FM33M0XX chip storage battery monitoring system based on sensing technology

Abstract

The invention relates to the technical field of storage battery monitoring, in particular to an FM33M0XX chip storage battery monitoring system based on a sensing technology. According to the invention, the FM33M0XX chip and the communication interface + LDO chip work cooperatively, so that the working current of the system is reduced, and the electric quantity loss of the storage battery is reduced; the working condition parameters of the storage battery are analyzed through the temperature compensation unit to obtain temperature compensation parameters, then the compensated temperature is obtained, and the temperature monitoring accuracy is improved; according to the invention, the internal resistance measurement unit analyzes the internal resistance multi-dimensional data and the temperature compensation parameters, that is, the internal resistance multi-dimensional data is identified and analyzed to output a plurality of characteristic variables, then the regression mapping output algorithm is used to analyze and output the internal resistance state value, and the internal resistance state value is further optimized, output and updated. And finally, outputting an internal resistance state grade through internal resistance measurement state evaluation, thereby realizing accurate grading of the internal resistance state of the storage battery, providing a more accurate health state analysis basis, and realizing more accurate calculation of the battery power SOC.

Description

Technical Field

[0001] This invention relates to the field of battery monitoring technology, specifically to a battery monitoring system based on the FM33M0XX chip using sensor technology. Background Technology

[0002] The FM33M0XX series chips are ARM Cortex-M0+ processor core chips developed by Shanghai Fudan Microelectronics. They are further subdivided into multiple models based on differences in internal FLASH and RAM capacity and external packaging. Currently, existing technologies for battery monitoring utilize single-chip technology (such as the ADUCM330WFS) for intelligent monitoring. However, these chips have excessively high operating current, leading to rapid battery drain during long-term battery power supply and application. Excessive self-heating causes significant discrepancies between the temperature measured by the chip's built-in temperature sensor and the ambient temperature, resulting in distorted temperature model data. Furthermore, the chips support only one communication method, making it difficult to meet the needs of practical applications requiring multiple communication methods. Summary of the Invention

[0003] This invention provides a battery monitoring system based on the FM33M0XX chip using sensor technology to solve the aforementioned technical problems.

[0004] The first aspect of the present invention provides a battery monitoring system based on the FM33M0XX chip using sensing technology, comprising:

[0005] The core control module includes a main control unit and a control auxiliary unit. The main control unit uses the FM33M0XX chip as the main controller and is responsible for data reception, command issuance, and resource scheduling of each module. The control auxiliary unit uses the communication interface + LDO chip as the auxiliary chip and is responsible for communication interface and power supply management.

[0006] The sensor detection module is used to collect battery monitoring parameters based on battery operating conditions and environment, and then send them to the detection data processing module.

[0007] As a further improvement of the present invention, battery monitoring parameters are obtained by collecting data on battery operating conditions and the environment, specifically as follows:

[0008] Battery monitoring data is obtained by detecting the operating conditions and environment of the battery itself and the environment through a preset sensor group; the sensor group includes a temperature detection group and an internal resistance detection group.

[0009] The temperature detection group collects data through pre-deployed temperature sensors and obtains the sensing temperature data of the current sampling window based on the preset sampling window. It also obtains the spatial and temporal coordinate parameters of each sensor in the temperature detection group based on a unified spatial coordinate system. The spatial and temporal coordinate parameters of each sensor are correlated with the corresponding sensing temperature data to obtain the battery operating condition parameters.

[0010] The internal resistance detection group monitors the current and response voltage of the battery under different operating conditions. After filtering, the signal is marked as a voltage response signal. It acquires internal resistance measurement data for a preset number of measurements under the same discharge state, calculates the standard deviation of the internal resistance value corresponding to each measurement data point, and records it as the repeatability error value. It also acquires the maximum deviation of the internal resistance measurement values ​​within a preset historical period and records it as the internal resistance drift. Finally, it aggregates the voltage response signal, repeatability error, and internal resistance drift to obtain multi-dimensional internal resistance data.

[0011] Battery operating parameters and internal resistance multidimensional data are recorded as battery monitoring data.

[0012] The detection data processing module includes a temperature compensation unit and an internal resistance measurement unit. It identifies the received battery monitoring parameters to obtain battery operating condition parameters and multi-dimensional internal resistance data. The temperature compensation unit analyzes the battery operating parameters to obtain temperature compensation parameters, which are then sent to the internal resistance measurement unit. The internal resistance measurement unit analyzes the multi-dimensional internal resistance data and temperature compensation parameters to obtain internal resistance measurement parameters. The compensated temperature is obtained using the temperature compensation parameters, and state optimization is performed based on the internal resistance state level.

