A Smart Battery Swapping Recommendation Method and System Based on Historical Battery Data

Abstract

This invention relates to the field of intelligent battery swapping recommendation technology, and discloses an intelligent battery swapping recommendation method and system based on historical battery data. The method includes: a vehicle central control system identifying the vehicle from a user-triggered battery swapping request signal to obtain parameter information of the current vehicle; querying historical battery matching data corresponding to the parameter information from a cloud database, and calculating the compatibility score between each battery model and the current vehicle; a battery swapping cabinet control system performing multi-objective collaborative recommendation decision-making with the compatibility score and the inventory batteries to obtain the optimal recommended battery identifier; and the battery swapping cabinet opening the corresponding compartment door according to the optimal recommended battery identifier to execute the battery swapping process and generate a battery swapping operation record. This invention improves the accuracy of battery compatibility assessment, can discover performance correlations between different battery models, and thus improves the automation and intelligence level of the entire battery swapping process.

Description

Technical Field

[0001] This invention relates to the field of intelligent battery swapping recommendation technology, and in particular to an intelligent battery swapping recommendation method and system based on historical battery data. Background Technology

[0002] Battery swapping services for two-wheeled vehicles can solve the problems of battery range and charging time. However, current battery swapping systems use simple first-in-first-out or random allocation strategies, lacking in-depth analysis of the compatibility between different battery models and vehicle models. This often leads to poor battery-vehicle matching during the swapping process. Traditional battery swapping systems typically rely on manual user operation and simple barcode scanning, failing to automatically obtain specific vehicle requirements or provide intelligent recommendations based on historical usage data. These technological limitations severely impact battery swapping efficiency and user experience. Summary of the Invention

[0003] This invention provides an intelligent battery swapping recommendation method and system based on historical battery data. This invention improves the accuracy of battery compatibility assessment, can discover the performance correlation between different battery models, and thus improves the automation and intelligence level of the entire battery swapping process.

[0004] In a first aspect, the present invention provides an intelligent battery swapping recommendation method based on historical battery data, the intelligent battery swapping recommendation method based on historical battery data comprising:

[0005] The vehicle's central control system identifies the vehicle by detecting the battery swapping request signal triggered by the user and obtains the current vehicle's parameter information.

[0006] The system retrieves historical battery matching data corresponding to the parameter information from the cloud database and calculates the compatibility score between each battery model and the current vehicle.

[0007] The battery swapping cabinet control system performs multi-objective collaborative recommendation decision-making with the compatibility score and the inventory batteries to obtain the optimal recommended battery identifier.

[0008] The battery swapping cabinet opens the corresponding compartment door based on the optimal recommended battery identifier to execute the battery swapping process and generate a battery swapping operation record.

[0009] In conjunction with the first aspect, in a first implementation of the first aspect of the present invention, the vehicle central control system performs vehicle identification on the user-triggered battery swapping request signal to obtain the current vehicle's parameter information, including:

[0010] Receive user key presses and generate a battery swap request signal;

[0011] Near-field communication scanning is performed on the vehicles in the battery swapping request signal to obtain the vehicle's unique identification code;

[0012] Using the CAN communication protocol, the vehicle model and battery interface specifications are read from the vehicle controller based on the vehicle's unique identification code to obtain the vehicle's basic configuration data;

[0013] The vehicle's central control system integrates the vehicle's basic configuration data with the current battery capacity requirements and rated power requirements to obtain the vehicle's parameter information.

[0014] In conjunction with the first aspect, in a second implementation of the first aspect of the present invention, the step of querying the historical battery matching data corresponding to the parameter information from the cloud database and calculating the compatibility score between each battery model and the current vehicle includes:

[0015] In the cloud database, the vehicle model in the parameter information is indexed and matched to obtain the battery usage record index table corresponding to the vehicle model;

[0016] Based on the battery usage record index table, the historical usage counts of each battery model on the vehicle model are traversed to obtain the battery usage record matrix;

[0017] The battery usage record matrix is ​​correlated with the performance degradation curve of each battery model on the vehicle model to obtain battery model performance correlation data.

[0018] Historical successful matching rate statistics and failure rate analysis were performed on the battery model performance correlation data to obtain historical battery matching data;

[0019] Based on the historical battery matching data, compatibility calculations are performed to obtain a compatibility score between each battery model and the current vehicle.

[0020] In conjunction with the first aspect, in a third implementation of the first aspect of the present invention, the step of performing compatibility calculations based on the battery historical matching data to obtain a compatibility score between each battery model and the current vehicle includes:

[0021] The usage frequency and success rate data in the battery historical matching data are encoded into feature vectors to obtain the historical performance feature vectors of each battery model.

[0022] Cross-model similarity calculation is performed on the historical performance feature vectors to obtain the performance similarity coefficients between battery models;

[0023] The risk coefficient for switching each battery model to the current vehicle is calculated based on the failure rate analysis in the battery historical matching data.

[0024] The performance similarity coefficient and the risk coefficient are weighted and fused to obtain the compatibility score between each battery model and the current vehicle.

[0025] In conjunction with the first aspect, in the fourth implementation of the first aspect of the present invention, the step of performing cross-model similarity calculation on the historical performance feature vector to obtain the performance similarity coefficient between battery models includes:

[0026] The historical performance feature vectors are hierarchically sorted according to the frequency of use of battery models from high to low to obtain a priority sequence of battery models.

[0027] The historical data of each battery model in the battery model priority sequence are weighted by a time decay factor to obtain a feature vector group with time-series weights.

[0028] Calculate the dynamic threshold parameter based on the similarity distribution data in the feature vector group with temporal weights;

[0029] Based on the dynamic threshold parameter, multiple rounds of similarity iteration calculations are performed between battery models until convergence, to obtain the performance similarity coefficient between battery models.

[0030] In conjunction with the first aspect, in a fifth implementation of the first aspect of the present invention, the battery swapping cabinet control system performs a multi-objective collaborative recommendation decision based on the compatibility score and the inventory batteries to obtain the optimal recommended battery identifier, including:

[0031] The battery swapping cabinet control system reads the compatibility score and obtains the health status and storage location data of the inventory batteries;

[0032] Based on the health status and the warehouse distribution data, a multi-objective optimization function set is constructed, which includes the objective function of battery remaining value utilization rate, the objective function of user waiting cost, and the objective function of inventory distribution equilibrium.

[0033] The crossover and mutation probabilities in the genetic algorithm are dynamically adjusted based on the convergence state of the multi-objective optimization function set.

[0034] Population evolution calculations are performed based on the crossover probability and the mutation probability to obtain the optimal recommended battery identifier.

[0035] In conjunction with the first aspect, in a sixth implementation of the first aspect of the present invention, the step of dynamically adjusting the crossover probability and mutation probability in the genetic algorithm according to the convergence state of the multi-objective optimization function set includes:

[0036] Fitness gradient monitoring is performed on the battery remaining value utilization objective function, user waiting cost objective function, and inventory distribution equilibrium objective function in the multi-objective optimization function group to obtain the convergence speed difference value;

[0037] The rate of change of genetic factors is obtained by comparing the convergence speed difference with the historical convergence pattern.

[0038] The correlation calculation between the rate of change of the genetic factors and the variance of the fitness distribution of individuals in the population is performed to obtain the crossover probability that matches the convergence state.

[0039] Complementarity calculations are performed based on the crossover probability and the degree of population evolution stagnation to obtain the mutation probability that works in conjunction with the crossover probability.

[0040] In conjunction with the first aspect, in the seventh implementation of the first aspect of the present invention, the battery swapping cabinet opens the corresponding compartment door according to the optimal recommended battery identifier to execute the battery swapping process and generates a battery swapping operation record, including:

[0041] The battery swapping cabinet parses the compartment number information in the optimal recommended battery identifier and drives the corresponding battery compartment door opening mechanism;

[0042] After the battery compartment is opened, the sensor array detects the user's sequence of removing the old battery and inserting the recommended battery.

