Battery emptying time prediction method and system based on working condition learning

Abstract

The embodiment of the application discloses a battery emptying time prediction method and system based on working condition learning; the method comprises the following steps: collecting real-time working condition characteristic parameters; matching a target working condition cluster and a target prediction model according to the real-time working condition characteristic parameters; inputting the real-time working condition characteristic parameters into the target prediction model to output the remaining emptying time of the battery under the current working condition. The embodiment of the application has the following advantages: the prediction deviation problem caused by the fixed discharge rate can be solved; the problem of insufficient working condition characteristic fusion and weak generalization ability can be solved; the problem of lacking a working condition self-adaptive learning mechanism can be solved; the contradiction between real-time performance and accuracy under dynamic working conditions can be solved; the precision of battery emptying time prediction can be significantly improved; the dynamic working condition adaptability is strong; the model iteration mechanism adapts to the whole life cycle of the battery, and the problem that the traditional model becomes less and less accurate with use can be solved; no additional sensors are needed, the computing power consumption is low, and the engineering practicability is outstanding.

Description

Technical Field

[0001] This invention relates to the field of battery management technology, and specifically to a method and system for predicting battery discharge time based on operating condition learning. Background Technology

[0002] Existing battery discharge time prediction methods have the following drawbacks:

[0003] 1. Fixed discharge rate assumption leads to large prediction deviation: Existing battery discharge time prediction is mostly based on the "fixed discharge rate" model (such as calculated by rated capacity ÷ current discharge current). However, in actual applications, the discharge current changes dynamically with the operating conditions (such as vehicle acceleration / deceleration, energy storage load fluctuation). The fixed rate assumption cannot adapt to the real operating conditions, resulting in a prediction error of more than 15%.

[0004] 2. Insufficient integration of operating condition characteristics and weak generalization ability: Existing technologies only consider discharge current or temperature parameters and have not formed a multi-dimensional operating condition characteristic system of "average discharge current + temperature + aging degree". They cannot cover complex scenarios such as low temperature, high rate, and battery aging, and the prediction results have low reliability under extreme operating conditions.

[0005] 3. Lack of adaptive learning mechanism for operating conditions: The existing prediction model has fixed parameters and has not established a mapping relationship between operating conditions and discharge time through offline learning. It cannot dynamically adapt to parameter drift (such as SOH decrease) and operating condition changes throughout the battery's entire life cycle, and the prediction accuracy gradually decreases with the use time.

[0006] 4. Conflict between real-time performance and accuracy under dynamic operating conditions: During dynamic charging and discharging, battery parameters fluctuate frequently. Traditional prediction methods rely on long-term data accumulation or complex calculations, resulting in a lag in prediction response (delay > 30s), which cannot meet the user's need to obtain the remaining discharge time in real time (such as vehicle range estimation and energy storage load scheduling). Summary of the Invention

[0007] In view of the deficiencies of the prior art mentioned in the background, the purpose of this invention is to provide a method and system for predicting battery discharge time based on operating condition learning.

[0008] To achieve the above objectives, in a first aspect, embodiments of the present invention provide a battery discharge time prediction method based on operating condition learning, comprising:

[0009] Collect real-time operating condition characteristic parameters; the real-time operating condition characteristic parameters include average discharge current, average temperature, battery aging degree and current state of charge;

[0010] The target operating condition cluster is matched based on the real-time operating condition feature parameters;

[0011] A target prediction model is matched based on the target working condition cluster;

[0012] The real-time operating condition characteristic parameters are input into the target prediction model, and the remaining discharge time of the battery under the current operating condition is output.

[0013] As a preferred implementation of this application, before collecting real-time operating condition feature parameters, the method further includes an offline training phase, specifically:

[0014] Under different operating conditions, the battery was subjected to a full charge-discharge cycle test, and the test data was recorded. The test data included average discharge current, average temperature, battery aging degree, initial state of charge, and actual time T_meas from discharge to cutoff voltage.

[0015] The average discharge current, average temperature, battery aging degree and initial state of charge are clustered using the K-means clustering algorithm, and the optimal number of clusters N is determined by the elbow rule, resulting in N operating condition clusters.

[0016] For each operating condition cluster, the average discharge current, average temperature, battery aging degree, and initial state of charge are used as input variables, and the actual time T_meas is used as the output variable. A lightweight machine learning algorithm is used to fit the model to obtain the battery discharge time prediction model corresponding to each operating condition cluster.

