A method and device for predicting the range of a new energy vehicle in a low battery state

Abstract

The application provides a method and device for predicting the mileage of a new energy vehicle in a low power state. The method comprises: determining that a target vehicle is in a low power state when it is determined that the remaining power of the target vehicle is lower than a preset power threshold; determining first running data, first battery state data and first environmental data of the target vehicle when the target vehicle is in the low power state; inputting the first running data, the first battery state data and the first environmental data into a pre-established prediction model to enable the prediction model to output a first predicted mileage of the target vehicle; wherein the prediction model is trained based on historical data of the same type of vehicle as the target vehicle, and the target vehicle is a new energy vehicle. The application realizes personalized and dynamic prediction of the cruising range of a new energy vehicle in a low power state, and improves the prediction accuracy of the remaining mileage of a new energy vehicle in a low power state.

Description

Technical Field

[0001] This application relates to the field of new energy vehicle technology, and in particular to a method and device for predicting the mileage of a new energy vehicle in a low battery state. Background Technology

[0002] The prospects for new energy development are very broad, especially in terms of environmental protection and sustainable economic development. New energy will gradually replace traditional energy and become the mainstream energy source in the future.

[0003] Batteries are crucial for the range of new energy vehicles, making energy storage technology a vital component in their development. Continuous technological innovation and improvement in batteries directly drive the advancement of new energy. However, when a new energy vehicle is low on battery power, the chemical reaction rate within the battery changes, affecting its energy output and consequently the rate at which the battery is consumed. Therefore, the remaining range displayed when a new energy vehicle is low on battery power is often inaccurate. Summary of the Invention

[0004] In view of this, embodiments of this application provide a method and apparatus for predicting the mileage of a new energy vehicle in a low-battery state, so as to solve the problem in the prior art that the remaining mileage of a new energy vehicle in a low-battery state is inaccurate.

[0005] A first aspect of this application provides a method for predicting the range of a new energy vehicle in a low-battery state, including:

[0006] When it is determined that the remaining battery power of the target vehicle is lower than the preset battery power threshold, the target vehicle is determined to be in a low battery state.

[0007] When the target vehicle is in a low battery state, determine the target vehicle's first operating data, first battery status data, and first environmental data; the first operating data includes speed, acceleration, driving mode, and air conditioning usage; the first battery status data includes remaining battery charge and battery health status; the first environmental data includes temperature and humidity.

[0008] The first operating data, the first battery status data, and the first environmental data are input into a pre-established prediction model so that the prediction model outputs the first predicted mileage of the target vehicle; wherein, the prediction model is trained based on historical data of similar vehicles to the target vehicle, and the target vehicle is a new energy vehicle.

[0009] A second aspect of this application provides a range prediction device for a new energy vehicle in a low-battery state, comprising:

[0010] The low battery state determination module is configured to determine that the target vehicle is in a low battery state when the remaining battery power of the target vehicle is determined to be lower than a preset battery threshold.

[0011] The first data determination module is configured to determine the target vehicle's first operating data, first battery status data, and first environmental data when the target vehicle is in a low battery state. The first operating data includes speed, acceleration, driving mode, and air conditioning usage. The first battery status data includes remaining battery power and battery health status. The first environmental data includes temperature and humidity.

[0012] The prediction model output module is configured to input the first operating data, the first battery state data, and the first environmental data into a pre-established prediction model so that the prediction model outputs the first predicted mileage of the target vehicle; wherein, the prediction model is trained based on historical data of similar vehicles to the target vehicle, and the target vehicle is a new energy vehicle.

[0013] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0014] A fourth aspect of this application provides a readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described method.

[0015] The beneficial effects of this application embodiment compared with the prior art are as follows: When the remaining battery power of the target vehicle is determined to be below a preset battery power threshold, the target vehicle is determined to be in a low battery state. When the target vehicle is in a low battery state, first operating data, first battery state data, and first environmental data of the target vehicle are determined. The first operating data, first battery state data, and first environmental data are input into a pre-established prediction model, so that the prediction model outputs a first predicted mileage for the target vehicle. The prediction model is trained based on historical data of similar vehicles to the target vehicle. The target vehicle is a new energy vehicle. This application embodiment realizes personalized and dynamic range prediction for new energy vehicles in a low battery state, improving the accuracy of remaining range prediction for new energy vehicles in a low battery state. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating a method for predicting the mileage of a new energy vehicle in a low-battery state, provided in an embodiment of this application.

[0018] Figure 2 This is a schematic diagram of the structure of a range prediction device for a new energy vehicle in a low battery state, provided in an embodiment of this application;

[0019] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0020] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0021] The following will describe in detail, with reference to the accompanying drawings, a method and apparatus for predicting the mileage of a new energy vehicle in a low-battery state according to an embodiment of this application.

[0022] Figure 1 This is a flowchart illustrating a method for predicting the mileage of a new energy vehicle in a low-battery state, as provided in an embodiment of this application. Figure 2 As shown, the method for predicting the range of this new energy vehicle in a low-battery state includes:

[0023] S101, when it is determined that the remaining battery power of the target vehicle is lower than the preset battery power threshold, the target vehicle is determined to be in a low battery state.