[0043] The operation sequence is timestamped and the battery identification verification results are collected to obtain battery swapping process data;

[0044] The battery swapping process data is associated and assembled with the optimal recommended battery identifier to generate a battery swapping operation record.

[0045] In conjunction with the first aspect, in the eighth implementation of the first aspect of the present invention, the intelligent battery swapping recommendation method based on battery history data further includes:

[0046] After receiving the battery swapping operation record, the cloud database extracts the vehicle model and the selected battery model to obtain the vehicle battery matching result for this battery swap.

[0047] The consistency between the vehicle battery pairing results and the previously recommended battery models is verified to obtain the recommendation matching accuracy evaluation value;

[0048] The battery historical matching data is updated based on the recommended matching accuracy evaluation value to refresh the historical successful matching rate of the corresponding vehicle model and battery model combination, and the battery historical matching data is updated accordingly.

[0049] Secondly, the present invention provides an intelligent battery swapping recommendation system based on historical battery data, the intelligent battery swapping recommendation system based on historical battery data comprising:

[0050] The vehicle central control system is used to identify the vehicle by the battery swapping request signal triggered by the user and obtain the parameter information of the current vehicle.

[0051] The calculation module is used to query the historical battery matching data corresponding to the parameter information from the cloud database, and calculate the compatibility score between each battery model and the current vehicle.

[0052] The battery swapping cabinet control system is used to perform multi-objective collaborative recommendation decision-making with the compatibility score and the inventory batteries to obtain the optimal recommended battery identifier;

[0053] The battery swapping cabinet is used to open the corresponding compartment door according to the optimal recommended battery identifier to execute the battery swapping process and generate a battery swapping operation record.

[0054] The technical solution provided by this invention accurately acquires key parameters such as vehicle model and interface specifications through the automatic identification mechanism of the vehicle's central control system, avoiding errors and incomplete information problems associated with traditional manual operations. Deep correlation analysis based on the battery usage record matrix and performance degradation curve significantly improves the scientific rigor of battery compatibility assessment, overcoming the limitations of relying solely on immediate conditions. Through cross-model similarity calculation of historical performance feature vectors and a dynamic threshold iteration algorithm, performance correlations between different battery models are discovered, overcoming the technical challenge of collaborative optimization of multiple battery models. A variable genetic factor multi-objective optimization algorithm is employed to simultaneously balance multiple objectives such as battery value utilization, user waiting costs, and inventory balance, achieving global optimization of battery swapping decisions. It possesses convergence state monitoring and dynamic genetic factor adjustment capabilities, avoiding the local optima problem of fixed-parameter algorithms. Through real-time feedback of battery swapping operation records and continuous updates of historical data, a complete closed-loop learning mechanism is established, achieving continuous improvement in recommendation accuracy and integrated linkage of vehicle, battery, and cabinet data. Attached Figure Description

[0055] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 This is a schematic diagram illustrating the steps of the intelligent battery swapping recommendation method based on historical battery data in an embodiment of the present invention;

[0057] Figure 2 This is a schematic diagram of the intelligent battery swapping recommendation system based on historical battery data in an embodiment of the present invention. Detailed Implementation

[0058] This invention provides an intelligent battery swapping recommendation method and system based on historical battery data. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0059] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 One embodiment of the intelligent battery swapping recommendation method based on historical battery data in this invention includes:

[0060] Step S1: The vehicle central control system performs vehicle identification on the battery swap request signal triggered by the user and obtains the parameter information of the current vehicle.

[0061] It is understood that the executing entity of this invention can be an intelligent battery swapping recommendation system based on historical battery data, or it can be a terminal or a server; no specific limitation is made here. This embodiment of the invention will be described using a server as an example.

[0062] Specifically, when a user presses a button to initiate the battery swapping process, the vehicle's central control system captures the user's button press in real time and parses the operation into a battery swapping request signal. The built-in near-field communication module then performs a short-range wireless scan of the vehicle to identify its unique identification code. This unique identification code corresponds one-to-one with the vehicle registration information stored in the cloud and the battery swapping cabinet system. The vehicle's central control system establishes a data interaction channel with the vehicle controller via the CAN communication protocol and directly reads key basic configuration data such as the vehicle model and battery interface specifications based on the vehicle's unique identification code. Through the vehicle controller's power management unit and battery monitoring module, it synchronously collects data on the battery capacity and rated power requirements of the vehicle under its current operating state. The vehicle model, battery interface specifications, capacity requirements, and rated power requirements are integrated and structured to form current vehicle parameter information directly used for cloud matching and battery swapping cabinet scheduling.

[0063] Step S2: Query the historical battery matching data corresponding to the parameter information from the cloud database, and calculate the compatibility score between each battery model and the current vehicle.

[0064] Specifically, the current vehicle parameter information transmitted from the vehicle's central control system is sent to the cloud database processing module. This module then establishes an index matching relationship based on the vehicle model field in the parameter information, quickly locating and retrieving the battery usage record index table corresponding to the vehicle model. Using this index table as the basis for retrieval, the module iterates through all historical usage counts of each battery model on the vehicle model, and constructs a battery usage record matrix based on the two-dimensional correspondence between battery model and vehicle model. This quantifies the application frequency of different battery models on the target vehicle model. The module then retrieves the performance degradation curves of each battery model on the vehicle model stored in the database, and correlates the usage counts in the battery usage record matrix with the degradation curves to generate battery model performance correlation data reflecting long-term performance. Historical successful matching rate statistics and failure rate analysis are performed on the performance correlation data, simultaneously incorporating the successful usage ratio and anomaly probability of each model on the vehicle model for evaluation, resulting in battery historical matching data verified for safety and reliability. Based on the battery historical matching data, a compatibility score between each battery model and the current vehicle is calculated through a weighted calculation considering multiple dimensions such as usage frequency, performance retention rate, successful matching rate, and failure risk.

[0065] Step S3: The battery swapping cabinet control system performs multi-objective collaborative recommendation decision-making based on the compatibility score and the inventory batteries to obtain the optimal recommended battery identifier;

[0066] Specifically, the battery swapping cabinet control system receives compatibility scores between each battery model and the current vehicle, and simultaneously calls the cabinet management unit to collect real-time health status data of the currently stocked batteries, including key operating parameters such as SOC, SOH, voltage, current, and temperature. It also obtains the battery's location distribution information within the cabinet, thus determining the physical location and availability of the available batteries. Combining the compatibility scores with the real-time status data of the stocked batteries, a multi-objective optimization function set is constructed. This function set includes a battery residual value utilization objective function aimed at extending usable lifespan and improving resource utilization; a user waiting cost objective function aimed at reducing the overall user operation time; and an inventory distribution equilibrium objective function aimed at balancing the inventory distribution of different locations and battery models. To improve the global search capability and convergence efficiency of the optimization solution, the crossover and mutation probabilities are dynamically adjusted based on the real-time convergence status of the multi-objective optimization function set during the genetic algorithm's operation. This maintains high diversity in the early stages to avoid local optima, while increasing the convergence speed in the later stages to quickly lock in high-quality solutions. Population evolution is performed using the adjusted crossover and mutation probabilities. In continuous iterations, excellent individuals that simultaneously satisfy multi-objective optimization constraints are continuously screened and retained. After the convergence condition is met, the individual with the highest comprehensive fitness is selected as the optimal solution. Battery identification information corresponding to the optimal individual is extracted to form the optimal recommended battery identification.

[0067] Step S4: The battery swapping cabinet opens the corresponding compartment door according to the optimal recommended battery identifier to execute the battery swapping process and generate a battery swapping operation record.

