A battery safety full-link intelligent management and control method and system for electric buses based on multi-working-condition data fusion

Abstract

This invention relates to the field of power battery safety management and intelligent control technology for electric vehicles, specifically a method and system for intelligent full-link control of electric bus battery safety based on multi-condition data fusion. The method includes collecting multi-condition data such as battery status, driving conditions, environmental conditions, and charging status; performing noise reduction, anomaly removal, time alignment, and normalization to form an input sequence; obtaining a fused feature vector through a multi-modal fusion model; estimating the equivalent internal resistance, reversible thermal coefficient, and thermal response time constant online to construct thermal balance constraints and calculate thermal safety indicators; using the above parameters and estimating rate stress parameters to construct lifetime consumption constraints and calculate degradation stress indicators; performing constraint optimization based on the two indicators, outputting an optimized set, and generating current limiting, power limiting, behavioral constraints, thermal management, and scheduling actions.

Description

Technical Field

[0001] This invention relates to the field of power battery safety management and intelligent control technology for electric vehicles, specifically a method and system for intelligent control of the entire battery safety chain of electric buses based on multi-condition data fusion. Background Technology

[0002] With the advancement of electrification in urban public transportation, electric buses commonly use lithium-ion batteries as their primary power source. Electric buses are characterized by significant fluctuations in passenger load, frequent starts and stops, varying route gradients, prolonged high-power operation, centralized fast charging / recharging at depots, and significant seasonal variations in temperature and humidity. These characteristics result in batteries exhibiting distinct multi-condition coupling features across vehicle operation, depot charging, and environmental exposure. During long-term service, power batteries may experience abnormal temperature rise, localized overheating, insulation failures, increased internal resistance, and capacity decay. Improper safety management could lead to malfunctions, shutdowns, or even thermal runaway. Therefore, comprehensive safety management of batteries across the entire service lifecycle—including operation, charging, and maintenance—is essential.

[0003] Existing electric bus power battery safety management technologies typically center on onboard battery management systems, using signals such as voltage, temperature, current, and fault codes for threshold monitoring, alarms, and protective control, or performing statistical analysis of historical data in the cloud to generate maintenance recommendations. While these solutions have a certain foundation in engineering applications, they still suffer from the following objective shortcomings under complex bus operating conditions: First, the input information is limited and fragmented. Many solutions rely primarily on battery-side parameters, rarely systematically incorporating information such as driving conditions (e.g., acceleration, braking frequency, gradient, operating mileage), environmental conditions (temperature, humidity, vibration, illumination), and charging status (charging voltage, current, duration, facility fault codes). This results in insufficient ability to interpret and locate battery risks, especially when "the same temperature / voltage performance but different risk sources" are difficult to distinguish. Second, the heterogeneity of multi-source data makes fusion difficult. Data from the vehicle, depot, and environment differ in sampling frequency, timestamps, units, and missing patterns. Without a unified preprocessing and alignment mechanism, the input to the evaluation model is prone to instability, leading to misjudgments, omissions, or unusability.

[0004] Furthermore, common risk assessment methods in existing technologies often employ fixed thresholds or empirical rules, or rely on data-driven models to output risk scores. These methods suffer from the following drawbacks under bus operating conditions: Firstly, they lack constraints consistent with the battery's thermo-electrophysical processes. For example, judging solely based on measurement point temperature or temperature rise rate fails to reflect the differences in mechanisms such as increased Joule heating due to rising internal resistance, reversible thermal changes with temperature / current direction, and varying thermal responses due to changes in heat dissipation conditions. This results in insufficient threshold generalization across vehicles, seasons, and cooling conditions. Secondly, they lack temporal foresight regarding risks. Many strategies only trigger current limiting or shutdown after temperature or voltage reaches the threshold, failing to promptly reflect trends approaching the threshold within a short period, especially in scenarios like fast charging and long-slope, high-load conditions, where response lag may exist.

[0005] Furthermore, there is a coupling relationship between battery safety and battery life degradation. Current technologies often use coarse-grained assessments of battery life based on indicators such as cycle count, cumulative energy, or average temperature, or treat capacity decay and internal resistance growth as long-term statistical results. These approaches present objective problems in the actual operation of electric buses: life assessment and safety assessment often employ different scales and calibers of indicator systems, making it difficult to coordinate and constrain them within the same strategic framework; in some cases, pursuing overly conservative current-limiting strategies for short-term safety may reduce operational efficiency, while solely pursuing operational efficiency may increase thermal risks and battery life depletion. Due to the lack of unified quantitative indicators and optimization mechanisms that simultaneously consider thermal safety and degradation risks, existing solutions often struggle to form a set of control actions. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for intelligent management and control of the entire battery safety chain of electric buses based on multi-condition data fusion, so as to solve the technical problems mentioned in the background art.

[0007] Based on the above ideas, the present invention provides the following technical solution:

[0008] A method for intelligent full-chain management and control of electric bus battery safety based on multi-condition data fusion includes:

[0009] S1. Obtain multi-condition data of the electric bus, wherein the multi-condition data includes at least battery status data, driving condition data, environmental condition data, and charging status data; wherein, the battery status data includes at least battery voltage, battery current, cell temperature, state of charge, health status, and fault codes; the driving condition data includes at least driving speed, acceleration or deceleration, braking frequency, climbing angle, and operating mileage; the environmental condition data includes at least ambient temperature, relative humidity, vibration intensity, and light intensity; and the charging status data includes at least charging voltage, charging current, charging time, number of charging cycles, and charging facility fault codes.

[0010] S2. Preprocessing and heterogeneous alignment of the multi-condition data includes at least: filtering and denoising the battery current and battery voltage, removing outliers in driving speed, aligning the timestamps of the multi-source data and normalizing the dimensions to obtain the fused input sequence.

