Electric quantity metering system with battery mifi
By identifying charging and discharging events and combining Thevenin equivalent circuit and Kalman filtering algorithm, high-precision SOC estimation of MIFI devices is achieved, solving the problems of high cost and poor accuracy in existing technologies, and possessing adaptability and reliability.
Patent Information
- Application Number
- CN202511449126.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-11
- Publication Date
- 2026-01-06
AI Technical Summary
Existing MIFI devices rely on dedicated fuel metering chips, which are costly and inaccurate, especially under dynamic loads where errors are significant. This can affect charging decisions, leading to inaccurate calibration timing and large cumulative errors.
An information acquisition unit is used to identify charging start and discharging termination events. Combined with the Thevenin equivalent circuit model and the extended Kalman filter algorithm, the SOC is estimated through voltage dynamic modeling, and calibration is performed in full-charge and zero-charge states to avoid relying on a dedicated fuel meter chip.
It achieves high-precision SOC estimation with an accuracy better than ±1%, reduces costs, has adaptability and reliability, avoids miscalibration of traditional interrupt signals, and is suitable for resource-constrained MIFI devices.
Smart Images
Figure CN121276367A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power management technology for mobile communication devices, and more specifically to a battery-powered MiFi power metering system. Background Technology
[0002] Existing MiFi devices mostly use dedicated fuel gauge ICs for SOC estimation. These chips rely on coulomb counting and complex calibration processes, resulting in high costs and sensitivity to battery aging. Some low-cost solutions rely solely on voltage lookup tables, which are inaccurate, especially under dynamic loads. Furthermore, current technologies often rely on the "charging complete" interrupt signal from the charging management chip to determine full charge, which is susceptible to voltage drop and temperature drift in the charging circuit, leading to inaccurate calibration timing and large cumulative errors.
[0003] Therefore, there is an urgent need for a power metering system that does not require a dedicated power meter chip, but combines charging start / stop behavior recognition and voltage dynamic modeling to achieve high-precision SOC estimation and reliable full-charge calibration. Summary of the Invention
[0004] In view of this, the present invention provides a battery-powered MiFi power metering system that does not rely on a fuel gauge. By identifying the voltage change trend during charging start-up, constant voltage phase, and discharge termination behavior, and combining extended Kalman filtering (EKF) and dynamic full-charge calibration mechanism, it achieves a highly reliable estimation of SOC accuracy better than ±1%.
[0005] To achieve the above objectives, the present invention adopts the following technical solution:
[0006] A battery-powered MiFi metering system includes:
[0007] The information acquisition unit is used to collect battery terminal voltage and identify charging start events and discharging termination events.
[0008] The computational processing unit is used to construct nonlinear state-space equations based on the Thevenin equivalent circuit model, use the extended Kalman filter algorithm and perform joint estimation of the battery state based on the battery terminal voltage and the charging start event, and output a real-time SOC estimate.
[0009] The calibration unit performs full-charge calibration or zero-charge calibration when preset conditions are met, and feeds back the full-charge and zero-charge calibration results to the calculation and processing unit to correct the SOC estimate; the zero-charge calibration is triggered by the discharge termination event.
[0010] Preferably, the computing processing unit includes:
[0011] Constructing nonlinear state-space equations based on Thevenin equivalent circuit model;
[0012] Establish the state transition equations and terminal voltage observation equations for the Thevenin equivalent circuit model;
[0013] Upon detecting a charging start event, initialize the state vector and covariance matrix, and perform the following steps within each sampling period:
[0014] Using the current charging and discharging current and the previous state vector and covariance matrix obtained through nonlinear state-space equations, the current state vector and covariance matrix are predicted based on the state transition equations.
[0015] The voltage at the current moment is predicted by the terminal voltage observation equation, and the residual is calculated based on the voltage after filtering by the median value of the battery terminal voltage.
[0016] Based on the residual and Kalman gain, the predicted current state vector and the predicted current covariance matrix are updated to complete the joint correction of the state of charge, ohmic internal resistance, polarization resistance, polarization capacitance, polarization voltage and charging / discharging current.
[0017] Output the real-time SOC estimate.
[0018] Preferably, the state vector at time k-1 in the nonlinear state-space equation is defined as follows:
[0019]
[0020] Among them, SOC k-1 Let R be the state of charge at time k-1. 0,k-1 R is the ohmic internal resistance at time k-1. 1,k-1 and C 1,k-1 V represents the polarization resistance and polarization capacitance of the polarization branch at time k-1. RC,k-1 Let I be the polarization voltage across the RC branch at time k-1. k-1 This represents the charging and discharging current at time k-1.