[0068] Specifically, the battery swapping cabinet parses the optimal recommended battery identifier, extracts the compartment number information contained therein, and uses this compartment number as a control signal input to the compartment door drive module. This drives the corresponding motor, latch, or electromagnetic mechanism to perform the opening action, ensuring the designated battery compartment opens smoothly when both mechanical and electrical unlocking conditions are met. After the compartment door opens, a multi-type sensor array (including infrared sensors, pressure sensors, Hall effect sensors, etc.) deployed inside the battery swapping cabinet simultaneously activates, monitoring the entire process of the user removing the old battery and inserting the recommended battery in real time, forming an operation sequence in the order of events. While collecting the operation sequence, a millisecond-level timestamp is added to each key action node, recording the time trajectory of the entire battery swapping process. RFID or QR code recognition modules are used to verify the identifier of the recommended battery placed by the user, ensuring it matches the optimal recommended battery identifier correctly. The battery swapping process data, including operation actions, time information, and identifier verification results, is associated and assembled with the original optimal recommended battery identifier, generating a battery swapping operation record through structured data storage. This record includes the compartment number, recommended battery model, user operation time and sequence, as well as verification status and execution results.

[0069] In one specific embodiment, the process of performing step S1 may specifically include the following steps:

[0070] Receive user key presses and generate a battery swap request signal;

[0071] Near-field communication scanning is performed on vehicles in the battery swapping request signal to obtain the vehicle's unique identification code;

[0072] Using the CAN communication protocol, the vehicle model and battery interface specifications are read from the vehicle controller based on the vehicle's unique identification code to obtain the vehicle's basic configuration data;

[0073] The vehicle's central control system integrates the vehicle's basic configuration data with the current battery capacity and rated power requirements to obtain the vehicle's current parameter information.

[0074] Specifically, user button operations are located in the human-machine interface module of the vehicle's central control system, completed through physical buttons or a touch interface. When a button operation is triggered, the vehicle's central control system parses the input signal into a standardized battery swapping request signal and generates a request event structure on the internal bus that matches the battery swapping business process. After the battery swapping request signal is generated, the vehicle's central control system initiates a near-field communication scanning process. Based on the onboard near-field communication module (such as NFC, Bluetooth Low Energy, or dedicated RF short-range communication equipment), it reads the vehicle identification unit by transmitting communication requests at a specific frequency and protocol format, obtaining a unique identification code bound to the vehicle hardware. This unique identification code is a unique digital identity marker for the vehicle throughout the entire battery swapping system, and it has a one-to-one mapping relationship with vehicle registration information, battery history records, and battery swapping cabinet scheduling data. The vehicle's central control system establishes a high-speed data interaction channel with the vehicle controller through the onboard CAN bus communication protocol. The CAN communication protocol is a real-time bus protocol suitable for the inboard environment, with strong anti-interference capabilities and low data transmission latency, ensuring the stability and accuracy of parameter readings. The central control system uses the vehicle's unique identification code as a query condition to send a parameter read command to the vehicle controller. Upon receiving the command, the controller retrieves and returns basic configuration data such as the vehicle model and battery interface specifications, which are bound to the vehicle's unique identification code, from local storage or the real-time status register. The vehicle model field indicates the vehicle's design category, power platform, and structural characteristics, while the battery interface specifications cover key information such as interface type, mechanical dimensions, pin definitions, and electrical compatibility parameters. Dynamic operating parameters, such as current battery capacity requirements and rated power requirements, are read in real-time from the vehicle's energy management unit and battery management module. Battery capacity requirements reflect the amount of electricity needed to meet the vehicle's range under current operating conditions, while rated power requirements represent the power output capacity required under normal driving or acceleration conditions. Dynamic operating parameters are affected by factors such as ambient temperature, load conditions, and vehicle operating modes, and are collected in real-time when a battery swap request is issued. The basic configuration data and real-time operating requirement parameters are uniformly formatted, redundancy removed, and logically verified. For example, the system checks the matching degree between the battery interface specifications and capacity requirements, confirms whether the rated power requirements are within the vehicle's design range, and adds timestamps and data acquisition accuracy indicators to the data. After the integration process, a set of structured current vehicle parameter information is formed.

[0075] In one specific embodiment, the process of performing step S2 may specifically include the following steps:

[0076] In the cloud database, the vehicle model in the parameter information is indexed and matched to obtain the battery usage record index table corresponding to the vehicle model.

[0077] Based on the battery usage record index table, the historical usage count of each battery model on the vehicle model is traversed to obtain the battery usage record matrix;

[0078] The battery usage record matrix is ​​correlated with the performance degradation curve of each battery model on the vehicle model to obtain battery model performance correlation data.

[0079] Historical successful matching rate statistics and failure rate analysis were performed on battery model performance correlation data to obtain historical battery matching data.

[0080] Based on historical battery matching data, compatibility calculations are performed to obtain a compatibility score between each battery model and the current vehicle.

[0081] Specifically, the cloud database receives current vehicle parameter information uploaded by the vehicle's central control system and uses the vehicle model field in the parameter information as the core input for the search index. The cloud database utilizes a pre-built index system of usage relationships for multiple vehicle models and battery types to match the vehicle model field, thereby quickly locating and calling the battery usage record index table bound to the vehicle model. The battery usage record index table stores detailed data from each battery swap according to time series and the vehicle model-battery model association, including the battery model installed in the current vehicle model each time, the runtime, the number of cycles, and the battery swap completion status. The cloud system uses the battery usage record index table as the traversal entry point, statistically analyzing the historical usage counts of different battery models on the target vehicle model, and arranging the statistical values ​​in a two-dimensional structure with battery model as rows and vehicle model as columns to generate a battery usage record matrix. It then calls the battery performance degradation curve data stored in the cloud. The battery performance degradation curve is fitted based on the cycle life test results, capacity retention rate change patterns, and failure probability of different battery models during long-term operation, and has been adapted for different vehicle models. The battery usage record matrix is ​​correlated with the performance degradation curves of corresponding battery models on target vehicle models. The direct impact of usage frequency on performance degradation is considered, and factors such as usage environment, load characteristics, and depth of discharge are incorporated to correct for degradation rates, generating battery model performance correlation data reflecting long-term operational performance. Historical successful matching rate statistics and failure rate analysis are performed on the battery model performance correlation data. The historical successful matching rate is calculated based on the operational stability and user feedback records after each battery swap, such as whether the current battery model can provide normal power within a specified time after a swap on the target vehicle model, and whether any abnormal alarms or premature retirement occur. The failure rate analysis is based on historical operation logs and maintenance records, calculating the proportion of various failures (such as overheating, overvoltage, and communication anomalies) that occurred during the use of the current battery model on the target vehicle model. By comprehensively analyzing the successful matching rate and failure rate, model combinations that perform stably in actual operation, beyond physical compatibility, are identified. Combinations that are compatible in terms of interface and specifications but have high operational risks are eliminated, forming verified historical battery matching data. Compatibility calculations are then performed based on the historical battery matching data. By integrating data from multiple dimensions such as historical usage frequency, performance retention rate, successful matching rate, and failure risk, a compatibility score between each battery model and the current vehicle is calculated through weighted scoring or machine learning models.

[0082] In this embodiment, after querying the historical battery matching data corresponding to the parameter information from the cloud database and calculating the compatibility score between each battery model and the current vehicle, the method further includes a step of verifying the compatibility score based on a multi-model battery compatibility evaluation model: inputting the historical battery matching data into the model cross-attention layer of the multi-model battery compatibility evaluation model for cross-model correlation analysis to obtain the attention weight matrix between battery models; performing time-series encoding on the long-term usage trajectory of each battery model on the current vehicle model in the historical battery matching data to obtain the historical trajectory feature sequence of the battery model; calculating the safety risk assessment value of each battery model switching to the current vehicle based on the fault records and performance degradation data in the historical battery matching data to obtain the safety boundary parameter of the battery model; performing weighted fusion calculation on the attention weight matrix, historical trajectory feature sequence and safety boundary parameter to obtain the compatibility confidence of each battery model with the current vehicle; and filtering the compatibility confidence based on a preset compatibility threshold to obtain a set of candidate battery models that pass the compatibility verification.