[0013] As a further improvement to the present invention, the specific analysis steps of the temperature compensation unit are as follows:

[0014] A1: Identify the battery operating parameters to obtain the spatial and temporal coordinate parameters corresponding to the built-in temperature of the main control chip and the ambient temperature;

[0015] A2: The temperature gradient vector is obtained by correlating the external temperature and the internal temperature through a preset thermal gradient calculation formula;

[0016] A3: Outputs the optimal ambient temperature estimate using a preset real ambient temperature estimation method;

[0017] A4: Obtain the thermal conductivity coefficient of the medium between the battery and the external environment. Input the thermal conductivity coefficient and temperature gradient vector into the preset heat flux density calculation formula to calculate and output the heat flux density. Obtain the thermal conductivity coefficient of the battery encapsulation material and the heat dissipation area of ​​the battery. Input the encapsulation thermal conductivity coefficient, heat flux density, and heat dissipation area into the preset dynamic temperature deviation calculation formula to calculate the deviation temperature caused by heat conduction. Compensate the built-in temperature with the built-in sensor deviation temperature to obtain the compensated temperature. Record the compensated temperature as the temperature compensation parameter.

[0018] Furthermore, the battery operating parameters are identified to obtain the spatial and temporal coordinate parameters corresponding to the built-in temperature and ambient temperature, specifically:

[0019] The battery operating parameters are identified to obtain the sensor temperature data and spatial-temporal coordinate parameters corresponding to the current sampling window. Based on the sensor temperature data, the built-in temperature and ambient temperature are obtained. The built-in temperatures of each sample window are added together to calculate the built-in average temperature. The built-in average temperature is then adjusted by adding or subtracting a preset temperature deviation threshold to obtain the built-in temperature distribution range. Built-in average temperatures exceeding the temperature distribution range are marked as temperature outliers. Similarly, based on the ambient temperature within the sampling window, the corresponding ambient average temperature and ambient temperature distribution range are obtained, thus identifying temperature outliers, which are then removed. The built-in temperature sensors corresponding to the built-in temperatures are obtained and numbered, with the total number of sensors recorded as N. The external temperature sensors corresponding to the ambient temperatures are obtained and numbered, with the total number of sensors recorded as M. The spatial-temporal coordinate parameters corresponding to the built-in temperature and ambient temperature are respectively marked as... , where i∈N and j∈M.

[0020] Furthermore, the optimal ambient temperature estimate is output through a preset real ambient temperature estimation method, specifically as follows:

[0021] Mark the current time as t, and the previous time as t-1. Obtain the estimated value of the real ambient temperature at the previous time from the database, and obtain the preset process noise. Input the estimated value of the real ambient temperature at the previous time and the process noise into the preset state equation to calculate and output the estimated value of the real ambient temperature at the current time.

[0022] By assuming a linear relationship between sensor measurements and the actual temperature, and defining the corresponding observation matrix, denoted as... ,in, Represented as a vector transpose, it is used to convert a row vector into a column vector. The first 1 indicates that the measured value of the built-in temperature sensor is linearly proportional to the actual ambient temperature in a 1:1 ratio; the second 1 indicates that the measured value of the external temperature sensor is linearly proportional to the actual ambient temperature in a 1:1 ratio.

[0023] The observation noise of the built-in temperature sensor and the external sensor at time t is obtained, and the observation noise vector is calculated by transposing the two observation noises.

[0024] The current observation noise vector, observation matrix, and estimated real ambient temperature are input into the preset sensor measurement relationship observation equation for calculation and output of the current observation vector.

[0025] Then, the optimal ambient temperature estimate is obtained by using the preset Kalman gain fusion formula to calculate the current ambient temperature estimate and the observation vector.

[0026] As a further improvement to the present invention, the specific analysis steps of the internal resistance measurement unit are as follows:

[0027] S1: Identify and analyze multidimensional internal resistance data to output multiple feature variables;

[0028] S2: Analyze and output the internal resistance state value through a regression mapping output algorithm;

[0029] S3: Optimize the output of the internal resistance state value and update the internal resistance state value;

[0030] S4: Internal resistance measurement status assessment outputs the internal resistance status level.

[0031] Furthermore, multiple feature variables are output through identification and analysis of the multidimensional internal resistance data. Specifically, the voltage response signal, repeatability error value, and internal resistance drift are obtained by identifying the multidimensional internal resistance data; the temperature compensation parameter is identified to obtain the compensated temperature; the voltage response signal, repeatability error value, internal resistance drift, and compensated temperature are normalized and mapped to [0, 1] and labeled as feature variables, and denoted as follows: .

[0032] Furthermore, the internal resistance state value is analyzed and output using a regression mapping output algorithm, specifically as follows:

[0033] Multiple data pairs are obtained by pairing each voltage response signal, repeatability error value, internal resistance drift, and compensated temperature normalized value. Each data pair is then input into a preset linear kernel function and an RBF kernel function. A preset mixed kernel function calculation formula is used to calculate the mixed kernel function for each data pair. The mixed kernel functions are then aggregated to obtain a mixed function similarity matrix, which is obtained through the formula... Perform calculations to output the hybrid kernel function. The similarity matrix of the mixing function is Hs; where, , These are the preset kernel weight coefficients; , Represented as vector transpose; , The preset RBF core width parameter;

[0034] Construct a single-hidden-layer neural network with four feature variables as input, L hidden nodes, and a single internal resistance state value as output; input the mixture function to the preset mixture output weight calculation formula. Calculate the mixed output weights ;in, The similarity matrix of the mixture function, calculated by the mixture kernel function, is a core component of the hidden layer output matrix H; Represented as vector transpose; The sample label is denoted by ; C is the preset regularization coefficient; the mixed output weights are input into the preset regression mapping output formula to calculate the output internal resistance state value.

