BMS optimization method and system based on shadow mode, medium and device

Abstract

A BMS optimization method and system based on a shadow mode, a medium and a device, relating to the technical field of BMS optimizations. The method comprises: when it is determined that a BMS detects a prediction event or a decision event, triggering a shadow mode; if a prediction event is detected, determining a first prediction model actually running on a battery device managed by the BMS, acquiring a first prediction result of the first prediction model, and simultaneously acquiring a second prediction result of a second prediction model virtually running in the shadow mode; optimizing the first prediction model or the second prediction model on the basis of the first prediction result and the second prediction result; if a decision event is detected, acquiring a first decision result of a first decision scheme actually executed on the battery device, and simultaneously acquiring a second decision result of a second decision scheme virtually executed in the shadow mode; and optimizing the first decision scheme or the second decision scheme on the basis of the first decision result and the second decision result. By implementing the technical solution provided in the present application, the overall performance of the BMS can be improved.

Description

BMS optimization method, system, medium and equipment based on shadow mode Technical Field

[0001] The present application relates to the technical field of BMS optimization, and in particular to a BMS optimization method, system, medium, and device based on a shadow mode. Background Art

[0002] With the popularity of electric vehicles, the Battery Management System (BMS), as one of the core components of electric vehicles, has the main function of accurately monitoring and protecting battery equipment through system algorithms. The system algorithms mainly include the prediction model algorithm of battery equipment and the decision-making algorithm in case of battery failure to ensure the safe and reliable operation of battery equipment.

[0003] Existing BMSs have certain limitations in battery status prediction and decision-making management during battery failures, which directly affect the performance of the BMS. For example, the prediction model algorithm and decision-making algorithm in the existing BMS still operate in a relatively fixed and traditional manner, resulting in a long feedback cycle for algorithm training and difficulty in adapting to the complex changes in the battery. This makes the algorithm accuracy in the BMS low, resulting in poor overall performance of the BMS.

[0004] Summary of the Invention

[0005] The present application provides a BMS optimization method, system, medium and device based on shadow mode, which can quickly optimize the prediction model and decision-making plan in the BMS, improve the accuracy of the algorithm in the BMS, and thus improve the overall performance of the BMS.

[0006] In a first aspect, the present application provides a BMS optimization method based on shadow mode, the method comprising:

[0007] determining that when a BMS detects a prediction event or a decision event, triggering a shadow mode corresponding to the BMS;

[0008] If the BMS detects the predicted event, obtaining first target data corresponding to the predicted event, determining a first prediction model actually running on the battery device managed by the BMS, obtaining a first prediction result predicted by the first prediction model based on the first target data, and simultaneously obtaining a second prediction result predicted by a second prediction model virtually running in the shadow mode based on the first target data;

[0009] Optimizing the first prediction model or the second prediction model based on the first prediction result and the second prediction result;

[0010] If the BMS detects the decision event, obtaining second target data corresponding to the decision event, determining a first decision solution actually executed on the battery device, obtaining a first decision result of executing the first decision solution based on the second target data, and simultaneously obtaining a second decision result of a second decision solution virtually executed in the shadow mode based on the second target data;

[0011] Based on the first decision result and the second decision result, the first decision solution or the second decision solution is optimized.

[0012] By adopting the above technical solution, when the BMS detects a prediction event or a decision event, the shadow mode is triggered. For the prediction event, the first prediction model actually running on the battery device obtains the first prediction result, and at the same time, a different second prediction model is run in parallel in the shadow mode to obtain the second prediction result. The shadow mode shortens the feedback cycle of model training, so that the BMS can adapt to the changing battery status more quickly, improves the accuracy of the prediction model, and realizes the rapid iterative optimization of the BMS prediction model. For the decision event, the first decision plan actually executed on the battery obtains the first decision result, and at the same time, the shadow mode simulates the parallel execution of different second decision plans to obtain the second decision result. By comparison, a plan with better decision effect is found, and the rapid iterative optimization of the decision plan is completed, which helps to quickly provide a better plan when making decisions. The comprehensive use of the shadow algorithm enables the BMS to be comprehensively optimized in the prediction algorithm and the decision algorithm, improves the accuracy of the algorithm in the BMS, and thus improves the overall performance of the BMS.

[0013] Optionally, determining whether the BMS has detected a prediction event or a decision event includes: calculating in real time the deviation value between the first prediction result and the second prediction result, the first prediction model and the second prediction model are in real-time operation; when the deviation value is greater than a preset value, determining that the BMS has detected a prediction event; obtaining various operating data of the battery equipment, and when it is detected that any of the operating data exceeds the corresponding preset range, determining that the BMS has detected a decision event.

[0014] By employing these technical solutions, prediction deviations and battery anomalies in the BMS are captured in real time, triggering corresponding event processing. The prediction model result deviation value determination method enables continuous monitoring of prediction accuracy, verifies the real-time and accuracy of the prediction model, and ensures the reliability of the prediction system. The battery hyperparameter detection method can quickly reflect abnormal battery conditions, promptly initiate decision-making plans, and implement battery protection and management.

[0015] Optionally, obtaining the first target data corresponding to the predicted event includes: during the BMS detection process, obtaining multimodal data collected by the BMS in real time; if it is determined that the BMS detects the predicted event, filtering out the first target data corresponding to the predicted event from the multimodal data.

[0016] By adopting the above technical solution, during operation, the BMS will continuously collect data on various parameters such as battery voltage, current, temperature, and environmental parameters, and store them to form a multimodal monitoring data set. The first target data related to the predicted event will be screened out from the multimodal data, which can improve the accuracy of the predicted event while avoiding interference from other irrelevant data.

[0017] Optionally, after obtaining the first target data corresponding to the predicted event, the method further includes: classifying the first target data based on a clustering algorithm to obtain the classified first target data, and uploading the classified first target data to the cloud.