[0083] The process involves several steps. First, after retrieving historical battery matching data from the cloud database and calculating the compatibility score between each battery model and the current vehicle, the battery swapping control system performs a multi-objective collaborative recommendation decision based on the compatibility score and the inventory batteries. This includes a multi-dimensional battery recommendation prediction control step triggered by dual events. The data-driven prediction framework constructs a spatiotemporal distribution prediction model for battery models based on historical battery matching data and the spatiotemporal distribution of each battery model in the current battery swapping network, obtaining the dynamic distribution prediction results of battery models across different swapping cabinets. Second, a dual-event trigger detector simultaneously monitors changes in the frequency of user battery swapping requests and the time-varying dynamic characteristics of battery performance. When a sudden increase in battery swapping demand or a significant degradation in battery performance is detected, the detector will respond accordingly. A trigger signal is generated to obtain a dual-event trigger status identifier; the battery model switching prediction model calculates the switching probability between different battery models based on the dual-event trigger status identifier to obtain the battery model switching prediction probability matrix; the adaptive model updater performs anomaly detection based on the deviation between the battery model switching prediction probability matrix and the actual battery swapping result. When the detected prediction deviation exceeds a preset threshold, the model parameters are adaptively updated to obtain the corrected prediction model parameters; the recommended stability controller calculates the stability evaluation value of the current recommended strategy based on the recommended stability function. When the stability evaluation value is lower than the recommended activation threshold, the predictive control mechanism is activated to predictively adjust the adaptability score to obtain the adaptive score after predictive control optimization.

[0084] The process includes retrieving historical battery matching data corresponding to parameter information from a cloud database and calculating the compatibility score between each battery model and the current vehicle. It also includes a step of multi-vehicle recommendation quality prediction based on interpretable decoupled transfer learning: a hierarchical sampling controller performs hierarchical sampling processing on the historical battery matching data according to different battery swapping frequencies and vehicle models, obtaining a multi-frequency battery history dataset containing high-frequency, medium-frequency, and low-frequency battery swapping sets; and a decoupled transfer learning module extracts vehicle-independent general battery performance features and vehicle-specific dedicated adaptation features from the multi-frequency battery history dataset to obtain a vehicle-independent feature representation. The system employs a vehicle-invariant feature representation and a domain representation learner. The domain representation learner learns common battery performance patterns across vehicle models based on vehicle-invariant feature representations and learns battery adaptation preference patterns for specific vehicle models based on vehicle-specific feature representations, resulting in domain-invariant and domain-specific representation vectors. A recommendation quality predictor fuses these vectors to construct a recommendation quality prediction model, yielding a recommendation quality prediction score for each battery model. An information-theoretic unwrapping optimizer performs unwrapping optimization on the domain-invariant and domain-specific representation vectors based on the principle of maximizing mutual information, obtaining an information-theoretic optimized feature representation and updating the recommendation quality prediction score.

[0085] In one specific embodiment, the process of performing compatibility calculations based on historical battery matching data to obtain compatibility scores between each battery model and the current vehicle can specifically include the following steps:

[0086] The usage frequency and success rate data in the historical battery matching data are encoded into feature vectors to obtain the historical performance feature vectors of each battery model.

[0087] Cross-model similarity calculations are performed on historical performance feature vectors to obtain performance similarity coefficients between battery models;

[0088] The risk coefficient for switching each battery model to the current vehicle is calculated based on the failure rate analysis in the battery historical matching data.

[0089] By weighting and fusing the performance similarity coefficient and the risk coefficient, a compatibility score between each battery model and the current vehicle is obtained.

[0090] Specifically, the cloud analytics module receives historical battery matching data from the historical data processing pipeline and extracts two core statistics: "usage frequency" and "success rate." These statistics are then normalized and weighted by time decay under a unified time scale and sample coverage to reduce the impact of earlier samples on the current assessment and highlight the latest performance. After time-series shaping, the cloud analytics module constructs historical performance feature vectors for each battery model. These vectors include fields such as cumulative usage counts on the target vehicle model, recent window success rate, cross-scenario stability markers, and power supply continuity labels under typical operating conditions. Each field is appended with a collection timestamp and data integrity identifier. Based on the historical performance feature vectors, the cloud analytics module performs cross-model similarity calculations for all battery models. The historical performance feature vectors of each battery model are mapped to a unified metric space, and a similarity criterion based on a combination of distance metric and historical success rate weighting is used to obtain the performance similarity coefficient between battery models. The more consistent the operating condition response, stability, and success rate of the two models on the target vehicle model, the higher the performance similarity coefficient. To enhance the sensitivity of similarity assessment to real-world operational effectiveness, the "closeness of feature vectors" is coupled with the "percentage of historical successful matches." This ensures that similarity reflects not only the similarity of statistical features but also the successful accumulation in real-world scenarios, thus avoiding the bias of ignoring usability based solely on similar feature shapes. The performance similarity coefficient dynamically reflects the substitutability and migration potential between models on the target vehicle model, and updates the similarity spectrum online with the arrival of new battery swapping activities. The cloud-based analysis module calculates the risk coefficient for switching battery models to the current vehicle based on fault records in historical battery matching data. The risk coefficient uses the target vehicle model as a reference, comprehensively statistically analyzing the proportion of events such as overheating, overvoltage, communication anomalies, premature power loss, and protection triggering per unit of runtime or unit of cycles after battery swapping. It also considers fault extrapolation and penalties for the same model in vehicles under similar operating conditions. When the fault proportion exceeds the operational threshold within the most recent time window, the risk coefficient increases, triggering a weight reduction strategy to lower the probability of subsequent recommendations. The performance similarity coefficient and risk coefficient are then weighted and fused. The fusion phase centers on the target vehicle model, expressing "substitutability and availability" information using performance similarity coefficients and "safety boundaries and maintenance costs" information using risk coefficients. A weighted fusion strategy outputs a compatibility score between each battery model and the current vehicle. This weighted fusion strategy employs hierarchical weights and dynamic temperature parameters: when data is sufficient and the success rate of the nearest window is stable, the weight of the performance similarity coefficient is increased to encourage cross-model migration and resource utilization; when an increased risk is detected for a particular model on the target vehicle model or similar models, the weight of the risk coefficient is increased to prioritize convergence to more robust model combinations, and the score of combinations with potential risks is reduced.During the fusion process, the frequency prior provided by the historical usage record matrix and the lifetime maintenance prior provided by the long-term decay record are used to perform boundary pruning and confidence labeling on the scores, so as to avoid the problem that unpopular models are overestimated due to sparse samples, or that popular models are insensitive to the latest risks due to large early samples.

[0091] In one specific embodiment, the process of performing cross-model similarity calculation on historical performance feature vectors to obtain performance similarity coefficients between battery models can specifically include the following steps:

[0092] The historical performance feature vectors are hierarchically sorted according to the frequency of use of battery models from high to low to obtain a priority sequence of battery models.

[0093] The historical data of each battery model in the battery model priority sequence are weighted by a time decay factor to obtain a set of feature vectors with time-series weights.

[0094] Calculate the dynamic threshold parameter based on the similarity distribution data in the feature vector group with time-series weights;

[0095] Based on dynamic threshold parameters, multiple rounds of similarity iterative calculations are performed between battery models until convergence, resulting in the performance similarity coefficient between battery models.