[0035] Furthermore, the internal resistance state value is optimized and updated by outputting the updated internal resistance state value, specifically as follows:

[0036] Obtain the four key parameters in S2: the number of hidden layer nodes, kernel weight coefficient, RBF kernel width, and regularization coefficient. Record these four key parameters as the parameter vector to be optimized. Obtain the number of internal resistance state values ​​corresponding to a preset number of samples, and record them as predicted internal resistance state values. Obtain the pre-designed and calibrated true internal resistance state values ​​corresponding to each predicted internal resistance state value. Calculate the mean absolute error of the parameter vector to be optimized using a preset fitness function. , Represented as a vector of parameters to be optimized;

[0037] The core search mechanism based on the GWO optimization algorithm filters the optimal parameter vector to obtain the optimal parameter vector with the smallest mean absolute error; obtains the current filtering iteration number and the global optimal parameter vector corresponding to each iteration number, and then marks it as the prey position vector; calculates and outputs the distance vector between the gray wolf and the prey using the pre-set distance calculation formula of the GWO optimization algorithm.

[0038] Substitute the distance vector between the gray wolf and its prey and the prey's position vector into the preset gray wolf next-generation position update formula to calculate and output the next-generation gray wolf position vector;

[0039] GWO retains the three optimal parameter vectors and records them as the optimal gray wolf positions. The optimal gray wolf positions are used to guide the calculation of the remaining gray wolves to obtain the candidate position vectors corresponding to the optimal gray wolf positions.

[0040] The final position vector of the remaining gray wolves for the next iteration is calculated using the final position update formula of the remaining gray wolves. The optimal parameter vector of the gray wolf with the best position and the final position vector of the remaining gray wolves are marked as the updated optimized parameter vector of the parameter vector to be optimized. The updated optimized parameter vector is input into the regression mapping output formula to obtain the corresponding internal resistance state value, and it is recorded as the updated internal resistance state value.

[0041] Furthermore, the internal resistance measurement status assessment outputs an internal resistance status level, which is specifically as follows: the updated internal resistance status value is divided into multiple internal resistance status intervals based on a preset status value interval. Each internal resistance status interval is assigned an internal resistance status level, namely, internal resistance status abnormal, internal resistance status warning, and internal resistance status normal. The higher the updated internal resistance status value, the more normal the corresponding internal resistance status level. The current updated internal resistance status value is matched with each internal resistance status interval to obtain the corresponding internal resistance status level.

[0042] The communication and power management module uses a communication interface and an LDO chip to convert the battery voltage into a preset voltage to independently power the FM33M0XX chip; and connects with different communication interface chips through the serial port of the FM33M0XX chip.

[0043] The beneficial effects of the technical solution provided by this invention compared with the prior art are as follows:

[0044] 1. This invention uses the FM33M0XX chip as the main controller and the communication interface + LDO chip as the auxiliary chip to reduce the system operating current and reduce battery power loss; and uses a temperature compensation unit to analyze the battery operating parameters to obtain temperature compensation parameters, thereby obtaining the compensated temperature and improving the accuracy of temperature monitoring.

[0045] 2. This invention analyzes multidimensional internal resistance data and temperature compensation parameters through an internal resistance measurement unit. Specifically, it identifies and analyzes the multidimensional internal resistance data to output multiple feature variables, then analyzes and outputs the internal resistance state value through a regression mapping algorithm, further optimizes and updates the internal resistance state value, and finally outputs the internal resistance state level through internal resistance measurement state evaluation. This achieves accurate grading of the battery's internal resistance state, provides a more accurate basis for State of Health (SOH) analysis, and thus obtains a more accurate calculation of the battery's State of Charge (SOC). Attached Figure Description

[0046] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. The following drawings are not deliberately drawn to scale according to the actual size, but are intended to show the main idea of ​​this application.

[0047] Figure 1 This is a schematic diagram of the principle of the present invention;

[0048] Figure 2 This is a flowchart of the temperature compensation unit of the present invention;

[0049] Figure 3 This is a flowchart of the internal resistance measurement unit of the present invention;

[0050] Figure 4 This is a schematic diagram of the installation of the main control unit and the control auxiliary unit of the present invention. Detailed Implementation

[0051] 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.

[0052] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1-4 In one embodiment of the present invention, the FM33M0XX chip battery monitoring system based on sensing technology includes: a core control module, a sensing detection module, a detection data processing module, and a communication and power management module.

[0053] The core control module includes a main control unit and a control auxiliary unit. The main control unit and the control auxiliary unit provide the battery monitoring system with the basis for data resources, communication and power scheduling. Specifically, the main control unit uses the FM33M0XX chip as the main controller and is responsible for data reception, command issuance and resource scheduling of each module; the control auxiliary unit uses the communication interface + LDO chip as the auxiliary chip and is responsible for communication interface and power supply management.

