CN-122017653-A - Electric bicycle battery state of health evaluation system based on data fusion

CN122017653ACN 122017653 ACN122017653 ACN 122017653ACN-122017653-A

Abstract

The invention relates to the technical field of battery health state evaluation, in particular to an electric bicycle battery health state evaluation system based on data fusion. The method comprises the steps of acquiring dynamic operation data and static characteristic data of a battery by a data acquisition module, aligning the two types of data according to time stamps by a data alignment module to generate a synchronous time-space associated data sequence, extracting capacity attenuation rate, internal resistance increase trend and voltage platform change characteristics from the sequence by a characteristic extraction module, carrying out multi-source evidence fusion analysis on the characteristics by a fusion evaluation module, and outputting comprehensive scores of the state of health of the battery and predicted values of the residual service life. According to the scheme, through accurate data fusion and multidimensional feature collaborative analysis, the accuracy of state evaluation and the reliability of life prediction are effectively improved.

Inventors

LI JIANHUA
GUO DONG
ZHANG GANG
Hao kaixuan
FANG YAN
SUN JUNLIN

Assignees

北京东方捷码科技开发有限公司

Dates

Publication Date: 20260512
Application Date: 20260413

Claims (9)

1. Electric bicycle battery state of health evaluation system based on data fusion, characterized in that, the system includes: the data acquisition module acquires dynamic operation data of the battery of the electric bicycle through the vehicle-mounted terminal and acquires static characteristic data of the battery through the battery management system, The data alignment module is used for creating a data fusion processing queue, performing alignment processing on the dynamic operation data and the battery static characteristic data according to time stamps to generate a synchronous time-space associated data sequence, wherein the synchronous time-space associated data sequence comprises a voltage value, a circulation number, a temperature value, a current value, a nominal capacity, a standard internal resistance and a material type corresponding to each group of time stamps; the feature extraction module is used for calling a preset health state evaluation model, inputting the synchronous space-time associated data sequence into a feature extraction layer of the health state evaluation model, and extracting a key parameter set representing the degradation feature of the battery by the feature extraction layer, wherein the key parameter set comprises a capacity attenuation rate, an actual internal resistance increase trend and a voltage platform curve change feature; The fusion evaluation module performs multi-source evidence fusion analysis on the key parameter set through a fusion evaluation layer of the health state evaluation model to generate a comprehensive score of the health state of the battery and a predicted value of the residual service life, and comprises the following steps: Assigning an initial confidence weight to each parameter in the set of key parameters, the initial confidence weights being adaptively adjusted based on the type identification of the battery manufacturing material; Adopting a fuzzy logic reasoning rule to process the association relation between the capacity fading rate and the voltage platform curve change characteristic, and obtaining a fuzzy evaluation value about the degradation degree of the battery performance; based on a particle filtering algorithm, taking the actual internal resistance increasing trend as a state observation value, carrying out iterative updating on equivalent circuit model parameters of the battery, and predicting internal resistance evolution paths after a plurality of charge and discharge cycles in the future; Inputting the fuzzy evaluation value and a performance degradation curve obtained based on internal resistance prediction into a preset score mapping function, and calculating to obtain the comprehensive score of the battery health state; and according to the cycle times when the performance degradation curve reaches a preset failure threshold value, the residual service life predicted value is calculated by combining the current charge and discharge cycle times.
2. The system for evaluating the state of health of a battery of an electric bicycle based on data fusion according to claim 1, wherein creating a data fusion processing queue, aligning the dynamic operation data with the battery static characteristic data according to a time stamp, and generating a synchronous time-space associated data sequence, comprises: The dynamic operation data comprises continuous real-time discharge voltage, charge-discharge cycle times, working environment temperature and instantaneous output current; Assigning a high-precision time stamp generated by the vehicle-mounted system to each frame of data in the dynamic operation data; Extracting fixed attribute information from the battery static characteristic data, and generating a constant time stamp sequence covering the whole evaluation period for the attribute information; establishing a time synchronization window, and carrying out matching association on the dynamic operation data frames with the same or the nearest adjacent time stamps and the static characteristic data attribute information of the battery; Carrying out moving average filtering processing on the working environment temperature and the instantaneous output current to eliminate high-frequency noise interference; And arranging the data subjected to matching association and filtering processing according to a time sequence to form the synchronous time-space association data sequence, wherein each record contains all data fields under the same time reference.
