Complaint prediction method, electronic equipment and vehicle

Abstract

The invention relates to the technical field of vehicles, and provides a complaint prediction method, electronic equipment and a vehicle. The method comprises the following steps: acquiring vehicle information of a vehicle and environment characteristics of an environment where the vehicle is located; determining an electricity utilization risk coefficient of the vehicle according to the electricity utilization behavior characteristics of the user after the vehicle is parked; determining a parking risk coefficient of the vehicle according to the start-stop time sequence characteristics of the vehicle; according to the health degree of the storage battery in the vehicle, the health degree attenuation slope of the storage battery is determined; and predicting a complaint probability value of the user for the power shortage of the storage battery based on the vehicle information, the environmental characteristics, the power utilization risk coefficient, the parking risk coefficient, the health degree and the health degree attenuation slope. The method can accurately predict the complaint probability value of the user for the power shortage of the storage battery.

Description

Technical Field

[0001] This application relates to the field of vehicle technology, specifically to a complaint prediction method, electronic equipment, and vehicle. Background Technology

[0002] Currently, a depleted vehicle battery can interfere with normal vehicle use. Related technologies typically employ physical parameter detection methods, collecting electrical signal data such as battery voltage, current, and temperature to determine the battery's charge level. However, these technologies cannot accurately predict the risk of a depleted battery, nor can they accurately anticipate potential customer complaints, thus failing to meet the actual needs of vehicle after-sales risk management and improved user experience. Summary of the Invention

[0003] This application provides a complaint prediction method, electronic device, and vehicle to solve the technical problem of being unable to accurately predict the probability of user complaints regarding a vehicle's battery being depleted.

[0004] A first aspect of this application provides a complaint prediction method, the method comprising: acquiring vehicle information and environmental characteristics of the environment in which the vehicle is located; determining an electricity consumption risk coefficient of the vehicle based on the user's electricity consumption behavior characteristics after the vehicle is parked; determining a parking risk coefficient of the vehicle based on the vehicle's start-stop timing characteristics; determining a battery health decay slope based on the health status of the battery in the vehicle; and predicting the probability value of a user's complaint regarding a depleted battery based on the vehicle information, the environmental characteristics, the electricity consumption risk coefficient, the parking risk coefficient, the health status, and the health decay slope.

[0005] In this embodiment, the vehicle's electricity consumption risk coefficient can be quantified in a fine-grained manner by the user's electricity consumption behavior characteristics after the vehicle is parked; the vehicle's parking risk coefficient can be quantified in a fine-grained manner by the vehicle's start-stop timing characteristics; and then, by combining the vehicle information, environmental characteristics, electricity consumption risk coefficient, parking risk coefficient, health status, and health status decay slope, the probability value of user complaints regarding the battery's low charge can be accurately predicted.

[0006] According to an embodiment of this application, the electricity consumption behavior characteristics include the on-time and off-time of the target electrical equipment in the vehicle after the vehicle is parked. The step of determining the electricity consumption risk coefficient of the vehicle based on the user's electricity consumption behavior characteristics after the vehicle is parked includes: determining the duration of a single electricity consumption of the target electrical equipment based on the off-time and the on-time; determining the frequency of abnormal electricity consumption of the vehicle based on a comparison between the duration of a single electricity consumption and a first preset duration; and calculating the electricity consumption risk coefficient based on the frequency of abnormal electricity consumption.

[0007] This application embodiment can accurately determine the duration of a single power consumption by using the on-time and off-time of the target electrical device after the vehicle is parked. Furthermore, by comparing the duration of a single power consumption with a first preset duration, the frequency of abnormal power consumption of the vehicle can be determined. Further, by analyzing the frequency of abnormal power consumption, the impact of the vehicle's power consumption on the battery can be quantified in a fine-grained manner.

[0008] According to an embodiment of this application, the start-stop timing characteristics include the vehicle's engine shutdown time within a preset time period. Determining the vehicle's parking risk coefficient based on the vehicle's start-stop timing characteristics includes: selecting a target time from the end time of the preset time period or the vehicle's ignition time within the preset time period; determining the vehicle's engine shutdown duration within the preset time period based on the engine shutdown time and the target time; determining the frequency of abnormal parking of the vehicle within the preset time period based on a comparison between the engine shutdown duration and a second preset duration; and calculating the parking risk coefficient based on the frequency of abnormal parking.

[0009] This application embodiment can reasonably determine the target time by using the end time of a preset time period or the ignition time of the vehicle within the preset time period. Then, by combining the vehicle's shutdown time within the preset time period with the target time, the vehicle's shutdown duration within the preset time period can be accurately determined. By comparing the vehicle's shutdown duration with a second preset duration, the frequency of abnormal parking within the preset time period can be determined. Furthermore, by analyzing the frequency of abnormal parking, the impact of the vehicle's ignition status on the battery can be quantified in a fine-grained manner.

[0010] According to an embodiment of this application, predicting the probability value of a user's complaint regarding the battery's low charge based on the vehicle information, environmental characteristics, electricity risk coefficient, parking risk coefficient, health status, and health status decay slope includes: using the vehicle information, environmental characteristics, electricity risk coefficient, parking risk coefficient, health status, and health status decay slope as input features of a prediction model, and using the prediction model to determine the complaint probability value.

[0011] This application embodiment, by combining multi-dimensional features such as vehicle information, environmental characteristics, electricity risk coefficient, parking risk coefficient, health status, and health status decay slope, can solve the problem of risk identification deviation caused by coarse and single granularity of input features. By combining multi-dimensional features such as static and dynamic features, it can accurately predict the probability value of user complaints about the battery being depleted.

[0012] According to an embodiment of this application, the method further includes: calculating an importance score for each input feature based on the Gini coefficient of each input feature; determining a target input feature from multiple input features based on the importance score; and outputting risk attribution information for the battery depletion based on the feature information of the vehicle on the target input feature.

[0013] This application embodiment uses the Gini coefficient of each input feature to accurately quantify the importance score of each input feature. Then, based on the importance score, target input features can be accurately selected. Using the vehicle's feature information on the target input features, matching risk attribution information can be output to the user, enabling the user to take appropriate strategies and improving the user experience.