[0054] The sensor detection module collects battery monitoring parameters based on battery operating conditions and the environment, and then sends these parameters to the detection data processing module. Specifically:

[0055] Battery monitoring data is obtained by detecting the operating conditions and environment of the battery itself and the environment through a preset sensor group; the sensor group includes a temperature detection group and an internal resistance detection group.

[0056] The temperature detection group collects data through pre-deployed temperature sensors (including but not limited to built-in temperature sensors and external temperature sensors), and obtains the sensing temperature data of the current sampling window based on a preset sampling window. Based on a unified spatial coordinate system (with the center point of the chip of the main control unit as the center coordinate), the spatial and temporal coordinate parameters of each sensor in the temperature detection group are obtained. The spatial and temporal coordinate parameters of each sensor are correlated with the corresponding sensing temperature data to obtain the battery operating condition parameters.

[0057] The internal resistance detection group monitors the current and response voltage of the battery under different operating conditions. After filtering, the signal is marked as a voltage response signal. The group acquires the internal resistance measurement data of the battery under the same discharge state for a preset number of measurements. The group then performs mathematical calculations on the internal resistance values ​​corresponding to each measurement to obtain the corresponding standard deviation, which is recorded as the repeatability error value. The group also acquires the maximum deviation of the internal resistance measurement values ​​within a preset historical period and records it as the internal resistance drift. Finally, the group combines the voltage response signal, repeatability error, and internal resistance drift to obtain multidimensional internal resistance data.

[0058] Battery operating parameters and internal resistance multidimensional data are recorded as battery monitoring data.

[0059] The detection data processing module includes a temperature compensation unit and an internal resistance measurement unit; it is used to identify the received battery monitoring parameters to obtain battery operating parameters and multi-dimensional internal resistance data, analyze the battery operating parameters through the temperature compensation unit to obtain temperature compensation parameters, and send the temperature compensation parameters to the internal resistance measurement unit; the internal resistance measurement unit analyzes the multi-dimensional internal resistance data and temperature compensation parameters to obtain internal resistance measurement parameters.

[0060] The compensated temperature is obtained through temperature compensation parameters, and the state is optimized based on the internal resistance state level. For example, if the current state is an internal resistance state warning (decreased accuracy, poor repeatability or increased drift), the corresponding state optimization is to adjust the current injection stability, update the internal resistance temperature coefficient of the temperature compensation model, and check the LDO power supply ripple.

[0061] The specific analysis steps for the temperature compensation unit are as follows:

[0062] A1. Temperature Dimension Data Identification: Identify battery operating parameters to obtain the sensor temperature data and spatial-temporal coordinate parameters corresponding to the current sampling window; obtain the built-in temperature and ambient temperature based on the sensor temperature data; sum the built-in temperatures within the sampling window to calculate the built-in average temperature; calculate the built-in temperature distribution range by adding or subtracting a preset temperature deviation threshold from the built-in average temperature; mark the built-in average temperature exceeding the temperature distribution range as an outlier; similarly, obtain the corresponding ambient average temperature and ambient temperature distribution range based on the ambient temperature within the sampling window, thereby obtaining temperature outliers, which are then removed; obtain the built-in temperature sensors corresponding to the built-in temperatures and number them, recording the total number of sensors as N; obtain the external temperature sensors corresponding to the ambient temperatures and number them, recording the total number of sensors as M; mark the spatial-temporal coordinate parameters corresponding to the built-in temperature and ambient temperature respectively as... , where i∈N and j∈M.

[0063] A2. Thermal gradient vector calculation: Calculated using a preset thermal gradient calculation formula. The temperature gradient vector is obtained by correlating the external temperature with the internal temperature. ;in, , , and , respectively, represent the temperature partial derivatives of the external sensor and the internal sensor in the x, y, and z axes; d is the corresponding linear spatial distance between the external sensor and the internal sensor; The unit vector of the main chip pointing to the corresponding external sensor is stored in the database, representing the direction of heat conduction. The thermal gradient calculation formula approximates the temperature gradient by the ratio of the temperature difference between the external sensor and the built-in sensor to the straight-line distance, and calculates the temperature gradient vector by reflecting the spatial characteristics of heat conduction in vector form, thereby quantifying the direction and intensity of heat conduction between the main chip and the environment.

[0064] A3. Real Ambient Temperature Estimation: Mark the current time as t, and the previous time as t-1. Obtain the estimated real ambient temperature value of the previous time from the database, and also obtain the preset process noise. Input the estimated real ambient temperature value and process noise of the previous time into the preset state equation. Calculate and output the estimated value of the current ambient temperature. ;in, This is the estimated value of the actual ambient temperature at time t-1; The process noise, whose value is obtained by fitting calibration data across the entire temperature range of the battery, is defined as the cumulative effect (unit: °C) of unpredictable minor disturbances experienced by the real ambient temperature at time t during dynamic changes. It is a key parameter ensuring that the state equation conforms to the actual physical scenario, rather than the zero-noise assumption under ideal conditions. Therefore, in battery monitoring scenarios, the real ambient temperature is not absolutely constant; even within a short period (e.g., a 30-second sampling interval), it can be subject to slight fluctuations due to weak disturbances. It is precisely this quantitative characterization of such fluctuations that ensures a close fit to the actual environment;

[0065] By assuming a linear relationship between sensor measurements and the actual temperature, and defining the corresponding observation matrix, denoted as... ,in, Represented as a vector transpose, it is used to convert a row vector into a column vector. The first 1 indicates that the measured value of the built-in temperature sensor is linearly proportional to the actual ambient temperature in a 1:1 ratio; the second 1 indicates that the measured value of the external temperature sensor is linearly proportional to the actual ambient temperature in a 1:1 ratio.