[0018] By adopting the above technical solution, different types of first target data have different influences and effects on the prediction results. The target data can be clustered into several categories according to their characteristics through a clustering algorithm, thereby distinguishing between data with greater and lesser influence on the prediction. This classification can treat data with lesser influence on the prediction as redundant data and filter it, thereby reducing the input amount of the prediction model and reducing the computational complexity. At the same time, it can also determine the most critical target data and improve the accuracy of the prediction.

[0019] Optionally, obtaining the second prediction result of the target data predicted by the second prediction model virtually running in the shadow mode includes: retrieving at least one second prediction model corresponding to the prediction event in a preset prediction model library; inputting the first target data into the second prediction model to obtain the second prediction result of the second prediction model virtually running in the shadow mode.

[0020] By adopting the above technical solution, a prediction model library is established because the optimal prediction models for different prediction events in the BMS are also different. The model library can provide storage and registration of multiple prediction models and match them according to the prediction event type, so that the BMS can quickly obtain the matching prediction model. By combining the precise retrieval of the prediction model with shadow operation, a second prediction result that is not affected by the actual system environment can be obtained. This result can be compared with the first prediction result to judge the advantages and disadvantages of the two models, so as to select a better model, or to inventory the causes of model deviation and achieve model optimization.

[0021] Optionally, the optimizing of the first prediction model or the second prediction model based on the first prediction result and the second prediction result includes: obtaining the actual operating result of the battery device under the predicted event after a preset time; calculating a first deviation value between the first prediction result and the operating result, and a second deviation value between the second prediction result and the operating result; if the first deviation value is less than the second deviation value, marking the first prediction model as the target prediction model corresponding to the predicted event, when the number of markings of the first prediction model reaches a preset number, using the first prediction model as the standard prediction model corresponding to the predicted event, and training and optimizing the second prediction model; if the second deviation value is less than the first deviation value, marking the second prediction model as the target prediction model corresponding to the predicted event, when the number of markings of the second prediction model reaches a preset number, using the second prediction model as the standard prediction model corresponding to the predicted event, and training and optimizing the first prediction model.

[0022] By adopting the above technical solution, the actual operating data of the battery under the predicted event is obtained as an accurate result after a certain period of time. This actual result is compared with the first and second prediction results, and the deviation between the two is calculated. The relative prediction accuracy of the two prediction models is determined based on the size of the deviation. This can realize the ranking and evaluation of the prediction capabilities of the two prediction models, and autonomously select, optimize and update the models, so that the prediction system can continuously improve itself.

[0023] Optionally, the optimizing the first decision plan or the second decision plan based on the first decision result and the second decision result includes: comparing the first decision result and the second decision result to obtain a comparison result; if the comparison result is that the first decision result is better than the second decision result, marking the first decision plan as the target decision plan corresponding to the decision event, and when the number of markings of the first decision plan reaches a preset number, using the first decision plan as the standard decision plan corresponding to the decision event, and optimizing the second decision plan; if the comparison result is that the second decision result is better than the first decision result, marking the second decision plan as the target decision plan corresponding to the decision event, and when the number of markings of the second decision plan reaches a preset number, using the second decision plan as the standard decision plan corresponding to the decision event, and optimizing the first decision plan.

[0024] By adopting the above technical solution, the processing results of the two decision-making solutions under the same decision event are directly compared. If the processing effect of the first solution is better, it will be marked as the current target solution. If the number of markings reaches the threshold, the solution will be set as the standard solution, and further optimization and adjustment of the second solution will be initiated at the same time to continuously approach the processing effect of the first solution. The effects of the two decision-making solutions can be ranked and evaluated, and the solutions can be selected and optimized and updated independently, so that the decision-making system can continuously improve itself. At the same time, the setting of the standard solution also provides reliability guarantees for subsequent disposal decisions.

[0025] In a second aspect of the present application, a BMS optimization system based on a shadow mode is provided, the system comprising:

[0026] a shadow mode triggering module, configured to trigger a shadow mode corresponding to the BMS when the BMS detects a prediction event or a decision event;

[0027] a prediction model running module, configured to, if the BMS detects the prediction event, obtain first target data corresponding to the prediction event, determine a first prediction model actually running on the battery device managed by the BMS, obtain a first prediction result predicted by the first prediction model based on the first target data, and simultaneously obtain a second prediction result predicted by a second prediction model virtually running in the shadow mode based on the first target data;

[0028] a prediction model optimization module, configured to optimize the first prediction model or the second prediction model based on the first prediction result and the second prediction result;

[0029] a decision scheme execution module, configured to, if the BMS detects the decision event, obtain second target data corresponding to the decision event, determine a first decision scheme actually executed on the battery device, obtain a first decision result of executing the first decision scheme based on the second target data, and simultaneously obtain a second decision result of a second decision scheme virtually executed in the shadow mode based on the second target data;

[0030] A decision solution optimization module is used to optimize the first decision solution or the second decision solution based on the first decision result and the second decision result.

[0031] In a third aspect of the present application, a computer storage medium is provided, wherein the computer storage medium stores a plurality of instructions, wherein the instructions are suitable for being loaded by a processor and executing the above method steps.

[0032] In a fourth aspect of the present application, an electronic device is provided, comprising: a processor and a memory; wherein the memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the above-mentioned method steps.