[0024] S102, when the target vehicle is in a low battery state, determine the target vehicle's first operating data, first battery status data, and first environmental data; the first operating data includes speed, acceleration, driving mode, and air conditioning usage; the first battery status data includes remaining battery power and battery health status; the first environmental data includes temperature and humidity;

[0025] S103, input the first operating data, the first battery state data and the first environmental data into the pre-established prediction model so that the prediction model outputs the first predicted mileage of the target vehicle; wherein, the prediction model is trained based on historical data of similar vehicles to the target vehicle, and the target vehicle is a new energy vehicle.

[0026] Specifically, in a low-battery state, if the remaining range prediction is inaccurate, drivers may face the risk of the battery running out, potentially causing the vehicle to stop on the road or even leading to a traffic accident. Accurate remaining range prediction can help drivers better plan their charging strategies to ensure driving safety.

[0027] Drivers typically rely on remaining range predictions to decide when to charge. Inaccurate predictions can lead to drivers charging when they don't need to, wasting time and resources, or failing to find a charging station when they do need to. Accurate predictions help optimize charging strategies to meet driver needs.

[0028] Accurate remaining range predictions can help drivers better plan their trips. If drivers know that the remaining range is very limited, they can choose shorter routes, avoid congestion, or find the nearest charging station. This helps improve driving efficiency and convenience.

[0029] The user experience of new energy vehicles is closely related to battery and remaining range predictions. If drivers frequently find these predictions inaccurate, they may become dissatisfied with new energy vehicles, potentially impacting their adoption and promotion. Accurate predictions help improve the user experience and enhance consumer confidence in new energy vehicles.

[0030] As batteries age over time and through repeated charge and discharge cycles, their capacity and performance may decline. This can result in the battery's actual energy storage capacity being lower than its initial specifications, thus affecting the accuracy of remaining range estimates. Furthermore, the discharge curve of a battery is typically non-linear, meaning that the battery voltage drops sharply when the battery level is low. This makes estimating remaining capacity and range based on battery voltage even more complex.

[0031] Taking all these factors into account, range prediction in low-battery conditions is often more challenging than prediction when the battery is fully charged. This is because at low battery levels, various uncertainties interact, making accurate prediction more difficult. Therefore, a method and device for predicting the range of new energy vehicles in low-battery conditions is needed to help solve the problem of accurate remaining range prediction.

[0032] When a new energy vehicle has a low battery level, simply displaying the estimated remaining range as a percentage of the remaining battery power often fails to accurately reflect the actual situation. For example, a vehicle with a claimed range of 500 kilometers might display an estimated remaining range of 100 kilometers when the battery is at 20%. However, in actual testing, the real remaining range may be far less than 100 kilometers, significantly less than the displayed value. Due to complex uncertainties, it is difficult to accurately estimate the actual remaining range. Therefore, the displayed range, simply converted from the battery percentage, has a significant error compared to the actual distance the vehicle can continue to travel.

[0033] This discrepancy between the displayed range and the actual range can easily lead to range anxiety for users, making it difficult to accurately determine how much further the remaining battery power can support, thus causing difficulties in planning subsequent trips and selecting charging stations. In the worst-case scenario, the displayed range may be insufficient to reach the destination or the next charging station, resulting in the battery running out while driving, causing significant inconvenience.

[0034] Furthermore, the battery capacity threshold is an important parameter set in the battery management system of new energy vehicles to determine the lower limit or critical value of the battery capacity. This threshold typically indicates how low the battery capacity is considered to be in a low-charge state. This threshold can be a fixed percentage, such as 20% of the battery capacity, or a fixed value, such as 20 kilowatt-hours.

[0035] Battery charge thresholds can typically be customized based on vehicle model, manufacturer, or driver preference. Different types of batteries and vehicles may have different battery charge threshold settings. Some drivers may prefer to continue driving when the battery is nearly depleted, while others may prefer to charge in advance to ensure sufficient charge.

[0036] Some vehicle battery management systems may take specific measures to optimize vehicle performance and range when the battery level approaches a threshold. For example, the system may reduce the vehicle's maximum power output to extend battery life. This optimization can help drivers better manage the vehicle when the battery is low.

[0037] The battery threshold setting can also be used to warn or remind the driver that the vehicle's battery level has dropped to a low level. This can be done through warning lights on the dashboard, audible alerts, on-screen messages, or a mobile app. This reminder helps the driver realize the battery is low and take steps to extend the driving range or find charging facilities.

[0038] The vehicle's battery management system continuously monitors the battery's charge level. This is typically achieved through sensors that measure parameters such as battery voltage, current, and temperature. The system tracks the battery charge in real time and compares it to preset charge thresholds.

[0039] When the battery level drops below a preset threshold, the vehicle's battery management system will confirm that the vehicle is in a low-battery state. At this point, the predicted remaining range may be inaccurate. A low-battery state is relative, not absolute, because the vehicle's battery management system typically sets a threshold to trigger warnings or remind the driver to take action.

[0040] When the battery is low, the predicted remaining range becomes more uncertain because many factors can affect the actual driving range, including driving behavior, environmental conditions, and battery status. This is why, when the battery is low, the battery management system usually prompts the driver to take actions such as finding a charging station or adopting energy-efficient driving modes to ensure the vehicle doesn't break down on the road.

[0041] When predicting remaining range, vehicles consider various factors and make estimates based on current conditions and data. However, due to uncertainties, this estimation is often more complex and may not be as accurate as when the battery is fully charged.