[0011] S3. Based on the multimodal fusion model, perform deep fusion on the fused input sequence and output a unified fusion feature vector;

[0012] The fused feature vector output by the multimodal data fusion module is used as the input feature of the parameter estimation module. Based on the fused feature vector, the parameter estimation module generates the working condition discrimination result and parameter adaptive configuration for parameter identification, and on this basis, completes the online estimation of equivalent internal resistance parameter, reversible thermal correlation coefficient parameter, thermal response time constant parameter and rate stress correlation parameter.

[0013] The fused feature vector is used to identify identifiable windows that meet preset conditions, including current transition windows, static windows, and steady-state heat dissipation windows; and is used to adaptively determine the time window length, filtering smoothing coefficient, and anomaly rejection threshold used for estimation. Furthermore, when multiple candidate estimation results exist, the fused feature vector is also used to generate estimation confidence scores to perform weighted fusion or selection of candidate results to obtain the final parameter estimates.

[0014] S4. Based on the fused feature vector, introduce thermal-electrophysical constraint parameters and calculate thermal safety indicators: online estimate the battery equivalent internal resistance parameter, reversible thermal correlation coefficient parameter and thermal response time constant parameter, and construct thermal balance constraint relationship based on the battery equivalent internal resistance parameter, the reversible thermal correlation coefficient parameter and the thermal response time constant parameter, and output thermal safety indicators characterizing the battery thermal runaway driving intensity and safety margin.

[0015] S5. Based on the fused feature vector, introduce degradation-temperature physical constraint parameters and calculate degradation stress index: use the battery equivalent internal resistance parameter and the reversible thermal correlation coefficient parameter, and estimate the rate stress correlation parameter online to construct a life consumption constraint relationship related to temperature, and output a degradation stress index characterizing the risk of capacity decay and internal resistance growth.

[0016] S6. Construct a strategy optimization problem using the thermal safety index and the degradation stress index as core variables, and output a final optimization set that satisfies safety constraints and operational constraints, which is used to generate at least one control action; the control action includes at least charging current limit, discharge power limit, driving behavior limit, thermal management trigger, and vehicle scheduling or site collaborative handling.

[0017] By introducing unified data collection across multiple operating conditions, including battery status, driving conditions, environmental conditions, and charging status, and completing preprocessing, fusion analysis, thermal safety assessment, degradation risk assessment, and strategy optimization within a single process, battery safety management is elevated from monitoring a single parameter or scenario to comprehensive intelligent management covering operation, environment, charging, and scheduling. This method enables systematic identification and comprehensive control of battery safety risks in complex public transportation operation scenarios, avoiding misjudgments or delayed responses caused by relying solely on temperature and voltage thresholds, thus improving the comprehensiveness and reliability of battery safety management.

[0018] Preferably, S4 further includes:

[0019] S41. When the preset current transition or operating condition change conditions are met, the equivalent internal resistance parameter of the battery is estimated based on the terminal voltage change and current change.

[0020] S42. Estimate the reversible thermal correlation coefficient parameters based on the relationship between the state of charge range, temperature range and open-circuit voltage as a function of temperature.

[0021] S43. Estimate the thermal response time constant parameter based on the response characteristics of cell temperature to current disturbance and ambient temperature disturbance;

[0022] S44. Substitute the battery equivalent internal resistance parameter, the reversible thermal correlation coefficient parameter, and the thermal response time constant parameter into the thermal balance constraint relationship to obtain the thermal safety index; wherein, the thermal safety index is used to characterize the trend strength or time margin of the cell temperature approaching the safety threshold under a given operating condition.

[0023] By incorporating battery equivalent internal resistance, reversible thermal correlation characteristics, and thermal response time characteristics into the thermal safety assessment process, thermal safety indicators are no longer static judgments based on empirical thresholds, but rather dynamic assessment results based on the actual thermoelectric behavior of the battery. The advantages of this technology are: it can distinguish the differences in battery thermal behavior under different vehicles, aging conditions, and environmental conditions, improving the targeting and accuracy of thermal risk assessment, and avoiding the problem of the same temperature threshold losing its discriminative ability under different operating conditions.

[0024] Preferably, the determination of the thermal safety index satisfies the following limitations:

[0025] The difference between the predicted core temperature of the battery cell within a future preset time window and the preset safety threshold temperature is used as the thermal safety deviation, and the rate of change of the thermal safety deviation over time or the equivalent approximation strength is used as the thermal safety index. The predicted core temperature of the battery cell is obtained recursively based on the thermal balance constraint relationship, which includes at least: the Joule heating term caused by the equivalent internal resistance of the battery, the reversible heat term caused by the reversible thermal correlation coefficient, and the heat dissipation to the environment characterized by the thermal response time constant, and the temperature rise process is constrained by the equivalent heat capacity of the battery cell.

[0026] By defining thermal safety indicators as the trend of the predicted core temperature of the battery cell approaching the safety threshold, rather than the temperature at a single moment, thermal safety assessment gains a forward-looking perspective. This approach can identify potential thermal runaway risks in advance and trigger control strategies before the battery cell reaches a dangerous temperature, buying valuable time for subsequent regulation and emergency response, thereby significantly improving the proactiveness and lead time of battery safety protection.

[0027] Preferably, S5 further includes:

[0028] S51. Using the battery equivalent internal resistance parameter and reversible thermal correlation coefficient parameter obtained in step S4, and estimating the rate stress correlation parameter based on the fused feature vector; wherein, the rate stress correlation parameter is at least related to the actual current, rated capacity and health status;

[0029] S52. Based on the battery equivalent internal resistance parameter, the reversible thermal correlation coefficient parameter, the rate stress correlation parameter, and the predicted value of the cell core temperature, a life consumption constraint relationship is constructed to obtain the degradation stress index.

[0030] S53. The degradation stress index is used to characterize the upward trend of the capacity decay rate or internal resistance growth rate under the current operating conditions and temperature.