[0014] According to an embodiment of this application, the training method of the prediction model includes: acquiring a historical training dataset, wherein each training sample in the historical training dataset includes: vehicle information, environmental features, electricity risk coefficient, parking risk coefficient, battery health and health decay slope, and a label corresponding to the training sample, the label being used to indicate whether the user corresponding to the training sample has complained about battery depletion; setting category weights for multiple training samples according to the labels, such that the weight of complaint-type samples is greater than the weight of non-complaint-type samples, the weight of complaint-type samples indicating that the user corresponding to the complaint-type samples has complained about battery depletion, and the weight of non-complaint-type samples indicating that the user corresponding to the non-complaint-type samples has not complained about battery depletion; and training an initial prediction model using the historical training dataset with set category weights, with the goal of maximizing recall, to obtain a trained prediction model.

[0015] In this embodiment, the corresponding category weights are determined by the labels corresponding to the training samples, and then the initial prediction model is trained using the historical training dataset with the category weights set. This allows the prediction model to focus on complaint samples with a small number of samples during the training process, thereby improving the recall rate of the prediction model for complaint samples with a small number of samples.

[0016] According to an embodiment of this application, the method further includes: determining the user's complaint risk level regarding the battery depletion based on multiple complaint probability values; determining the user's influence score based on the user's user characteristics; and determining a response strategy for the battery depletion based on the complaint risk level and the influence score.

[0017] This application embodiment can accurately determine the risk level of a user's complaint regarding the battery's low power by using multiple complaint probability values. By quantifying the user's influence score through user characteristics, and then combining the complaint risk level and the influence score, an appropriate response strategy can be determined to improve response timeliness and reduce the spread rate of negative public opinion.

[0018] According to an embodiment of this application, the user characteristics include the number of followers of the user on the information publishing platform, the amount of interaction of the user with the target content, and the dissemination of the information published by the user regarding the battery being low on power. The step of determining the user's influence score based on the user characteristics includes: determining the influence score based on the ratio of the number of followers to a first preset value, the ratio of the amount of interaction to a second preset value, and the ratio of the dissemination to a third preset value, wherein the first preset value is greater than the second preset value, and the second preset value is greater than the third preset value.

[0019] This application's embodiments integrate the number of fans, the amount of interaction, and the amount of dissemination, which can accurately quantify the user's influence score.

[0020] A second aspect of this application provides a complaint prediction device, comprising: an acquisition unit for acquiring vehicle information and environmental characteristics of the environment in which the vehicle is located; a determination unit for determining an electricity consumption risk coefficient of the vehicle based on the user's electricity consumption behavior characteristics after the vehicle is parked; the determination unit is further configured to determine a parking risk coefficient of the vehicle based on the vehicle's start-stop timing characteristics; the determination unit is further configured to determine a health degradation slope of the battery based on the health status of the battery in the vehicle; and a prediction unit for predicting the probability value of a user's complaint regarding the battery being depleted, based on the vehicle information, the environmental characteristics, the electricity consumption risk coefficient, the parking risk coefficient, the health status, and the health degradation slope.

[0021] A third aspect of this application provides an electronic device, the electronic device comprising: a memory for storing computer programs; and a processor for executing the computer programs stored in the memory to implement the complaint prediction method.

[0022] A fourth aspect of this application provides a vehicle equipped with electronic devices for performing the complaint prediction method.

[0023] A fifth aspect of this application provides a computer-readable storage medium storing a computer program that is executed by a processor in an electronic device to implement the complaint prediction method. Attached Figure Description

[0024] Figure 1 This is an application scenario diagram of the complaint prediction method provided in one embodiment of this application.

[0025] Figure 2 This is a flowchart of a complaint prediction method provided in an embodiment of this application.

[0026] Figure 3 This is a flowchart of a training method for a prediction model provided in an embodiment of this application.

[0027] Figure 4 This is a flowchart of a complaint prediction method provided in another embodiment of this application.

[0028] Figure 5 This is a functional block diagram of a complaint prediction device provided in an embodiment of this application.

[0029] Figure 6 This is a schematic diagram of the structure of an electronic device for implementing a complaint prediction method according to an embodiment of this application. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this application clearer, the application will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0031] It should be noted that in this application, "at least one" means one or more, and "more than one" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and drawings of this application are used to distinguish similar objects, not to describe a specific order or sequence.

[0032] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a specific manner. Unless otherwise specified, the following embodiments and features described herein can be combined with each other.

[0033] Currently, a depleted vehicle battery can disrupt normal vehicle use. Related technologies typically employ physical parameter detection methods, collecting electrical signal data such as battery voltage, current, and temperature to determine the battery's discharge status. However, these technologies rely on a single parameter to predict the risk of battery depletion, ignoring the impact of user habits on battery discharge, resulting in low prediction accuracy. Furthermore, these technologies cannot accurately predict potential customer complaints, failing to meet the practical needs of after-sales risk management and improved user experience.

[0034] Based on the above problems, this application provides a complaint prediction method that can accurately predict the probability of user complaints about vehicle battery depletion.

[0035] like Figure 1 The diagram shown is an application scenario diagram of the complaint prediction method provided in an embodiment of this application.

[0036] In some embodiments of this application, the complaint prediction method can be applied to electronic device 100. Electronic device 100 can be installed in vehicle 10; for example, electronic device 100 can be an in-vehicle terminal in vehicle 10. Vehicle 10 may also include an in-vehicle telematics box (T-Box) 101, a target electrical device 102, and a battery 103. Target electrical device 102 may include electrical devices in the vehicle that do not affect vehicle operation; for example, target electrical device 102 may include, but is not limited to, electrical devices in an in-vehicle entertainment system, ambient lighting, and USB charging ports.

[0037] In some embodiments of this application, the electronic device 100 can collect the stationary voltage of the vehicle 10 via the vehicle-mounted telematics processor 101. The stationary voltage can represent the battery voltage measured after the vehicle is turned off and left to stand still for a first preset time. The first preset time can be set and adjusted according to actual needs; for example, the first preset time can be set to 2 hours.