[0017] As a specific implementation of this application, the target operating condition cluster is matched based on the real-time operating condition feature parameters, specifically as follows:

[0018] Based on the real-time operating condition characteristic parameters, calculate the Euclidean distance to the center of each operating condition cluster;

[0019] The cluster of operating conditions with the smallest Euclidean distance is taken as the target cluster of operating conditions.

[0020] As a preferred implementation of this application, before outputting the remaining discharge time of the battery under the current operating condition, the method further includes:

[0021] The process of real-time operating condition acquisition, operating condition cluster matching, and venting time calculation is repeated to update the predicted venting time based on the latest operating condition parameters.

[0022] As a preferred implementation of this application, the method further includes an offline training update phase, which includes re-collecting the test data, re-dividing the operating condition clusters, and updating the parameters of the battery discharge time prediction model.

[0023] Secondly, embodiments of this application also provide a battery discharge time prediction system based on operating condition learning, comprising:

[0024] The data acquisition module, integrated into the BMS, is used to collect real-time operating condition characteristic parameters, including average discharge current, average temperature, battery aging degree, and current state of charge.

[0025] The online matching module is used to match the target working condition cluster based on the real-time working condition feature parameters, and to match the target prediction model based on the target working condition cluster.

[0026] The discharge time calculation module is used to input the real-time operating condition characteristic parameters into the target prediction model and output the remaining discharge time of the battery under the current operating condition.

[0027] Furthermore, the data acquisition module is also used to: perform full charge-discharge cycle tests on the battery under different operating conditions and record test data; the test data includes average discharge current, average temperature, battery aging degree, initial state of charge, and actual time T_meas from discharge to cutoff voltage;

[0028] The system also includes an offline training module for dividing the operating condition clusters and building the prediction model, specifically:

[0029] The average discharge current, average temperature, battery aging degree and initial state of charge are clustered using the K-means clustering algorithm, and the optimal number of clusters N is determined by the elbow rule, resulting in N operating condition clusters.

[0030] For each operating condition cluster, the average discharge current, average temperature, battery aging degree, and initial state of charge are used as input variables, and the actual time T_meas and measured temperature are used as output variables. A lightweight machine learning algorithm is used to fit the model to obtain the battery discharge time prediction model corresponding to each operating condition cluster.

[0031] As one specific implementation of this application, the online matching module is specifically used for:

[0032] Based on the real-time operating condition characteristic parameters, calculate the Euclidean distance to the center of each operating condition cluster;

[0033] The cluster of operating conditions with the smallest Euclidean distance is taken as the target cluster of operating conditions.

[0034] As a preferred implementation of this application, the system further includes a model iteration module, used for:

[0035] Based on the new full charge-full discharge cycle data, the operating condition cluster division and prediction model parameters are optimized periodically.

[0036] Thirdly, embodiments of the present invention also provide another battery discharge time prediction system based on operating condition learning, including a processor, an input device, an output device, and a memory, wherein the processor, input device, output device, and memory are interconnected, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the method of the first aspect described above.

[0037] The advantages of implementing the embodiments of the present invention are as follows:

[0038] 1. It can solve the prediction deviation problem caused by a fixed discharge rate;

[0039] 2. It can solve the problems of insufficient fusion of working condition features and weak generalization ability;

[0040] 3. It can solve the problem of lacking an adaptive learning mechanism for working conditions;

[0041] 4. It can resolve the conflict between real-time performance and accuracy under dynamic operating conditions;

[0042] 5. It can significantly improve the accuracy of battery discharge time prediction;

[0043] 6. Strong adaptability to dynamic operating conditions;

[0044] 7. The model iteration mechanism adapts to the entire battery lifecycle, which can solve the problem of traditional models becoming less accurate with use;

[0045] 8. No new sensors are required, and the computing power consumption is low, making it highly practical for engineering applications. Attached Figure Description

[0046] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below.