[0033] In summary, one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0034] When the BMS detects a prediction event or a decision event, the shadow mode is triggered. For a prediction event, the first prediction model actually running on the battery device obtains a first prediction result, and at the same time, a different second prediction model is run in parallel in the shadow mode to obtain a second prediction result. The shadow mode shortens the feedback cycle of model training, so that the BMS can adapt to the changing battery status more quickly, improves the accuracy of the prediction model, and realizes rapid iterative optimization of the BMS prediction model. For a decision event, the first decision plan actually executed on the battery obtains a first decision result, and at the same time, different second decision plans are simulated and executed in parallel in the shadow mode to obtain a second decision result. By comparison, a plan with better decision effect is found, and rapid iterative optimization of the decision plan is completed. At the same time, it helps to quickly provide a better plan when making decisions. The comprehensive use of the shadow algorithm enables the BMS to be comprehensively optimized in the prediction algorithm and decision algorithm, improves the accuracy of the algorithm in the BMS, and thus improves the overall performance of the BMS. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] FIG1 is a flow chart of a BMS optimization method based on a shadow mode provided in an embodiment of the present application;

[0036] FIG2 is a module diagram of a BMS optimization system based on a shadow mode provided in an embodiment of the present application;

[0037] FIG3 is a schematic structural diagram of an electronic device provided in an embodiment of the present application.

[0038] Description of reference numerals: 300, electronic device; 301, processor; 302, communication bus; 303, user interface; 304, network interface; 305, memory. DETAILED DESCRIPTION

[0039] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments.

[0040] In the description of the embodiments of this application, words such as "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "for example" or "for instance" in the embodiments of this application should not be construed as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "for example" or "for instance" is intended to present the relevant concepts in a concrete manner.

[0041] In the description of the embodiments of the present application, the term "multiple" means two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals. In addition, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized.

[0042] The following will provide a clear and complete description of the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments.

[0043] Referring to FIG1 , a flowchart of a BMS optimization process based on a shadow mode is proposed. The method can be implemented by a computer program, a single-chip microcomputer, or run on a BMS optimization system based on a shadow mode. The computer program can be integrated into a smart terminal or run as an independent tool application. Specifically, the method includes steps 10 to 50, which are as follows:

[0044] Step 10: When the BMS detects a prediction event or a decision event, a shadow mode corresponding to the BMS is triggered.

[0045] The embodiments of the present application can be applied to all fields that require optimization of the battery management system, including but not limited to all battery systems that use BMS, such as electric vehicles, storage batteries, electronic devices, etc., for example, battery status evaluation and power management can be performed.

[0046] The Battery Management System (BMS) refers to an electronic system used to manage, protect and monitor battery equipment. It is a core component to ensure the safe and reliable operation of batteries. The BMS can monitor the voltage, current, temperature and other parameters of battery equipment in real time, and can perform functions such as charge and discharge management, battery balancing, and status assessment.

[0047] Shadow mode refers to an emerging system modeling method. Its core concept is to dynamically simulate the various states and operating scenarios of the real system by establishing a virtual "shadow" system to achieve performance evaluation and optimization of the real system. The shadow system can execute multiple solutions in parallel without interfering with the normal operation of the real system. In the embodiments of this application, shadow mode refers to the establishment of a virtual BMS corresponding to the real BMS when the BMS is operating normally.

[0048] In the embodiments of the present application, prediction events and decision events may be referred to as trigger events, wherein a prediction event refers to an event in which, when performing battery status prediction, there is a large deviation between two prediction results of multiple prediction models, which is determined to be an abnormal prediction event; a decision event refers to an event in which, during battery charge and discharge management, it is detected that the actual operating parameters of the battery exceed the normal range, which is determined to be a decision event.

[0049] Specifically, to optimize and upgrade the BMS, it's necessary to be able to rapidly iterate the system's prediction models and decision-making solutions. This solution proposes that during the BMS monitoring process, when abnormal events such as abnormal prediction results or abnormal decision-making effects are detected, a shadow BMS is triggered to perform parallel calculations. This is because large prediction deviations and decision failures indicate room for BMS optimization. Shadow mode can obtain result data from different operating conditions in parallel, which is conducive to finding a more optimal prediction model or decision-making solution.

[0050] Based on the above embodiment, as an optional embodiment, the step of determining whether the BMS detects a prediction event or a decision event may further include the following steps:

[0051] Step 101: Calculate the deviation value between the first prediction result and the second prediction result in real time, and the first prediction model and the second prediction model are in real-time operation state.

[0052] Specifically, the prediction models in the BMS are all in real-time operation. These prediction models can be the first prediction model actually running on the battery device, or at least one second prediction model running virtually in shadow mode. The system allows these prediction models to use the same set of real-time collected battery data for prediction, thereby obtaining the battery health status results output by each prediction model in parallel. By allowing multiple models to predict in parallel, the prediction capabilities of each model can be directly compared. In the embodiment of the present application, all prediction models corresponding to the prediction event are regarded as consisting of a first prediction model and a second prediction model, and can also include multiple other prediction models. Therefore, in order to determine whether the prediction model is abnormal, it is necessary to quantitatively analyze the differences between the prediction results of different models. After obtaining the prediction results of each prediction model for the same data set, that is, after obtaining the first prediction result corresponding to the first prediction model and the second prediction result corresponding to the second prediction model, the system will use methods such as root mean square error to calculate the deviation between the first prediction result and the second prediction result in real time. The reason for calculating the prediction result deviation is that by directly comparing the prediction values ​​of different prediction models, it is possible to quantitatively determine whether there are significant differences in the prediction effects of each model, reflecting the abnormality of the prediction.

[0053] Step 102: When the deviation value is greater than the preset value, it is determined that the BMS detects a prediction event.

[0054] Specifically, after obtaining the deviation value between the prediction outputs of each prediction model, the deviation value is compared with a preset value. When the deviation between two prediction models exceeds the preset value, it is determined that the BMS has detected a prediction event at that moment. Since the deviation between different prediction models is too large, it indicates that there is a prediction model in the system that needs to be optimized.