[0021] Preferably, the state transition equation is:
[0022]
[0023] I k =I k-1 +ω I
[0024] R 0,k =R 0,k-1
[0025] R 1,k =R 1,k-1
[0026] C 1,k =C 1,k-1
[0027] Among them, SOC k SOC represents the state of charge at time k. k-1 I represents the state of charge at time k-1. k Let Q represent the predicted current at time k, where Δt is the sampling period, and Q is the current at time k. nom This refers to the battery's nominal capacity; V RC,k V represents the polarization voltage across the RC branch at time k. RC,k-1 R represents the polarization voltage of the RC branch at time k-1. 1,k-1 and C 1,k-1 Let η represent the resistance and capacitance of the polarization branch at time k-1, and let η represent the Coulomb efficiency, ω I R represents the noise during the current process. 0,k Let R be the ohmic internal resistance at time k. 1,k and C 1,k Let be the polarization resistance and polarization capacitance of the polarization branch at time k;
[0028] The terminal voltage observation equation is:
[0029] V terminal,k =OCV(SOC) k ,T k )-I k ·R 0,k -V RC,k
[0030] Among them, OCV(SOC) k ,T k Based on the current state of charge and battery temperature T k The open-circuit voltage function, R 0,k Let V be the ohmic internal resistance at time k. terminal,k This represents the terminal voltage at time k.
[0031] Preferably, the residual calculation formula is as follows:
[0032] y k =V filtered -V terminal,k
[0033] Among them, V filtered This represents the filtered voltage value after median voltage analysis at the battery terminal. k Represents the residual.
[0034] Preferably, the calibration unit performs the following calibration logic:
[0035] Full charge calibration: When the device is in the constant voltage charging stage and the battery terminal voltage change rate is continuously monitored to be less than 5mV / min for more than a preset time, it is determined that the battery has reached a true full charge state, the current output SOC estimate is forcibly corrected to 100%, and the battery parameters at this time are used as the new full charge reference point.
[0036] Zero-voltage calibration: When the discharge termination event is detected, the current output SOC estimate is corrected to 0%, and the open-circuit voltage corresponding to 0% SOC is used as the new zero-voltage reference point;
[0037] The full-charge and zero-charge calibration results are used to update the OCV-SOC lookup table and fed back to the calculation and processing unit to reset the updated covariance matrix.
[0038] Preferably, the information acquisition unit performs the following steps:
[0039] The battery terminal voltage is periodically collected by the built-in ADC module of the main control chip, with a sampling frequency of 1Hz to 10Hz.
[0040] Perform median filtering on N consecutive voltage samples and output the filtered voltage.
[0041] When the main control chip receives an interrupt signal from the charging management IC, the GPIO detects a change in the USB or power adapter connection status, or the system service detects the start of the charging process, it is identified as a charging start event.
[0042] When the device is powered off or enters deep sleep mode, and the battery terminal voltage is lower than the preset cutoff voltage, it is identified as a discharge termination event.
[0043] Preferably, the preset duration is greater than or equal to 5 minutes.
[0044] As can be seen from the above technical solution, compared with the prior art, the present invention discloses a battery-powered MiFi metering system, which has the following advantages:
[0045] Hardware-free: No dedicated fuel meter chip is required, reducing costs and simplifying the BOM;
[0046] High precision: By integrating voltage dynamics and charging behavior through EKF, SOC accuracy can be achieved within ±1%;
[0047] Reliable calibration: Full charge determination based on voltage change rate is superior to traditional interrupt signals, avoiding miscalibration;
[0048] Strong Adaptability: Through dual-point calibration at full charge / zero charge, the OCV-SOC meter is dynamically updated to achieve adaptive compensation for battery aging.
[0049] In summary, without increasing hardware costs, it provides SOC estimation capabilities for resource-constrained devices such as MIFI that are close to those of professional power meters. It combines low cost, high robustness, and strong adaptability, and has significant engineering application value and market competitiveness. Attached Figure Description
[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0051] Figure 1 This is a schematic diagram of a battery-powered MIFI power metering system provided by the present invention.
[0052] Figure 2 The flowchart of the computational processing unit provided by the present invention.
[0053] Figure 3 The Thevenin equivalent circuit model provided for this invention. Detailed Implementation
[0054] 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.