[0066] The observation noise from the built-in temperature sensor and the external sensor at time t is obtained, and the observation noise vector is calculated by transposing the two observation noises. ,in, The noise levels of the built-in temperature sensor and the external sensor at time t are respectively.

[0067] Input the current observation noise vector, observation matrix, and estimated real ambient temperature into the preset sensor measurement relationship observation equation. Perform calculations and output the observation vector at the current time. ;

[0068] Then, using a pre-defined Kalman gain fusion formula, the optimal ambient temperature estimate is calculated from the current real ambient temperature estimate and the observation vector, i.e., by... The optimal ambient temperature estimate is obtained through calculation. ;in, The predicted value at time t, that is, the estimated optimal ambient temperature at time t-1, is used as the initial reference value for the current time. The Kalman gain value ranges from [0,1] and is determined based on the observation noise. The lower the noise, the higher the weight. The calculation logic is: optimal estimate = predicted value at the previous time step + Kalman gain × observation residual. If the observation residual is small (the observed value is close to the predicted value), the correction magnitude is small and mainly depends on the predicted value. If the observation residual is large (the observed value deviates greatly from the predicted value), the correction magnitude is determined by the Kalman gain (the lower the observation noise, the larger the gain, and more dependent on the observation vector correction).

[0069] A4. Temperature Compensation Correction: Obtain the thermal conductivity coefficient of the medium between the battery and the external environment, and input the thermal conductivity coefficient and temperature gradient vector into the preset heat flux density calculation formula. Calculate and output heat flux density ;in, The thermal conductivity coefficient of the medium is determined by the installation environment (e.g., 2.6 × 10−4 W / (cm·°C) for air).

[0070] Obtain the thermal conductivity coefficient of the battery encapsulation material (e.g., 1.8 × 10−3 W / (cm·°C) for epoxy resin) and the heat dissipation area of ​​the battery. Input the encapsulation thermal conductivity coefficient, heat flux density, and heat dissipation area into a preset dynamic temperature deviation calculation formula. The deviation temperature caused by heat conduction was calculated. This refers to the additional temperature generated by the main chip's self-heating through heat conduction; among which, These are the thermal conductivity coefficient and heat dissipation area of ​​the package, respectively. The preset compensation weighting coefficient is obtained by fitting calibration data of the entire temperature range of the battery, and is used to balance the influence of thermal conduction deviation and original temperature deviation.

[0071] The compensated temperature is obtained by calculating the difference between the built-in temperature and the built-in sensor temperature. ,Right now Record the compensated temperature as the temperature compensation parameter; first calculate the deviation using the heat conduction model, then use... Fine-tuning was performed on the preliminary compensation results to ensure... It always revolves around the actual ambient temperature fluctuations, avoiding over- or under-compensation.

[0072] The specific analysis steps for the internal resistance measurement unit are as follows:

[0073] S1. Internal Resistance Analysis Data Identification: Multidimensional internal resistance data is identified to obtain the voltage response signal, repeatability error value, and internal resistance drift; temperature compensation parameters are identified to obtain the compensated temperature; the voltage response signal, repeatability error value, internal resistance drift, and compensated temperature are normalized and mapped to [0, 1] and labeled as characteristic variables, and denoted as follows: ;

[0074] S2. Multidimensional Data Analysis: Multiple data pairs are obtained by pairing each voltage response signal, repeatability error value, internal resistance drift, and compensated temperature normalized value. Each data pair is then input into a preset linear kernel function and an RBF kernel function. A preset mixed kernel function calculation formula is used to calculate the mixed kernel function for each data pair. The mixed kernel functions are then aggregated to obtain a mixed function similarity matrix, which is obtained through the formula... Perform calculations to output the hybrid kernel function. The similarity matrix of the mixing function is Hs; where, , The preset kernel weight coefficient takes values ​​in the range (0, 1); , Represented as vector transpose; , The preset RBF kernel width parameter can be determined by GWO (Grey Wolf Optimization Algorithm, an intelligent optimization algorithm that simulates the hunting behavior of grey wolves in nature. Its core logic is to find the optimal solution through group cooperative search, and its function is to automatically optimize the RBF kernel width parameter) and is used to control the nonlinear fitting capability.

[0075] Construct a single-hidden-layer neural network with four feature variables as input, L hidden nodes, and a single internal resistance state value as output; input the mixture function to the preset mixture output weight calculation formula. Calculate the mixed output weights ;in, The similarity matrix of the mixture function, calculated using the mixture kernel function, is a core component of the hidden layer output matrix H. ; Represented as vector transpose; is the sample label, i.e., the pre-calibrated measured internal resistance state value; C is the pre-calibrated regularization coefficient, used to balance model complexity and generalization ability; I is the identity matrix, a square matrix of dimension L×L, with diagonal elements of 1 and the rest of the elements of 0, and is related to C. −1 Combined with regularization terms, this ensures that the matrix Reversible; the mixed output weights are input into a preset regression mapping output formula. Calculate the output internal resistance state value ;in, This is the hidden layer output matrix, with dimensions N×L, where N is the preset number of samples.