[0031] By using the battery's equivalent internal resistance and reversible thermal-related characteristics employed in the thermal safety assessment phase in the degradation risk assessment, and further introducing the rate stress factor, the battery degradation assessment and thermal safety assessment are established on a consistent physical basis. This design avoids the problem of the safety risk assessment and life assessment being disconnected, enabling the risk assessment of capacity decay and internal resistance growth to truly reflect the comprehensive impact of actual operating conditions such as high-load operation and high-temperature environments on battery life.

[0032] Preferably, the determination of the degradation stress index satisfies the following limitations:

[0033] Within a preset time window, a cumulative lifespan consumption is constructed based on a temperature acceleration mechanism, wherein the temperature acceleration mechanism is at least characterized by an exponential growth trend in lifespan consumption caused by increased temperature; and a rate-sensitive mechanism is constructed based on rate stress-related parameters, wherein the rate-sensitive mechanism is at least characterized by a power-law growth trend in lifespan consumption caused by increased discharge or charge rates; the temperature acceleration mechanism and the rate-sensitive mechanism are cumulatively integrated or equivalently cumulative within the time window to obtain the degradation stress index; wherein the temperature input used for the temperature acceleration mechanism is taken from the predicted core temperature value of the battery cell obtained from the thermal balance constraint relationship, and the construction of the cumulative lifespan consumption explicitly uses the battery equivalent internal resistance parameter and the reversible thermal correlation coefficient parameter to ensure that the degradation stress index and the thermal safety index are at the same physical constraint scale.

[0034] By simultaneously considering both temperature-accelerated and rate-sensitive mechanisms, this method models battery life degradation over time, enabling the degradation stress index to reflect the different impacts of short-term high stress and long-term moderate stress on battery life. The beneficial effect is that, compared to assessment methods based solely on cycle count or average temperature, this method can more precisely characterize the actual impact of typical operating conditions in buses, such as high-frequency start-stop cycles and high-rate discharge, on battery life, thus improving the scientific rigor of lifespan management and maintenance decisions.

[0035] Preferably, S6 further includes:

[0036] S61. Construct a set of control variables, which includes at least the upper limit of charging current, the upper limit of discharging power, the acceleration constraint threshold, the thermal management trigger threshold, and the coordinated handling actions of vehicles or stations.

[0037] S62. Based on the set of control variables, perform feasibility prediction on the current trajectory and temperature trajectory in the future time domain to obtain the corresponding thermal safety index sequence and decay stress index sequence; S63. Under the condition of satisfying safety constraints and operational constraints, solve to obtain the final optimization set and output at least one corresponding control action.

[0038] By introducing a set of control variables and making feasibility predictions about the future evolution of current and temperature, battery safety management shifts from reactive response to proactive, prediction-based control. This approach can pre-select feasible combinations of control strategies while meeting both safety and operational constraints, avoiding new risks or impacts on operational efficiency caused by single control actions, thereby improving the overall coordination and feasibility of control decisions.

[0039] Preferably, the final optimized set is obtained by satisfying the following constraints:

[0040] Using the set of control variables as decision input, the deviation of the thermal safety index sequence is taken as the first optimization objective, and the cumulative consumption of the degradation stress index sequence is taken as the second optimization objective. The two are jointly minimized within a preset prediction time domain. At the same time, the following constraints are applied: the predicted value of the cell core temperature does not exceed the safety threshold temperature, the charging and discharging current does not exceed the allowable upper limit, and the state of charge is kept within the allowable range. Furthermore, operational feasibility constraints are applied to meet vehicle scheduling, timetable, and station resource limitations. Within the feasible domain that satisfies all constraints, the set of control variables that makes the above joint objective optimal or near optimal is obtained, which is taken as the final optimization set.

[0041] By taking thermal safety risk and degradation risk as joint optimization objectives and solving the final optimization set under the combined effect of safety constraints and operational constraints, the control strategy can achieve a balance between immediate safety and long-term lifespan.

[0042] To avoid accelerating battery aging in pursuit of short-term safety or reducing safety redundancy in pursuit of lifespan, we must achieve the best overall management and control results for the long-term stable operation of public transportation.

[0043] A multi-condition data fusion-based intelligent management and control system for the entire battery safety chain of electric buses, used to implement the aforementioned multi-condition data fusion-based intelligent management and control method for the entire battery safety chain of electric buses, includes:

[0044] The multi-condition data acquisition module is used to collect multi-condition data during the operation of the electric bus. The multi-condition data includes at least battery status data, driving condition data, environmental condition data, and charging status data.

[0045] The data preprocessing and heterogeneous alignment module is used to filter and denoise the multi-condition data, remove abnormal data, align the time and unify the units, so as to form a data sequence for fusion analysis.

[0046] A multimodal data fusion module is used to perform multi-source feature fusion on the data sequence to generate a unified fusion feature representation;

[0047] The thermal safety assessment module is used to construct battery thermal behavior constraints based on the fused feature representation, introduce battery equivalent internal resistance related parameters, reversible thermal related parameters and thermal response characteristic parameters, and output thermal safety indicators characterizing the battery thermal safety status.

[0048] The degradation risk assessment module is used to construct a battery life consumption constraint relationship by introducing rate stress parameters based on the battery equivalent internal resistance parameters and the reversible thermal parameters, and output a degradation stress index that characterizes the battery degradation risk.

[0049] The strategy optimization and control module is used to jointly optimize the battery operation strategy with the thermal safety index and the degradation stress index as the core constraints, and generate control instructions that meet the battery safety requirements and vehicle operation requirements.

[0050] The control commands include at least charging current control, discharge power control, driving behavior constraints, thermal management control, and vehicle or station collaborative handling commands.