[0038] In some embodiments of this application, the electronic device 100 can also collect the device status of the target electrical device 102 after the vehicle is parked via the vehicle-mounted telematics processor 101. The device status may include an active state and an off state. The electronic device 100 can collect the time when the target electrical device 102 enters the active state as the on time of the target electrical device 102, and the electronic device 100 can also collect the time when the target electrical device 102 enters the off state as the off time of the target electrical device 102.

[0039] In some embodiments, the electronic device 100 may collect the on-time and off-time of the target electrical device based on a preset period. The preset period can be set and adjusted according to actual needs; for example, the preset period may be set to 10 minutes.

[0040] In some embodiments of this application, the electronic device 100 can also collect the location information and ignition status of the vehicle 10 via the vehicle-mounted telematics processor 101. The ignition status can include an ignition state and an off state. The electronic device 100 can collect the time when the vehicle 10 enters the ignition state as the ignition time, and can also collect the time when the vehicle 10 enters the off state as the off time.

[0041] In some embodiments of this application, the electronic device 100 can also collect charge-discharge cycle data of the battery 103 via the vehicle-mounted telematics processor 101. This charge-discharge cycle data may include the charging current of the battery 103 during charging, the total charging time of the battery 103, the discharging current of the battery 103 during discharging, and the total discharging time of the battery 103. The electronic device 100 can also collect the total mileage of the vehicle 10 within a second preset time period via the vehicle-mounted telematics processor 101. This second preset time period can be set and adjusted according to actual needs; for example, it may be set to 7 days.

[0042] In some embodiments of this application, the electronic device 100 can also communicate with the server 200. The server 200 can be configured with a Customer Relationship Management (CRM) system. The electronic device 100 can obtain user information from the CRM system, including but not limited to: whether vehicle 10 is the user's first car, whether the user is the first owner of vehicle 10, gender, age, number of followers on the information publishing platform, user interaction with target content, and the dissemination of information posted by the user regarding a dead battery. The target content includes vehicle-related content, and the dissemination of information posted by the user regarding a dead battery can include the total number of times the user's information regarding a dead battery is forwarded, commented on, and liked.

[0043] In some embodiments of this application, server 200 may also be configured with a Dealer Management System (DMS). Electronic device 100 can obtain the vehicle model of vehicle 10, battery model, and battery health at the time of manufacture from the dealer management system.

[0044] In some embodiments of this application, the electronic device 100 may also obtain the environmental characteristics of the environment in which the vehicle 10 is located from the server 20. The environmental characteristics of the environment in which the vehicle is located may include the average temperature of the environment in which the vehicle is located within a third preset time period. The third preset time period can be set and adjusted according to actual needs, for example, the third preset time period is set to 30 days.

[0045] In another embodiment, the electronic device 100 may also be an electronic product that communicates with the vehicle 200. For example, the electronic device 100 may be a personal computer, tablet computer, smartphone, personal digital assistant (PDA), game console, interactive network television (Internet Protocol Television, IPTV), smart wearable device, etc.

[0046] Electronic device 100 may include network devices and / or user devices. Among them, network devices include, but are not limited to, a single network electronic device, a group of electronic devices consisting of multiple network electronic devices, or a cloud based on cloud computing consisting of a large number of hosts or network electronic devices.

[0047] The network where electronic device 100 is located may include, but is not limited to: the Internet, wide area network, metropolitan area network, local area network, and virtual private network (VPN).

[0048] like Figure 2 The diagram shown is a flowchart of a complaint prediction method provided in one embodiment of this application. The order of the steps in this flowchart can be changed, and some steps can be omitted, depending on different requirements.

[0049] S201, Obtain vehicle information and environmental characteristics of the vehicle's surroundings.

[0050] In at least one embodiment of this application, the electronic device can acquire the vehicle's static voltage, vehicle model, battery model, battery health at the time of manufacture, and average daily mileage as vehicle information. The static voltage can represent the battery voltage measured after the vehicle is turned off and left idle for a first preset time. The average daily mileage can be determined based on the vehicle's total mileage within a second preset time period. The first and second preset times can be set and adjusted according to actual needs; for example, the first preset time can be set to 2 hours, and the second preset time to 7 days.

[0051] In at least one embodiment of this application, the electronic device can acquire the environmental characteristics of the vehicle's environment. The environmental characteristics of the vehicle's environment may include the average temperature of the vehicle's environment over a third preset time period. The third preset time period can be set and adjusted according to actual needs, for example, the third preset time period is set to 30 days.

[0052] S202, determine the vehicle's electricity consumption risk coefficient based on the user's electricity consumption behavior characteristics after the vehicle is parked.

[0053] In at least one embodiment of this application, the electronic device can be set with a preset period, which can be set and adjusted according to actual needs. For example, the preset period can be set to 10 minutes. Based on the preset period, the electronic device can acquire the user's power consumption behavior characteristics after the vehicle is parked. The power consumption behavior characteristics include the on-time and off-time of target electrical devices in the vehicle after the vehicle is parked. The target electrical devices can include electrical devices that do not affect vehicle operation. For example, the target electrical devices may include, but are not limited to, electrical devices in the vehicle entertainment system, ambient lights, and USB charging ports.

[0054] In at least one embodiment of this application, the electronic device determines the vehicle's power consumption risk coefficient based on the user's power consumption behavior characteristics after the vehicle is parked, including: determining the duration of a single power consumption of the target electrical device based on the shutdown time and the startup time; determining the frequency of abnormal power consumption of the vehicle based on a comparison between the duration of a single power consumption and a first preset duration; and calculating the power consumption risk coefficient based on the frequency of abnormal power consumption.

[0055] In some embodiments of this application, the electronic device calculates the difference between the off time and the on time of the target electrical device as the single power consumption duration of the target electrical device.

[0056] In some embodiments of this application, a first preset duration can be set according to actual needs, for example, the first preset duration can be set to 1 hour. The electronic device can count the total number of times the single power consumption duration is greater than or equal to the first preset duration within the configuration period as the abnormal power consumption frequency of the vehicle. The configuration period can be set and adjusted according to actual needs, for example, the configuration period can be set to weekly, monthly, etc.