[0076] S3, Internal Resistance Optimization Output: Obtain the four key parameters from S2: number of hidden layer nodes, kernel weight coefficient, RBF kernel width, and regularization coefficient; denote these four key parameters as the parameter vector to be optimized, and obtain the number of internal resistance state values ​​corresponding to the preset sample size, and denote them as predicted internal resistance state values; obtain the pre-designed and calibrated true internal resistance state values ​​corresponding to each predicted internal resistance state value; and apply the preset fitness function... Calculate the mean absolute error of the parameter vector to be optimized. The smaller the value, the higher the model accuracy; among which, Represented as the first The internal resistance state value and the true internal resistance state value of each sample; Represented as a vector of parameters to be optimized; The sample size is the number of samples taken.

[0077] Based on the core search mechanism of the GWO optimization algorithm (i.e., balancing "getting closer to the prey (local optimization)" and "moving away from the prey (global exploration)" through coefficient adjustment), the optimal parameter vector is screened to obtain the optimal parameter vector with the smallest mean absolute error; the current screening iteration number is obtained, and the globally optimal parameter vector corresponding to each iteration number (i.e., the parameter combination with the smallest mean absolute error at the current time) is obtained, and then it is marked as the prey position vector; the distance calculation formula between the gray wolf and the prey is preset by the GWO optimization algorithm. Calculate and output the distance vector between the gray wolf and its prey. ; This is the prey position vector corresponding to the d-th iteration. This represents the position vector of the gray wolf corresponding to the d-th iteration, i.e., the corresponding parameter combination, such as the position of the third gray wolf. ; The preset random weight coefficient vector takes values ​​obtained from experimental fitting. Its main purpose is to randomly amplify or reduce the influence of the prey's position to avoid the algorithm getting stuck in local optima.

[0078] The distance vector between the gray wolf and its prey, and the prey's position vector, are substituted into a pre-defined formula for updating the gray wolf's next-generation position to calculate and output the next-generation gray wolf's position vector. ,in, This is the gray wolf position vector corresponding to the (d+1)th iteration. The core coefficients used to control the search direction and step size are preset convergence factor vectors, whose values ​​are obtained from experimental fitting.

[0079] GWO retains the three optimal parameter vectors and denotes them as the optimal gray wolf positions. Using these optimal gray wolf positions as a guide, it calculates the candidate position vectors corresponding to the optimal gray wolf positions for the remaining gray wolves. ,in For candidate position vectors, The optimal gray wolf is defined as the parameter vector to be optimized. ; These are the convergence factor vectors corresponding to each optimal gray wolf. These are the random weight coefficient vectors corresponding to each optimal gray wolf; where... Let d be the current position vector of the remaining gray wolves a4 corresponding to the d-th iteration.

[0080] Update the formula based on the final positions of the remaining gray wolves. The final position vector of the gray wolf in the next iteration a4 is calculated; the optimal parameter vector of the gray wolf in the optimal position and the final position vectors of the other gray wolves are marked as the updated optimized parameter vector of the parameter vector to be optimized. The updated optimized parameter vector is input into the regression mapping output formula to obtain the corresponding internal resistance state value, and it is recorded as the updated internal resistance state value.

[0081] S4. Internal resistance measurement status assessment: The updated internal resistance status value is divided into multiple internal resistance status intervals based on the preset status value interval. Each internal resistance status interval is set with an internal resistance status level, namely, internal resistance status abnormal, internal resistance status warning, and internal resistance status normal. The higher the updated internal resistance status value, the more normal the corresponding internal resistance status level. The current updated internal resistance status value is matched with each internal resistance status interval to obtain the corresponding internal resistance status level.

[0082] The communication and power management module, based on the communication interface of the auxiliary chip and the power supply of the LDO chip, converts the battery voltage into a preset voltage to independently power the FM33M0XX chip corresponding to the main controller; and connects with different communication interface chips through the serial port of the FM33M0XX chip to realize multiple communication methods.

[0083] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A battery monitoring system based on the FM33M0XX chip using sensor technology, characterized in that, include: The core control module includes a main control unit and a control auxiliary unit. The main control unit uses the FM33M0XX chip as the main controller and is responsible for data reception, command issuance, and resource scheduling of each module. The control auxiliary unit uses the communication interface + LDO chip as the auxiliary chip and is responsible for communication interface and power supply management. The sensor detection module is used to collect battery monitoring parameters by measuring the battery's operating conditions and environment, and then send them to the detection data processing module. The detection data processing module includes a temperature compensation unit and an internal resistance measurement unit. It identifies the received battery monitoring parameters to obtain battery operating condition parameters and multi-dimensional internal resistance data. The temperature compensation unit analyzes the battery operating parameters to obtain temperature compensation parameters and sends these parameters to the internal resistance measurement unit. The internal resistance measurement unit analyzes the multi-dimensional internal resistance data and temperature compensation parameters to obtain internal resistance measurement parameters. The module then obtains the compensated temperature based on the temperature compensation parameters and performs state optimization based on the internal resistance state level. The communication and power management module uses a communication interface and an LDO chip to convert the battery voltage into a preset voltage to independently power the FM33M0XX chip; and connects with different communication interface chips through the serial port of the FM33M0XX chip.