[0055] This invention discloses a battery-powered MiFi metering system, such as... Figure 1 As shown, it includes:
[0056] The information acquisition unit is used to collect battery terminal voltage and identify charging start events and discharging termination events;
[0057] The computational processing unit is used to construct nonlinear state-space equations based on the Thevenin equivalent circuit model, use the extended Kalman filter algorithm and perform joint estimation of battery state based on battery terminal voltage and charging start event, and output real-time SOC estimate.
[0058] The calibration unit performs full-charge calibration or zero-charge calibration when preset conditions are met, and feeds back the full-charge and zero-charge calibration results to the calculation and processing unit to correct the SOC estimate and update the battery model parameters; zero-charge calibration is triggered by the discharge termination event.
[0059] In this embodiment, the information acquisition unit performs the following steps:
[0060] The battery terminal voltage is periodically collected by the built-in ADC module of the main control chip, with a sampling frequency of 1Hz to 10Hz.
[0061] The median value is filtered for N consecutive voltage samples (N≥3), and the filtered voltage Vfiltered is used as the input for subsequent processing to suppress ADC quantization noise and transient interference.
[0062] The charging start event is detected, and the triggering conditions include: the main control chip receiving the "CHRG_START" interrupt signal from the charging management IC, the GPIO detecting a change in the USB or power adapter connection status, or the system service detecting the start of the charging process;
[0063] The discharge termination event is detected when the device is powered off or enters a deep sleep state, and the battery terminal voltage is lower than the preset cutoff voltage.
[0064] In this embodiment, a nonlinear state-space equation is constructed based on the Thevenin equivalent circuit model. An extended Kalman filter algorithm is used, and the battery state is jointly estimated based on the battery terminal voltage and the charging start event. A real-time SOC estimate is then output, such as... Figure 2 As shown, it specifically includes:
[0065] (1) Establishing the Thevenin equivalent circuit model: The battery is equivalent to a circuit model consisting of open-circuit voltage OCV, ohmic internal resistance R0, an RC parallel branch R1, C1, and load current. The RC branch is used to characterize the dynamic polarization characteristics of the battery, such as... Figure 3 As shown;
[0066] (2) Constructing nonlinear state-space equations:
[0067] Define the state vector at time k-1 as:
[0068]
[0069] Among them, SOC k-1 Let R be the state of charge at time k-1. 0,k-1 R is the ohmic internal resistance at time k-1. 1,k-1 and C 1,k-1 V represents the polarization resistance and polarization capacitance of the polarization branch at time k-1. RC,k-1 Let I be the polarization voltage across the RC branch at time k-1. k-1 This represents the charging and discharging current at time k-1;
[0070] The state transition equations for establishing the Thevenin equivalent circuit model are as follows:
[0071]
[0072] I k =I k-1 +ω I
[0073] R 0,k =R 0,k-1
[0074] R 1,k =R 1,k-1
[0075] C 1,k =C 1,k-1
[0076] Among them, SOC k SOC represents the state of charge at the current time k. k-1 I represents the state of charge of k-1 at the previous time step. k Q represents the charging / discharging current at time k, where Δt is the sampling period. nom This refers to the battery's nominal capacity; V RC,k-1 R represents the polarization voltage of the RC branch at time k-1. 1,k-1 and C 1,k-1 Let η represent the resistance and capacitance of the polarization branch at time k-1, and let η represent the Coulomb efficiency, ω I R represents the noise during the current process. 0,k Let R be the ohmic internal resistance at time k. 1,k and C 1,k Let be the polarization resistance and polarization capacitance of the polarization branch at time k.
[0077] Establish the terminal voltage observation equation:
[0078] V terminal,k =OCV(SOC) k ,T k )-I k ·R 0,k -V RC,k
[0079] Among them, OCV(SOC) k ,T k Based on the current SOC and battery temperature T k The open-circuit voltage function is experimentally calibrated and stored in the OCV-SOC-T lookup table. That is, the open-circuit voltage is calculated by interpolating the current SOC from the OCV-SOC lookup table. The OCV-SOC table is constructed as follows:
[0080] In a standard laboratory environment, slow charge and discharge tests were performed on the target battery model at multiple temperature points (including at least 0°C, 25°C, and 45°C) to obtain high-resolution open-circuit voltage (OCV) and SOC corresponding data.
[0081] Establish a two-dimensional or multi-dimensional lookup table to store the OCV values corresponding to each 1% SOC interval at different temperatures, with a resolution of not less than 1mV.