[0047] Figure 1 This is a flowchart of the battery discharge time prediction method based on operating condition learning provided in an embodiment of the present invention;

[0048] Figure 2 and Figure 3 This is a flowchart of the offline training phase;

[0049] Figure 4 and Figure 5 This is a flowchart of the online operating condition identification phase;

[0050] Figure 6 This is a structural diagram of the battery discharge time prediction system based on operating condition learning provided in an embodiment of the present invention;

[0051] Figure 7 yes Figure 6 Another structural diagram of the system shown. Detailed Implementation

[0052] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0054] This invention provides a battery discharge time prediction method and system based on operating condition learning. The core of the method is to learn a discharge time prediction model specific to each operating condition by dividing the operating conditions into multiple dimensions (average discharge current + temperature + state of equilibrium, SOH), thereby achieving accurate and real-time prediction under dynamic operating conditions. The relevant core definitions involved in this technical solution are as follows:

[0055] 1. Operating condition characteristic parameters: including average discharge current (I_avg, calculated by 10-minute moving average), average temperature (T_avg, calculated by 5-minute average), battery aging degree (SOH), and current state of charge (SOC_current), which constitute a four-dimensional operating condition characteristic system;

[0056] 2. Working condition clusters: Scenes with similar working condition features are grouped together using a clustering algorithm. The optimal number of clusters is determined by the elbow rule combined with the profile coefficient (preferably 10-15).

[0057] 3. Operating condition-specific discharge time prediction model: The prediction model corresponding to each operating condition cluster takes I_avg, T_avg, SOH, and SOC_current as inputs and outputs the remaining time for the battery to discharge from the current SOC to the preset cutoff voltage (e.g., 2.7V / 3.0V) under that operating condition.

[0058] 4. Discharge time: The cumulative time (in minutes / hour) during which the battery continues to discharge from its current state of charge (SOC) under current operating conditions until the voltage drops to the preset cutoff voltage.

[0059] Please refer to Figure 1 This invention provides a battery discharge time prediction method based on operating condition learning, which mainly includes two parts: an offline training stage and an online prediction stage.

[0060] Among them, such as Figure 2 and Figure 3 As shown, the offline training phase includes:

[0061] S101 collects data under multiple operating conditions.

[0062] The battery was subjected to full charge-discharge cycle tests under different operating conditions, covering the following test conditions:

[0063] - Temperature range: -20℃ to 60℃ (in increments of 5℃);

[0064] - Average discharge current range: 0.2C~3C (each 0.3C increment);

[0065] - SOH range: 70%~100% (in increments of 5%)

[0066] - Initial SOC range: 30%~100% (each 10% represents a gradient);

[0067] Record I_avg, T_avg, SOH, SOC_initial (initial SOC), and the actual time from discharge to cutoff voltage (T_meas) for each test group.

[0068] S102, cluster the collected data to obtain multiple operating condition clusters.

[0069] The I_avg, T_avg, and SOH collected in step S101 are used as the core clustering features and input into the K-means algorithm. The optimal number of clusters N (N=10-15) is determined by the elbow rule, and N working condition clusters are obtained to ensure that the similarity of working condition features within each working condition cluster is high and the differences between clusters are significant.

[0070] S103 constructs multiple dedicated prediction models based on multiple operating condition clusters.

[0071] For each operating condition cluster, with I_avg, T_avg, SOH, and SOC_current as input variables and T_meas as output variable, a lightweight machine learning algorithm is used to fit the prediction model, with the multiple linear regression model being the preferred choice (balancing accuracy and real-time performance). The model expression is as follows:

[0072] Tpred = a * I_avg + b * T_avg + c * SOH + d * SOC_current + e

[0073] Where a, b, c, d, and e are the specific fitting coefficients for this working condition cluster, and the model parameters of all working condition clusters are stored in the model library of the offline training module.

[0074] Please refer to this again. Figure 4 and Figure 5 The online prediction phase includes:

[0075] S201 collects real-time operating condition characteristic parameters.

[0076] In practice, the system continuously collects battery current, temperature, and voltage data to calculate the current I_avg (10-minute moving average) and T_avg (5-minute average). Simultaneously, it acquires the battery's SOH and SOC_current (obtained through ampere-hour integration and OCV correction). In other words, the collected real-time operating condition characteristics include average discharge current, average temperature, battery aging level, and current state of charge.

[0077] S202, Match the target operating condition cluster according to the real-time operating condition feature parameters.

[0078] In practice, the current four-dimensional working condition feature parameters (I_avg, T_avg, SOH, SOC_current) are extracted, the Euclidean distance between the parameter and the center of each working condition cluster is calculated, and the working condition cluster with the smallest distance is matched (matching time < 5ms) and used as the target working condition cluster.