[0051] By implementing the above methods in a systematic manner and dividing them into functional modules such as data acquisition, data preprocessing, fusion analysis, safety assessment, degradation assessment, and strategy optimization, the complex multi-condition battery safety management method has a clear system structure and engineering implementation path. This system can be flexibly deployed in vehicle-side, depot-side, or cloud environments, supporting collaborative operation of multiple vehicles and depots, and providing a scalable, maintainable, and implementable intelligent management platform for electric bus battery safety.

[0052] The technical solution of the present invention may include the following beneficial effects:

[0053] This invention constructs an intelligent management and control method and system for the entire battery safety chain, covering "monitoring-evaluation-decision-execution," by uniformly collecting, preprocessing, and fusing data from multiple operating conditions, including battery status, driving conditions, environmental conditions, and charging status. This overcomes the limitations of existing technologies that rely solely on single parameters or localized scenarios for safety monitoring. This technical solution enables comprehensive perception and control of battery safety risks under complex operating conditions such as high-frequency start-stop cycles, long-term operation, and multi-station charging in electric buses, significantly improving the completeness, stability, and applicability of battery safety management.

[0054] Based on fused data, parameters with clear physical meaning, such as battery equivalent internal resistance, reversible thermal characteristics, thermal response characteristics, and rate stress, are introduced to construct thermal safety assessment and degradation risk assessment mechanisms, respectively. Key physical parameters are used in both assessments, ensuring that thermal risk judgment and lifespan degradation judgment are based on consistent physical constraints. This avoids the problem of disconnect or contradiction between safety assessment and lifespan assessment, making battery thermal runaway risk identification more forward-looking, while making the assessment of capacity decay and internal resistance growth closer to real-world operating conditions, improving the accuracy and interpretability of the assessment results.

[0055] A joint optimization mechanism is constructed based on thermal safety assessment results and degradation risk assessment results. This mechanism satisfies battery safety constraints while also considering vehicle operational constraints. Through comprehensive optimization of charging and discharging strategies, driving behavior, thermal management, and coordinated handling by vehicles or depots, it generates directly executable control commands. This approach transforms battery safety management from a passive response to a predictive, proactive control method. It ensures immediate operational safety while slowing down battery aging, effectively reducing the risk of operational interruptions caused by battery failures. It demonstrates strong engineering feasibility and significant potential for widespread application. Attached Figure Description

[0056] Figure 1 This is a flowchart illustrating the overall workflow of an intelligent management and control method for the entire battery safety chain of electric buses based on multi-condition data fusion, as described in this invention.

[0057] Figure 2 This is a flowchart of the process of S4, a method for intelligent full-link management and control of electric bus battery safety based on multi-condition data fusion, according to the present invention.

[0058] Figure 3 This is a flowchart of the process of S5, a method for intelligent full-link control of electric bus battery safety based on multi-condition data fusion, according to the present invention.

[0059] Figure 4 This is a flowchart of the process of S6, a method for intelligent full-link management and control of electric bus battery safety based on multi-condition data fusion, according to the present invention.

[0060] Figure 5 This is a system block diagram of S4, an intelligent control system for the entire battery safety chain of electric buses based on multi-condition data fusion, according to the present invention. Detailed Implementation

[0061] Example 1

[0062] like Figure 1In this embodiment, multi-condition data is collected by both the vehicle-mounted and the site-based terminals and aggregated using a unique vehicle identifier and a unified timestamp. Vehicle-mounted data collection includes battery voltage, battery current, cell temperature, state of charge, health status, and fault codes output by the battery management system; vehicle communication / positioning and chassis signal collection includes driving speed, acceleration or deceleration, braking signals (used to statistically analyze braking frequency), climbing angle (which can be estimated by an inertial measurement unit or elevation difference), and operating mileage; environmental sensor data collection includes ambient temperature, relative humidity, vibration intensity, and light intensity; site-based data collection includes charging voltage, charging current, charging duration, number of charging cycles, and charging facility fault codes. The sampling frequencies used in this embodiment are: voltage / current / temperature 1 Hz (optional 2–10 Hz), vehicle speed / acceleration 1 Hz, ambient temperature, humidity, and light intensity 0.5 Hz (optional 0.1–1 Hz), vibration intensity 1 Hz, with the root mean square calculated using a 5-second sliding window; and charging voltage / current at 1 Hz during the charging phase. The short window length Δt for online rolling calculation is 10 s (optional 5–30 s), and the time domain H for policy optimization prediction is 300 s (optional 120–900 s).

[0063] In this embodiment, when preprocessing and heterogeneous alignment of multi-source data, time alignment is performed using a 1-second unified time grid; nearest neighbor or linear interpolation alignment is used for data from different frequencies; event-type fault codes are converted into 0 / 1 trigger identifiers or category IDs within the alignment window. Battery voltage and battery current are denoised using first-order low-pass or Kalman filtering, with the observation noise variance set according to sensor accuracy; abnormal driving speed values ​​are eliminated using a threshold; if the absolute value of the speed difference between two adjacent seconds is greater than 20 km / h, it is judged as abnormal and replaced with the previous time-stress hold or local median; abnormal temperature transitions are identified using a threshold. For short-term missing values ​​(e.g., <60 s), hold or interpolate. For long-term missing values ​​(e.g., continuous temperature missing value >30 s, continuous current missing value >10 s), trigger a degradation strategy: hold the most recent valid value and use conservative parameter boundaries (e.g., make the thermal response time constant larger or the heat dissipation term smaller to make the predicted temperature rise more conservative). At the same time, shrink the strategy optimization candidate set to a more conservative range of control variables to ensure that the online calculation is continuous and controllable.