[0042] When a target vehicle has a low battery, the system collects comprehensive data to better understand the vehicle's operating status and environmental conditions, thereby improving the accuracy of remaining range prediction. This data includes primary operating data such as vehicle speed, air conditioning / heater usage, driving mode (e.g., Normal, Eco, Sport), acceleration and braking, as well as charging and discharging rates. These factors directly affect battery energy consumption. Different driving behaviors lead to different battery depletion rates. For example, rapid acceleration and hard braking typically increase battery energy consumption, while steady driving is more energy-efficient. Therefore, understanding driving behavior helps to more accurately estimate remaining range.

[0043] At the same time, different driving modes may adjust the vehicle's performance and power output, thus affecting battery usage. For example, Sport mode may provide higher power output, but it may also cause the battery to deplete faster. Therefore, understanding driving modes can help better predict remaining range.

[0044] When the target vehicle is in a low-battery state, the system also collects initial battery status data, such as the actual remaining charge and battery health status, as well as initial environmental data, such as temperature and humidity. The actual remaining charge is a key factor directly reflecting the battery's current energy storage status, while the battery health status indicates whether there are issues such as capacity degradation or performance decline. Environmental data such as temperature and humidity have a significant impact on battery performance; low temperatures may lead to performance degradation, while high temperatures may make the battery more susceptible to heat-related effects. By comprehensively considering these factors, the system can more accurately estimate the target vehicle's potential remaining driving range in a low-battery state, thereby improving driver convenience and safety.

[0045] The first operating data, the first battery status data, and the first environmental data are input into the pre-established prediction model in order to use this data to predict the remaining range.

[0046] To build an accurate remaining range prediction model, historical data from other vehicles of similar type to the target vehicle is needed. This historical data typically includes operational data, battery status data, and environmental data of new energy vehicles of the same model or type under various driving conditions. This data can include various factors such as different driving modes, driving conditions, and weather conditions, so that the model can more comprehensively understand the behavior and performance of new energy vehicles.

[0047] The process of building a predictive model typically includes steps such as data collection, feature selection, model training, and evaluation. In this case, the model needs to be fully trained to understand the characteristics and behavior of new energy vehicles and adapt to historical data from similar vehicles. This enables the model to better capture the consumption patterns of new energy vehicles when the battery is low.

[0048] Predictive models are typically dynamic systems that can be continuously improved and optimized. By constantly collecting new data and driving experience, models can gradually improve their accuracy and reliability to meet the needs of drivers. This can be achieved by updating model parameters, adding new features, or employing more advanced machine learning techniques.

[0049] By inputting initial operational data, initial battery state data, and initial environmental data into the predictive model, a more accurate remaining range prediction is generated. This data-driven approach helps to better understand battery usage and vehicle performance, thereby improving the accuracy and reliability of remaining range predictions and enhancing driver convenience and safety.

[0050] According to the technical solution provided in this application, when the remaining battery power of the target vehicle is determined to be lower than a preset battery power threshold, the target vehicle is determined to be in a low battery state. When the target vehicle is in a low battery state, first operating data, first battery state data, and first environmental data of the target vehicle are determined. The first operating data, first battery state data, and first environmental data are input into a pre-established prediction model so that the prediction model outputs a first predicted mileage of the target vehicle. The prediction model is trained based on historical data of similar vehicles to the target vehicle. The target vehicle is a new energy vehicle. This application embodiment realizes personalized and dynamic range prediction for new energy vehicles in a low battery state, improving the accuracy of remaining range prediction for new energy vehicles in a low battery state.

[0051] In some embodiments, the method further includes: plotting a discharge curve of the target vehicle based on historical driving data of the target vehicle; determining a release critical point of the battery of the target vehicle based on the inflection point of the discharge curve; determining a first critical charge value corresponding to the release critical point; and determining the first critical charge value as a charge threshold.

[0052] Specifically, the discharge critical point refers to the point where the charge of a battery undergoes a sudden change during discharge. The discharge characteristics of a battery differ depending on its state of charge. As the battery charge decreases, a breakpoint appears in the discharge curve; after this point, the remaining capacity of the battery is rapidly released. This point is called the discharge critical point. The discharge critical point reflects the nonlinear characteristics of battery discharge.

[0053] When the battery charge is above the release threshold, its discharge is relatively stable. However, when the charge drops below the release threshold, the battery discharge curve changes abruptly, and the charge is released rapidly. For new energy vehicles, the release threshold indicates a point where the battery enters a low-charge state. When the charge drops to the release threshold, the vehicle's driving range decreases rapidly. Therefore, intelligently determining the release threshold can help to more accurately judge the battery's state of charge, thereby optimizing the setting of the charge threshold and improving the accuracy of mileage prediction in the low-charge state of new energy vehicles.

[0054] Furthermore, historical driving data of the target vehicle is used to plot a discharge curve. This curve reflects how the battery charge changes as the mileage decreases. In this way, the battery's discharge behavior under different states of charge can be visualized. The discharge curve plotted by analyzing the target vehicle's historical driving data (including mileage and remaining charge) shows the trend of the vehicle's charge decreasing as the mileage decreases.

[0055] On the discharge curve, the discharge critical point is a crucial inflection point, marking the transition of the battery from a stable discharge state to a rapid discharge state. At this point, the battery's discharge rate changes significantly. Identifying this point is vital because it reveals the battery's discharge characteristics at low charge levels. The remaining charge value corresponding to the discharge critical point is defined as the first critical charge value. This value is the boundary between the battery's transition from a stable discharge state to a rapid discharge state and is a key indicator for measuring the battery's state of discharge.