[0079] S203, Match the target prediction model according to the target working condition cluster.

[0080] S204, input the real-time operating condition characteristic parameters into the target prediction model, and output the remaining discharge time of the battery under the current operating condition.

[0081] In practice, the dedicated prediction model for the matching working condition cluster is called, and the current I_avg, T_avg, SOH, and SOC_current data are substituted to calculate the remaining venting time (T_pred) under the current working condition.

[0082] S205, the prediction results are updated dynamically.

[0083] For example, steps S201-S204 are repeated every 30 seconds, that is, the real-time operating condition acquisition, operating condition cluster matching and venting time calculation process are repeatedly executed, and the venting time prediction results are updated in combination with the latest operating condition parameters to ensure the real-time prediction under dynamic operating conditions.

[0084] S206, Model Iterative Optimization.

[0085] For example, after every 8 full charge-discharge cycles, the historical data of the data acquisition module is updated, the offline training module re-optimizes the operating condition cluster division and prediction model parameters, compensates for parameter drift caused by battery aging, and continuously improves prediction accuracy.

[0086] The advantages of implementing the battery discharge time prediction method based on operating condition learning provided in this embodiment of the invention are as follows:

[0087] 1. It can solve the prediction deviation problem caused by a fixed discharge rate.

[0088] By dividing operating conditions into multiple dimensions, dynamically changing operating conditions are classified into clusters with similar characteristics, and each cluster corresponds to a dedicated prediction model. The model learns the discharge time pattern based on actual characteristics such as average discharge current and temperature under the operating condition, replacing the traditional fixed rate assumption. For example, under high rate operating conditions, the model automatically corrects the current coefficient to avoid "overestimating the discharge time". The core logic is "operating condition adapts to the model, and the model accurately predicts the time".

[0089] 2. It can solve the problems of insufficient fusion of working condition characteristics and weak generalization ability.

[0090] A four-dimensional operating condition feature system was constructed, comprising "average discharge current + temperature + SOH + SOC_current," covering a temperature range of -20℃ to 60℃, a current range of 0.2C to 3C, and a SOH range of 70% to 100%. K-means clustering was used to achieve full operating condition coverage. The model for each operating condition cluster was specifically fitted to the discharge time pattern under that scenario, significantly improving generalization ability compared to single-parameter prediction.

[0091] 3. It can solve the problem of lacking an adaptive learning mechanism for working conditions.

[0092] The design employs a closed-loop mechanism of "offline training + online matching + iterative optimization". In the offline phase, an initial set of operating conditions and the model are established. In the online phase, the operating conditions are matched and the model is called in real time. After every 8 full charge-discharge cycles, the model parameters are automatically iterated and optimized to adapt to battery aging (SOH decrease) and changes in operating conditions, ensuring stable prediction accuracy throughout the entire life cycle.

[0093] 4. It can resolve the contradiction between real-time performance and accuracy under dynamic operating conditions.

[0094] The offline phase completes the division of working conditions and model training. In the online phase, it only needs to quickly match the working condition clusters through Euclidean distance (time < 5ms). The prediction model adopts a lightweight multiple linear regression algorithm (computation time < 1ms), combined with a dynamic update mechanism every 30 seconds, which balances prediction accuracy and real-time performance, with a response latency ≤ 30s, meeting the real-time requirements under dynamic working conditions.

[0095] 1. Prediction accuracy has been significantly improved.

[0096] Under complex operating conditions, the prediction error of venting time has been reduced from more than 15% in traditional solutions to less than 5%. Among them, the error in low temperature (-10℃) and high rate (2.5C) operating conditions has been reduced by more than 65%, meeting the high-precision requirements of electric vehicle range estimation and energy storage load scheduling.

[0097] 2. Strong adaptability to dynamic working conditions.

[0098] It supports full operating condition coverage of 0.2C~3C discharge current and -20℃~60℃ temperature range, adapts to dynamic scenarios such as vehicle acceleration and deceleration and energy storage load fluctuations, and the prediction results are updated in real time with changes in operating conditions without lag.

[0099] 3. Stable and reliable throughout the entire life cycle.

[0100] The model iteration mechanism is adapted to the entire battery life cycle (SOH 70%~100%), and the prediction accuracy decreases by ≤3% during the aging process, solving the problem of traditional models becoming less and less accurate with use.