[0055] Step 103: Acquire various operating data of the battery device, and when it is detected that any operating data exceeds the corresponding preset range, determine that the BMS detects a decision event.

[0056] Specifically, in order to optimize the decision-making scheme of the BMS, it is necessary to monitor whether the real-time operating data of the battery equipment is within the normal range. When it is detected that a certain battery operating parameter exceeds the corresponding preset range, it can be determined as a decision event and the decision-making scheme needs to be optimized. The BMS in this application will continuously detect the operating indicators of the battery equipment, such as voltage, current, temperature and other data, and compare them with the corresponding preset range. These operating parameters of the battery reflect the effectiveness of the power management decision strategy. When it is detected that parameters such as the single cell voltage and the battery operating temperature exceed the normal preset range configured in advance, it can be judged that the battery operating status is abnormal, indicating that the decision-making scheme needs to be activated to deal with the abnormal state of the battery. At this time, it will be determined that the BMS has detected a decision event.

[0057] Based on the above embodiment, as an optional embodiment, after determining that the BMS detects a prediction event or a decision event and triggers the shadow mode corresponding to the BMS, the process of collecting and uploading data corresponding to the prediction event or the decision event is also included. The specific process is as follows:

[0058] Specifically, in order to make full use of the data collected during the operation of the BMS, it is necessary to obtain the multimodal monitoring data of the BMS in real time. When a prediction event or a decision event is detected, these multimodal data can provide additional effective information for analyzing the cause of the abnormality and assisting in decision-making. Multimodal data refers to data collected by the BMS during operation that contains multiple modes or forms, mainly including but not limited to environmental data and battery monitoring data. Environmental data refers to monitoring data of the battery working environment, such as sensor data of parameters such as temperature, humidity, and vibration, which can reflect the environmental conditions of the battery. Battery data refers to direct monitoring data of battery performance, such as data of parameters such as voltage, current, temperature, and internal resistance, which can directly reflect the working condition of the battery.

[0059] When it is determined that the BMS has detected a prediction event or a decision event, it is necessary to enable shadow mode for model or strategy optimization. The BMS will first filter out the target data corresponding to the time period of the prediction event or decision event from the collected historical multimodal data. This data is more valuable and contains richer information about anomalies. For example, when it is determined that the BMS has detected a prediction event, the first target data corresponding to the time period of the prediction event will be filtered out from the multimodal data. The first target data includes real-time battery data and environmental data corresponding to the time period of the prediction event. When it is determined that the BMS has detected a decision event, the second target data corresponding to the time period of the decision event will be filtered out from the multimodal data.

[0060] Furthermore, BMS will use clustering algorithms, such as K-Means, to classify and analyze the filtered abnormality-related data. For example, events of the type of temperature anomaly will be grouped together. Because different types of abnormal situations have corresponding data patterns that are also different, classification can help to more clearly analyze the data characteristics of each type of abnormality. The classified data will be uploaded to the cloud server to utilize the massive storage and computing resources of cloud computing to carry out in-depth data training and mining, assist in quickly determining the optimal prediction model or decision-making plan, and complete the online optimization and upgrade of the BMS system. At the same time, it provides the system with more refined data support and optimizes the data upload efficiency. It should be noted that multimodal data in a normal state does not need to be uploaded to the cloud, which also improves the efficiency of data upload.

[0061] Step 20: If the BMS detects a prediction event, obtain the first target data corresponding to the prediction event, determine the first prediction model actually running on the battery device managed by the BMS, obtain the first prediction result predicted by the first prediction model based on the first target data, and simultaneously obtain the second prediction result predicted by the second prediction model virtually running in the shadow mode based on the first target data.

[0062] Specifically, when a BMS detects a predicted event, to evaluate the effectiveness of different prediction models, it is necessary to obtain the output of a first prediction model running on a real battery device, namely the first prediction result. Simultaneously, different second models are run in a shadow system to obtain second prediction results for comparative analysis. In specific implementations, if a prediction anomaly occurs after the actual BMS detects a predicted event, to locate the root cause of the problem, it is necessary to identify the first prediction model currently running on the battery device managed by the BMS. This first prediction model can be the default or manually selected prediction model for the battery device, such as an LSTM network-based model. The system then selects the data at the time of the predicted anomaly from the stored multimodal data as the first target data corresponding to the anomaly. The first target data includes battery operating parameters such as voltage and temperature. The corresponding input data used by the first prediction model at the time of the anomaly, namely the first target data, is then used to predict the first target data, resulting in a first prediction result. The first prediction result can include battery life, remaining charge, and battery status, among other factors.

[0063] At the same time, the BMS in shadow mode invokes a pre-configured second prediction model, such as one based on a random forest algorithm, using the same set of first target data as input. This prediction is then performed in the shadow system to generate a second prediction result. By obtaining the prediction outputs of the two models in parallel on the same dataset, it is possible to directly compare the differences in their prediction performance and determine which model's prediction results are more accurate. This allows the optimal model to be determined in the current state, which can then be used to adjust and optimize the prediction model of the actual BMS system to correct for prediction errors and improve model prediction accuracy.

[0064] Based on the above embodiment, as an optional embodiment, the step of obtaining the second prediction result of the second prediction model running virtually in the shadow mode for predicting the target data may further include the following steps:

[0065] Step 201: Retrieve at least one second prediction model corresponding to a prediction event from a preset prediction model library.

[0066] Step 202: Input the first target data into the second prediction model to obtain a second prediction result of virtually running the second prediction model in the shadow mode.