[0056] This first critical battery level is set as the battery threshold. This means that when the battery level drops below this threshold, the vehicle is considered to be in a low-battery state. Setting this threshold is crucial for predicting the driving range of new energy vehicles, as it directly affects the calculation and display of remaining range.

[0057] The method described in this application enables the system to more accurately identify and predict the actual state and performance of the battery, thereby providing drivers with more accurate and reliable information on remaining driving range. This not only improves driving safety but also optimizes overall energy management efficiency, enhancing user trust and satisfaction with new energy vehicles.

[0058] In some embodiments, plotting a discharge curve of a target vehicle based on its historical driving data includes: determining the driving mileage and remaining battery power corresponding to the historical driving data; and plotting a discharge curve based on the driving mileage and remaining battery power.

[0059] Specifically, each driver has different driving habits, which affect vehicle energy consumption. By analyzing an individual's historical driving data, it's possible to better understand how a particular driver influences battery discharge. For example, some drivers may accelerate more frequently or use the air conditioning more often, leading to faster battery depletion. The vehicle's operating environment, such as city traffic and highway driving, as well as different climate conditions, all affect battery discharge characteristics. Individual historical driving data can capture battery performance under these conditions. Mileage and remaining battery charge can encompass an individual's driving habits and the vehicle's operating environment.

[0060] As batteries age, their capacity and discharge efficiency gradually decrease. Using an individual's historical driving data can more accurately reflect the current state of the battery, providing a more accurate discharge curve compared to using standardized or averaged data.

[0061] Furthermore, in order to plot the discharge curve of the target vehicle, it is first necessary to collect the historical driving data of the target vehicle. This historical driving data contains relevant information about the target vehicle in different driving cycles, mainly including two key parameters: mileage and remaining battery power.

[0062] Mileage reflects the cumulative distance traveled by the target vehicle within a certain driving cycle. This can be obtained directly through devices such as an onboard odometer. Remaining battery charge indicates the remaining battery charge of the target vehicle at a specific moment during driving. The battery management system can monitor the remaining battery charge in real time and display it as a percentage.

[0063] By analyzing historical mileage and remaining battery charge, it's possible to understand the battery's discharge patterns at different mileage levels. This analysis helps reveal how battery performance changes with usage time and conditions.

[0064] Based on this data, a discharge curve can be plotted. This chart visually shows how the remaining battery charge changes as driving mileage increases. The shape and slope of the curve provide important information about the battery's discharge rate and efficiency.

[0065] One of the important uses of discharge profiles is to identify key characteristics of the battery discharge process, such as the steady discharge phase and the rapid discharge phase. Especially at low charge levels, this analysis helps to understand how the battery reacts and predict its future discharge behavior.

[0066] In some embodiments, when the target vehicle cannot determine the release critical point based on the discharge curve, the method further includes: recording the voltage and current curves during the discharge process based on the discharge process of the sample vehicle; analyzing the voltage and current curves to determine the inflection point in the voltage and current curves as the release critical point; determining the second critical charge value corresponding to the release critical point; and determining the second critical charge value as the charge threshold.

[0067] Specifically, for new vehicles, the battery may not have undergone enough charge-discharge cycles to establish a stable discharge pattern. In this case, the discharge curve may not yet show a clear enough trend or characteristic point, making it difficult to identify the release critical point.

[0068] If the driver's previous driving habits did not allow the battery charge to drop to a sufficiently low level, then a clear discharge threshold may not appear on the discharge curve. This is because the discharge threshold typically occurs when the battery charge is low.

[0069] Therefore, in these cases, other methods, such as analyzing voltage-current curves, may be needed to help determine the battery's release threshold, thereby improving the accuracy of charge threshold prediction.

[0070] Furthermore, using a sample vehicle of the same type as the target vehicle, the operating parameters of the sample vehicle's battery pack during the discharge process were measured in an experimental environment. First, it was necessary to monitor and record the voltage and current data of the sample vehicle throughout the discharge process, which provides detailed information about how the battery discharges under different charge states.

[0071] By testing the battery discharge process under different initial charge conditions, the relationship curves of the battery's terminal voltage and current versus time during discharge can be recorded. Analysis of the voltage and current curves reveals that when a significant inflection point or change in curvature appears, it usually indicates that the battery pack is transitioning from stable discharge to a stage of significant degradation. The point corresponding to this state of charge is the release critical point.

[0072] Based on the battery status at the point of release threshold, the remaining battery level is converted into a corresponding numerical value and determined as the second critical battery level. This second critical battery level is then set as the battery threshold. This means that when the battery level drops below this threshold, the vehicle is considered to be in a low-battery state. Setting this threshold is crucial for predicting the driving range of new energy vehicles, as it directly affects the calculation and display of the remaining range.

[0073] In some embodiments, the method further includes: determining an initial model; constructing a training dataset for the initial model based on historical data; the historical data includes second operating data, second battery state data, and second environmental data of vehicles of the same type; determining the annotation information corresponding to the training dataset; the annotation information includes the actual remaining mileage corresponding to the historical data; inputting the second operating data, second battery state data, and second environmental data into the initial model to enable the initial model to output an initial prediction result; using the initial prediction result and the annotation information, determining the loss index of the initial model based on a preset loss function; adjusting the hyperparameters of the initial model when the loss index does not meet the training conditions; and determining the initial model as a prediction model when the loss index meets the training conditions.