[0101] 4. The project has outstanding practicality.

[0102] No new sensors are needed; the existing voltage, current, and temperature data acquisition capabilities of the BMS can be used, resulting in low modification costs. The lightweight algorithm is compatible with embedded BMS systems, consumes little computing power (occupying less than 10% of the BMS's idle computing power), and is easy to industrialize.

[0103] Based on the same inventive concept, embodiments of the present invention provide a battery discharge time prediction system based on operating condition learning, comprising:

[0104] The data acquisition module, integrated into the BMS, is used to collect real-time operating condition characteristic parameters, including average discharge current, average temperature, battery aging degree, and current state of charge.

[0105] Offline training module, used for condition cluster partitioning and prediction model construction;

[0106] The online matching module is used to match the target working condition cluster based on the real-time working condition feature parameters, and to match the target prediction model based on the target working condition cluster.

[0107] The discharge time calculation module is used to input the real-time operating condition characteristic parameters into the target prediction model and output the remaining discharge time of the battery under the current operating condition.

[0108] The model iteration module is used to periodically optimize the operating condition cluster division and prediction model parameters based on new full charge-full discharge cycle data.

[0109] In specific implementation, the data acquisition module is also used to: perform full charge-discharge cycle tests on the battery under different operating conditions and record the test data; the test data includes average discharge current, average temperature, battery aging degree, initial state of charge, and the actual time T_meas from discharge to cutoff voltage; the offline training module is specifically used for:

[0110] The average discharge current, average temperature, battery aging degree and initial state of charge were clustered using the K-means clustering algorithm, and the optimal number of clusters N was determined by the elbow rule, resulting in N operating condition clusters.

[0111] For each operating condition cluster, the average discharge current, average temperature, battery aging degree, and initial state of charge are used as input variables, and the actual time T_meas and measured temperature are used as output variables. A lightweight machine learning algorithm is used to fit the model to obtain the battery discharge time prediction model corresponding to each operating condition cluster.

[0112] In specific implementation, the online matching module is used for:

[0113] Based on the real-time operating condition characteristic parameters, calculate the Euclidean distance to the center of each operating condition cluster;

[0114] The cluster of operating conditions with the smallest Euclidean distance is taken as the target cluster of operating conditions.

[0115] It should be noted that the specific workflow of this embodiment is described in the foregoing method embodiment section, and will not be repeated here.

[0116] Furthermore, such as Figure 7 As shown, another embodiment of the present invention also provides a battery discharge time prediction system based on operating condition learning, which may include: one or more processors 101, one or more input devices 102, one or more output devices 103, and a memory 104. The processors 101, input devices 102, output devices 103, and memory 104 are interconnected via a bus 105. The memory 104 is used to store a computer program, the computer program including program instructions, and the processor 101 is configured to invoke the program instructions to execute the method described in the above-described method embodiment.

[0117] It should be understood that, in this embodiment of the invention, the processor 101 may be a central processing unit (CPU), but it may also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0118] Input device 102 may include a keyboard, etc., and output device 103 may include a display (LCD, etc.), a speaker, etc.

[0119] The memory 104 may include read-only memory and random access memory, and provides instructions and data to the processor 101. A portion of the memory 104 may also include non-volatile random access memory. For example, the memory 104 may also store device type information.

[0120] In specific implementations, the processor 101, input device 102, and output device 103 described in the embodiments of the present invention can execute the implementation methods described in the embodiments of the battery discharge time prediction method based on operating condition learning provided in the embodiments of the present invention, which will not be repeated here.

[0121] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0122] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices or units, or may be electrical, mechanical or other forms of connection.

[0123] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of the present invention, depending on actual needs.

[0124] Furthermore, the functional units in the various embodiments of this invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units can be implemented in hardware or as software functional units. When using each module, user information is collected and stored only with the user's full authorization and in compliance with relevant laws and regulations, protecting the security and privacy of user data, and strictly prohibiting unauthorized access; data processing will be conducted within the scope stipulated by law and will not exceed the purpose and scope authorized by the user; at the same time, users have the rights to access, correct, delete, restrict processing, and refuse their personal data; and must strictly comply with applicable laws and regulations and conduct compliance reviews.

