jumps, this application provides a PC battery life intelligent management method and system.

[0006] Firstly, this application provides a PC battery life intelligent management method, which adopts the following technical solution: A PC battery life intelligent management method, comprising: The battery's terminal voltage and discharge current are collected at a preset period to obtain the battery's dynamic internal resistance; wherein, when the difference in discharge current between adjacent periods is greater than a preset step threshold, the dynamic internal resistance is calculated based on the difference and the drop difference corresponding to the terminal voltage. Monitor the memory scheduling stack, and when the loading signal of the preset model is captured, extract the peak current corresponding to the preset model and calculate the user weight based on the input interrupt frequency; Based on the terminal voltage, the discharge current, the dynamic internal resistance, and the peak current, a drop-proof calculation is performed to obtain the predicted drop voltage. The predicted voltage drop is compared with the preset protection voltage; If the predicted drop voltage is greater than the preset protection voltage, the remaining battery power is obtained, and a joint scheduling strategy including a power consumption limit is issued based on the remaining battery power and the user weight. If the predicted drop voltage is not greater than the preset protection voltage, peak-shaving control is performed to reduce the display backlight and switch the radio frequency communication to sleep mode.

[0007] Optionally, obtaining the dynamic internal resistance of the battery includes: When the difference between the discharge currents in adjacent cycles is not greater than the preset step threshold, the difference is determined to be thermal noise, and the division operation for calculating the dynamic internal resistance is blocked. Extract the preset historical internal resistance and use the historical internal resistance as the dynamic internal resistance to perform the drop protection calculation.

[0008] Optionally, when the monitored memory scheduling stack captures a loading signaling for a preset model, it extracts the peak current corresponding to the preset model, including: Obtain the bus throughput of the memory scheduling stack mapping; If the bus throughput is greater than a preset mutation value and the loading signaling is captured synchronously, the calibration current matching the preset model is extracted, and the calibration current is used as the peak current.

[0009] Optionally, the calculation of user weights based on input interruption frequency includes: The frequency of input interruptions within a preset time window is obtained, and the gaze duration is extracted simultaneously. If the frequency of input interruption is greater than a preset frequency threshold and the duration of gaze is greater than a preset focus threshold, the user weight is calculated by weighting the frequency of input interruption and the duration of gaze by a preset coefficient.

[0010] Optionally, the step of performing drop-proof calculations based on the terminal voltage, the discharge current, the dynamic internal resistance, and the peak current to calculate the predicted drop-proof voltage includes: Subtracting the discharge current from the peak current yields the transient differential current, and multiplying the transient differential current by the dynamic internal resistance yields the ohmic voltage divider. The predicted voltage drop is calculated by subtracting the ohmic voltage divider from the terminal voltage.

[0011] Optionally, the step of subtracting the ohmic voltage divider from the terminal voltage to calculate the predicted dropout voltage includes: The surface temperature of the battery is obtained, and the polarization threshold and temperature compensation value are extracted. If the surface temperature is less than the polarization threshold, the ohmic voltage divider is multiplied by the temperature compensation value to obtain the corrected voltage divider, and the terminal voltage is subtracted from the corrected voltage divider to calculate the predicted drop voltage.

[0012] Optionally, comparing the predicted drop voltage with the preset protection voltage includes: Obtain the baseline voltage and the margin voltage, and add the baseline voltage and the margin voltage to obtain the preset protection voltage; If the predicted drop voltage is greater than the preset protection voltage, a permission signal is output to trigger the joint scheduling strategy. If the predicted voltage drop is not greater than the preset protection voltage, an interception signal is output to trigger the peak shaving control.

[0013] Optionally, in response to the permission signaling, the issuance of a joint scheduling policy including a power consumption cap based on the remaining power and the user weight includes: Extract the preset power distribution network and encapsulate the remaining power and the user weight into a state sequence; The state sequence is input into the power distribution network to deduce the joint scheduling strategy, which includes the power consumption limit, the display duty cycle, and the radio frequency status. The joint scheduling strategy is issued, feedback status is obtained synchronously, and the feedback status is input into the power distribution network for iteration.