[0067] Specifically, to evaluate the effectiveness of different prediction models under current conditions, in addition to obtaining the results of the first prediction model, it is necessary to call on other models for prediction to obtain a second prediction result for comparative analysis. During implementation, the system pre-establishes a prediction model library of various prediction models, including those based on LSTM, random forest, and other methods. When a prediction anomaly occurs, the system selects one or more alternative prediction models related to the predicted event from the prediction model library as the second prediction model. Taking the current battery health prediction as an example, another battery health status prediction model is retrieved from the prediction model library as the second prediction model. The first target data that triggered this prediction event is then input, and the second prediction model is run in the shadow BMS system to obtain the second prediction model's prediction output for the same input as the second prediction result. By using the prediction outputs of the two models for the same set of input data, the prediction robustness and effectiveness of the two models under such abnormal conditions can be directly compared, which can help determine whether the first prediction model needs to be adjusted and optimized to improve the BMS's prediction capabilities.

[0068] Step 30: Optimize the first prediction model or the second prediction model based on the first prediction result and the second prediction result.

[0069] Specifically, after obtaining two sets of prediction results from the first and second prediction models for the same input data, a comparison can be used to determine the optimal prediction model within the BMS and improve prediction performance. The first prediction result can reflect any issues or deficiencies in the current prediction model used in the BMS, including prediction errors or anomalies. The second prediction result, however, can serve as a benchmark for comparison under the same input, determining whether the first model's predictions are excessively biased. In practice, the system statistically compares the error magnitude and trend deviation of the two results to identify issues with the first prediction model, such as insufficient training or overly complex models. If the error of the first prediction result is significantly greater than that of the second, the structure of the first prediction model or the training parameters should be adjusted for further optimization, such as increasing the training dataset or employing regularization to prevent overfitting. If the errors of the two results are similar, the data may contain anomalies. In this case, the anomaly can be identified and the first prediction model retrained. By comparing the first prediction result with the second prediction result as a reference, the effectiveness of the first prediction model in addressing the current situation can be determined, allowing targeted model optimization to reduce model errors and improve the accuracy and robustness of the BMS's battery state predictions.

[0070] Based on the above embodiment, as an optional embodiment, the step of optimizing the prediction model in the BMS based on the first prediction result and the second prediction result may further include the following steps:

[0071] Step 301: Acquire the actual operating result of the battery device under the predicted event after a preset time.

[0072] Step 302: Determine a first deviation value between the first prediction result and the operation result, and a second deviation value between the second prediction result and the operation result.

[0073] Specifically, to evaluate the effectiveness of different prediction models, it is necessary to obtain the deviation between the model prediction results and the actual battery operating results. After performing two model predictions, the system will continue to detect the actual operating data of the battery equipment within a preset time period, such as voltage, current, temperature and other parameters, and use these operating parameters as the actual operating results under the predicted event. The operating results are then compared with the data in the first prediction result to calculate the first deviation value. At the same time, the second deviation value between the second prediction result and the operating result is also calculated. Calculating the deviation value between the model prediction results and the actual results can intuitively reflect the prediction effect of each prediction model.

[0074] Step 303: If the first deviation value is less than the second deviation value, the first prediction model is marked as the target prediction model corresponding to the prediction event. When the number of marking times of the first prediction model reaches a preset number, the first prediction model is used as the standard prediction model corresponding to the prediction event, and the second prediction model is trained and optimized.

[0075] Specifically, in order to continuously optimize the prediction capability of BMS, it is necessary to determine the best standard prediction model based on the prediction effects of different models and actual conditions. The system will count the deviation value comparisons of the prediction results of each prediction model each time. For example, after comparison, if the first deviation value is smaller than the second deviation value, it means that the first prediction model is more accurate, and the first prediction model is marked as the target prediction model corresponding to the predicted event. When the number of times the first model is better than the second model reaches a preset threshold, for example, more than 5 times, the first model is confirmed to be the standard prediction model corresponding to the current predicted event. The comparison of the first deviation value and the second deviation value can directly judge which model's prediction result is more accurate and reliable in a certain type of predicted anomaly detected. If the deviation of the first prediction model continues to be small, it fully proves that the first prediction model is more suitable for this type of predicted event. So that in subsequent practical applications, when a similar predicted event is detected again, the system will directly call the standard prediction model for prediction. At the same time, the first prediction model is used as the optimization training target, and more relevant data is used to conduct targeted tuning of the second prediction model, gradually approaching the prediction results of the standard model. This also shortens the feedback cycle of model training, enabling the system to adapt more quickly to the changing battery status and improving the accuracy of the model. This can continuously improve the BMS's adaptability to specific prediction scenarios, making its prediction results more accurate and reliable.

[0076] Step 304: If the second deviation value is less than the first deviation value, the second prediction model is marked as the target prediction model corresponding to the prediction event. When the number of marking times of the second prediction model reaches a preset number, the second prediction model is used as the standard prediction model corresponding to the prediction event, and the first prediction model is trained and optimized.

[0077] Specifically, if the second deviation value is smaller than the first deviation value, the training and optimization of the first prediction model may be implemented by referring to the process of step 303 above, which will not be described in detail here.

[0078] Step 40: If the BMS detects a decision event, obtain the second target data corresponding to the decision event, determine the first decision plan actually executed on the battery device, obtain a first decision result of executing the first decision plan based on the second target data, and simultaneously obtain a second decision result of the second decision plan virtually executed in the shadow mode based on the second target data.

[0079] When the BMS detects a decision event, to evaluate the effectiveness of different decision scenarios, it is necessary to obtain the results of the decision scenario executed on the real system (the first decision result). Simultaneously, different decision scenarios are run on the shadow system to obtain the second decision results for comparative analysis. By directly comparing the management effects of the two decision scenarios on the same battery, it is possible to determine which decision strategy is more reasonable, thereby optimizing the existing decision mechanism in a targeted manner.