[0064] In this embodiment, the aligned multi-condition data is constructed into an input sequence for fusion. A multi-source feature vector is generated per second, containing at least voltage, current, temperature, state of charge, health status, vehicle speed, acceleration, ambient temperature, humidity, vibration intensity, light intensity, charging voltage / current, and charging status indicators. Optional derived features include: current change rate, temperature change rate, braking frequency (10-second sliding window statistics), and fault code embedding vectors. The numerical features are then subjected to dimensional unification and normalization (optional standardization or Min-Max) to form the input tensor Z∈R.L×F Where L is the number of time steps within the window (in this embodiment, Δt = 10 s and sampling 1 Hz, L = 10), and F is the feature dimension (optional 30–120, depending on the number of temperature measurement points, whether to include individual voltage, etc.).

[0065] In this embodiment, the multimodal fusion model adopts an implementable multi-head attention temporal fusion structure, which is linearly mapped to the model dimension d. model (Optional 64–256), employing multi-head attention (number of heads optional 4–8) and feedforward network (hidden dimension optional 2 times d) model The system outputs a unified fused feature vector. This output vector serves as the fused feature representation for subsequent calculations of thermal safety and decay stress indices; optional attention contribution weights are retained to explain the contributions of features under different operating conditions.

[0066] In this embodiment, the fused feature vector serves as the input feature of the parameter estimation module. Based on the fused feature vector, the parameter estimation module outputs operating condition discrimination results and adaptive configuration parameters for parameter identification. Accordingly, it completes online estimation of battery equivalent internal resistance parameters, reversible thermal correlation coefficient parameters, thermal response time constant parameters, and rate stress correlation parameters within a time window that meets preset conditions such as current transition, static or steady-state heat dissipation. Optionally, the fused feature vector is also used to generate estimation confidence levels to select or weight and fuse multiple candidate estimation results.

[0067] In this embodiment, thermal safety index and degradation stress index are calculated based on fused feature vectors. In the thermal safety index calculation, the equivalent internal resistance parameter, reversible thermal correlation coefficient parameter, and thermal response time constant parameter of the battery are estimated online, and substituted into the thermal balance constraints to recursively obtain the predicted value of the cell core temperature within a future preset time window, thereby calculating the thermal safety index X. The degradation stress index calculation uses the battery equivalent internal resistance parameter and reversible thermal correlation coefficient parameter, and estimates the rate stress correlation parameter online. Combining the temperature acceleration mechanism and rate sensitivity mechanism, the degradation stress index Y is obtained by accumulating the lifespan consumption. Subsequently, a strategy optimization problem is constructed using X and Y as core variables, forming a final optimization set that satisfies both safety and operational constraints, and outputting at least one control action, including charging current limit, discharge power limit, driving behavior restriction, thermal management triggering, and vehicle scheduling or site collaborative handling. During the online inference stage, if the input is missing or the confidence level decreases, the candidate control set is shrunk according to a degradation strategy, and the safety constraint screening is enhanced to ensure that the output strategy is executable and does not exceed the safety boundary.

[0068] like Figure 2In this embodiment, when the current jumps or the operating conditions change, the equivalent internal resistance is estimated by the change in terminal voltage and the change in current; the reversible thermal coefficient is estimated based on the SOC range, temperature range and open circuit voltage temperature characteristics; the thermal time constant is estimated based on the cell temperature response to current and ambient temperature disturbances; the above parameters are substituted into the thermal balance constraint to calculate the thermal safety index, which characterizes the trend or margin of approaching the threshold.

[0069] In this embodiment, the online estimation of the battery's equivalent internal resistance parameter employs short-window identification triggered by current transitions or changes in operating conditions. When a preset current transition condition is met (e.g., |I(t)−I(t−1)|≥30A, with a threshold selectable from 10–80 A) or an operating condition change condition is met (e.g., sudden acceleration or braking trigger), the average voltage and average current values ​​are taken for 1–3 seconds before and after the transition. The ratio of the voltage change to the current change is calculated as an instantaneous estimate of the equivalent internal resistance, and an exponential moving average is used for updating to suppress noise. The smoothing coefficient can be selected from 0.05–0.30. The unit of this equivalent internal resistance parameter is ohms (or milliohms), used to characterize the Joule heating intensity generated by current passing through the battery. That is, under the same current, the larger the equivalent internal resistance, the stronger the Joule heating and the faster the temperature rise.

[0070] In this embodiment, the reversible thermal correlation coefficient parameter is used to characterize the reversible thermal effect caused by the entropy change of the electrochemical reaction. Two estimation methods are available: First, based on battery calibration data (e.g., the relationship between open-circuit voltage, state of charge, and temperature), the rate of change of open-circuit voltage with temperature is obtained from a table within the current state of charge range and converted into a reversible thermal correlation coefficient. Second, when the vehicle is stationary and the current is close to 0 (e.g., |I| < 5 A for > 60 s), the reversible thermal correlation coefficient is obtained by fitting the voltage-temperature change trend during the stationary period. This coefficient, along with current and temperature, is used to determine the magnitude and direction of the reversible thermal term, ensuring that the thermal balance constraint reflects the reversible thermal contribution under different temperatures and current directions.

[0071] In this embodiment, the thermal response time constant parameter is used to characterize the heat dissipation response speed between the battery cell and the environment, and can be estimated online from the dynamic characteristics of temperature regression to the environment. Optionally, the decay curve of the temperature difference between the battery cell and the ambient temperature is collected in a relatively stable operating condition (e.g., |I| < 20 A and vehicle speed changes little), and the time constant is obtained online by using exponential decay fitting or recursive least squares update; its typical range can be from 10 s to 2000 s, depending on the battery pack structure and cooling conditions. A larger time constant indicates slower heat dissipation and more significant short-term temperature rise.

[0072] In this embodiment, the battery equivalent internal resistance parameter, reversible thermal correlation coefficient parameter and thermal response time constant parameter are substituted into the thermal balance constraint relationship to recursively obtain the predicted value of the cell core temperature within a future preset time window. Based on the relationship between the predicted value and the safety threshold, the thermal safety index is calculated to characterize the trend strength or time margin of the cell temperature approaching the safety threshold under a given operating condition.