3. The system for estimating the state of health of a battery of an electric bicycle based on data fusion according to claim 2, wherein said inputting the synchronized spatiotemporal correlated data sequence into the feature extraction layer of the state of health estimation model, extracting by the feature extraction layer a set of key parameters characterizing the degradation features of the battery, comprises: The battery static characteristic data comprises a battery factory nominal capacity, a battery standard internal resistance value and a type identifier of a battery manufacturing material; Intercepting voltage and current data of a complete discharge period from the synchronous time-space associated data sequence, calculating the actual discharge capacity of the complete discharge period, comparing the actual discharge capacity with the nominal capacity of the battery corresponding to the complete discharge period, and calculating the capacity attenuation rate; Selecting a specific state of charge point in the complete discharge period, calculating the actual internal resistance of the battery at the state of charge point based on the discharge voltage, current and temperature corresponding to the state of charge point by using a direct current internal resistance test method, and comparing the actual internal resistance with the standard internal resistance value of the battery to generate the actual internal resistance increasing trend; drawing a change curve of voltage relative to discharge capacity in a complete discharge period, and identifying the change characteristic of the voltage plateau curve, wherein the change characteristic of the voltage plateau curve comprises a reduction value of plateau voltage, a shortening length of the plateau curve and an increasing amount of the plateau slope; And carrying out normalized packaging on the calculated capacity attenuation rate, the actual internal resistance increasing trend and the voltage platform curve change characteristic, and outputting the obtained normalized package to be the key parameter set.
4. The system for evaluating the state of health of a battery of an electric bicycle based on data fusion according to claim 1, wherein the processing the association between the capacity fade rate and the voltage plateau curve change characteristic by using fuzzy logic inference rules to obtain a fuzzy evaluation value about the degree of degradation of the battery performance comprises: defining a fuzzy linguistic variable for the capacity fade rate and a fuzzy linguistic variable for the voltage plateau curve variation characteristic, respectively, each fuzzy linguistic variable comprising a plurality of sub-states; Establishing a fuzzy rule base, wherein each fuzzy rule takes a fuzzy sub-state of the capacity attenuation rate and a fuzzy sub-state of the curve change characteristic of the voltage platform as preconditions, and takes a fuzzy state of the battery performance degradation degree as a conclusion; converting the calculated specific capacity attenuation rate and the quantized voltage platform curve change characteristics into membership distribution of the specific capacity attenuation rate and the quantized voltage platform curve change characteristics on respective fuzzy language variables by using a triangular membership function; based on the membership distribution and the fuzzy rule base, performing fuzzy reasoning operation to obtain activation intensities of a plurality of states of the battery performance degradation degree; and performing defuzzification processing on the activation intensity of all states of the degradation degree of the battery performance to obtain a numerical value output as the fuzzy evaluation value.
5. The system for evaluating the state of health of a battery of an electric bicycle based on data fusion according to claim 4, wherein the particle filter algorithm-based method for predicting the internal resistance evolution path after a plurality of charge and discharge cycles in the future by iteratively updating the equivalent circuit model parameters of the battery with the actual internal resistance increase trend as a state observation value comprises: Initializing a set of particles, each particle representing a state vector of equivalent circuit model parameters of the battery; taking the historical data of the actual internal resistance increasing trend as an observation sequence, and calculating likelihood probability of each particle for each new observation value, wherein the likelihood probability represents the possibility that the equivalent circuit model parameters of the particle representing the battery can generate the observation value; Resampling the particle set according to the likelihood probability of each particle, increasing the number of high likelihood probability particles and reducing the number of low likelihood probability particles; Predicting the state of each resampled particle by using a battery aging mechanism model, and simulating the change of the equivalent circuit model parameters in the next charge-discharge cycle; Repeating the steps of observation updating, resampling and state prediction, and finally preserving the statistical characteristics of the particle set, namely, representing the optimal estimation of the battery equivalent circuit model parameters, and extrapolating based on the optimal estimation to generate the internal resistance evolution path.