[0096] Specifically, historical performance feature vectors are hierarchically sorted according to the frequency of battery model usage, from high to low. High-frequency models are prioritized at the beginning of the sequence, while low-frequency models are ranked at the end, forming a battery model priority sequence. Since the data in the historical performance feature vectors comes from different time periods, early operating data can have an excessive impact on current evaluation results, thus masking the trend changes in the latest operating status. The cloud analysis module assigns a time decay factor to the historical data of each battery model in the priority sequence. The time decay factor is calculated exponentially or linearly according to the interval between the data collection time and the current time, giving higher weight to data more recent than the current time and gradually decreasing weight to earlier data. This dynamically emphasizes the importance of recent operating performance and reduces the proportion of outdated data contributing to the overall feature vector, resulting in a set of feature vectors with time-series weights. Similarity distribution data is calculated based on the time-series weighted feature vector sets. For each pair of battery models, weighted similarity indices are calculated across multiple dimensions such as usage frequency, success rate, and stability, generating a full similarity matrix. Statistical analysis is performed on the similarity matrix to identify its distribution characteristics, such as central tendency, dispersion, and the range of maximum and minimum values. Based on the analysis of distribution characteristics, a dynamic threshold parameter is extracted that can adapt to the current data state. For example, when the overall similarity distribution is high, the dynamic threshold parameter is increased accordingly to improve discriminative power; conversely, when the overall distribution is low, the dynamic threshold is decreased to retain a sufficient number of candidate pairs for iterative calculation. The dynamic threshold parameter is applied to multiple rounds of iterative similarity calculations between battery models. The cloud analysis module uses a set of feature vectors with time-series weights as input to perform preliminary similarity calculations between battery models and selects model pairs with similarity higher than the dynamic threshold parameter as the retention objects for the first round of iteration. Feature fusion and similarity updates are performed among the retention objects. The high similarity relationships discovered in the first round are used to adjust the weight distribution of the feature vectors, and new similarity values ​​are calculated again. In each round of iteration, the dynamic threshold parameter is applied for filtering, while the field weights and time decay parameters in the feature vector set are continuously updated to reflect new patterns and correlations discovered during the iteration process. As the number of iterations increases, the similarity relationships between models gradually converge, manifested as a stabilization of the similarity matrix values ​​and a lack of significant changes in the ranking of highly similar model pairs. When the convergence condition is met, such as when the change in the similarity matrix is ​​less than a preset convergence threshold in two consecutive iterations, the iteration process terminates, and the final stable similarity matrix is ​​output as a set of performance similarity coefficients between battery models.

[0097] The process includes, after performing multiple rounds of iterative similarity calculations between battery models based on dynamic threshold parameters until convergence, a step of dynamically calculating the performance similarity coefficients of multiple battery models is also included: the cloud analysis system calculates the feature vector distance between the performance similarity coefficients of battery models to obtain the feature space distribution matrix of battery models; the historical success rate statistician calculates the ratio of the number of successful matches for each battery model combination to the total number of matches in the historical battery matching data to obtain the historical successful match rate value of the battery model; the similarity coefficient calculator performs distance weighting processing on the feature space distribution matrix through an exponential decay function to obtain the distance weighting coefficient; the multi-model performance similarity coefficient generator multiplies the distance weighting coefficient with the historical successful match rate value to obtain the performance similarity coefficients of multiple battery models; and the similarity coefficient update module corrects the performance similarity coefficients of multiple battery models online based on real-time battery swapping feedback data to obtain the dynamically updated battery performance similarity coefficient matrix.

[0098] The process involves cross-model similarity calculation of historical performance feature vectors to obtain performance similarity coefficients between battery models. This includes heterogeneous data processing steps based on an improved Battery-Transformer model: dividing historical performance feature vectors into three heterogeneous data subsets: BMS time-series data, maintenance record data, and performance test data; a heterogeneous data standardization layer uses sliding window resampling on BMS time-series data to unify different sampling frequencies to a fixed time interval, one-hot encoding on maintenance record data to convert text format into numerical vectors, and maximum-minimum value normalization on performance test data to unify different dimensions to the 0-1 interval, resulting in a standardized heterogeneous data matrix; The battery model normalization layer includes a model embedding module and a feature mapping module. The model embedding module converts the battery model identifier into a 128-dimensional embedding vector. The feature mapping module contains three fully connected layers. Each fully connected layer uses the ReLU activation function and Dropout regularization to concatenate the normalized heterogeneous data matrix with the 128-dimensional embedding vector and map it to a 512-dimensional unified feature space to obtain the model normalized feature matrix. The battery charge / discharge sequence attention layer includes a timing encoder and a charge / discharge pattern recognizer. The timing encoder uses a position encoding mechanism to add position information to each time step in the sequence. The charge / discharge pattern recognizer contains four multi-head attention sub-layers, each containing eight attention heads. The self-attention mechanism is used to calculate the dependencies within the charge-discharge cycle, obtaining the sequence features perceived by the charge-discharge mode. The multi-model cross-attention layer includes a cross-model query generator and a cross-attention calculator. The cross-model query generator generates query vectors, key vectors, and value vectors for each battery model. The cross-attention calculator uses a scaled dot product attention mechanism to calculate the attention weight matrix between different battery models and performs feature fusion through a weighted summation operation to obtain the multi-model cross-association feature representation. The improved Transformer encoder includes six encoder layers, each containing a multi-head self-attention sublayer and a feedforward neural network sublayer. The multi-head self-attention sublayer uses 16 attention... The E-head, a feedforward neural network sublayer, contains two linear transformation layers and a GELU activation function. Each sublayer is followed by residual connections and layer normalization to perform deep encoding of the cross-correlation feature representations of multiple models, resulting in high-level semantic feature representations. The performance prediction output layer includes a global pooling layer and a multi-task prediction head. The global pooling layer performs average pooling on the high-level semantic feature representations. The multi-task prediction head contains three parallel fully connected branches that predict battery capacity decay, internal resistance growth, and cycle life, respectively. Each branch contains two fully connected layers and one sigmoid activation function to obtain the performance prediction results of each battery model for future use and calculate the performance similarity coefficient between battery models.

[0099] In one specific embodiment, the process of performing step S3 may specifically include the following steps:

[0100] The battery swapping cabinet control system reads the compatibility score and obtains the health status and storage location data of the inventory batteries;

[0101] Based on health status and warehouse distribution data, a multi-objective optimization function set is constructed, which includes the objective function of battery remaining value utilization rate, the objective function of user waiting cost, and the objective function of inventory distribution equilibrium.

[0102] The crossover and mutation probabilities in the genetic algorithm are dynamically adjusted based on the convergence state of the multi-objective optimization function set.

[0103] Population evolution calculations are performed based on crossover and mutation probabilities to obtain the optimal recommended battery identifier.

[0104] Specifically, the battery swapping cabinet control system establishes a real-time data communication channel with the cloud data processing module to receive the compatibility score calculated for the current vehicle. Simultaneously, it calls the cabinet's local inventory management module to read the health status data and storage location information of all stored batteries from the internal database and sensor acquisition system. Health status data includes key indicators such as SOC (State of Charge), SOH (State of Health), current voltage, current, temperature, charge / discharge cycle count, and recent operational anomaly records. Storage location information includes the specific location of each battery within the swapping cabinet, its distance from the cabinet door, the access path of the robotic arm or transmission mechanism, and the availability status of the entry / exit channels. Based on the health status data, a battery remaining value utilization objective function is established. This objective function optimizes for maximizing the usable lifespan utilization rate, assigning higher target values ​​to batteries with higher SOH, lower cycle counts, and gentler performance degradation curves to maximize the battery's full lifespan value. Based on the location of battery swapping bays and the current location of user battery swapping requests, a user waiting cost objective function is established. This objective function comprehensively considers the path length of the mechanical actuators of the battery swapping cabinets, the door opening time, user queuing time, and possible internal scheduling delays within the battery swapping cabinets, aiming to minimize the total waiting cost and thus improve user experience. Based on the overall inventory's bay occupancy rate and model distribution, an inventory distribution equilibrium objective function is established. This objective function aims to avoid prolonged idleness or overuse of certain battery models or bays, reducing long-term maintenance risks caused by uneven battery performance degradation through a balanced inventory structure. These three objective functions together constitute a multi-objective optimization function set, and a multi-objective solution framework is constructed using weighted or Pareto optimal methods, enabling the optimization process to achieve a dynamic balance between performance utilization, user experience, and inventory equilibrium. The battery swapping cabinet control system uses a genetic algorithm as the optimization engine and dynamically adjusts the crossover and mutation probabilities of the genetic algorithm based on the convergence state of the optimization function set during the iteration process. Monitoring the convergence state includes analyzing the variance of the fitness distribution, the maximum fitness improvement rate, and population diversity indicators in the current iteration population. When slow convergence and declining population diversity are detected, the mutation probability is appropriately increased to introduce more random perturbations and expand the search space coverage. When convergence approaches stability and fitness improvement slows down, the crossover probability is appropriately increased to accelerate the recombination and propagation of superior genes, driving the population to quickly converge to a better solution region. After the adjustment strategy takes effect, the genetic algorithm uses the currently set crossover and mutation probabilities to perform population evolution calculations. The initial population consists of a set of battery candidates that meet physical compatibility and safety constraints, with each individual corresponding to a specific battery compartment and its matching battery model. The algorithm iteratively generates a new generation of the population through genetic operations such as selection, crossover, and mutation, and calculates the fitness values ​​of all individuals on the three objective functions in each generation.For multi-objective optimization problems, fitness evaluation employs non-dominated sorting and crowding distance calculation methods to retain uniformly distributed and high-performing solution sets. During iteration, solutions that satisfy the optimal conditions for multi-objective trade-offs are gradually retained and strengthened. When the convergence condition is met, i.e., when the increase in population fitness and the change in diversity are below preset thresholds, the algorithm selects the individual with the highest overall fitness from the current optimal solution set. The corresponding battery compartment number and model are then used as the optimal recommended battery identifier.