[0073] Specifically, the thermal safety index X satisfies the following equation:

[0074]

[0075] in, It is derived recursively from the following heat balance equation:

[0076]

[0077] To predict the core temperature of the battery cell at a given time; T safe Δt is the preset cell safety threshold temperature; Δt is the prediction time window length; t is the current time; T(t) is the cell measurement temperature at the current time; I(t) is the battery operating current at the current time; a is the battery equivalent internal resistance parameter, used to characterize the Joule heating effect caused by current; b is the reversible thermal correlation coefficient parameter, used to characterize the reversible thermal effect related to temperature and current; C th The equivalent heat capacity of the battery cell is used to characterize the heat absorbed per unit temperature rise; c is the thermal response time constant parameter, used to characterize the heat dissipation response characteristics between the battery cell and the environment; T amb (t) represents the ambient temperature at the current moment.

[0078] like Figure 3 In this embodiment, equivalent internal resistance and reversible thermal coefficient are used, and the rate stress parameters related to current, rated capacity and health status are estimated by combining fused feature vectors. Based on the equivalent internal resistance, reversible thermal coefficient, rate stress parameters and core temperature prediction, a lifetime consumption constraint is constructed to obtain the degradation stress index. The degradation stress index is used to characterize the upward trend of capacity decay or internal resistance growth under the current operating conditions and temperature.

[0079] Specifically, the degradation stress index Y satisfies the following equation:

[0080]

[0081] in, ;and satisfy Constraint embedding for the temperature rise term;

[0082] Y is the degradation stress index, used to characterize the cumulative intensity of battery life depletion within a preset time window; k0 is the life depletion pre-exponential factor, used to characterize the base degradation rate; E a Apparent activation energy parameter characterizing battery degradation mechanism; R g Gas constant; γ is the rate sensitivity coefficient, used to characterize the impact of rate changes on battery life; τ is the integration time variable; [t, t+Δt] is the time window for cumulative battery life; d(τ) is the rate stress-related parameter, used to characterize the equivalent rate load borne by the battery at time τ; I(τ) is the battery operating current at time τ; Q nom τ represents the battery's rated capacity; SOH(τ) represents the battery's state of health at time τ.

[0083] like Figure 4 This embodiment constructs a set of control variables including the upper limit of charging current, the upper limit of discharging power, the acceleration threshold, the thermal management trigger threshold, and the coordinated handling actions; based on this set, it predicts the future current and temperature trajectories, and generates a thermal safety index sequence and a degradation stress index sequence; under safety and operational constraints, it solves the final optimization set and outputs the corresponding control actions.

[0084] Specifically, the final optimized set Ω * The following optimization problem was solved:

[0085]

[0086] Satisfy constraints:

[0087] ,

[0088] And operational constraints:

[0089]

[0090] Ω * The final optimization set is used to represent the set of optimal or near-optimal control strategies that satisfy safety and operational constraints.

[0091] u is a set of control variables, including at least the upper limit of charging current, the upper limit of discharging power, driving behavior constraints, and thermal management triggering strategies;

[0092] Ω represents the feasible region of the control variable;

[0093] X(u,τ) is the thermal safety index under the control variable u;

[0094] Y(u,τ) is the decay stress index under the control variable u;

[0095] H represents the prediction time domain length;

[0096] λ is the trade-off coefficient used to balance thermal safety objectives with lifespan degradation objectives;

[0097] argmin represents the set of control variables that minimizes the objective function;

[0098] This is the predicted core temperature of the battery cell under the influence of the control variable u;

[0099] T safe This refers to the safe threshold temperature for the battery cell.

[0100] I(u,τ) is the battery operating current under the control variable u;

[0101] I max The maximum allowable charge and discharge current;

[0102] SOC(u,τ) is the state of charge of the battery under the control variable u.

[0103] g(u) is the operational feasibility constraint function, which describes the restrictions imposed on control variables by operational conditions such as vehicle scheduling, departure intervals, and depot resources.

[0104] In this embodiment, the set of control variables u for strategy optimization includes at least: upper limit of charging current, upper limit of discharge power, acceleration constraint threshold, thermal management trigger threshold, and coordinated handling actions of vehicles or depots. To ensure feasibility, the value range of the control variables can be set according to the capabilities of the vehicle and depot, for example, upper limit of charging current 50–250A, upper limit of discharge power 50–300kW, acceleration constraint threshold 0.5–2.0 m / s², and thermal management trigger threshold 40–60℃; the coordinated handling actions of vehicles or depots can be selected as actions such as parking nearby, returning to the depot for maintenance, switching charging piles, and dispatching backup vehicles (executed by the dispatching system). When generating the candidate control variable set, the above continuous interval can be discretized into several grids (for example, each variable takes 5–10 candidate values), combined to form a candidate strategy set, which is used for subsequent screening and solving of the final optimization set.

[0105] In this embodiment, given a candidate control variable u, the feasibility of predicting the current trajectory and temperature trajectory in the future time domain is performed. If the device is in a charging state, the current trajectory is constrained by the upper limit of the charging current; if it is in an operating state, the current can be estimated from the required power and battery voltage, and is jointly limited by the upper limit of the discharge power and the acceleration constraint threshold (the acceleration threshold affects the power demand, thus affecting the current). Subsequently, the predicted current trajectory is substituted into the thermal balance constraint to recursively obtain the core temperature prediction sequence in the future time domain, and the corresponding thermal safety index sequence and degradation stress index sequence are formed accordingly. For each candidate u, the objective is calculated under the condition of satisfying the safety constraints and operational constraints, and the optimal or near-optimal set of control variables is selected as the final optimization set; the preferred strategy in the set is converted into a control action and issued for execution. If the execution fails or communication is abnormal, the suboptimal strategy in the set is downgraded and switched to ensure executability and continuity.