[0014] Optionally, in response to the interception signaling, triggering the peak-shaving control includes: If the user weight is greater than the preset immersion value, the display backlight is reduced to a minimum value and the radio frequency communication is switched to the sleep state; Extract the compensation current released during the peak shift control, and subtract the compensation current from the peak current to obtain the updated current; The peak current is replaced by the updated current to perform a single drop test, resulting in the recalculated predicted drop voltage; If the recalculated predicted drop voltage is not greater than the preset protection voltage, the loading signaling is intercepted.

[0015] Secondly, the PC battery life intelligent management system provided in this application adopts the following technical solution: A PC battery life intelligent management system, comprising: An internal resistance calculation module is used to collect the battery's terminal voltage and discharge current at preset cycles to obtain the battery's dynamic internal resistance; wherein, when the difference in discharge current between adjacent cycles is greater than a preset step threshold, the dynamic internal resistance is calculated based on the difference and the drop difference corresponding to the terminal voltage. The load awareness module is used to monitor the memory scheduling stack. When it captures the loading signal of the preset model, it extracts the peak current corresponding to the preset model and calculates the user weight based on the input interrupt frequency. The safety calculation module is used to perform anti-drop calculations based on the terminal voltage, the discharge current, the dynamic internal resistance, and the peak current, and calculate the predicted drop voltage. The power distribution scheduling module is used to compare the predicted voltage drop with the preset protection voltage; if the predicted voltage drop is greater than the preset protection voltage, the remaining power of the battery is obtained, and a joint scheduling strategy including a power consumption limit is issued based on the remaining power and the user weight; if the predicted voltage drop is not greater than the preset protection voltage, peak-shaving control is performed to reduce the display backlight and switch the radio frequency communication to a sleep state.

[0016] In summary, this application includes the following beneficial technical effects: 1. By collecting terminal voltage and discharge current at a preset cycle and calculating dynamic internal resistance in real time when the current difference is greater than the step threshold, and monitoring the memory scheduling stack to pre-fetch peak current when capturing model loading signaling, the predicted drop voltage is obtained by performing anti-drop trial calculation based on the current electrical parameters and peak current. This effectively avoids the risk of undervoltage power failure caused by the internal resistance calculation lagging behind the current jump, and realizes the prediction of power supply safety before the actual voltage drop.

[0017] 2. By acquiring the battery surface temperature and introducing a temperature compensation value when it is below the polarization threshold to correct the ohmic voltage divider, the drop voltage is calculated and predicted using the corrected voltage divider. This effectively reduces the impact of increased polarization resistance on voltage prediction accuracy in low-temperature environments and improves the adaptability of intelligent battery range management under different temperature conditions.

[0018] 3. In response to the interception signal and when the user weight is greater than the preset immersion value, peak control is performed. The display backlight is reduced to the minimum value and the radio frequency communication is switched to sleep mode. The released compensation current is extracted to replace the peak current and the anti-drop test is re-executed. When the recalculated voltage is still not greater than the protection voltage, the model loading signal is intercepted, thus realizing fine adjustment of transient power consumption and on-demand blocking of loading operation. Attached Figure Description

[0019] Figure 1 A flowchart of a PC battery life intelligent management method provided in an embodiment of the present invention; Figure 2This is a closed-loop hardware architecture diagram provided in an embodiment of the present invention; Figure 3 The anti-fall calculation and interception control curve diagram provided in the embodiments of the present invention; Figure 4 This is a block diagram of a PC battery life intelligent management system provided in an embodiment of the present invention. Detailed Implementation

[0020] The following combination Figures 1-4 This application will be described in further detail.

[0021] This application discloses a PC battery life intelligent management method, combined with... Figure 1 As shown in the flowchart, the method proceeds step by step downwards in the following order: collecting the voltage and discharge current at the acquisition end, capturing the loading signal to extract the peak current, calculating the dynamic internal resistance when the current difference is greater than the step threshold, and entering the anti-drop test to predict the drop voltage. There is a loop path that backtracks upwards to the anti-drop test at the judgment branch where the recalculated voltage is greater than the protection voltage. The method is executed by a closed-loop hardware loop consisting of a battery management microcontroller unit, an embedded power management controller, and an edge intelligent strategy engine.