[0074] Specifically, predictive models can comprehensively analyze multiple factors, including vehicle operating data, battery status, and environmental conditions, to more accurately estimate the remaining range of an electric vehicle when the battery is low. This helps drivers better plan their trips and take appropriate measures to avoid running out of power. Each vehicle's battery status and performance vary and are influenced by various factors. Predictive models can be customized based on the target vehicle's historical data and characteristics to better adapt to the vehicle's battery performance and improve personalized prediction accuracy. Electric vehicles with reliable remaining range predictions provide a better driving experience. Drivers can better manage the battery status, thus enjoying a smooth driving experience even when the battery is low.

[0075] Furthermore, the initial model is a foundational model established at the start of the prediction task. It provides a starting point and baseline performance, laying the foundation for subsequent data training and model improvement. As data accumulates and the model is continuously optimized, the initial model will gradually adapt to new conditions to provide more accurate predictions.

[0076] The initial model is typically a simple model or initial parameter settings. Over time, the vehicle's battery performance may change, driving habits may change, and environmental conditions may also change. Through continuous training, the model can adapt to these changes to maintain accuracy.

[0077] The initial model's training dataset is constructed based on historical data from similar vehicles. Historical data refers to information already collected and recorded regarding the target vehicle and similar vehicles' past driving history and battery status. This data is typically used to train and improve the prediction model, enhancing the accuracy and reliability of mileage predictions under low battery conditions.

[0078] Historical data comes from the actual operating conditions of the target vehicle and similar vehicles. This data can be acquired through onboard sensors, recorders, vehicle management systems, or other data collection methods. Historical data typically includes information at a series of points in time. It may cover different time periods, often including a longer historical data set for training and validation.

[0079] Historical data includes secondary operational data, secondary battery state data, and secondary environmental data from similar vehicles. This data should represent the vehicle's driving and battery status under different conditions to ensure the model's diversity and generalization ability.

[0080] The annotation information in the historical data includes the actual remaining mileage corresponding to the historical data points. This is a crucial supervised learning label, representing the true output or target value of the historical data points. The accuracy of the annotation information is essential because it will be used to train the model, enabling it to learn how to predict remaining mileage.

[0081] By inputting second operational data, second battery state data, and second environmental data into the initial model, the initial model generates initial prediction results. These initial prediction results represent the initial model's predictions for the target vehicle under the current parameter settings.

[0082] The loss function is calculated using labeled information and initial predictions. The loss function is typically used to measure the difference between the model's predictions and the actual observations. It can take various forms, depending on the task and model type.

[0083] In machine learning, the value of the loss function is used as an important metric. It's used to check whether the loss exponent meets the training conditions. These conditions typically include a threshold or a set of convergence criteria. If the loss exponent does not meet these conditions, it indicates that the model's performance has not yet reached the expected level.

[0084] If the loss exponent does not meet the training requirements, it may be necessary to adjust the hyperparameters of the initial model. These hyperparameters include the learning rate, regularization coefficient, and model structure. The goal of hyperparameter tuning is to find better model parameter settings to reduce the loss function value.

[0085] After multiple iterations, the initial model was determined as the final prediction model when the loss exponent met the training conditions. This model has been trained through supervised learning to better predict the remaining range of the target vehicle when the battery is low.

[0086] In summary, the prediction model is trained through iterative iterations, loss function optimization, and hyperparameter tuning to ensure it provides accurate mileage predictions under low battery conditions. The model training process ensures the model's performance and reliability, thereby providing accurate information for driving new energy vehicles.

[0087] In some embodiments, the method further includes: determining third operating data, third battery state data, and third environmental data of the target vehicle; inputting the third operating data, third battery state data, and third environmental data into a pre-established prediction model so that the prediction model outputs a second predicted mileage of the target vehicle; and determining a third predicted mileage based on the first predicted mileage and the second predicted mileage.

[0088] Specifically, the reason for measuring the third predicted mileage is to improve the accuracy and reliability of the remaining mileage prediction. If the vehicle experiences high power demands while in a low-battery state, such as rapid acceleration, high-speed driving, or climbing hills, the battery may deplete more quickly. The first predicted mileage fails to account for these rapid battery depletion scenarios in time, and therefore may underestimate the remaining mileage.

[0089] In adverse road conditions, such as muddy, rough, or icy roads, vehicles may require more energy to cope with the challenges, thus reducing battery range. First-predicted mileage typically cannot anticipate these conditions and may therefore overestimate the remaining range.

[0090] A driver's driving habits and patterns can change over time. For example, high-speed driving on highways and frequent stopping in the city can have different effects on battery range. First-predicted range may not capture these changes, leading to inaccurate estimates.

[0091] Taking all the above factors into account, range prediction under low battery conditions is highly uncertain. Even under similar conditions, different vehicles may behave differently, so the accuracy of the initial range prediction may vary depending on the circumstances.

[0092] Therefore, to more accurately estimate remaining range, it is often necessary to use more recent data and predictive models to provide second and third predicted ranges, better reflecting current driving conditions and battery status. Such a multi-level prediction approach can provide more reliable information when the battery is low.