2. The FM33M0XX chip battery monitoring system based on sensor technology according to claim 1, characterized in that, The battery monitoring parameters obtained by collecting data on battery operating conditions and the environment are as follows: Battery monitoring data is obtained by detecting the operating conditions and environment of the battery itself and the environment through a preset sensor group; the sensor group includes a temperature detection group and an internal resistance detection group. The temperature detection group collects data and obtains the sensing temperature data of the current sampling window based on the preset sampling window. It also obtains the spatial and temporal coordinate parameters of each sensor in the temperature detection group based on a unified spatial coordinate system. The spatial and temporal coordinate parameters of each sensor are correlated with the corresponding sensing temperature data to obtain the battery operating condition parameters. The internal resistance detection group monitors the current and response voltage of the battery under different operating conditions. After filtering, the signal is marked as a voltage response signal. The group also acquires the internal resistance measurement data of the battery under the same discharge state for a preset number of measurements. The group then performs mathematical calculations on the internal resistance values ​​corresponding to each internal resistance measurement data to obtain the corresponding standard deviation, which is recorded as the repeatability error value. Obtain the maximum deviation of the internal resistance measurement value within a preset historical period and record it as the internal resistance drift; combine the voltage response signal, repeatability error, and internal resistance drift to obtain multidimensional internal resistance data; Battery operating parameters and internal resistance multidimensional data are recorded as battery monitoring data.

3. The FM33M0XX chip battery monitoring system based on sensor technology according to claim 2, characterized in that, The specific analysis steps for the temperature compensation unit are as follows: A1: Identify the battery operating parameters to obtain the spatial and temporal coordinate parameters corresponding to the built-in temperature of the main control chip and the ambient temperature; A2: The temperature gradient vector is obtained by correlating the external temperature and the internal temperature through a preset thermal gradient calculation formula; A3: Outputs the optimal ambient temperature estimate using a preset real ambient temperature estimation method; A4: Obtain the thermal conductivity coefficient of the medium between the battery and the external environment, input the thermal conductivity coefficient and temperature gradient vector into the preset heat flux density calculation formula to calculate and output the heat flux density; The thermal conductivity coefficient of the battery packaging material and the heat dissipation area of ​​the battery are obtained. The thermal conductivity coefficient, heat flux density and heat dissipation area are input into the preset dynamic temperature deviation calculation formula to calculate the deviation temperature caused by heat conduction. The temperature built into the main control chip and the deviation temperature are compensated to obtain the compensated temperature. The compensated temperature is recorded as the temperature compensation parameter.

4. The FM33M0XX chip battery monitoring system based on sensor technology according to claim 3, characterized in that, The process of identifying battery operating parameters to obtain the spatial-temporal coordinate parameters corresponding to the built-in temperature of the main control chip and the ambient temperature is as follows: The battery operating parameters are identified to obtain the sensor temperature data and spatial-temporal coordinate parameters corresponding to the current sampling window. Based on the sensor temperature data, the built-in temperature and ambient temperature are obtained. The built-in temperatures of each sample window are added together to calculate the built-in average temperature. The built-in average temperature is then adjusted by adding or subtracting a preset temperature deviation threshold to obtain the built-in temperature distribution range. Built-in average temperatures exceeding the temperature distribution range are marked as temperature anomalies. Similarly, based on the ambient temperature within the sampling window, the corresponding ambient average temperature and ambient temperature distribution range are obtained, thus generating temperature anomalies, which are then removed. The built-in temperature sensors corresponding to the built-in temperatures are obtained and numbered, with the total number of sensors recorded as N. The external temperature sensors corresponding to the ambient temperatures are obtained and numbered, with the total number of sensors recorded as M. The spatial-temporal coordinate parameters corresponding to the built-in temperature and the ambient temperature are respectively labeled as follows: , where i∈N and j∈M.

5. The FM33M0XX chip battery monitoring system based on sensor technology according to claim 4, characterized in that, The step of outputting the optimal ambient temperature estimate using a preset real ambient temperature estimation method specifically involves: Mark the current time as t, and the previous time as t-1. Obtain the estimated value of the real ambient temperature at the previous time from the database, and obtain the preset process noise. Input the estimated value of the real ambient temperature at the previous time and the process noise into the preset state equation to calculate and output the estimated value of the real ambient temperature at the current time. By assuming a linear relationship between sensor measurements and actual temperature, and defining the corresponding observation matrix, denoted as... ,in, Represented as a vector transpose, it is used to convert a row vector into a column vector. The first 1 indicates that the measured value of the built-in temperature sensor is linearly proportional to the actual ambient temperature in a 1:1 ratio; the second 1 indicates that the measured value of the external temperature sensor is linearly proportional to the actual ambient temperature in a 1:1 ratio. The observation noise of the built-in temperature sensor and the external sensor at time t is obtained, and the observation noise vector is calculated by transposing the two observation noises. The current observation noise vector, observation matrix, and estimated real ambient temperature are input into the preset sensor measurement relationship observation equation for calculation and output of the current observation vector. Then, the optimal ambient temperature estimate is obtained by using the preset Kalman gain fusion formula to calculate the current ambient temperature estimate and the observation vector.