[0080] Specifically, after the BMS detects a decision event, that is, when a certain fault occurs in the battery device, in order to locate the root cause of the problem, it needs to clarify the first decision plan currently running on the battery device, such as a rule-based charging strategy. The system then filters out the data at the moment of the abnormality from the stored multimodal data as the second target data, which may include battery voltage, SOC and other data. The first decision plan is actually run on the battery device, using the second target data as input to generate a decision output for controlling the battery, such as a charging instruction to output a certain current. If the battery pressure changes, the changed battery pressure is used as the first decision result, such as the battery's health status, voltage change, etc.

[0081] To evaluate the effectiveness of different decision plans under current conditions, in addition to executing the first decision plan, other decision plans are invoked for comparison to obtain a second decision result. The system has a pre-set library of various decision plans, including rule-based and reinforcement learning-based strategies. When a decision event is detected, the system selects at least one alternative decision plan related to the decision event from the library as the second decision plan. For the current battery charging decision, for example, the system retrieves another charging decision plan from the library, such as an adaptive charging strategy based on Q-learning optimization, as the second decision plan. Then, the acquired second target data is input and the shadow BMS simulates the execution of the second decision plan, simulating charging according to that strategy. The data results of the virtual execution of the second decision plan are used as the second decision result. By comparing the performance of the two decision plans on the same set of input data, it is possible to determine which decision plan is more suitable for the current battery state and thus determine whether the first decision plan needs to be optimized to improve the BMS's decision-making effectiveness.

[0082] Step 50: Optimize the first decision solution or the second decision solution based on the first decision result and the second decision result.

[0083] After obtaining the execution results of the first and second decision-making plans, the results can be compared to determine the optimal decision-making plan in the BMS and improve the system's decision-making capabilities. The first decision result can reflect the problems and limitations of the current decision-making strategy used in the BMS, which may have caused decision deviations or anomalies. The second decision result can be used as a comparison benchmark to determine whether the first decision-making plan is reasonable.

[0084] Specifically, the system will statistically compare the differences in the effects of the two decision results on changes in battery health status, battery life loss, etc. to determine the shortcomings of the first decision plan, such as improper handling of a certain situation. If the first result is significantly worse than the second result, it is necessary to adjust the first decision plan, modify its decision log or rules, and optimize the decision result for this situation. If the two results are similar, further analyze the data, optimize the decision basis, and add beneficial decision factors in the second plan to the decision model. By comparing the effects of the second decision plan as a reference, the decision-making mechanism of the BMS can be continuously improved to make it more widely adaptable to various battery health conditions and optimize the battery management effect.

[0085] Based on the above embodiment, as an optional embodiment, the step of optimizing the decision solution in the BMS based on the first decision result and the second decision result may further include the following steps:

[0086] Step 501: Compare the first decision result and the second decision result to obtain a comparison result.

[0087] Specifically, the result data generated by the first and second decision plans under the same second target data input is obtained. The first decision result may include battery health, voltage curve, etc.; the second decision result also includes the corresponding battery status data. Then, the system will quantitatively or qualitatively compare the two sets of decision results and generate a comparative analysis report. For example, comparing the impact of two decision results on battery health status, the first plan causes the health to drop by 10%, while the second plan only drops by 5%, or directly giving a ranking conclusion of the advantages and disadvantages of the two. Such a comparison result can intuitively show which decision plan can bring better battery management effects in the current scenario, providing a reference for the subsequent determination of optimization targets.

[0088] Step 502: If the comparison result shows that the first decision result is better than the second decision result, the first decision plan is marked as the target decision plan corresponding to the decision event. When the number of markings of the first decision plan reaches a preset number, the first decision plan is used as the standard decision plan corresponding to the decision event, and the second decision plan is optimized.

[0089] Specifically, to continuously optimize the BMS's decision-making capabilities, it is necessary to compare different decision solutions and determine the standard decision solution that performs best in the current decision scenario. The system will then compile statistics on each comparison of the two decision results. For example, if the first decision solution outperforms the second, this indicates that the first decision solution has better battery management capabilities for the corresponding decision event. The first decision solution will be marked as the target decision solution for the decision event. When the number of times the first decision solution outperforms the second reaches a preset threshold, for example, more than five times, the first decision solution is designated as the standard decision solution for the specific decision-making anomaly. In actual application, when a similar decision-making event is detected, the system will directly invoke this standard decision solution for decision management. Simultaneously, using the first decision solution as the optimization target, the system will use more relevant data to train and modify the second decision solution, gradually approaching the performance of the first decision solution. This approach continuously improves the BMS's adaptability to specific decision scenarios, ensuring that its decision results are more aligned with the actual needs of the current battery, thereby achieving more optimal battery management.

[0090] Step 503: If the comparison result shows that the second decision result is better than the first decision result, the second decision plan is marked as the target decision plan corresponding to the decision event. When the number of markings of the second decision plan reaches a preset number, the second decision plan is used as the standard decision plan corresponding to the decision event, and the first decision plan is optimized.

[0091] Specifically, if the comparison result shows that the second decision result is better than the first decision result, the training and optimization of the first decision solution can be implemented by referring to the process of step 502 above, which will not be described in detail here.

[0092] Please refer to FIG2 , which is a schematic diagram of a BMS optimization system module based on a shadow mode provided in an embodiment of the present application. The BMS optimization system based on a shadow mode may include: a shadow mode triggering module, a prediction model operation module, a prediction model optimization module, a decision solution execution module, and a decision solution optimization module, wherein:

[0093] a shadow mode triggering module, configured to trigger a shadow mode corresponding to the BMS when the BMS detects a prediction event or a decision event;

[0094] a prediction model running module, configured to, if the BMS detects the prediction event, obtain first target data corresponding to the prediction event, determine a first prediction model actually running on the battery device managed by the BMS, obtain a first prediction result predicted by the first prediction model based on the first target data, and simultaneously obtain a second prediction result predicted by a second prediction model virtually running in the shadow mode based on the first target data;

[0095] a prediction model optimization module, configured to optimize the first prediction model or the second prediction model based on the first prediction result and the second prediction result;

[0096] a decision scheme execution module, configured to, if the BMS detects the decision event, obtain second target data corresponding to the decision event, determine a first decision scheme actually executed on the battery device, obtain a first decision result of executing the first decision scheme based on the second target data, and simultaneously obtain a second decision result of a second decision scheme virtually executed in the shadow mode based on the second target data;

[0097] A decision solution optimization module is used to optimize the first decision solution or the second decision solution based on the first decision result and the second decision result.