[0093] Furthermore, third-party operational data typically includes vehicle speed, driving mode, and whether air conditioning is used, reflecting the vehicle's driving behavior. Third-party battery status data includes information such as actual battery charge and battery health status, describing the battery's current condition. Third-party environmental data includes environmental factors such as temperature, humidity, and wind speed, which can affect battery performance and energy consumption.

[0094] The acquired third operational data, third battery status data, and third environmental data are input into a pre-built predictive model. This model can be a machine learning model, a neural network, a statistical model, or other predictive methods. The model will generate a prediction result based on these input data, namely the second predicted mileage.

[0095] The first and second predicted mileages are combined to determine the final third predicted mileage. This process may involve correcting, adjusting, or combining the methods used for the first and second predictions to provide a more accurate forecast.

[0096] In summary, the first predicted mileage is an initial prediction based on vehicle data when the battery is initially low. The second predicted mileage is a second prediction made using updated data after the target vehicle has continued driving for a certain period and distance. The third predicted mileage analyzes and compares the first two predictions, taking into account the current driving conditions, and makes adjustments and corrections to provide the most accurate predicted mileage result.

[0097] This multi-level prediction process aims to provide more accurate and reliable range forecasts in low-battery conditions to help drivers make informed decisions, such as whether to charge, adjust trip plans, or take other measures to ensure a safe arrival at their destination. It utilizes a combination of real-time data and predictive models to continuously improve prediction accuracy, adapting to different driving situations and changing environmental conditions.

[0098] In some embodiments, the time when the first operating data, the first battery state data, and the first environmental data are determined is a first moment; the time when the third operating data, the third battery state data, and the third environmental data are determined is a second moment; the first moment is earlier than the second moment; then determining the third predicted mileage based on the first predicted mileage and the second predicted mileage includes: determining the first driving distance of the target vehicle between the first moment and the second moment; determining the first remaining distance based on the first predicted mileage and the first driving distance; determining a distance compensation value when the difference between the first remaining distance and the second predicted mileage is greater than a preset reference threshold; and determining the third predicted mileage based on the distance compensation value and the second predicted mileage.

[0099] Specifically, the first moment refers to the time point at which the first set of data is recorded when the battery is low, including the first operational data, the first battery status data, and the first environmental data. The second moment is the time point at which the second set of data is recorded at a later time, including the third operational data, the third battery status data, and the third environmental data. The first moment is earlier than the second moment, which means that the data recording at the second moment occurs after the first moment.

[0100] The first driving distance refers to the actual distance traveled by the target vehicle between the first and second moments. This distance is calculated based on the vehicle's operating data and driving records, reflecting the vehicle's actual usage when the battery is low.

[0101] The vehicle is equipped with various sensors and monitoring devices that record operational data such as speed, driving mode, acceleration, and braking. Based on this data, the system analyzes the vehicle's actual driving conditions when the battery is low. This may include considerations such as urban roads, highways, and traffic conditions. Based on the operational data and driving records, the system calculates the actual distance traveled by the target vehicle within a specific time period; this is the initial driving distance.

[0102] The first remaining distance refers to the estimated remaining distance the vehicle can travel when its battery is low, after considering the initial distance already traveled by the target vehicle during the initial prediction. The predicted mileage output by the model at the first moment is the first predicted mileage. The actual distance the vehicle travels from the first moment to the second moment is defined as the first travel distance. Subtracting the first travel distance from the first predicted mileage gives the remaining value as the first remaining distance.

[0103] Furthermore, when the difference between the first remaining distance and the second predicted mileage exceeds a preset reference threshold, distance compensation is required to more accurately estimate the vehicle's range in a low-battery state. When there is a significant error between the first remaining distance (first predicted mileage minus the first traveled distance) and the second predicted mileage, it indicates that the two different calculation methods are inconsistent in predicting the remaining mileage at the second time point, and adjustments are necessary.

[0104] In theory, both the first remaining distance and the second predicted mileage represent the estimated remaining mileage at the second time point, and their values ​​should be close. If the second predicted mileage is significantly lower than the first remaining distance, it indicates that the actual power consumption between the first and second time points is greater than the prediction at the first time point, and the consumption rate exceeds expectations.

[0105] Considering the battery discharge characteristics, this situation may mean that the rate of power consumption will continue to increase. In other words, the actual rate of power consumption will be higher than the predicted rate at the second stage. Therefore, the second predicted range may still overestimate the actual driving range. To address this, a distance compensation mechanism needs to be introduced to correct the prediction. The compensation value is determined based on the difference between the first and second predictions, appropriately reducing the prediction in the second stage to obtain a more accurate and reliable final prediction result.

[0106] For example, suppose the predicted remaining battery range at the first moment is 100 kilometers, and you actually travel 20 kilometers during that time. According to the prediction, you should still have 80 kilometers left. However, when you make a prediction at the second moment, the result shows that the remaining battery range is only 50 kilometers, instead of the expected 80 kilometers.

[0107] This indicates that the actual rate of battery consumption from the first to the second moment was greater than the initially predicted rate. Considering the characteristics of battery discharge, this situation may mean that the rate of battery consumption will continue to increase thereafter. Therefore, based solely on the prediction at the second moment, 50 kilometers may be overly optimistic. Compensation is needed at this point, appropriately reducing the predicted 50-kilometer range to obtain a more accurate result. By subtracting the distance compensation value, the vehicle's range in a low-battery state can be estimated more accurately, allowing for more reliable trip planning and avoiding inconvenience and risks caused by insufficient battery power.