[0057] In some embodiments of this application, the electronic device calculates a vehicle power consumption risk coefficient based on the frequency of abnormal power consumption. Each vehicle corresponds to a power consumption risk coefficient within each configuration cycle. This coefficient can be used to measure the impact of the vehicle's power consumption on the battery. The power consumption risk coefficient can be expressed as: ,in, It can represent the vehicle's electricity consumption risk coefficient. This can indicate the frequency of abnormal power consumption by a vehicle. In this case, ;exist , , It can be set and adjusted according to actual needs, for example, . ,For example, .

[0058] This application embodiment can accurately determine the duration of a single power consumption by the target electrical device after the vehicle is parked by measuring the on and off times of the target electrical device. Furthermore, by comparing the duration of a single power consumption with a first preset duration, the frequency of abnormal power consumption of the vehicle can be determined. Moreover, by analyzing the frequency of abnormal power consumption, the impact of the vehicle's power consumption on the battery can be quantified in a fine-grained manner.

[0059] S203 determines the parking risk coefficient of a vehicle based on its start-stop timing characteristics.

[0060] In at least one embodiment of this application, the electronic device can set a preset time period, which can be used to indicate the nighttime period. For example, the preset time period can be from 18:00 to 6:00 the next day. The embodiments of this application can be set and adjusted according to actual needs. The electronic device can collect the start-stop timing characteristics of the vehicle within the preset time period, which can include the vehicle's shutdown time within the preset time period.

[0061] In at least one embodiment of this application, the electronic device determines the parking risk coefficient of the vehicle based on the start-stop timing characteristics of the vehicle, including: selecting a target time from the end time of a preset time period or the ignition time of the vehicle within the preset time period; determining the engine shutdown duration of the vehicle within the preset time period based on the engine shutdown time and the target time; determining the frequency of abnormal parking of the vehicle within the preset time period based on a comparison between the engine shutdown duration and a second preset duration; and calculating the parking risk coefficient based on the frequency of abnormal parking.

[0062] In some embodiments of this application, if the vehicle does not enter the ignition state within a preset time period, the electronic device can use the end time of the preset time period as the target time. If the vehicle enters the ignition state within the preset time period, the electronic device can use the ignition time of the last time the vehicle entered the ignition state within the preset time period as the target time.

[0063] In some embodiments of this application, the electronic device calculates the difference between the target time and the time when the vehicle first enters the shutdown state within a preset time period as the shutdown duration.

[0064] In some embodiments of this application, the total number of times the electronic device's shutdown duration is greater than or equal to a second preset duration within a statistical configuration cycle is taken as the abnormal parking frequency. The second preset duration can be set and adjusted according to actual needs; for example, the second preset duration can be set to 8 hours.

[0065] In some embodiments of this application, the electronic device calculates a parking risk coefficient for the vehicle based on the frequency of abnormal parking. Each vehicle corresponds to a parking risk coefficient within each configuration cycle. This parking risk coefficient can be used to measure the impact of the vehicle's ignition status on the battery. The parking risk coefficient can be expressed as: ,in, It can represent the parking risk factor of a vehicle. It can indicate the frequency of abnormal parking of vehicles. and It can be set and adjusted according to actual needs, for example, , ,For example, .

[0066] This application embodiment can reasonably determine the target time by using the end time of a preset time period or the ignition time of the vehicle within the preset time period. Then, by combining the vehicle's shutdown time within the preset time period with the target time, the vehicle's shutdown duration within the preset time period can be accurately determined. By comparing the vehicle's shutdown duration with a second preset duration, the frequency of abnormal parking within the preset time period can be determined. Furthermore, by analyzing the frequency of abnormal parking, the impact of the vehicle's ignition status on the battery can be quantified in a fine-grained manner.

[0067] S204, determine the battery health degradation slope based on the health status of the battery in the vehicle.

[0068] In at least one embodiment of this application, the electronic device can also collect charge-discharge cycle data of the battery and determine the actual full-charge capacity of the battery based on the charge-discharge cycle data. The electronic device determines the battery health based on the actual full-charge capacity and the rated capacity of the battery. The battery health can be expressed as SOH = (actual full-charge capacity / rated capacity) × 100%.

[0069] In at least one embodiment of this application, the battery health degradation slope can be expressed as: , This can represent the slope of battery health degradation. It can indicate the health status of the battery at the first moment. This can indicate the battery's health status at a second moment, with the first moment preceding the second moment. It can indicate the first moment. It can represent the second moment.

[0070] S205 predicts the probability of user complaints about battery depletion based on vehicle information, environmental characteristics, electricity risk coefficient, parking risk coefficient, health status, and health status decay slope.

[0071] In at least one embodiment of this application, an electronic device predicts the probability value of a user's complaint about a dead battery based on vehicle information, environmental characteristics, electricity risk coefficient, parking risk coefficient, health status and health status decay slope, including: using vehicle information, environmental characteristics, electricity risk coefficient, parking risk coefficient, health status and health status decay slope as input features of the prediction model, and using the prediction model to determine the probability value of the complaint.

[0072] In some embodiments of this application, the prediction model can be a model based on the random forest algorithm, or it can be a model based on other algorithms. This application does not impose specific limitations on the prediction model. The prediction model can be trained based on historical training datasets, and the training method for the prediction model can be referred to... Figure 3 The process is shown below.

[0073] In some embodiments of this application, the electronic device can encode the vehicle information, environmental features, electricity risk coefficient, parking risk coefficient, health status and health status decay slope respectively to obtain corresponding encoding vectors, and use multiple encoding vectors as input features of the prediction model.

[0074] This application embodiment combines multiple dimensions of features, such as vehicle information, environmental characteristics, electricity risk coefficient, parking risk coefficient, health status, and health status decay slope, to solve the problem of bias in risk identification caused by the coarse and singular granularity of input features. By combining multiple dimensions of features, such as static and dynamic features, it can accurately predict the probability value of user complaints about battery depletion.

[0075] In at least one embodiment of this application, the electronic device calculates an importance score for each input feature based on its Gini coefficient, and determines a target input feature from multiple input features based on the importance score. The electronic device then outputs risk attribution information for battery depletion based on the vehicle's feature information on the target input feature.