[0022] The battery management microcontroller has a built-in analog-to-digital conversion circuit. The sampling input is coupled to the positive and negative terminals of the battery. It collects the cell terminal voltage, transient discharge current and surface temperature according to the polling cycle determined by the physical bus bandwidth. The sampling frames are reported to the edge intelligent strategy engine via the system management bus SMBus.

[0023] The embedded power management controller has write access to the system hardware registers and acts as the underlying power distribution actuator to drive the power wall register, the PWM duty cycle of the screen backlight, and the power activity level of the wireless RF front end.

[0024] The edge intelligence strategy engine internally cascades three logical sub-modules: a load sensing unit, a power supply security calculation unit, and a power distribution network. For example... Figure 2 As shown in the hardware interaction architecture, in addition to receiving data from the microcontroller unit and outputting joint scheduling strategies, the edge intelligent strategy engine directly aggregates downlink signals from the memory scheduling stack or PCIe bus at its top, and synchronously connects the memory scheduling stack signals and the feedback status reported by the embedded power management controller at its bottom, thus entering the actual physical loop through this multi-source topology; this hardware loop enters a continuous closed-loop following when the PC is disconnected from the external power supply and the battery cell is chemically aged, without setting a final judgment to terminate power distribution.

[0025] The polling period is constrained by the bus transmit and receive time constant and ranges from 10ms to 100ms. A period of less than 10ms causes the sampling frame density to exceed the SMBus arbitration bandwidth and crowd out the main control computing power. A period of more than 100ms causes the sampling to lag behind the millisecond-level computing power step and miss the interception window. A period of 50ms is preferred to balance bus occupancy and real-time tracking.

[0026] S1 completes the extraction of microscopic electrical state and physical calibration of instantaneous dynamic internal resistance to obtain the dynamic internal resistance of the battery.

[0027] The battery management microcontroller collects transient discharge current during the current polling cycle. With terminal voltage The data is then reported to the edge intelligent policy engine, which caches the sampled values ​​of two adjacent cycles and calculates the difference between the discharge currents of the two adjacent cycles (i.e., the current jump). The voltage drop difference between the terminal voltage and the voltage sag (i.e., the voltage sag amount) Specifically, the difference in discharge current between adjacent cycles must be greater than a preset step threshold (i.e., a preset current step threshold). When the difference is calculated, the dynamic internal resistance is obtained based on the difference between the voltage difference and the voltage drop corresponding to the terminal voltage. That is, the engine performs a division logic to calculate the instantaneous dynamic internal resistance. And write it to the status register to overwrite the historical value. The range is 2A to 3A. Below 2A, the cell self-discharge and quantization noise will be misjudged as a step in computing power, which will frequently trigger invalid solutions. Above 3A, the moderate-intensity inference load will be missed. 2.5A is preferred to filter out the background thermal noise and retain the sensitivity to the loading current of large language models.

[0028] The status register uses double buffering partitioning, and new data is written to it. With reading old To avoid cross-cycle read / write contention, the engine assigns the current to different partitions. When the difference in discharge current between adjacent cycles is not greater than a preset step threshold, the engine classifies the difference (i.e., the jump variable) as thermal noise (i.e., classifies it as background thermal noise) and blocks the division operation for calculating dynamic internal resistance to avoid division-to-zero overflow. Instead, it extracts a preset historical internal resistance and uses the historical internal resistance as the dynamic internal resistance for drop test calculation, i.e., reads the most recent valid history from the status register. Proceed to subsequent trial calculations, historical Set a validity window; if the window is not refreshed, set the sampling failure flag and maintain the previous joint scheduling strategy.

[0029] S2 completes cross-layer feature pre-fetching and user habit quantification.