[0105] Before performing population evolution calculations based on crossover and mutation probabilities, the process includes an initialization step based on a battery-vehicle-battery swapping cabinet collaborative optimization algorithm: The objective function weight allocator calculates the dynamic weight coefficients of each objective function based on the current user battery swapping frequency and the distribution of inventory batteries, obtaining a time-varying weight parameter set; the population initialization module generates initial population individuals that meet the constraints based on fitness scores and inventory battery health status data, obtaining a feasible solution population set; the fitness evaluator performs a weighted summation calculation of the time-varying weight parameter set with the battery remaining value utilization rate objective function, the user waiting cost objective function, and the inventory distribution equilibrium objective function, obtaining the individual's comprehensive fitness value; the constraint handling mechanism penalizes and corrects individuals in the feasible solution population set that violate the battery health status threshold and warehouse availability constraints, obtaining an effective population that satisfies the constraints; and the elite retention strategy selects the individual with the highest fitness from the effective population for retention and propagation, obtaining an elite individual set for subsequent genetic operations.

[0106] In one specific embodiment, the process of dynamically adjusting the crossover and mutation probabilities in the genetic algorithm based on the convergence state of the multi-objective optimization function set can specifically include the following steps:

[0107] Fitness gradient monitoring was performed on the objective functions of battery residual value utilization, user waiting cost, and inventory distribution equilibrium in the multi-objective optimization function set to obtain the convergence speed difference value.

[0108] The rate of change of genetic factors was obtained by comparing the difference in convergence speed with historical convergence patterns.

[0109] The crossover probability matching the convergence state is obtained by correlating the rate of change of genetic factors with the variance of the fitness distribution of individuals in the population.

[0110] Complementary calculations are performed based on crossover probability and the degree of population evolutionary stagnation to obtain mutation probability that works in conjunction with crossover probability.

[0111] Specifically, in each iteration of the multi-objective solution, the battery swapping cabinet control system establishes monitoring sequences for the fitness changes of the battery remaining value utilization objective function, the user waiting cost objective function, and the inventory distribution equilibrium objective function, respectively. By jointly analyzing the fitness improvement of adjacent generations, the improvement slope within the sliding window, and the changes in the non-dominated frontier density, the instantaneous convergence speed and stability markers of the three objective curves are extracted. The convergence speed difference value is obtained by comparing the three curves pairwise, thereby identifying whether the current optimization process is dominated by value utilization, waiting cost, or inventory equilibrium, so that the search direction can be accelerated around the trade-off of multiple objectives rather than a single objective. Based on the convergence speed difference value and historical convergence pattern comparison analysis, the current difference map is compared with the typical convergence trajectory under the same load, similar vehicle-battery-cabinet combination, and similar inventory ratio in the historical iteration log. Through nearest neighbor pattern or time attention weighted retrieval, the most likely convergence evolution direction and the expected plateau period length are inferred, and the genetic factor change rate is calculated. The genetic factor change rate describes the extent to which crossover and mutation operations need to be increased or decreased in the next cycle. If the historical model shows that the multi-objective frontier has approached stability but individual objects still exhibit repeated oscillations, a positive change rate of "increasing crossover and moderately suppressing mutation" is output. If the historical model shows insufficient diversity in the early stages and stagnation at the frontier, a corrected change rate of "increasing mutation and maintaining crossover" is output. The historical model library is continuously expanded based on the behavioral trajectories and online learning capabilities of the cloud platform, enabling continuous calibration of convergence priors as data accumulates, shifting the parameter tuning logic from static experience to data-driven adaptive regulation. The genetic factor change rate is correlated with the variance of the fitness distribution of individuals in the current population to generate a crossover probability that matches the convergence state. The variance of the fitness distribution is considered an immediate measure of population diversity: when the variance is large and the genetic factor change rate indicates accelerated convergence, the crossover probability is increased accordingly to amplify the diffusion effect of superior gene recombination; when the variance is small and the genetic factor change rate suggests the need to restore exploration capabilities, the crossover probability is moderately decreased. The crossover probability is updated simultaneously with the multi-objective fitness assessment, ensuring that the scale of crossover operations in each generation matches the current generation's "recombinability value," rather than using fixed empirical values, thereby reducing the risk of premature convergence and invalid recombination. Based on the crossover probability, the mutation probability, which is complementary to the crossover probability, is calculated in conjunction with the degree of population evolutionary stagnation. The degree of stagnation is characterized by several criteria, such as the length of generations in which the global optimum remains unimproved, the duration for which the supervolume increment of the non-dominated front approaches zero, and the rate of decline of the mean crowding distance. When the degree of stagnation increases and the crossover probability is in the upward adjustment range, to avoid repeated recombination in local neighborhoods, the mutation probability is moderately increased by the complementarity coefficient, injecting perturbation into the uncovered regions of the solution space. When the degree of stagnation is low and the crossover probability is in the downward adjustment range, the mutation probability remains at a moderate level, mainly serving as "background noise" to maintain diversity.Complementarity calculation uses crossover probability as the dominant variable and stagnation degree as the correction variable to output the mutation probability that works in conjunction with the crossover strategy, thus forming a dynamic balance between "accelerating the utilization of known superior structures" and "opening up unknown potential optimal solution regions". Because the system maintains an online update mechanism for performance similarity coefficients and risk weights in the cloud for a long time, the population evaluation incorporates the latest model preferences and safety boundaries in each generation. Therefore, the mutation intensity can be calibrated in a timely manner, avoiding overexploration of high-risk combinations and ensuring the consistency of recommendation results in terms of usability and safety.

[0112] In one specific embodiment, the process of performing step S4 may specifically include the following steps:

[0113] The battery swapping cabinet analyzes the compartment number information in the optimal recommended battery identifier and drives the corresponding battery compartment door opening mechanism.

[0114] After the battery compartment is opened, the sensor array detects the user's sequence of removing the old battery and inserting the recommended battery.

[0115] The operation sequence is timestamped and the battery identification verification results are collected to obtain battery swapping process data;

[0116] The battery swapping process data is associated and assembled with the optimal recommended battery identifier to generate a battery swapping operation record.