[0082] (3) Initialize the extended Kalman filter (EKF) algorithm: When the information acquisition unit detects the charging start event, it triggers the initialization process of the EKF algorithm. The initial state of charge (SOC) is obtained by looking up the OCV-SOC table through the static voltage. The initial state vector X0 and covariance matrix P0 are initialized. The initial ohmic resistor, polarization resistor and polarization capacitor use the factory calibration value or typical value.
[0083] (4) Perform the EKF prediction and update steps sequentially within each sampling period:
[0084] Using the state vector and covariance matrix from the previous time step, combined with the charging / discharging current I at the current time step k The current state vector and prior covariance are predicted through the state transition equation;
[0085] The voltage at the current moment is predicted using the terminal voltage observation equation, and the voltage V after filtering the median value of the pool terminal voltage is then calculated. filtered As the actual measured value, calculate the residual:
[0086] y k =V filtered -V terminal,k
[0087] Among them, y k V represents the residual. filtered The filtered terminal voltage represents the actual battery terminal voltage obtained by the information acquisition unit through ADC sampling and median filtering, which has suppressed noise and transient interference and serves as the actual observation input for EKF.
[0088] The predicted current state vector and the posterior covariance obtained based on the residual and Kalman gain are updated to complete the joint correction of the state of charge, ohmic internal resistance, polarization resistance, polarization capacitance, polarization voltage, and charging and discharging current.
[0089] (5) Output real-time SOC estimate: Output the updated state of charge as the current state of charge.
[0090] Even better, during the initial charging phase, the system current I... kA rough estimate is made based on the known charging level of the system, or a reverse calculation is performed based on an empirical model of voltage float difference, which serves as the input excitation for the EKF state equation. During the initial discharge phase, the load power is estimated. The specific process of reverse calculation based on the empirical model of voltage float difference is as follows:
[0091]
[0092] Where ΔV represents the voltage float difference, and I0 represents the initial charging current. The open-circuit voltage obtained from the OCV-SOC meter is approximately the resting voltage, R. eff Indicates the equivalent internal resistance, factory calibration.
[0093] In this embodiment, the calibration unit performs the following calibration logic:
[0094] Full charge calibration: (a) When the device is in the constant voltage (CV) charging stage and the terminal voltage change rate is continuously monitored to be <5mV / min for more than 5 minutes, (b) when the charging current is less than the preset termination current threshold, it is determined that the battery has reached a true full charge state, the current output SOC estimate is forcibly corrected to 100%, and the battery parameters, including ohmic internal resistance, polarization resistance and polarization capacitance and polarization voltage, are updated. The battery parameters at this time are used as the new full charge reference point.
[0095] The dual judgment mechanism of this invention significantly improves the reliability of full charge identification and avoids premature calibration caused by contact resistance, temperature drift or erroneous signals from the charging management chip.
[0096] Zero-voltage calibration: When a discharge termination event is detected and the terminal voltage is lower than the preset cutoff voltage, the current output SOC estimate is corrected to 0%, and the open-circuit voltage corresponding to 0% SOC is used as the new zero-voltage reference point.
[0097] The full-charge and zero-charge calibration results are used to update the OCV-SOC lookup table and fed back to the computational processing unit to reset the covariance matrix of the EKF, preventing state estimation divergence.
[0098] Dynamic reconstruction of OCV-SOC curve:
[0099] After each complete cycle from full charge to zero charge or vice versa, the OCV-SOC relationship is reconstructed using data from both endpoints via linear or piecewise interpolation to compensate for battery capacity degradation. After full-charge calibration, the current battery parameters are used as a new reference point for subsequent OCV-SOC mapping updates during discharge, achieving adaptive compensation for battery aging.
[0100] This invention, EKF, is not only used for SOC estimation, but also enables online identification of battery internal resistance and polarization parameters. It works in conjunction with a full charge judgment mechanism based on voltage change rate and a two-point calibration strategy, significantly improving the stability and accuracy of long-term use.
[0101] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.