[0030] The application load-aware unit monitors the bus throughput mapped by the operating system's memory scheduling stack. If the throughput exceeds a preset mutation value and the local large language model loading signaling is captured synchronously, the calibration current matching the model is extracted from the calibration table and used as the peak current. This ensures that the peak current enters the prediction channel before the actual physical discharge occurs. The calibration table is established by calibrating the full-load current of each model individually using discrete power consumption data acquisition. In addition to memory scheduling stack monitoring, the application load sensing unit can be replaced with a prefetch that monitors the abrupt edges of PCIe bus data throughput. Both constitute an equivalent feedforward triggering path.

[0031] The application load sensing unit synchronously counts the frequency of external input interruptions within a preset time window and extracts the gaze duration captured by the camera. If the input interruption frequency is greater than a preset frequency threshold and the gaze duration is greater than a preset focus threshold, the two are weighted and summed according to a preset coefficient to calculate the user immersion weight coefficient. fixation duration can be used independently as a metric. The deformation calculation is based on adapting to the terminal form without an external keyboard.

[0032] S3 performs a preliminary anti-fall calculation based on Ohm's law of physics, forming the core interception point.

[0033] Power supply safety calculation unit reads terminal voltage Transient discharge current Dynamic internal resistance With peak current Subtracting the transient discharge current from the peak current yields the transient differential current. Multiplying the transient differential current by the dynamic internal resistance gives the ohmic voltage divider. Subtracting the ohmic voltage divider from the terminal voltage calculates the transient sag prediction voltage. ,Right now The power supply safety calculation unit will use the baseline voltage. With safety margin voltage The sum is used to obtain the preset protection voltage. Compare the absolute value with the preset protection voltage. If the voltage exceeds the preset protection voltage, a safety clearance signal will be output, such as... Figure 3 China and Israel For the vertical axis, The physical timing display is on the horizontal axis. During the stage where the terminal voltage marked by the solid line drops steadily and does not touch the horizontal dotted line, the predicted drop voltage marked by the dashed line experiences a transient dip and breaks through the preset protection voltage mark. Based on this penetration intersection, the hardware determines that it is not greater than the preset protection voltage and outputs a red line interception signal. Based on the cell series connection system, a voltage of 6.0V to 6.2V is chosen for a 2-cell lithium battery system. This value is higher than the sum of the undervoltage lockout points of the two single cells, thus allowing for a bus ripple margin. Substituting... The transient differential current is 9A, and the ohmic voltage divider is 1.35V. The calculated voltage is 5.85V, which is below the 6.0V protection threshold, triggering a red-line interception signal. This confirms that aging battery cells do indeed have a risk of voltage drop during the moment of loading a large language model. Low temperatures increase the cell's polarization resistance, thus affecting the calibrated voltage. Insufficient to cover low-temperature transient impedance, the battery management microcontroller synchronously reports the surface temperature. Once the temperature drops below the preset low-temperature polarization threshold, the power supply safety calculation unit extracts the system's preset nonlinear temperature compensation coefficient. , If the value is greater than 1, multiply the ohmic voltage divider by 1. The corrected voltage divider is obtained, and the predicted voltage drop is calculated by subtracting the corrected voltage divider from the terminal voltage. , It is obtained by piecewise linear interpolation of the cell's temperature resistance curve at the factory, with a value of 1.2 to 1.5 near 0℃.

[0034] S4 performs de-blackboxing reinforcement learning dynamic scheduling and physical loop closure.

[0035] After the power supply security calculation unit outputs a security permission signal, the power distribution network is activated, and the current remaining power is compared with the user immersion weighting coefficient. Encapsulated as a state sequence, the state sequence is input into the power distribution network to deduce a joint scheduling strategy including the target power consumption wall limit, the target backlight PWM duty cycle, and the power state of the target wireless module. The power consumption wall limit is a continuous variable. The power distribution network uses a deep deterministic strategy gradient DDPG or a near-end strategy optimization PPO algorithm to carry the continuous action space rather than discrete Q-learning. Its reward function has the battery life as a positive term and the number of drops that cross the line as a penalty term. The weights converge through positive and negative samples during the offline training phase.