6. The FM33M0XX chip battery monitoring system based on sensor technology according to claim 5, characterized in that, The specific analysis steps for the internal resistance measurement unit are as follows: S1: Identify and analyze multidimensional internal resistance data to output multiple feature variables; S2: Analyze and output the internal resistance state value through a regression mapping output algorithm; S3: Optimize the output of the internal resistance state value and update the internal resistance state value; S4: Internal resistance measurement status assessment outputs the internal resistance status level.

7. The FM33M0XX chip battery monitoring system based on sensor technology according to claim 6, characterized in that, The process of identifying and analyzing multidimensional internal resistance data to output multiple feature variables specifically involves: identifying the voltage response signal, repeatability error value, and internal resistance drift from the multidimensional internal resistance data; identifying the temperature compensation parameter to obtain the compensated temperature; normalizing the voltage response signal, repeatability error value, internal resistance drift, and compensated temperature to [0, 1] and labeling them as feature variables, and denoting them as follows: .

8. The FM33M0XX chip battery monitoring system based on sensor technology according to claim 7, characterized in that, The analysis and output of the internal resistance state value through the regression mapping output algorithm is as follows: Multiple data pairs are obtained by pairing each voltage response signal, repeatability error value, internal resistance drift, and compensated temperature normalized value. Each data pair is then input into a preset linear kernel function and an RBF kernel function. A preset mixed kernel function calculation formula is used to calculate the mixed kernel function for each data pair. The mixed kernel functions are then aggregated to obtain a mixed function similarity matrix, which is obtained through the formula... Perform calculations to output the hybrid kernel function The similarity matrix of the mixing function is Hs; where, , These are the preset kernel weight coefficients; , Represented as vector transpose; , The preset RBF core width parameter; Construct a single-hidden-layer neural network with four feature variables as input, L hidden nodes, and a single internal resistance state value as output; input the mixture function to the preset mixture output weight calculation formula. Calculate the mixed output weights ;in, The similarity matrix of the mixture function, calculated by the mixture kernel function, is a core component of the hidden layer output matrix H; Represented as vector transpose; The sample label is denoted by ; C is the preset regularization coefficient; the mixed output weights are input into the preset regression mapping output formula to calculate the output internal resistance state value.

9. The FM33M0XX chip battery monitoring system based on sensor technology according to claim 1, characterized in that, The optimization and output update of the internal resistance state value specifically involves: Obtain the four key parameters in S2: the number of hidden layer nodes, kernel weight coefficient, RBF kernel width, and regularization coefficient. Record these four key parameters as the parameter vector to be optimized. Obtain the number of internal resistance state values ​​corresponding to a preset number of samples, and record them as predicted internal resistance state values. Obtain the pre-designed and calibrated true internal resistance state values ​​corresponding to each predicted internal resistance state value. Calculate the mean absolute error of the parameter vector to be optimized using a preset fitness function. , Represented as a vector of parameters to be optimized; The core search mechanism based on the GWO optimization algorithm filters the optimal parameter vector to obtain the optimal parameter vector with the smallest mean absolute error; obtains the current filtering iteration number and the global optimal parameter vector corresponding to each iteration number, and then marks it as the prey position vector; calculates and outputs the distance vector between the gray wolf and the prey using the pre-set distance calculation formula of the GWO optimization algorithm. Substitute the distance vector between the gray wolf and its prey and the prey's position vector into the preset gray wolf next-generation position update formula to calculate and output the next-generation gray wolf position vector; GWO retains the three optimal parameter vectors and records them as the optimal gray wolf positions. The optimal gray wolf positions are used to guide the calculation of the remaining gray wolves to obtain the candidate position vectors corresponding to the optimal gray wolf positions. The final position vector of the remaining gray wolves for the next iteration is calculated using the final position update formula of the remaining gray wolves. The optimal parameter vector of the gray wolf with the best position and the final position vector of the remaining gray wolves are marked as the updated optimized parameter vector of the parameter vector to be optimized. The updated optimized parameter vector is input into the regression mapping output formula to obtain the corresponding internal resistance state value, and it is recorded as the updated internal resistance state value.

10. The FM33M0XX chip battery monitoring system based on sensor technology according to claim 9, characterized in that, The internal resistance measurement status assessment outputs an internal resistance status level, which specifically involves: dividing the updated internal resistance status value into multiple internal resistance status intervals based on a preset status value interval; each internal resistance status interval is assigned an internal resistance status level, namely, internal resistance status abnormal, internal resistance status warning, and internal resistance status normal; the higher the updated internal resistance status value, the more normal the corresponding internal resistance status level; and matching the current updated internal resistance status value with each internal resistance status interval to obtain the corresponding internal resistance status level.

Images