[0076] In some embodiments of this application, the Gini coefficient of each input feature can be determined during the training process of the prediction model. Exemplarily, the electronic device can calculate the Gini impurity of any input feature at a split node, where the Gini impurity can be expressed as… , This can represent the proportion of non-complaint samples in the historical training dataset. This can represent the proportion of complaint samples in the historical training dataset. Based on the Gini impurity of any input feature at the split node, the electronic device determines the contribution value of any input feature at the split node; the contribution value can be represented as... , It can represent the contribution value of any input feature at the current split node. It can represent the Gini impurity of any input feature at the current split node. It can represent any input feature at the current split node. The impurity of the ginni This can represent the number of samples in the first child node. It can represent any input feature at the current split node. The impurity of the ginni This can represent the number of samples in the second child node. This can represent the sum of the number of samples in the first child node and the number of samples in the second child node. The electronic device calculates the sum of the contributions of any input feature across all split nodes to obtain the Gini coefficient of any input feature.

[0077] In some embodiments of this application, the electronic device uses the Gini coefficient of any input feature as the importance score of that input feature. The electronic device determines the M input features with the highest importance scores as target input features. M can be set and adjusted according to actual needs; for example, M can be set to 3.

[0078] In some embodiments of this application, the electronic device sets a corresponding attribution template for the target input features. The attribution template can be set and adjusted according to actual needs. For example, assuming that the target input features include the health decay slope, parking risk coefficient and environmental features, if the environmental feature is temperature, the attribution template can be set as: "Your vehicle battery health has recently deteriorated rapidly, and coupled with long-term overnight parking and low temperature, the risk of battery depletion is high."

[0079] In some embodiments of this application, the electronic device can output risk attribution information for battery depletion based on the vehicle's feature information on the target input features and the attribution template corresponding to the target input features.

[0080] Continuing with the example above, if the battery health degradation slope is 0.6% and the environmental characteristic is -5℃, the risk attribution information output to high-risk users can be set as follows: "Your vehicle battery health has recently degraded rapidly (0.6% per day), coupled with recent long-term overnight parking (≥8 hours / day) and low temperature in the area (-5℃), resulting in faster power consumption and a higher risk of battery depletion. It is recommended to schedule an on-site inspection within 4 hours."

[0081] For example, assuming the target input features include electricity risk coefficient and average daily mileage, if the vehicle's abnormal electricity usage frequency is 3 times / hour per day and the average daily mileage is 4 kilometers, the risk attribution information output to the user for a medium-risk user can be set as follows: "You have recently been using the in-car entertainment system frequently after parking (3 times / hour per day), and the average daily mileage is only 4 kilometers (insufficient short-distance charging). As a new car owner, please note: drive at least once a week for a long distance (≥20 kilometers), and turn off unnecessary electrical equipment promptly after parking. A free inspection coupon is attached (valid for 7 days)."

[0082] In some embodiments of this application, the electronic device may also output charts related to complaint risk trends.

[0083] This application embodiment uses the Gini coefficient of each input feature to accurately quantify the importance score of each input feature. Then, based on the importance score, target input features can be accurately selected. By using the vehicle's feature information on the target input features, matching risk attribution information can be output to the user, enabling the user to take appropriate strategies and improve the user experience.

[0084] In several embodiments of this application, the vehicle's electricity consumption risk coefficient can be quantified in a fine-grained manner by the user's electricity consumption behavior characteristics after the vehicle is parked; the vehicle's parking risk coefficient can be quantified in a fine-grained manner by the vehicle's start-stop timing characteristics; and then, by combining fine-grained dynamic characteristics such as vehicle information, environmental characteristics, electricity consumption risk coefficient, parking risk coefficient, health status and health status decay slope, the probability value of user complaints about battery depletion can be accurately predicted.

[0085] like Figure 3 The diagram shown is a flowchart of a training method for a prediction model provided in one embodiment of this application. The order of the steps in this flowchart can be changed, and some steps can be omitted, depending on different requirements.

[0086] S301, Obtain the historical training dataset.

[0087] In at least one embodiment of this application, the electronic device acquires multiple sample data, each sample data including vehicle information and the battery health degradation slope. Vehicle information may include, but is not limited to, the vehicle's static voltage and average daily mileage. Based on the vehicle information and the battery health degradation slope, the electronic device selects training samples from the multiple sample data. These training samples are used to construct a historical training dataset. Each training sample in the historical training dataset includes: vehicle information, environmental characteristics, electricity risk coefficient, parking risk coefficient, battery health and health degradation slope, and a label corresponding to the training sample. The label corresponding to the training sample indicates whether the user corresponding to that training sample has filed a battery depletion complaint. The label corresponding to the training sample may include: complained, not complained.

[0088] In at least one embodiment of this application, the electronic device can be set with a voltage threshold, a mileage threshold, and a health decay threshold. The voltage threshold, mileage threshold, and health decay threshold can be set and adjusted according to actual needs. For example, the voltage threshold can be set to 12.5V, the mileage threshold can be set to 10 kilometers, and the health decay threshold can be set to 0.3%. This application does not limit the specific values ​​of the voltage threshold, mileage threshold, and health decay threshold.

[0089] In at least one embodiment of this application, the electronic device can compare the static voltage with a voltage threshold, the average daily mileage with a mileage threshold, and the battery health degradation slope with a health degradation threshold in each sample data set. The electronic device can determine sample data where the static voltage is consistently greater than or equal to the voltage threshold, the average daily mileage is greater than or equal to the mileage threshold, and the battery health degradation slope is less than or equal to the health degradation threshold as risk-free samples. The electronic device determines the sample data excluding the risk-free samples from a plurality of sample data sets as training samples.

[0090] This application embodiment selects training samples from multiple sample data by using vehicle information and the health degradation slope of the battery. This can filter out risk-free samples, reduce the amount of training data for the prediction model, and improve the training efficiency of the prediction model.

[0091] S302, set class weights for multiple training samples based on the labels corresponding to the training samples in the historical training dataset, so that the weight of complaint samples is greater than the weight of non-complaint samples.

[0092] In at least one embodiment of this application, the weight of complaint-type samples indicates that the user corresponding to the complaint-type sample has filed a complaint about battery depletion, and the weight of non-complaint-type samples indicates that the user corresponding to the non-complaint-type sample has not filed a complaint about battery depletion.