[0036] The embedded power management controller receives the joint scheduling strategy via I2C, SPI, or eSPI bus, concurrently sends execution instructions to the system bus, collects the physical state after execution as feedback state to feed back to the power distribution network for iteration, and returns to S1 to enter the next round of following. After the power supply safety calculation unit outputs the red line interception signal, the power distribution network is forcibly suspended and taken over by the compensation logic. The power supply safety calculation unit determines the user immersion weight coefficient. , If the value exceeds the preset immersion value, an emergency peak-shaving command is sent to the embedded power management controller. The embedded power management controller reduces the backlight PWM duty cycle voltage to the basic minimum threshold and drives the wireless RF front end into sleep mode.

[0037] The power supply safety calculation unit quantifies the compensation current released during peak shifting. The peak current is subtracted from the compensation current to obtain the updated current. The updated current is used to replace the peak current to perform a single anti-dropout calculation, resulting in a recalculated predicted dropout voltage. If the recalculated predicted dropout voltage is greater than the preset protection voltage, the backlight and RF currents released during peak shifting are horizontally backfilled into the computing power bus, allowing the neural network processing unit to smoothly pass through the computing power peak without triggering hardware undervoltage lockout. If the recalculated predicted dropout voltage is not greater than the preset protection voltage, the loading signal is intercepted, effectively reducing the probability of forced power outage and system crash.

[0038] This application also discloses a PC battery life intelligent management system. This system is used to execute the aforementioned method. The overall system includes an internal resistance calculation module, a load sensing module, a safety calculation module, and a power distribution scheduling module. Figure 4 As can be seen from the hierarchical topology, the internal resistance calculation module and the load sensing module constitute the input stage that flows into the safety calculation module in parallel. The load sensing module has a feedforward branch that directly reaches the power distribution dispatch module, while the power distribution dispatch module has a reverse connection that leads back to the safety calculation module.

[0039] The internal resistance calculation module uses the battery management microcontroller unit as its physical carrier. The sampling input terminal of its analog-to-digital conversion circuit is coupled to the positive and negative terminals of the battery and is configured to collect the cell terminal voltage, transient discharge current, and surface temperature according to a preset polling cycle to obtain the dynamic internal resistance of the battery. The logic operation unit of the internal resistance calculation module is configured to calculate the difference between the discharge currents of adjacent cycles. When the difference between the discharge currents of adjacent cycles is greater than a preset step threshold, the dynamic internal resistance is calculated based on the voltage drop difference corresponding to the difference and the terminal voltage. That is, when the current difference is greater than the preset current step threshold, the dynamic internal resistance is determined by the difference between the voltage drop difference and the current difference. When the difference between the discharge currents of adjacent cycles is not greater than the preset step threshold, the difference is determined as thermal noise, and the division operation for calculating the dynamic internal resistance is blocked. The preset historical internal resistance is extracted and used as the dynamic internal resistance for drop test calculation. That is, when the current difference is not greater than the preset current step threshold, the division operation is blocked and the historical dynamic internal resistance in the status register is kept valid.

[0040] The load-aware module is communicatively coupled to the operating system's memory scheduling stack. It is configured to extract a matching calibration current from the calibration table as the peak current when the bus throughput mapped by the memory scheduling stack exceeds a preset mutation value and local large language model loading signaling is simultaneously captured. The statistics unit of the load-aware module is configured to calculate the user immersion weight coefficient by weighted summation of input interrupt frequency and gaze duration within a preset time window. The characteristic input terminal of the load-aware module can optionally be reconnected to a PCIe bus throughput monitoring point.