[0117] Specifically, the battery swapping cabinet control system receives the optimal recommended battery identifier and performs semantic parsing to extract the compartment number information. This compartment number information is then input into the compartment door drive control link, driving the corresponding motor, latch, or electromagnetic actuator to complete both mechanical and electrical unlocking, thereby opening the target battery compartment door. Once the door is open, the sensor array activates. Infrared beam sensors, pressure sensors, Hall effect position detection, and digital inputs work together to form a continuous monitoring mechanism for the "remove old battery - insert recommended battery" action. The sensor array generates event flags for each key action and binds them to the compartment number information, ensuring seamless action sequence over time. The system design emphasizes real-time communication and remote control capabilities between the battery swapping cabinet and the vehicle's central control system and battery management system (BMS), ensuring that sensor acquisition, status verification, and command execution operate in a closed loop within a unified channel, reducing the window period for action recognition and the probability of misjudgment. A millisecond-level timestamp is generated for each action node, and the unique identifier of the battery placed in the process is verified based on RFID identification or QR code scanning, completing a three-dimensional verification of "action-time-identity". The timestamp is used to construct the time-series trajectory of the entire battery swapping process, and the identity verification is used to ensure that the recommended battery is consistent with the actual battery placed. If the consistency verification fails, the anomaly is recorded locally and the anomaly event is synchronized to the cloud. The cloud platform implements full-link tracking and traceable evidence storage of the battery swapping behavior trajectory. Therefore, the timestamp and identifier verification results generated on the battery swapping cabinet side can form consistent behavioral evidence together with the recommendation record and equipment status record on the cloud side. The battery swapping cabinet control system writes "action sequence, time information, identifier verification result, compartment number information, recommended battery model, and user operation time index" into a structured data container to form battery swapping process data, and assembles it with the optimal recommended battery identifier as the primary key. A battery swapping operation record is generated.

[0118] In one specific embodiment, the method for performing intelligent battery swapping recommendations based on historical battery data further includes the following steps:

[0119] After receiving the battery swapping operation record, the cloud database extracts the vehicle model and the selected battery model to obtain the vehicle battery matching result for this battery swapping.

[0120] The consistency between the vehicle battery pairing results and the previously recommended battery models is verified to obtain the recommendation matching accuracy evaluation value;

[0121] The battery historical matching data is updated based on the recommended matching accuracy evaluation value to refresh the historical successful matching rate of the corresponding vehicle model and battery model combination, and the battery historical matching data is updated.

[0122] Specifically, the cloud database receives battery swapping operation records uploaded by the battery swapping cabinets through a message channel. During the parsing phase, it performs validation and standardization transformation on structured fields, extracting key elements such as vehicle model, selected battery model, optimal recommended battery identifier output during the recommendation phase, bay number, timestamp, and transaction number to form the vehicle battery matching result for each battery swap. The cloud database establishes a one-to-one comparison relationship between the vehicle battery matching result and the target model in the recommendation phase. Based on the matching result, it generates a consistency marker and records it in the metric set of the sample dimension. At the same time, it accumulates consistent samples according to a sliding time window or natural period, calculates the recommendation matching accuracy evaluation value, and attaches side information such as abnormal alarms, battery swapping duration, and equipment failure codes as context to the evaluation value. This allows for the acquisition of online indicators that can directly quantify the effectiveness of recommendations without increasing the complexity of the terminal side. The cloud database uses the combination key of "vehicle model + battery model" to locate historical statistical entries, performs atomic-level addition operations on the total number of matches and the number of successful samples under the combination key, refreshes the historical successful matching rate based on the matching results of new samples, and writes the new successful matching rate, recent window performance, and update time back to the battery historical matching data view. To avoid duplicate entries and statistical biases, the cloud database uses timestamps and unique transaction numbers for deduplication verification before writing back, and employs distributed transactions or eventually consistent queues to ensure that details and summaries are synchronously written to disk, thereby guaranteeing the stability and traceability of statistical standards. After completing the basic statistical update, the cloud database triggers a linkage calibration process based on the latest trend of the recommendation matching accuracy evaluation value. This process performs partial or batch recalculation on relevant model pairs involving the target vehicle in cross-model similarity calculations, enabling the similarity spectrum to quickly absorb new success or failure evidence. Simultaneously, it penalizes and reduces risk control weights on combinations with increased failure rates and limits the priority of subsequent recommendations, thus strengthening the safety boundary while maintaining availability. When the recommendation matching accuracy declines significantly within a short window, the cloud database simultaneously initiates a dual-path adaptive adjustment on both the strategy and model sides. The strategy side dynamically adjusts the weighting ratio and candidate selection threshold of multi-objective optimization to reduce the exposure probability of low-performing combinations. The model side increases the sampling weight of problematic combination samples in the training batch or triggers a rapid retraining process, ensuring that parameter convergence reflects the latest successful matching rate and anomaly distribution. To ensure that the edge device immediately benefits from the latest statistics upon the next request, the cloud database pushes the updated battery historical matching data, the historical successful matching rate corresponding to the combination key, and the related similarity and weight matrix to the battery swapping cabinet control system and the vehicle-side cache after the refresh is completed. It also writes a snapshot of the indicators "vehicle battery pairing results, consistency mark, recommended matching accuracy evaluation value and historical successful matching rate comparison" into a visual dashboard for operation-side monitoring and audit playback.

[0123] The process includes several steps. After updating the historical successful matching rate of the corresponding vehicle model and battery model combination based on the recommendation matching accuracy evaluation value, the next step is to optimize the multi-site collaborative recommendation model based on federated learning. The local model trainer trains a local recommendation model at each battery swapping station using the station's historical battery matching data and extracts the model gradient parameters to obtain the station's local model parameter vector. A privacy-preserving aggregator performs differential privacy processing and secure aggregation calculation on the station's local model parameter vector uploaded by each station to obtain the global model parameter update vector. The federated model update mechanism distributes the global model parameter update vector to each battery swapping station and updates the local recommendation model parameters, resulting in a federated collaboratively optimized recommendation model. A cross-site knowledge transfer module learns the battery adaptation experience of other stations based on the federated collaboratively optimized recommendation model and integrates it into the station's recommendation strategy, resulting in a knowledge-enhanced local recommendation strategy. Finally, a collaborative effect evaluator calculates the improvement in recommendation accuracy before and after federated learning and dynamically adjusts the federated participation weights to obtain an adaptive multi-site collaborative recommendation system.

[0124] The above describes the intelligent battery swapping recommendation method based on historical battery data in the embodiments of the present invention. The following describes the intelligent battery swapping recommendation system based on historical battery data in the embodiments of the present invention. Please refer to [link / reference]. Figure 2 One embodiment of the intelligent battery swapping recommendation system based on historical battery data in this invention includes:

[0125] The vehicle central control system 001 is used to identify the vehicle by the battery swapping request signal triggered by the user and obtain the parameter information of the current vehicle.

[0126] The calculation module 002 is used to query the historical battery matching data corresponding to the parameter information from the cloud database and calculate the compatibility score between each battery model and the current vehicle.

[0127] The battery swapping cabinet control system 003 is used to perform multi-objective collaborative recommendation decision-making by combining the compatibility score with the inventory batteries to obtain the optimal recommended battery identifier.

[0128] Battery swapping cabinet 004 is used to open the corresponding compartment door based on the optimal recommended battery identifier to execute the battery swapping process and generate a battery swapping operation record.