[0093] In at least one embodiment of this application, during the process of setting class weights for multiple training samples, the electronic device counts the number of complaint-type samples as a first quantity and the number of non-complaint-type samples as a second quantity. Based on the first and second quantities, the electronic device determines the class weight (class_weight) of the training samples.

[0094] For example, if the first number of complaint samples is less than or equal to the second number of non-complaint samples, a first preset ratio can be set as the category weight for complaint samples, and a second preset ratio can be set as the category weight for non-complaint samples. The first preset ratio is greater than the second preset ratio, and the sum of the first and second preset ratios can be equal to a preset value, for example, the preset value can be set to 1. For example, assuming the first number of complaint samples is 100 and the second number of non-complaint samples is 900, the first number is less than the second number. The first preset ratio can be set to 5 / 6, and the second preset ratio can be set to 1 / 6. Then the category weight for complaint samples is 5 / 6, and the category weight for non-complaint samples is 1 / 6.

[0095] The embodiments of this application can set category weights for multiple training samples based on labels, so that the weight of complaint samples is greater than the weight of non-complaint samples, thereby improving the rationality of category weights.

[0096] S303, with the optimization objective of maximizing recall, uses a historical training dataset with class weights set to train the initial prediction model, resulting in a trained prediction model.

[0097] In at least one embodiment of this application, the electronic device may divide a training set based on a historical training dataset. The electronic device then trains an initial prediction model using the training set based on class weights.

[0098] In some embodiments of this application, the initial prediction model can be a model based on the random forest algorithm. The initial prediction model may include multiple decision trees. During the training process of the initial prediction model, model parameters such as the number of decision trees n_estimators, the maximum depth of each decision tree max_depth, the minimum number of samples required for internal node splits min_samples_split, the minimum number of samples required for leaf nodes min_samples_leaf, and the maximum number of features to be considered at each split max_features can be set.

[0099] In some embodiments of this application, during the training of the initial prediction model, the Randomized SearchCV (Randomized Parameter Search) method or the Grid SearchCV (Grid SearchCV) method can be used to train the initial prediction model based on 5-fold or 10-fold cross-validation. During the training of the initial prediction model, the model parameters can be optimized to obtain a trained prediction model.

[0100] In one example, the initial number of decision trees can be set to 200, and can be adjusted between 100 and 500 during training. The number of decision trees is related to the performance and training cost of the prediction model. Setting a larger number of decision trees in the prediction model can improve the stability and performance of the initial prediction model, but the training time and computational cost of the prediction model will also increase.

[0101] In another example, the maximum depth of each decision tree can be set between 10 and 20; it can also be set to None, in which case the nodes will continue to expand until all leaf nodes are pure nodes. To avoid overfitting in the prediction model, electronic devices can employ a strategy of limiting the depth of the decision tree.

[0102] In another example, the minimum number of samples required for internal node splitting can be set between 2 and 10. Setting the minimum number of samples required for internal node splitting can prevent the prediction model from overlearning local features and can also improve the generalization ability of the prediction model.

[0103] In another example, the minimum number of samples required for a leaf node can be set between 1 and 4. Setting the minimum number of samples required for a leaf node to a smaller value helps the predictive model capture fine-grained features of the data, but may make the predictive model more sensitive to data noise.

[0104] In another example, the initial value of the maximum number of features to be considered at each split can be set to the square root of the total number of features or log2. By setting the maximum number of features to be considered at each split, the diversity and generalization ability of the prediction model can be effectively enhanced.

[0105] In at least one embodiment of this application, the electronic device dynamically divides the historical training dataset into test sets according to a preset ratio. The preset ratio can be set and adjusted according to actual needs; for example, the preset ratio can be set to 20%. This embodiment dynamically divides the test set from the historical training dataset, and the test set can be used to evaluate the performance of the prediction model and prevent the prediction model from aging.

[0106] In at least one embodiment of this application, the electronic device can use training samples in a test set to determine the model metrics of an initial prediction model. If the model metrics of the initial prediction model meet preset conditions, a trained prediction model can be obtained. Model metrics may include, but are not limited to: Area Under the Curve of the Receiver Operating Characteristic (AUC-ROC), recall, precision, and F1 score.

[0107] In one example, AUC-ROC can be used to measure the classification ability of the initial prediction model. A higher AUC-ROC value indicates a higher classification ability of the initial prediction model for the training samples. The default condition can be set to AUC-ROC greater than or equal to 0.85.

[0108] In another example, recall can be used to measure the proportion of complaint samples that the predictive model successfully predicts. A higher recall value indicates a lower probability of the predictive model missing complaint samples. A preset condition can be set that the recall is greater than or equal to 0.75. For example, if there are 100 complaint samples, the predictive model's output can represent the probability value of the predicted user's complaint about a dead battery. If the number of complaints with a probability of 1 is 20 and the number of complaints with a probability of 0 is 80, then the predictive model's recall can be expressed as: 20 / (20+80) = 0.2.

[0109] In another example, precision can be used to measure the proportion of complaint samples among those identified as complaints by the predictive model. The preset condition can be that precision is greater than or equal to 0.5. For example, if there are 100 complaint samples and 900 non-complaint samples, and the predictive model outputs a complaint probability value of 1 for 20 out of the 100 samples, and 180 out of the 900 samples, then 20 / (20 + 180) = 0.1.

[0110] In another example, the F1 score is a model metric that comprehensively evaluates the performance of the predictive model. The F1 score can be the harmonic mean of precision and recall. A preset condition can be set that the F1 score is greater than or equal to 0.6.

[0111] The embodiments of this application set the training stopping conditions of the initial prediction model through multiple model indicators, which can address the problem caused by sample imbalance and improve the training accuracy of the prediction model.

[0112] In at least one embodiment of this application, the electronic device periodically reacquires the training set and uses the reacquired training set to train a prediction model.

[0113] In other embodiments of this application, the electronic device reacquires the test set and uses the reacquired test set to determine the model metrics of the prediction model. If the rate of decline of the model metrics of the prediction model is greater than or equal to a set metric threshold, the electronic device can retrain the prediction model. For example, if the rate of decline of the AUC-ROC of the prediction model is greater than or equal to 0.05, or the rate of decline of the recall rate of the prediction model is greater than or equal to 0.1, the electronic device can retrain the prediction model.