[0106] The reproducible method for obtaining λ is as follows: First, the thermal safety target and the degradation target are scaled and normalized by dividing them by the median or 95th percentile of the corresponding indicators in historical data to make the two numerically comparable; second, λ=1 is taken as the initial value after normalization; third, a grid search is performed based on historical playback data (e.g., λ∈{0.1,0.3,1,3,10}, using the Pareto compromise point with the fewest safety constraint violations and the smallest cumulative degradation as the selection criterion to determine the final λ and solidify it into the parameter package; if seasonal adaptation is required, λ can be adjusted to a safer value during the high-temperature season (equivalent to reducing the weight of the lifetime term or strengthening the safety term), but version management is still performed using a fixed grid and calibration criteria.

[0107] like Figure 5In this embodiment, the vehicle-side includes a BMS, an onboard communication and positioning unit, environmental and vibration sensors, and an execution controller; the station-side includes a charging facility data acquisition interface and a station environment acquisition interface; the edge-side deploys real-time inference and strategy optimization services; and the cloud is used for long-term storage, model training / calibration, version management, and auditing. The multi-condition data acquisition module acquires vehicle-side data via CAN / Ethernet, acquires charging data via charging facility protocols or manufacturer interfaces, and outputs a unified message format; the data preprocessing and heterogeneous alignment module subscribes to the acquired data and performs filtering, anomaly removal, missing data handling, time alignment, and normalization, outputting a sequence tensor for fusion; the multimodal data fusion module loads the fixed fusion model weights (which can be inference formats such as ONNX / TensorRT), inputs the sequence, and outputs a fusion feature vector; the thermal safety assessment module estimates the battery's equivalent internal resistance parameters, reversible thermal correlation coefficient parameters, and thermal response time constant parameters online, and calculates the thermal safety index X by recursively extrapolating the core temperature prediction value according to thermal balance constraints; and the decay... The degradation risk assessment module uses equivalent internal resistance parameters and reversible thermal correlation coefficient parameters, and calculates rate stress-related parameters online. It then discretizes and accumulates these parameters using temperature acceleration and rate sensitivity models to obtain the degradation stress index Y. The strategy optimization and control module generates a set of candidate control variables, predicts the X / Y sequence over the next H seconds, and obtains the final optimized set by constraint filtering and objective function solving. The preferred strategy within the set is converted into control commands and sent to the vehicle controller or driver terminal and dispatch platform. These control commands include at least charging current control, discharging power control, driving behavior constraints, thermal management control, and vehicle or depot collaborative handling commands. Execution receipts and key status feedback are used for the next cycle of rolling solution to update initial values. During the training / calibration phase, the cloud generates a model weight parameter package, which includes at least equivalent heat capacity, temperature acceleration parameters, rate sensitivity parameters, tradeoff coefficients, and thresholds. The edge device loads these parameters according to version and supports rollback, ensuring continuous online execution without external network access.

Claims

1. A multi-working-condition data fusion-based intelligent management and control method for the whole link of electric bus battery safety, characterized in that, include: S1. Obtain multi-condition data of the electric bus, wherein the multi-condition data includes at least battery status data, driving condition data, environmental condition data, and charging status data; wherein, the battery status data includes at least battery voltage, battery current, cell temperature, state of charge, health status, and fault codes; the driving condition data includes at least driving speed, acceleration or deceleration, braking frequency, climbing angle, and operating mileage; the environmental condition data includes at least ambient temperature, relative humidity, vibration intensity, and light intensity; and the charging status data includes at least charging voltage, charging current, charging time, number of charging cycles, and charging facility fault codes. S2. Preprocessing and heterogeneous alignment of the multi-condition data includes at least: filtering and denoising the battery current and battery voltage, removing outliers in driving speed, aligning the timestamps of the multi-source data and normalizing the dimensions to obtain the fused input sequence. S3. Based on the multimodal fusion model, perform deep fusion on the fused input sequence and output a unified fusion feature vector; S4. Based on the fused feature vector, introduce thermal-electrophysical constraint parameters and calculate thermal safety indicators: online estimate the battery equivalent internal resistance parameter, reversible thermal correlation coefficient parameter and thermal response time constant parameter, and construct thermal balance constraint relationship based on the battery equivalent internal resistance parameter, the reversible thermal correlation coefficient parameter and the thermal response time constant parameter, and output thermal safety indicators characterizing the battery thermal runaway driving intensity and safety margin. S5. Based on the fused feature vector, introduce degradation-temperature physical constraint parameters and calculate degradation stress index: use the battery equivalent internal resistance parameter and the reversible thermal correlation coefficient parameter, and estimate the rate stress correlation parameter online to construct a life consumption constraint relationship related to temperature, and output a degradation stress index characterizing the risk of capacity decay and internal resistance growth. S6. Construct a strategy optimization problem using the thermal safety index and the degradation stress index as core variables, and output a final optimization set that satisfies safety constraints and operational constraints, which is used to generate at least one control action; the control action includes at least charging current limit, discharge power limit, driving behavior limit, thermal management trigger, and vehicle scheduling or site collaborative handling.

2. The method for intelligent full-chain control of electric bus battery safety based on multi-condition data fusion according to claim 1, characterized in that, S4 further includes: S41. When the preset current transition or operating condition change conditions are met, the equivalent internal resistance parameter of the battery is estimated based on the terminal voltage change and current change. S42. Estimate the reversible thermal correlation coefficient parameters based on the relationship between the state of charge range, temperature range and open-circuit voltage as a function of temperature. S43. Estimate the thermal response time constant parameter based on the response characteristics of cell temperature to current disturbance and ambient temperature disturbance; S44. Substitute the battery equivalent internal resistance parameter, the reversible thermal correlation coefficient parameter, and the thermal response time constant parameter into the thermal balance constraint relationship to obtain the thermal safety index; wherein, the thermal safety index is used to characterize the trend strength or time margin of the cell temperature approaching the safety threshold under a given operating condition.