[0102] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A battery powered mifi power metering system, comprising: The application relates to a battery state of charge (SOC) estimation method and device. The application comprises: An information acquisition unit is configured to collect battery terminal voltage and identify charging start events and discharging termination events; A calculation processing unit is configured to construct a nonlinear state space equation based on a Thevenin equivalent circuit model, jointly estimate battery states by using an extended Kalman filter algorithm and based on the battery terminal voltage and the charging start events, and output real-time SOC estimation values; A correction unit is configured to perform full-charge calibration or zero-charge calibration when preset conditions are met, and feed back full-charge and zero-charge calibration results to the calculation processing unit to correct the SOC estimation values; 2. The battery powered Mifi power metering system of claim 1, wherein, The zero-charge calibration is triggered by the discharging termination event. The calculation processing unit comprises: constructing a nonlinear state space equation based on a Thevenin equivalent circuit model; establishing a state transition equation and a terminal voltage observation equation of the Thevenin equivalent circuit model; after detecting the charging start event, initializing a state vector and a covariance matrix, and performing the following steps in each sampling period: using current charging and discharging currents, a last-time state vector and a last-time covariance matrix obtained by the nonlinear state space equation, and predicting a current-time state vector and a current-time covariance matrix based on the state transition equation; predicting a current-time voltage by the terminal voltage observation equation, and calculating a residual error based on a voltage filtered by a median value of the battery terminal voltage; updating the predicted current-time state vector and the predicted current-time covariance matrix based on the residual error and a Kalman gain, and completing joint correction of a state of charge, an ohmic resistance, a polarization resistance, a polarization capacitance, a polarization voltage and charging and discharging currents; 3. The battery powered Mifi power metering system of claim 2, wherein, outputting real-time SOC estimation values. wherein SOC k-1 is the state of charge at the k-1th moment, R 0,k-1 is the ohmic internal resistance at the k-1th moment, R 1,k-1 and C 1,k-1 are the polarization resistance and the polarization capacitance of the polarization branch at the k-1th moment, V RC,k-1 is the polarization voltage across the RC branch at the k-1th moment, I k-1 denotes the charge and discharge current at the k-1th moment.
4. The battery powered Mifi power metering system of claim 2, wherein, In the nonlinear state space equation, a state vector at a k-1 time is defined as: Among them, SOC k SOC represents the state of charge at time k. k-1 I represents the state of charge at time k-1. k Q represents the predicted charging / discharging current at time k, where Δt is the sampling period. nom This refers to the battery's nominal capacity; V RC,k V represents the polarization voltage across the RC branch at time k. RC,k-1 R represents the polarization voltage of the RC branch at time k-1. 1,k-1 and C 1,k-1 Let η represent the resistance and capacitance of the polarization branch at time k-1, and let η represent the Coulomb efficiency, ω I I represents the noise during the current process. k-1 This represents the charging and discharging current at time k-1; the state transition equation is: V terminal,k = OCV(SOC k ,T k ) - I k · R 0,k - V RC,k where OCV(SOC k , T k ) is an open circuit voltage function according to the current state of charge and the battery temperature T k , R 0,k is the ohmic internal resistance at the kth moment, and V terminal,k represents the terminal voltage at the kth moment.
5. The battery powered Mifi power metering system of claim 4, wherein, the terminal voltage observation equation is: y k = V filtered - V terminal,k wherein V filtered represents the filtered voltage value of the battery terminal voltage, y k represents the residual.
6. The battery powered Mifi power metering system of claim 2, wherein, a residual error calculation formula is: The correction unit performs the following calibration logic: full-charge calibration: when a device is in a constant-voltage charging stage, and a battery terminal voltage change rate is continuously monitored to be less than 5 mV / min for more than a preset time length, it is determined that the battery reaches a real full-charge state, a current output SOC estimation value is forcibly corrected to 100%, and battery parameters at this time are taken as new full-charge reference points; zero-charge calibration: when the discharging termination event is detected, the current output SOC estimation value is corrected to 0%, and an open-circuit voltage corresponding to the 0% SOC is taken as new zero-charge reference points; 7. The battery powered Mifi power metering system of claim 1, wherein, full-charge and zero-charge calibration results are used to update an OCV-SOC lookup table, and are fed back to the calculation processing unit to reset an updated covariance matrix. The information acquisition unit performs the following steps: periodically collecting battery terminal voltage by using an ADC module built in a master control chip, and the sampling frequency is 1 Hz-10 Hz; performing median value filtering on N continuous voltage sampling values, and outputting filtered voltage; identifying a charging start event when the master control chip receives an interrupt signal sent by a charging management IC, a GPIO detects USB or power adapter access state change, or a system service detects charging process start. When the device is powered off or enters a deep sleep state, and the terminal voltage of the battery is lower than the preset cutoff voltage, the discharge termination event is identified.
8. The battery powered Mifi power metering system of claim 6, wherein, The preset time length is greater than or equal to 5 minutes.