[0041] The data input terminal of the safety calculation module is coupled to the output terminals of the internal resistance calculation module and the load sensing module, respectively. It is configured to obtain an ohmic voltage divider by multiplying the difference between the peak current and the transient discharge current by the dynamic internal resistance, and to obtain the transient drop prediction voltage by subtracting the ohmic voltage divider from the terminal voltage. The comparison unit of the safety calculation module is configured to compare the transient drop prediction voltage with the sum of the baseline voltage and the safety margin voltage. If the former is greater than the latter, a safety permission signal is set; if the former is not greater than the latter, a red line interception signal is set. The safety calculation module is further configured to introduce a nonlinear temperature compensation coefficient greater than 1 to weight the ohmic voltage divider when the surface temperature is below the low-temperature polarization threshold.

[0042] The power distribution scheduling module uses an embedded power management controller as its actuator. Its command output is coupled to the system power wall register, backlight PWM driver, and wireless RF front-end via I2C, SPI, or eSPI bus. The power distribution scheduling module has an embedded power distribution network and is configured to deduce a joint scheduling strategy including the power wall limit, backlight duty cycle, and wireless module power status based on the remaining power and user immersion weight coefficient as a state sequence when the safety permission signal is valid. It is also configured to suspend the power distribution network, reduce the backlight duty cycle voltage to the basic minimum threshold, and put the wireless RF front-end into sleep mode when the red line interception signal is valid. At the same time, it quantifies the released compensation current, subtracts the compensation current from the peak current to obtain the updated current, and uses the updated current to replace the peak current to feed back the safety calculation module to perform a single drop test, obtain the recalculated predicted drop voltage, and intercept the loading signal when the recalculated predicted drop voltage is not greater than the preset protection voltage.

[0043] The internal resistance calculation module, load sensing module, safety calculation module, and power distribution scheduling module can be physically housed in different hardware partitions of the same integrated main control chip, or they can be placed separately in the battery management microcontroller unit, neural network processing unit, and embedded power management controller, interconnected via a board-level bus. The virtual logic functions of each module are driven by the processor reading instruction sequences from the memory, and the memory and processor are coupled via a data bus to form the physical hardware architecture. The microscopic details of internal resistance calibration, calculation formulas, and temperature correction have been described in detail in the aforementioned method embodiments and will not be repeated here.

[0044] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for intelligent management of PC battery life, characterized in that, include: The battery's terminal voltage and discharge current are collected at a preset period to obtain the battery's dynamic internal resistance; wherein, when the difference in discharge current between adjacent periods is greater than a preset step threshold, the dynamic internal resistance is calculated based on the difference and the drop difference corresponding to the terminal voltage. Monitor the memory scheduling stack, and when the loading signal of the preset model is captured, extract the peak current corresponding to the preset model and calculate the user weight based on the input interrupt frequency; Based on the terminal voltage, the discharge current, the dynamic internal resistance, and the peak current, a drop-proof calculation is performed to obtain the predicted drop voltage. The predicted voltage drop is compared with the preset protection voltage; If the predicted drop voltage is greater than the preset protection voltage, the remaining battery power is obtained, and a joint scheduling strategy including a power consumption limit is issued based on the remaining battery power and the user weight. If the predicted drop voltage is not greater than the preset protection voltage, peak-shaving control is performed to reduce the display backlight and switch the radio frequency communication to sleep mode.

2. The PC battery life intelligent management method according to claim 1, characterized in that, The process of obtaining the dynamic internal resistance of the battery includes: When the difference between the discharge currents in adjacent cycles is not greater than the preset step threshold, the difference is determined to be thermal noise, and the division operation for calculating the dynamic internal resistance is blocked. Extract the preset historical internal resistance and use the historical internal resistance as the dynamic internal resistance to perform the drop protection calculation.

3. The PC battery life intelligent management method according to claim 1, characterized in that, When the monitoring memory scheduling stack captures the loading signaling of a preset model, it extracts the peak current corresponding to the preset model, including: Obtain the bus throughput of the memory scheduling stack mapping; If the bus throughput is greater than a preset mutation value and the loading signaling is captured synchronously, the calibration current matching the preset model is extracted, and the calibration current is used as the peak current.