[0108] From the first moment to the second, the actual rate of battery consumption was significantly higher than the initial prediction. This means that the rate of battery consumption may continue to increase, and relying solely on the second predicted mileage may be overly optimistic. To improve prediction accuracy, distance compensation is necessary. Specifically, the system calculates the difference between the first remaining distance and the second predicted mileage, and then compares this difference with a pre-set reference threshold. This reference threshold is typically set based on the vehicle model, battery performance, and other factors, representing an acceptable range of prediction error. If this difference exceeds the preset reference threshold, distance compensation is required. The distance compensation value is a correction factor used to adjust the prediction of the first remaining distance to more accurately reflect the vehicle's actual range.

[0109] Once the distance compensation value is obtained, it is applied to the second predicted mileage. This means that the second predicted mileage is adjusted based on the distance compensation value to generate the third predicted mileage. The third predicted mileage represents a more accurate driving range for the vehicle when the battery is low.

[0110] In summary, the purpose of distance compensation is to provide more accurate range predictions when the battery is low. By monitoring and correcting the difference between the first remaining distance and the second predicted range, drivers can better understand the vehicle's battery status and range, enabling them to make more informed driving and charging decisions.

[0111] By analyzing the operational data between the first and second moments, the vehicle's average speed during this period can be calculated. Then, multiplying the average speed by the time difference yields the vehicle's first distance traveled. This distance represents the actual distance the vehicle traveled while the battery was low.

[0112] The first predicted range is an estimate generated based on the vehicle's initial data and the model, representing the distance the vehicle is expected to travel when the battery is low. However, since the actual driving conditions of the vehicle may differ from the model's estimate, the first remaining range needs to be calculated based on the actual distance traveled by the vehicle between the first and second time points.

[0113] All of the above-mentioned optional technical solutions can be combined in any way to form the optional embodiments of this application, and will not be described in detail here.

[0114] The following are embodiments of the apparatus described in this application, which can be used to execute the embodiments of the method described in this application. For details not disclosed in the apparatus embodiments of this application, please refer to the embodiments of the method described in this application.

[0115] Figure 2 This is a schematic diagram of a range prediction device for a new energy vehicle in a low-battery state, provided in an embodiment of this application. Figure 2 As shown, the range prediction device for this new energy vehicle in a low battery state includes:

[0116] The low battery state determination module 201 is configured to determine that the target vehicle is in a low battery state when the remaining battery power of the target vehicle is determined to be lower than a preset battery threshold.

[0117] The first data determination module 202 is configured to determine the first operating data, first battery status data, and first environmental data of the target vehicle when the target vehicle is in a low battery state. The first operating data includes speed, acceleration, driving mode, and air conditioning usage. The first battery status data includes the remaining battery charge and battery health status. The first environmental data includes temperature and humidity.

[0118] The prediction model output module 203 is configured to input the first operating data, the first battery state data and the first environmental data into a pre-established prediction model so that the prediction model outputs the first predicted mileage of the target vehicle; wherein, the prediction model is trained based on historical data of similar vehicles to the target vehicle, and the target vehicle is a new energy vehicle.

[0119] In some embodiments, Figure 2 The low battery state determination module 201 draws a discharge curve of the target vehicle based on the historical driving data of the target vehicle; determines the release critical point of the target vehicle's battery based on the inflection point of the discharge curve; determines the first critical charge value corresponding to the release critical point; and determines the first critical charge value as the charge threshold.

[0120] In some embodiments, Figure 2 The low battery status determination module 201 determines the driving mileage and remaining battery value corresponding to historical driving data; and draws a discharge curve based on the driving mileage and remaining battery value.

[0121] In some embodiments, Figure 2 The low-charge state determination module 201 records the voltage and current curves during the discharge process of the sample vehicle; analyzes the voltage and current curves to determine the inflection point in the voltage and current curves as the release critical point; determines the second critical charge value corresponding to the release critical point; and determines the second critical charge value as the charge threshold.

[0122] In some embodiments, Figure 2The prediction model output module 203 determines the initial model; constructs a training dataset for the initial model based on historical data; the historical data includes second operating data, second battery state data, and second environmental data of similar vehicles; determines the annotation information corresponding to the training dataset; the annotation information includes the actual remaining mileage corresponding to the historical data; inputs the second operating data, second battery state data, and second environmental data into the initial model so that the initial model outputs an initial prediction result; uses the initial prediction result and annotation information to determine the loss index of the initial model based on a preset loss function; when the loss index does not meet the training conditions, adjusts the hyperparameters of the initial model; when the loss index meets the training conditions, the initial model is determined as the prediction model.

[0123] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0124] Figure 3 This is a schematic diagram of the electronic device 3 provided in an embodiment of this application. Figure 3 As shown, the electronic device 3 of this embodiment includes: a processor 301, a memory 302, and a computer program 303 stored in the memory 302 and executable on the processor 301. When the processor 301 executes the computer program 303, it implements the steps in the various method embodiments described above. Alternatively, when the processor 301 executes the computer program 303, it implements the functions of each module / unit in the various device embodiments described above.

[0125] Electronic device 3 can be a desktop computer, laptop, handheld computer, cloud server, or other electronic device. Electronic device 3 may include, but is not limited to, processor 301 and memory 302. Those skilled in the art will understand that... Figure 3 This is merely an example of electronic device 3 and does not constitute a limitation on electronic device 3. It may include more or fewer components than shown, or different components.