[0125] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for predicting battery discharge time based on operating condition learning, characterized in that, include: Collect real-time operating condition characteristic parameters; the real-time operating condition characteristic parameters include average discharge current, average temperature, battery aging degree and current state of charge; The target operating condition cluster is matched based on the real-time operating condition feature parameters; A target prediction model is matched based on the target working condition cluster; The real-time operating condition characteristic parameters are input into the target prediction model, and the remaining discharge time of the battery under the current operating condition is output.

2. The battery discharge time prediction method based on operating condition learning as described in claim 1, characterized in that, Before collecting real-time operating condition characteristic parameters, the method also includes an offline training phase, specifically: Under different operating conditions, the battery was subjected to a full charge-discharge cycle test, and the test data was recorded. The test data included average discharge current, average temperature, battery aging degree, initial state of charge, and actual time T_meas from discharge to cutoff voltage. The average discharge current, average temperature, battery aging degree and initial state of charge are clustered using the K-means clustering algorithm, and the optimal number of clusters N is determined by the elbow rule, resulting in N operating condition clusters. For each operating condition cluster, the average discharge current, average temperature, battery aging degree, and initial state of charge are used as input variables, and the actual time T_meas is used as the output variable. A lightweight machine learning algorithm is used to fit the model to obtain the battery discharge time prediction model corresponding to each operating condition cluster.

3. The battery discharge time prediction method based on operating condition learning as described in claim 2, characterized in that, The target operating condition cluster is matched based on the real-time operating condition feature parameters, specifically as follows: Based on the real-time operating condition characteristic parameters, calculate the Euclidean distance to the center of each operating condition cluster; The cluster of operating conditions with the smallest Euclidean distance is taken as the target cluster of operating conditions.

4. The battery discharge time prediction method based on operating condition learning as described in claim 3, characterized in that, Before outputting the remaining discharge time of the battery under the current operating condition, the method further includes: The process of real-time operating condition acquisition, operating condition cluster matching, and venting time calculation is repeated to update the predicted venting time based on the latest operating condition parameters.

5. The battery discharge time prediction method based on operating condition learning as described in claim 2, characterized in that, The method also includes updating the offline training phase, including re-collecting the test data, re-dividing the operating condition clusters, and updating the parameters of the battery discharge time prediction model.

6. A battery discharge time prediction system based on operating condition learning, characterized in that, include: The data acquisition module, integrated into the BMS, is used to collect real-time operating condition characteristic parameters, including average discharge current, average temperature, battery aging degree, and current state of charge. The online matching module is used to match the target working condition cluster based on the real-time working condition feature parameters, and to match the target prediction model based on the target working condition cluster. The discharge time calculation module is used to input the real-time operating condition characteristic parameters into the target prediction model and output the remaining discharge time of the battery under the current operating condition.

7. The battery discharge time prediction system based on operating condition learning as described in claim 6, characterized in that, The data acquisition module is also used to: perform full charge-discharge cycle tests on the battery under different operating conditions and record the test data; the test data includes average discharge current, average temperature, battery aging degree, initial state of charge, and actual time T_meas from discharge to cutoff voltage; The system also includes an offline training module for dividing the operating condition clusters and building the prediction model, specifically: The average discharge current, average temperature, battery aging degree and initial state of charge are clustered using the K-means clustering algorithm, and the optimal number of clusters N is determined by the elbow rule, resulting in N operating condition clusters. For each operating condition cluster, the average discharge current, average temperature, battery aging degree, and initial state of charge are used as input variables, and the actual time T_meas is used as the output variable. A lightweight machine learning algorithm is used to fit the model to obtain the battery discharge time prediction model corresponding to each operating condition cluster.

8. The battery discharge time prediction system based on operating condition learning as described in claim 6, characterized in that, The online matching module is specifically used for: Based on the real-time operating condition characteristic parameters, calculate the Euclidean distance to the center of each operating condition cluster; The cluster of operating conditions with the smallest Euclidean distance is taken as the target cluster of operating conditions.

9. The battery discharge time prediction system based on operating condition learning as described in any one of claims 6-8, characterized in that, The system also includes a model iteration module, used for: Based on the new full charge-full discharge cycle data, the operating condition cluster division and prediction model parameters are optimized periodically.

10. A battery discharge time prediction system based on operating condition learning, characterized in that, The system includes a processor, an input device, an output device, and a memory, which are interconnected. The memory is used to store a computer program, which includes program instructions. The processor is configured to invoke the program instructions to perform the method as described in any one of claims 1-5.

Images