[0098] Optionally, the shadow mode trigger module is also used to calculate in real time the deviation value between the first prediction result and the second prediction result, and the first prediction model and the second prediction model are in real-time operation state; when the deviation value is greater than a preset value, it is determined that the BMS has detected a prediction event; and various operating data of the battery equipment are obtained, and when it is detected that any of the operating data exceeds the corresponding preset range, it is determined that the BMS has detected a decision event.

[0099] Optionally, the prediction model running module is also used to obtain the multimodal data collected by the BMS in real time during the BMS detection process; if it is determined that the BMS detects the predicted event, the first target data corresponding to the predicted event is screened out from the multimodal data.

[0100] Optionally, the BMS optimization system based on the shadow mode further includes a data upload module, which is used to classify the first target data based on a clustering algorithm to obtain the classified first target data, and upload the classified first target data to the cloud.

[0101] Optionally, the prediction model running module is also used to call at least one second prediction model corresponding to the prediction event in the preset prediction model library; input the first target data into the second prediction model to obtain a second prediction result of virtually running the second prediction model in the shadow mode.

[0102] Optionally, the prediction model optimization module is also used to obtain the actual operating results of the battery device under the prediction event after a preset time; calculate the first deviation value between the first prediction result and the operating result, and the second deviation value between the second prediction result and the operating result; if the first deviation value is less than the second deviation value, the first prediction model is marked as the target prediction model corresponding to the prediction event, and when the number of markings of the first prediction model reaches a preset number, the first prediction model is used as the standard prediction model corresponding to the prediction event, and the second prediction model is trained and optimized; if the second deviation value is less than the first deviation value, the second prediction model is marked as the target prediction model corresponding to the prediction event, and when the number of markings of the second prediction model reaches a preset number, the second prediction model is used as the standard prediction model corresponding to the prediction event, and the first prediction model is trained and optimized.

[0103] Optionally, the decision scheme optimization module is also used to compare the first decision result and the second decision result to obtain a comparison result; if the comparison result is that the first decision result is better than the second decision result, the first decision scheme is marked as the target decision scheme corresponding to the decision event, and when the number of markings of the first decision scheme reaches a preset number, the first decision scheme is used as the standard decision scheme corresponding to the decision event, and the second decision scheme is optimized; if the comparison result is that the second decision result is better than the first decision result, the second decision scheme is marked as the target decision scheme corresponding to the decision event, and when the number of markings of the second decision scheme reaches a preset number, the second decision scheme is used as the standard decision scheme corresponding to the decision event, and the first decision scheme is optimized.

[0104] It should be noted that the above embodiments provide systems that implement their functions using only the division of the above functional modules as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiments are based on the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0105] An embodiment of the present application further provides a computer storage medium, which can store multiple instructions. The instructions are suitable for being loaded by a processor and executing a BMS optimization method based on a shadow mode of the above embodiment. The specific execution process can be found in the specific description of the above embodiment and will not be repeated here.

[0106] Referring to FIG3 , this application also discloses an electronic device. FIG3 is a schematic diagram of the structure of an electronic device disclosed in an embodiment of this application. The electronic device 300 may include: at least one processor 301 , at least one network interface 304 , a user interface 303 , a memory 305 , and at least one communication bus 302 .

[0107] The communication bus 302 is used to implement the connection and communication between these components.

[0108] The user interface 303 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 303 may also include a standard wired interface and a wireless interface.

[0109] The network interface 304 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0110] The processor 301 may include one or more processing cores. The processor 301 utilizes various interfaces and circuits to connect various components within the server. By running or executing instructions, programs, code sets, or instruction sets stored in the memory 305, and accessing data stored in the memory 305, the processor 301 performs various server functions and processes data. Optionally, the processor 301 may be implemented using at least one hardware form selected from the group consisting of digital signal processing (DSP), field-programmable gate array (FPGA), and programmable logic array (PLA). The processor 301 may integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. The CPU primarily processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; and the modem is responsible for handling wireless communications. It is understood that the modem may not be integrated into the processor 301 and may be implemented separately on a single chip.

[0111] Among them, the memory 305 may include a random access memory (RAM) or a read-only memory (ROM). Optionally, the memory 305 includes a non-transitory computer-readable storage medium. The memory 305 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 305 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 305 may also be at least one storage device located away from the aforementioned processor 301. Referring to Figure 3, the memory 305 as a computer storage medium may include an operating system, a network communication module, a user interface module and an application program based on a shadow mode BMS optimization method.

[0112] In the electronic device 300 shown in FIG3 , the user interface 303 is mainly used to provide an input interface for the user and obtain the data input by the user; and the processor 301 can be used to call an application program stored in the memory 305 that stores a BMS optimization method based on the shadow mode. When executed by one or more processors 301, the electronic device 300 executes one or more of the methods described in the above embodiments. It should be noted that for the aforementioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should know that this application is not limited to the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required for this application.

[0113] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0114] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are merely schematic, such as the division of units, which is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interface, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0115] Units described as separate components may or may not be physically separate, and 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 these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0116] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0117] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a memory and includes several instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned memory includes various media that can store program codes, such as USB flash drives, mobile hard drives, magnetic disks or optical disks.