6. The data fusion-based electric bicycle battery state of health assessment system of claim 1, further comprising: The model self-adaptive calibration module is used for carrying out self-adaptive calibration on the health state evaluation model based on the comprehensive scores of the health states of the batteries, and specifically comprises the following steps: After the electric bicycle completes a complete full charge and discharge cycle, recording the actual measured total discharge capacity and the average working temperature of the cycle; comparing the measured total discharge capacity with the factory nominal capacity of the battery, and calculating the actual capacity retention rate of the cycle; comparing the actual capacity retention rate with a predicted capacity retention rate output by the health state evaluation model in the last cycle to obtain a prediction error; According to the magnitude and the direction of the prediction error, reversely adjusting the parameter weight of the fusion evaluation layer in the health state evaluation model, in particular adjusting the confidence weight related to processing capacity attenuation and the parameters of a fuzzy rule base; and processing the synchronous time-space associated data sequence input in the next round by using the adjusted parameter weight, so as to realize the online self-adaptive calibration of the model.
7. The data fusion-based battery state of health evaluation system of an electric bicycle of claim 6, wherein said calculating an actual capacity retention for the cycle comparing the measured total discharge capacity to the battery factory nominal capacity comprises: controlling the battery of the electric bicycle to perform constant power discharge from a full-charge state to a preset termination voltage, and recording the discharge duration of the whole discharge process; Calculating the actual measurement total discharge capacity of the current cycle according to the product of the constant discharge power and the discharge duration; Acquiring the factory nominal capacity of the battery corresponding to the battery; Dividing the measured total discharge capacity by the factory nominal capacity of the battery, and multiplying the measured total discharge capacity by a percentage base to obtain the actual capacity retention rate; and storing the calculation result, the average working temperature of the current cycle and the charge and discharge cycle times together as a true value data point for model calibration.
8. The data fusion-based electric bicycle battery state of health evaluation system of claim 2, further comprising: the maintenance suggestion module is used for receiving the comprehensive scores of the battery health states and the predicted values of the residual service lives, which are output by the health state evaluation model; Setting a plurality of health state scoring thresholds, comparing the comprehensive health state score of the battery with the health state scoring thresholds, and determining the health state grade of the battery; According to the health state level of the battery, matching a corresponding basic maintenance suggestion template from a preset maintenance strategy knowledge base; combining the historical statistical characteristics of the working environment temperature and the frequency of the charge-discharge cycle times, and performing fine adjustment on parameters in the basic maintenance suggestion template; And adding the residual service life predicted value with the current date, calculating the estimated battery expiration date, combining the date information with the maintenance advice after parameter fine adjustment, and generating a final personalized maintenance advice report.
9. The system for evaluating the health status of a battery of an electric bicycle based on data fusion according to claim 8, wherein the matching the corresponding basic maintenance advice template from a preset maintenance policy knowledge base according to the health status level of the battery comprises: the maintenance strategy knowledge base comprises a plurality of preset health state grade intervals, and each grade interval is associated with one or more basic maintenance suggestion templates; The basic maintenance suggestion template comprises a text description and adjustable parameters, wherein the text description comprises a charging frequency suggestion, a discharging depth suggestion and a working temperature range suggestion, and the adjustable parameters comprise a charging current multiplying power and a suggested idle time; inquiring the maintenance strategy knowledge base to find a class interval of the health state class of the battery; extracting all the basic maintenance suggestion templates associated with the level interval; and screening templates matched with materials from all the extracted basic maintenance recommended templates according to the type identification of the battery manufacturing materials of the battery, and taking the templates as candidate basic maintenance recommended templates.