[0114] This application embodiment solves the performance degradation problem caused by the lack of dynamic temporal adaptation of the prediction model by periodically retraining the prediction model, or by retraining the prediction model when the decline slope of the model index is greater than or equal to a set index threshold, thereby ensuring the performance of the prediction model.

[0115] In at least one embodiment of this application, after retraining the prediction model, the electronic device can redetermine the importance score of each input feature.

[0116] In several embodiments of this application, the corresponding category weights are determined by the labels corresponding to the training samples, and then the initial prediction model is trained using the historical training dataset with the category weights set. This allows the prediction model to focus on complaint samples with a small number of samples during the training process, thereby improving the recall rate of the prediction model for complaint samples with a small number of samples.

[0117] like Figure 4 The diagram shown is a flowchart of a complaint prediction method provided in another embodiment of this application. Depending on different needs, the order of the steps in this flowchart can be changed, and some steps can be omitted.

[0118] S401, obtain vehicle information and environmental characteristics of the vehicle's surroundings.

[0119] S402 determines the vehicle's electricity risk coefficient based on the user's electricity consumption behavior characteristics after the vehicle is parked.

[0120] S403 determines the parking risk coefficient of a vehicle based on its start-stop timing characteristics.

[0121] S404 determines the battery health degradation slope based on the battery health status in the vehicle.

[0122] S405 predicts the probability of user complaints about battery depletion based on vehicle information, environmental characteristics, electricity risk coefficient, parking risk coefficient, health status, and health status decay slope.

[0123] For details on steps S401 to S405, please refer to the above text. Figure 2 The detailed descriptions of steps S201 to S205 are not repeated here.

[0124] S406 determines the risk level of user complaints regarding battery depletion based on multiple complaint probability values.

[0125] In at least one embodiment of this application, the electronic device can obtain the complaint probability value corresponding to each configuration cycle based on vehicle information, environmental characteristics, electricity risk coefficient, parking risk coefficient, health status and health status decay slope within each configuration cycle using a prediction model.

[0126] In at least one embodiment of this application, the electronic device determines the risk trend of user complaints regarding battery depletion based on complaint probability values ​​corresponding to multiple configuration cycles.

[0127] In some embodiments of this application, the electronic device calculates a complaint trend slope based on complaint probability values ​​corresponding to multiple configuration periods. For example, the complaint trend slope can be expressed as: , It can represent the slope of the complaint trend. These can represent the complaint probability values ​​corresponding to different configuration periods. It can represent the number of days in the configuration period. The corresponding configuration cycle is Before the corresponding configuration cycle, for example, This can represent the probability value of complaints from September 1st to September 7th. This can represent the probability value of complaints from September 2nd to September 8th. For example, This can represent the probability value of complaints from September 1st to September 7th. This can represent the probability value of complaints from September 8th to September 14th.

[0128] In some embodiments of this application, the electronic device may also set a first threshold and a second threshold. The first threshold and the second threshold can be set and adjusted according to actual needs. The first threshold is less than the second threshold. For example, the first threshold can be set to -0.02, the second threshold can be set to 0.02, or the second threshold can be set to 0.03. This application does not limit this.

[0129] In some embodiments of this application, if the slope of the complaint trend is less than a first threshold, the complaint risk trend can be determined to be a decreasing trend. If the slope of the complaint trend is greater than or equal to the first threshold and less than or equal to a second threshold, the complaint risk trend can be determined to be a stable trend. If the slope of the complaint trend is greater than the second threshold, the complaint risk trend can be determined to be an increasing trend.

[0130] In some embodiments of this application, the electronic device can also set a number of times threshold. The number of times threshold can be set and adjusted according to actual needs. For example, the number of times threshold can be set to 3. If the number of consecutive times the slope of the complaint trend is greater than the second threshold is greater than or equal to the number of times threshold, it can be determined that the complaint risk trend is a continuously rising risk trend.

[0131] In at least one embodiment of this application, the electronic device can determine the recall and precision of the prediction model. The methods for determining recall and precision can be found in [reference needed]. Figure 3 The description of recall and precision in step S304. The electronic device can also construct a curve corresponding to the recall and precision of the prediction model, which can also be called the PR curve. The electronic device can select a threshold from the PR curve that simultaneously satisfies preset conditions for both recall and precision as a first decision threshold. The electronic device can also set a second decision threshold, which is greater than the first decision threshold. For example, if the first decision threshold is 0.3, the second decision threshold can be set to 0.7; or if the first decision threshold is 0.4, the second decision threshold can be set to 0.6.

[0132] In at least one embodiment of this application, if the complaint probability value is less than a first decision threshold and the complaint risk trend is decreasing, the user's complaint risk level can be determined to be low risk. If the complaint probability value is greater than or equal to the first decision threshold, and if the complaint probability value is less than a second decision threshold and the complaint risk trend is increasing, the user's complaint risk level can be determined to be medium risk. If the complaint probability value is greater than or equal to the second decision threshold and the complaint risk trend is continuously increasing, the user's complaint risk level can be determined to be high risk.

[0133] S407 determines a user's influence score based on their user characteristics.

[0134] In at least one embodiment of this application, user characteristics may include, but are not limited to: the number of followers a user has on an information publishing platform, the amount of interaction a user has with target content, and the dissemination volume of information posted by a user regarding a dead battery. The information publishing platform can be a car-related social media platform, or other social media platforms; this application does not impose any restrictions on this. Target content may refer to car-related content posted by a user. The amount of interaction a user has with the target content may represent the cumulative number of various interactive behaviors triggered by other users within a set period after the user posts the target content. These interactive behaviors may include, but are not limited to, liking, commenting, forwarding, saving, and clicking. The dissemination volume of information posted by a user regarding a dead battery may represent the cumulative number of dissemination behaviors triggered by other users within a set period after the user posts information related to a dead battery. These dissemination behaviors may include, but are not limited to, liking, commenting, and forwarding.

[0135] In at least one embodiment of this application, the electronic device determines the user's influence score based on the user's user characteristics, including: determining the influence score based on the ratio of the number of fans to a first preset value, the ratio of the amount of interaction to a second preset value, and the ratio of the amount of dissemination to a third preset value.