[0129] Through the collaborative efforts of the aforementioned components and the automatic identification mechanism of the vehicle's central control system, key parameters such as vehicle model, battery interface specifications, and power requirements can be accurately obtained. Compared to traditional manual operation, this avoids user input errors and incomplete information. Based on the correlation analysis of battery usage record matrices and performance degradation curves, it can deeply explore the historical performance patterns of different battery models on specific vehicle models, improving the accuracy of battery compatibility assessment compared to traditional methods that rely solely on real-time conditions. Through cross-model similarity calculation of historical performance feature vectors and dynamic threshold iteration algorithms, performance correlations between different battery models can be discovered, solving the technical challenge of traditional systems failing to handle the collaborative optimization of multiple battery models. Employing a variable genetic factor multi-objective optimization algorithm, it can simultaneously balance multiple objectives such as battery residual value utilization, user waiting costs, and inventory distribution equilibrium, achieving global optimization of battery swapping decisions compared to single-objective optimization strategies. Through convergence state monitoring and dynamic adjustment mechanisms of genetic factors, algorithm parameters can be adaptively adjusted according to the real-time state of the optimization process, avoiding the problem of traditional fixed-parameter algorithms easily getting trapped in local optima. By leveraging real-time feedback from battery swapping operation records and continuous updates to historical matching data, a complete closed-loop learning mechanism has been established. This enables the system to continuously optimize recommendation strategies based on actual battery swapping results, thereby continuously improving recommendation accuracy. Real-time synchronization and collaborative operation of data from vehicles, batteries, and battery swapping cabinets have been achieved, breaking down the technical barriers of isolated components in traditional systems and enhancing the automation and intelligence of the entire battery swapping process.

[0130] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0131] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0132] 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 smart battery swapping recommendation method based on historical battery data, characterized in that, include: The vehicle's central control system identifies the vehicle by detecting the battery swapping request signal triggered by the user and obtains the current vehicle's parameter information. The system retrieves historical battery matching data corresponding to the parameter information from the cloud database and calculates the compatibility score between each battery model and the current vehicle. Specifically, this includes: in the cloud database, performing index matching based on the vehicle model in the parameter information to obtain a battery usage record index table corresponding to the vehicle model; traversing the historical usage counts of each battery model on the vehicle model based on the battery usage record index table to obtain a battery usage record matrix; correlating the battery usage record matrix with the performance degradation curves of each battery model on the vehicle model to obtain battery model performance correlation data; performing historical successful matching rate statistics and failure rate analysis on the battery model performance correlation data to obtain historical battery matching data; and calculating compatibility based on the historical battery matching data to obtain a compatibility score between each battery model and the current vehicle. The battery swapping cabinet control system performs multi-objective collaborative recommendation decision-making with the compatibility score and the inventory batteries to obtain the optimal recommended battery identifier. The battery swapping cabinet opens the corresponding compartment door based on the optimal recommended battery identifier to execute the battery swapping process and generate a battery swapping operation record.

2. The intelligent battery swapping recommendation method based on historical battery data according to claim 1, characterized in that, The vehicle central control system performs vehicle identification on the user-triggered battery swapping request signal to obtain the current vehicle's parameter information, including: Receive user key presses and generate a battery swap request signal; Near-field communication scanning is performed on the vehicles in the battery swapping request signal to obtain the vehicle's unique identification code; Using the CAN communication protocol, the vehicle model and battery interface specifications are read from the vehicle controller based on the vehicle's unique identification code to obtain the vehicle's basic configuration data; The vehicle's central control system integrates the vehicle's basic configuration data with the current battery capacity requirements and rated power requirements to obtain the vehicle's parameter information.

3. The intelligent battery swapping recommendation method based on historical battery data according to claim 1, characterized in that, The compatibility calculation based on the battery's historical matching data yields a compatibility score for each battery model and the current vehicle, including: The usage frequency and success rate data in the battery historical matching data are encoded into feature vectors to obtain the historical performance feature vectors of each battery model. Cross-model similarity calculation is performed on the historical performance feature vectors to obtain the performance similarity coefficients between battery models; The risk coefficient for switching each battery model to the current vehicle is calculated based on the failure rate analysis in the battery historical matching data. The performance similarity coefficient and the risk coefficient are weighted and fused to obtain the compatibility score between each battery model and the current vehicle.

4. The intelligent battery swapping recommendation method based on historical battery data according to claim 3, characterized in that, The step of performing cross-model similarity calculation on the historical performance feature vector to obtain the performance similarity coefficient between battery models includes: The historical performance feature vectors are hierarchically sorted according to the frequency of use of battery models from high to low to obtain a priority sequence of battery models. The historical data of each battery model in the battery model priority sequence are weighted by a time decay factor to obtain a feature vector group with time-series weights. Calculate the dynamic threshold parameter based on the similarity distribution data in the feature vector group with temporal weights; Based on the dynamic threshold parameter, multiple rounds of similarity iteration calculations are performed between battery models until convergence, to obtain the performance similarity coefficient between battery models.

5. The intelligent battery swapping recommendation method based on historical battery data according to claim 1, characterized in that, The battery swapping cabinet control system performs multi-objective collaborative recommendation decisions based on the compatibility score and the inventory of batteries to obtain the optimal recommended battery identifier, including: The battery swapping cabinet control system reads the compatibility score and obtains the health status and storage location data of the inventory batteries; Based on the health status and the warehouse distribution data, a multi-objective optimization function set is constructed, which includes the objective function of battery remaining value utilization rate, the objective function of user waiting cost, and the objective function of inventory distribution equilibrium. The crossover and mutation probabilities in the genetic algorithm are dynamically adjusted based on the convergence state of the multi-objective optimization function set. Population evolution calculations are performed based on the crossover probability and the mutation probability to obtain the optimal recommended battery identifier.

6. The intelligent battery swapping recommendation method based on historical battery data according to claim 5, characterized in that, The step of dynamically adjusting the crossover and mutation probabilities in the genetic algorithm based on the convergence state of the multi-objective optimization function set includes: Fitness gradient monitoring is performed on the battery remaining value utilization objective function, user waiting cost objective function, and inventory distribution equilibrium objective function in the multi-objective optimization function group to obtain the convergence speed difference value; The rate of change of genetic factors is obtained by comparing the convergence speed difference with the historical convergence pattern. The correlation calculation between the rate of change of the genetic factors and the variance of the fitness distribution of individuals in the population is performed to obtain the crossover probability that matches the convergence state. Complementarity calculations are performed based on the crossover probability and the degree of population evolution stagnation to obtain the mutation probability that works in conjunction with the crossover probability.

7. The intelligent battery swapping recommendation method based on historical battery data according to claim 1, characterized in that, The battery swapping cabinet opens the corresponding compartment door according to the optimal recommended battery identifier to execute the battery swapping process and generates a battery swapping operation record, including: The battery swapping cabinet parses the compartment number information in the optimal recommended battery identifier and drives the corresponding battery compartment door opening mechanism; After the battery compartment is opened, the sensor array detects the user's sequence of removing the old battery and inserting the recommended battery. The operation sequence is timestamped and the battery identification verification results are collected to obtain battery swapping process data; The battery swapping process data is associated and assembled with the optimal recommended battery identifier to generate a battery swapping operation record.

8. The intelligent battery swapping recommendation method based on historical battery data according to claim 1, characterized in that, The intelligent battery swapping recommendation method based on historical battery data also includes: After receiving the battery swapping operation record, the cloud database extracts the vehicle model and the selected battery model to obtain the vehicle battery matching result for this battery swap. The consistency between the vehicle battery pairing results and the previously recommended battery models is verified to obtain the recommendation matching accuracy evaluation value; The battery historical matching data is updated based on the recommended matching accuracy evaluation value to refresh the historical successful matching rate of the corresponding vehicle model and battery model combination, and the battery historical matching data is updated accordingly.

9. A smart battery swapping recommendation system based on historical battery data, characterized in that, A method for performing intelligent battery swapping recommendation based on battery history data as described in any one of claims 1-8, comprising: The vehicle central control system is used to identify the vehicle by the battery swapping request signal triggered by the user and obtain the parameter information of the current vehicle. The calculation module is used to query the historical battery matching data corresponding to the parameter information from the cloud database, and calculate the compatibility score between each battery model and the current vehicle. The battery swapping cabinet control system is used to perform multi-objective collaborative recommendation decision-making with the compatibility score and the inventory batteries to obtain the optimal recommended battery identifier; The battery swapping cabinet is used to open the corresponding compartment door according to the optimal recommended battery identifier to execute the battery swapping process and generate a battery swapping operation record.

Images