Description

Electric bicycle battery state of health evaluation system based on data fusion Technical Field The invention relates to the technical field of battery health state evaluation, in particular to an electric bicycle battery health state evaluation system based on data fusion. Background The existing electric bicycle battery health state evaluation technology mostly relies on limited static data of a battery management system or isolated dynamic data flow of a vehicle-mounted terminal for analysis. Conventional schemes typically process these two types of data separately, or only do rough time period matching, resulting in a lack of accurate timing correlation between dynamic operating parameters and the inherent static properties of the battery. The data-level fracture ensures that the input information received by the model is misplaced in time and state, so that the real-time degradation behavior of the battery under specific inherent characteristics is difficult to accurately reflect, and the accuracy of feature extraction and the reliability of model evaluation are restricted. In terms of the evaluation method, the prior art generally focuses on the health index of a single dimension, such as judging mainly by depending on a capacity attenuation curve or an internal resistance change trend. The evaluation mode based on the single evidence source cannot comprehensively reflect the complex state of the interleaving influence of multiple physical and chemical processes in the battery aging process. Because the aging rates and the sensitivities of different characteristic parameters are different, the parameter measurement noise or accidental fluctuation interference is easily caused by only depending on a single parameter, so that the misjudgment of the health state and the prediction deviation of the residual life are larger, and the robustness and the comprehensiveness of the evaluation result are insufficient. The technical scheme is needed, the problem of accurate space-time synchronization of multi-source heterogeneous data can be fundamentally solved, and a comprehensive evaluation model capable of fusing multi-dimensional degradation evidence is constructed, so that the overall accuracy of battery health state evaluation and the robustness of prediction are improved. Disclosure of Invention The invention aims to solve the defects in the prior art, and provides an electric bicycle battery health state evaluation system based on data fusion. In order to achieve the purpose, the invention adopts the following technical scheme that the electric bicycle battery health state evaluation system based on data fusion comprises: the data acquisition module acquires dynamic operation data of the battery of the electric bicycle through the vehicle-mounted terminal and acquires static characteristic data of the battery through the battery management system, The data alignment module is used for creating a data fusion processing queue, performing alignment processing on the dynamic operation data and the battery static characteristic data according to time stamps to generate a synchronous time-space associated data sequence, wherein the synchronous time-space associated data sequence comprises a voltage value, a circulation number, a temperature value, a current value, a nominal capacity, a standard internal resistance and a material type corresponding to each group of time stamps; the feature extraction module is used for calling a preset health state evaluation model, inputting the synchronous space-time associated data sequence into a feature extraction layer of the health state evaluation model, and extracting a key parameter set representing the degradation feature of the battery by the feature extraction layer, wherein the key parameter set comprises a capacity attenuation rate, an actual internal resistance increase trend and a voltage platform curve change feature; The fusion evaluation module performs multi-source evidence fusion analysis on the key parameter set through a fusion evaluation layer of the health state evaluation model to generate a comprehensive score of the health state of the battery and a predicted value of the residual service life, and comprises the following steps: Assigning an initial confidence weight to each parameter in the set of key parameters, the initial confidence weights being adaptively adjusted based on the type identification of the battery manufacturing material; Adopting a fuzzy logic reasoning rule to process the association relation between the capacity fading rate and the voltage platform curve change characteristic, and obtaining a fuzzy evaluation value about the degradation degree of the battery performance; based on a particle filtering algorithm, taking the actual internal resistance increasing trend as a state observation value, carrying out iterative updating on equivalent circuit model parameters of the battery, and predicting internal resistance evolution path