[0118] The foregoing is merely an exemplary embodiment of the present disclosure and is not intended to limit the scope of the present disclosure. In other words, any equivalent variations and modifications made in accordance with the teachings of the present disclosure are still within the scope of the present disclosure. Those skilled in the art will readily conceive of other embodiments of the present disclosure after considering the disclosure and the practical implications thereof.

[0119] This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art not described herein. The description and examples are to be considered as exemplary only, and the scope and spirit of the present disclosure are to be defined by the claims.

Claims

1. A BMS optimization method based on shadow mode, characterized in that: The method comprises: calculating in real time a deviation between a first prediction result and a second prediction result, wherein the first prediction model and the second prediction model are in a real-time operating state; determining that the BMS has detected a prediction event when the deviation is greater than a preset value; acquiring various operating data of the battery device, and determining that the BMS has detected a decision event when it is detected that any of the operating data exceeds a corresponding preset range; When it is determined that the BMS detects a prediction event or a decision event, triggering a shadow mode corresponding to the BMS; If the BMS detects the predicted event, during the BMS detection process, multimodal data collected by the BMS is obtained in real time; first target data corresponding to the predicted event is screened out from the multimodal data, a first prediction model actually running on the battery device managed by the BMS is determined, a first prediction result of the first prediction model based on the first target data is obtained, and at the same time, at least one second prediction model corresponding to the predicted event is retrieved from a preset prediction model library; the first target data is input into the second prediction model to obtain a second prediction result of the second prediction model being virtually run in the shadow mode; Optimizing the first prediction model or the second prediction model based on the first prediction result and the second prediction result; If the BMS detects the decision event, obtaining second target data corresponding to the decision event, determining a first decision solution actually executed on the battery device, obtaining a first decision result of executing the first decision solution based on the second target data, and simultaneously obtaining a second decision result of a second decision solution virtually executed in the shadow mode based on the second target data; Based on the first decision result and the second decision result, the first decision solution or the second decision solution is optimized.

2. The BMS optimization method based on shadow mode according to claim 1, characterized in that: After the first target data corresponding to the predicted event is screened out from the multimodal data, the method further includes: The first target data is classified based on a clustering algorithm to obtain classified first target data, and the classified first target data is uploaded to the cloud.

3. The BMS optimization method based on shadow mode according to claim 1, characterized in that: The optimizing the first prediction model or the second prediction model based on the first prediction result and the second prediction result includes: Obtaining the actual operating result of the battery device under the predicted event after a preset time; Calculating a first deviation value between the first prediction result and the operating result, and a second deviation value between the second prediction result and the operating result; If the first deviation value is less than the second deviation value, the first prediction model is marked as the target prediction model corresponding to the prediction event. When the number of times the first prediction model is marked reaches a preset number, the first prediction model is used as the standard prediction model corresponding to the prediction event, and the second prediction model is trained and optimized. If the second deviation value is smaller than the first deviation value, the second prediction model is marked as the target prediction model corresponding to the prediction event. When the second prediction model is marked a preset number of times, the second prediction model is used as the standard prediction model corresponding to the prediction event, and the first prediction model is trained and optimized.

4. The BMS optimization method based on shadow mode according to claim 1, characterized in that: The optimizing the first decision solution or the second decision solution based on the first decision result and the second decision result includes: Comparing the first decision result and the second decision result to obtain a comparison result; If the comparison result shows that the first decision result is better than the second decision result, the first decision solution is marked as the target decision solution corresponding to the decision event. When the number of times the first decision solution is marked reaches a preset number, the first decision solution is used as the standard decision solution corresponding to the decision event, and the second decision solution is optimized. If the comparison result shows that the second decision result is better than the first decision result, the second decision plan is marked as the target decision plan corresponding to the decision event. When the number of markings of the second decision plan reaches a preset number, the second decision plan is used as the standard decision plan corresponding to the decision event, and the first decision plan is optimized.

5. A BMS optimization system based on shadow mode, characterized in that: The system comprises: a shadow mode triggering module, configured to calculate in real time a deviation value between a first prediction result and a second prediction result, wherein the first prediction model and the second prediction model are in a real-time operating state; determine that the BMS has detected a prediction event when the deviation value is greater than a preset value; obtain various operating data of the battery device, and determine that the BMS has detected a decision event when it is detected that any of the operating data exceeds a corresponding preset range; and trigger a shadow mode corresponding to the BMS when it is determined that the BMS has detected a prediction event or a decision event; The prediction model operation module is used to obtain the multimodal data collected by the BMS in real time during the BMS detection process if the BMS detects the predicted event; selecting first target data corresponding to the predicted event, determining a first prediction model actually running on the battery device managed by the BMS, obtaining a first prediction result of the first prediction model based on the first target data, and simultaneously retrieving at least one second prediction model corresponding to the predicted event from a preset prediction model library; inputting the first target data into the second prediction model to obtain a second prediction result of the second prediction model virtually running in the shadow mode; a prediction model optimization module, configured to optimize the first prediction model or the second prediction model based on the first prediction result and the second prediction result; a decision scheme execution module, configured to, if the BMS detects the decision event, obtain second target data corresponding to the decision event, determine a first decision scheme actually executed on the battery device, obtain a first decision result of executing the first decision scheme based on the second target data, and simultaneously obtain a second decision result of a second decision scheme virtually executed in the shadow mode based on the second target data; A decision solution optimization module is used to optimize the first decision solution or the second decision solution based on the first decision result and the second decision result.

6. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the method according to any one of claims 1 to 4.

7. An electronic device, characterized in that: It includes a processor, a memory, a user interface and a network interface, the memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device executes the method according to any one of claims 1 to 4.

Images