3. The method for intelligent full-chain control of electric bus battery safety based on multi-condition data fusion according to claim 2, characterized in that, The determination of the thermal safety index satisfies the following constraints: The difference between the predicted core temperature of the battery cell within a future preset time window and the preset safety threshold temperature is used as the thermal safety deviation, and the rate of change of the thermal safety deviation over time or the equivalent approximation strength is used as the thermal safety index. The predicted core temperature of the battery cell is obtained recursively based on the thermal balance constraint relationship, which includes at least: the Joule heating term caused by the equivalent internal resistance of the battery, the reversible heat term caused by the reversible thermal correlation coefficient, and the heat dissipation to the environment characterized by the thermal response time constant, and the temperature rise process is constrained by the equivalent heat capacity of the battery cell.

4. The method for intelligent full-chain control of electric bus battery safety based on multi-condition data fusion according to claim 3, characterized in that, The S5 also includes: S51. Using the battery equivalent internal resistance parameter and reversible thermal correlation coefficient parameter obtained in step S4, and estimating the rate stress correlation parameter based on the fused feature vector; wherein, the rate stress correlation parameter is at least related to the actual current, rated capacity and health status; S52. Based on the battery equivalent internal resistance parameter, the reversible thermal correlation coefficient parameter, the rate stress correlation parameter, and the predicted core temperature of the cell, a life consumption constraint relationship is constructed to obtain the degradation stress index. S53. The degradation stress index is used to characterize the upward trend of the capacity decay rate or internal resistance growth rate under the current operating conditions and temperature.

5. The method and system for intelligent full-link management and control of electric bus battery safety based on multi-condition data fusion as described in claim 4, characterized in that, The determination of the degradation stress index satisfies the following constraints: Within a preset time window, a cumulative lifespan consumption is constructed based on a temperature acceleration mechanism, wherein the temperature acceleration mechanism is at least characterized by an exponential growth trend in lifespan consumption caused by increased temperature; and a rate-sensitive mechanism is constructed based on rate stress-related parameters, wherein the rate-sensitive mechanism is at least characterized by a power-law growth trend in lifespan consumption caused by increased discharge or charge rates; the temperature acceleration mechanism and the rate-sensitive mechanism are cumulatively integrated or equivalently cumulative within the time window to obtain the degradation stress index; wherein the temperature input used for the temperature acceleration mechanism is taken from the predicted core temperature of the battery cell obtained from the thermal balance constraint relationship, and the equivalent internal resistance parameter and reversible thermal correlation coefficient parameter of the battery are explicitly used in the construction of the cumulative lifespan consumption to ensure that the degradation stress index and the thermal safety index are at the same physical constraint scale.

6. The method and system for intelligent full-link management and control of electric bus battery safety based on multi-condition data fusion as described in claim 5, characterized in that, S6 further includes: S61. Construct a set of control variables, which includes at least the upper limit of charging current, the upper limit of discharging power, the acceleration constraint threshold, the thermal management trigger threshold, and the coordinated handling actions of vehicles or stations. S62. Based on the set of control variables, perform feasibility prediction on the current trajectory and temperature trajectory in the future time domain to obtain the corresponding thermal safety index sequence and decay stress index sequence; S63. Under the condition of satisfying safety constraints and operational constraints, solve to obtain the final optimization set and output at least one corresponding control action.

7. The method and system for intelligent full-link management and control of electric bus battery safety based on multi-condition data fusion as described in claim 6, characterized in that, The final optimized set is obtained while satisfying the following constraints: Using the set of control variables as decision input, the deviation of the thermal safety index sequence is taken as the first optimization objective, and the cumulative consumption of the degradation stress index sequence is taken as the second optimization objective. The two are jointly minimized within a preset prediction time domain. At the same time, the following constraints are applied: the predicted value of the cell core temperature does not exceed the safety threshold temperature, the charging and discharging current does not exceed the allowable upper limit, and the state of charge is kept within the allowable range. Furthermore, operational feasibility constraints are applied to meet vehicle scheduling, timetable, and station resource limitations. Within the feasible domain that satisfies all constraints, the set of control variables that makes the above joint objective optimal or near optimal is obtained, which is taken as the final optimization set.

8. A multi-working-condition data fusion-based intelligent whole-link management and control system for battery safety of an electric bus, characterized in that, The method for implementing the intelligent full-chain control of electric bus battery safety based on multi-condition data fusion as described in any one of claims 1-7 includes: The multi-condition data acquisition module is used to collect multi-condition data during the operation of the electric bus. The multi-condition data includes at least battery status data, driving condition data, environmental condition data, and charging status data. The data preprocessing and heterogeneous alignment module is used to filter and denoise the multi-condition data, remove abnormal data, align the time and unify the units, so as to form a data sequence for fusion analysis. A multimodal data fusion module is used to perform multi-source feature fusion on the data sequence to generate a unified fusion feature representation; The thermal safety assessment module is used to construct battery thermal behavior constraints based on the fused feature representation, introduce battery equivalent internal resistance related parameters, reversible thermal related parameters and thermal response characteristic parameters, and output thermal safety indicators characterizing the battery thermal safety status. The degradation risk assessment module is used to construct a battery life consumption constraint relationship by introducing rate stress parameters based on the battery equivalent internal resistance parameters and the reversible thermal parameters, and output a degradation stress index that characterizes the battery degradation risk. The strategy optimization and control module is used to jointly optimize the battery operation strategy with the thermal safety index and the degradation stress index as the core constraints, and generate control instructions that meet the battery safety requirements and vehicle operation requirements. The control commands include at least charging current control, discharge power control, driving behavior constraints, thermal management control, and vehicle or station collaborative handling commands.

Images