4. The PC battery life intelligent management method according to claim 3, characterized in that, The calculation of user weights based on input interruption frequency includes: The frequency of input interruptions within a preset time window is obtained, and the gaze duration is extracted simultaneously. If the frequency of input interruption is greater than a preset frequency threshold and the duration of gaze is greater than a preset focus threshold, the user weight is calculated by weighting the frequency of input interruption and the duration of gaze by a preset coefficient.

5. The PC battery life intelligent management method according to claim 1, characterized in that, The step of performing drop-proof calculations based on the terminal voltage, the discharge current, the dynamic internal resistance, and the peak current to calculate the predicted drop-proof voltage includes: Subtracting the discharge current from the peak current yields the transient differential current, and multiplying the transient differential current by the dynamic internal resistance yields the ohmic voltage divider. The predicted voltage drop is calculated by subtracting the ohmic voltage divider from the terminal voltage.

6. The PC battery life intelligent management method according to claim 5, characterized in that, The step of subtracting the ohmic voltage divider from the terminal voltage to calculate the predicted voltage drop includes: The surface temperature of the battery is obtained, and the polarization threshold and temperature compensation value are extracted. If the surface temperature is less than the polarization threshold, the corrected voltage is obtained by multiplying the ohmic voltage divider by the temperature compensation value, and the predicted drop voltage is calculated by subtracting the corrected voltage divider from the terminal voltage.

7. The PC battery life intelligent management method according to claim 6, characterized in that, The step of comparing the predicted voltage drop with the preset protection voltage includes: Obtain the baseline voltage and the margin voltage, and add the baseline voltage and the margin voltage to obtain the preset protection voltage; If the predicted drop voltage is greater than the preset protection voltage, a permission signal is output to trigger the joint scheduling strategy. If the predicted voltage drop is not greater than the preset protection voltage, an interception signal is output to trigger the peak shaving control.

8. The PC battery life intelligent management method according to claim 7, characterized in that, In response to the permission signaling, the issuance of a joint scheduling policy including a power consumption cap based on the remaining power and the user weight includes: Extract the preset power distribution network and encapsulate the remaining power and the user weight into a state sequence; The state sequence is input into the power distribution network to deduce the joint scheduling strategy, which includes the power consumption limit, the display duty cycle, and the radio frequency status. The joint scheduling strategy is issued, feedback status is obtained synchronously, and the feedback status is input into the power distribution network for iteration.

9. The PC battery life intelligent management method according to claim 7, characterized in that, In response to the interception signaling, triggering the peak-shaving control includes: If the user weight is greater than the preset immersion value, the display backlight is reduced to a minimum value and the radio frequency communication is switched to the sleep state; Extract the compensation current released during the peak shift control, and subtract the compensation current from the peak current to obtain the updated current; The peak current is replaced by the updated current to perform a single drop test, resulting in the recalculated predicted drop voltage; If the recalculated predicted drop voltage is not greater than the preset protection voltage, the loading signaling is intercepted.

10. A PC battery life intelligent management system, used to execute the PC battery life intelligent management method according to any one of claims 1-9, characterized in that, include: An internal resistance calculation module is used to collect the battery's terminal voltage and discharge current at preset cycles to obtain the battery's dynamic internal resistance; wherein, when the difference in discharge current between adjacent cycles is greater than a preset step threshold, the dynamic internal resistance is calculated based on the difference and the drop difference corresponding to the terminal voltage. The load awareness module is used to monitor the memory scheduling stack. When it captures the loading signal of the preset model, it extracts the peak current corresponding to the preset model and calculates the user weight based on the input interrupt frequency. The safety calculation module is used to perform drop test based on the terminal voltage, the discharge current, the dynamic internal resistance and the peak current, and calculate the predicted drop voltage. The power distribution scheduling module is used to compare the predicted voltage drop with the preset protection voltage; if the predicted voltage drop is greater than the preset protection voltage, the remaining power of the battery is obtained, and a joint scheduling strategy including a power consumption limit is issued based on the remaining power and the user weight; if the predicted voltage drop is not greater than the preset protection voltage, peak-shaving control is performed to reduce the display backlight and switch the radio frequency communication to a sleep state.

Images