[0126] The processor 301 may be a central processing unit (CPU), or 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.

[0127] The memory 302 can be an internal storage unit of the electronic device 3, such as a hard disk or RAM. The memory 302 can also be an external storage device of the electronic device 3, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. The memory 302 can also include both internal and external storage units of the electronic device 3. The memory 302 is used to store computer programs and other programs and data required by the electronic device.

[0128] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments 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 unit can be implemented in hardware or as a software functional unit.

[0129] If integrated modules / units are implemented as software functional units and sold or used as independent products, they can be stored in a readable storage medium (e.g., a computer-readable storage medium). Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program may include computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. A computer-readable storage medium may include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0130] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for predicting the range of a new energy vehicle in a low-battery state, characterized in that, The method includes: Based on the historical driving data of the target vehicle, determine the driving mileage and remaining battery power corresponding to the historical driving data, and plot the discharge curve of the target vehicle based on the driving mileage and remaining battery power. Based on the inflection point of the discharge curve, determine the release critical point of the target vehicle's battery, where the remaining battery capacity will be rapidly released after the inflection point. Determine the first critical battery power value corresponding to the release critical point, and set the first critical battery power value as the battery power threshold. When it is determined that the remaining battery power of the target vehicle is lower than the preset battery power threshold, it is determined that the target vehicle is in a low battery state. When the target vehicle is in the low battery state, the following first operating data, first battery status data, and first environmental data of the target vehicle are determined: the first operating data includes speed, acceleration, driving mode, and air conditioning usage; the first battery status data includes remaining battery power and battery health status; and the first environmental data includes temperature and humidity. The first operating data, the first battery status data, and the first environmental data are input into a pre-established prediction model so that the prediction model outputs the first predicted mileage of the target vehicle; wherein, the prediction model is trained based on historical data of similar vehicles to the target vehicle, and the target vehicle is a new energy vehicle.

2. The method according to claim 1, characterized in that, When the target vehicle cannot determine the release critical point based on the discharge curve, the method further includes: Based on the discharge process of the sample vehicle, record the voltage and current curves during the discharge process; Analyze the voltage-current curves and determine the inflection points in the voltage-current curves as the release critical points; Determine the second critical charge value corresponding to the release critical point; The second critical charge value is determined as the charge threshold.

3. The method according to claim 1, characterized in that, Also includes: Determine the initial model; The training dataset for the initial model is constructed based on the historical data; the historical data includes the second operating data, the second battery status data, and the second environmental data of the same type of vehicle. Determine the annotation information corresponding to the training dataset; The annotation information includes the actual remaining mileage corresponding to the historical data; The second operating data, the second battery status data, and the second environmental data are input into the initial model so that the initial model outputs an initial prediction result. Using the initial prediction results and the annotation information, the loss index of the initial model is determined based on a preset loss function; If the loss index does not meet the training conditions, adjust the hyperparameters of the initial model; When the loss index meets the training conditions, the initial model is determined as the prediction model.

4. The method according to any one of claims 1 to 3, characterized in that, Also includes: The third operating data, third battery status data, and third environmental data of the target vehicle are determined. The third operating data, the third battery status data, and the third environmental data are input into a pre-established prediction model so that the prediction model outputs a second predicted mileage for the target vehicle. A third predicted mileage is determined based on the first predicted mileage and the second predicted mileage.

5. The method according to claim 4, characterized in that, The time when the first operating data, first battery status data, and first environmental data are determined is designated as the first time point; the time when the third operating data, third battery status data, and third environmental data are determined is designated as the second time point; the first time point is earlier than the second time point; then, determining the third predicted mileage based on the first predicted mileage and the second predicted mileage includes: Determine the first travel distance of the target vehicle between the first time point and the second time point; The first remaining distance is determined based on the first predicted mileage and the first travel distance; When the difference between the first remaining distance and the second predicted mileage is greater than a preset reference threshold, a distance compensation value is determined; The third predicted mileage is determined based on the distance compensation value and the second predicted mileage.

6. A range prediction device for a new energy vehicle in a low battery state, characterized in that, include: The low battery state determination module is configured to determine the mileage and remaining battery value corresponding to the historical driving data of the target vehicle, and to plot the discharge curve of the target vehicle based on the mileage and remaining battery value; to determine the release critical point of the battery of the target vehicle based on the inflection point of the discharge curve, wherein the remaining battery capacity will be rapidly released after the inflection point; to determine a first critical battery value corresponding to the release critical point, and to set the first critical battery value as a battery threshold; and to determine that the target vehicle is in a low battery state when the remaining battery capacity of the target vehicle is determined to be lower than the preset battery threshold. The first data determination module is configured to determine the first operating data, the first battery status data, and the first environmental data of the target vehicle when the target vehicle is in the low battery state. The first set of operational data includes speed, acceleration, driving mode, and air conditioning usage. The first battery status data includes remaining battery power and battery health status; the first environmental data includes temperature and humidity. The prediction model output module is configured to input the first operating data, the first battery state data, and the first environmental data into a pre-established prediction model, so that the prediction model outputs the first predicted mileage of the target vehicle; wherein the prediction model is trained based on historical data of similar vehicles to the target vehicle, and the target vehicle is a new energy vehicle.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 5.

8. A readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 5.

Images