[0136] In some embodiments of this application, a user's influence score can be expressed as: , It can represent an influence score. It can represent the first preset value. It can represent a second preset value. It can represent a third preset value, where the first preset value is greater than the second preset value, and the second preset value is greater than the third preset value. , and It can represent weights in different dimensions. + + =1. , , , and It can be set and adjusted according to actual needs. For example, It can be set to 10000. It can be set to 1000. It can be set to 100; It can be set to 0.3. It can be set to 0.4. It can be set to 0.3.

[0137] This application's embodiments integrate the number of fans, the amount of interaction, and the amount of dissemination, which can accurately quantify the user's influence score.

[0138] S408 determines the response strategy for battery depletion based on the complaint risk level and impact score.

[0139] In at least one embodiment of this application, if the user's complaint risk level is low, the electronic device can continue monitoring using a complaint prediction method. If the user's complaint risk level is medium, the electronic device can select an appropriate response strategy for the user based on the user's influence score.

[0140] In at least one embodiment of this application, if a user's complaint risk level is high-risk, the electronic device can acquire rescue vehicles within a preset distance from the vehicle. Based on the idle status of multiple rescue vehicles, the distance between multiple rescue vehicles and the vehicle, and the estimated idle time of multiple rescue vehicles, the electronic device selects a target vehicle from the multiple rescue vehicles. The electronic device then calls upon the target vehicle to provide service to the user corresponding to the high-risk level.

[0141] In other embodiments of this application, the electronic device performs a weighted summation of the influence score and the complaint probability value to obtain a priority score. Based on the user's priority score, the electronic device sequentially calls upon vehicles belonging to users corresponding to high-risk levels to provide services.

[0142] In at least one embodiment of this application, the electronic device may set a score threshold. If a user's influence score is greater than or equal to the score threshold, the user can be identified as a user with high public opinion influence. If a user's influence score is less than the score threshold, the user can be identified as an ordinary user.

[0143] In at least one embodiment of this application, the electronic device can set different service priorities based on the complaint risk level and influence score. For example, for high-risk users, the service priority corresponding to high-public opinion users is higher than that corresponding to ordinary users. This embodiment of the application dynamically sets the corresponding service priority by combining the user's influence score, which can improve resource matching.

[0144] Different service priorities correspond to different response strategies. For example, the response strategy for high-risk users with high public opinion risk could be: a dedicated customer service representative would call within 30 minutes, and a roadside assistance vehicle would arrive within 2 hours. For high-risk users with average risk risk, the response strategy could be: a dealer would call within 1 hour, and on-site inspection would be provided within 4 hours. Similarly, the response strategy for medium-risk users with high public opinion risk could be: sending risk attribution information and an on-site inspection coupon, and prioritizing appointments within 48 hours. For medium-risk users with average risk risk, the response strategy could be: sending risk attribution information, general battery maintenance information, and an in-store inspection coupon.

[0145] In at least one embodiment of this application, the electronic device can obtain the user's after-sales results, which may include, but are not limited to, the user's acceptance of the response strategy, the execution time of the response strategy, and the resolution rate of the response strategy for battery depletion. The electronic device can adjust the response strategy based on the user's after-sales results. For example, if the user has a low acceptance of coupons, the response strategy can be adjusted to: free on-site inspection.

[0146] For example, if user A's vehicle information includes: vehicle model Haobo HL, mileage of 30,000 km, battery model 70Ah AGM, average static voltage of 12.2V, environmental characteristics of the vehicle's environment including: average monthly temperature of -5℃, vehicle power consumption risk coefficient including: in-vehicle entertainment used 4 times / hour per day with each usage duration greater than or equal to 1 hour, parking risk coefficient including: long-term overnight parking 6 days per week, battery health degradation slope of 0.6% / day, user A's user information including: the vehicle is user A's first car, the vehicle is user A's first owner, age 25, social media followers 50,000, and the electronic device's predictive model estimates user A's complaint probability value for battery depletion at 0.85, and user A's complaint risk trend is continuously increasing. Based on user A's complaint probability value and complaint risk trend, the electronic device can determine user A's complaint risk level as high risk. Based on user information, the electronic device can determine user A as a high-publicity user. The response strategy selected by the electronic device for user A is as follows: send risk attribution information, have a dedicated customer service representative call within 30 minutes, match an available rescue vehicle within 50 kilometers, and complete on-site inspection and charging services within 1.5 hours.

[0147] In several embodiments of this application, multiple complaint probability values ​​can be used to accurately determine the risk level of a user's complaint regarding battery depletion. User influence scores can be quantified using user characteristics. By combining the complaint risk level and influence score, appropriate response strategies can be determined for proactive maintenance, improving response timeliness and reducing the spread of negative public opinion and service costs.

[0148] like Figure 5 The diagram shown is a functional block diagram of a complaint prediction device provided in an embodiment of this application. The complaint prediction device 51 includes an acquisition unit 510, a determination unit 511, a prediction unit 512, a calculation unit 513, an output unit 514, a setting unit 515, and a training unit 516. The module / unit referred to in this application refers to a module / unit that can be processed by a processor (e.g., ...). Figure 6 A series of computer program segments acquired by the processor 1101 shown, and capable of performing a fixed function, which are stored in memory (e.g., memory). Figure 6 In the memory 1102 shown.

[0149] In one embodiment, the acquisition unit 510 is used to acquire vehicle information and environmental characteristics of the vehicle's environment; the determination unit 511 is used to determine the vehicle's electricity consumption risk coefficient based on the user's electricity consumption behavior characteristics after the vehicle is parked; the determination unit 511 is also used to determine the vehicle's parking risk coefficient based on the vehicle's start-stop timing characteristics; the determination unit 511 is also used to determine the battery's health degradation slope based on the battery's health status; and the prediction unit 512 is used to predict the probability value of a user's complaint about a dead battery based on vehicle information, environmental characteristics, electricity consumption risk coefficient, parking risk coefficient, health status, and health degradation slope.

[0150] In one embodiment, the electricity consumption behavior characteristics include the on-time and off-time of the target electrical equipment in the vehicle after the vehicle is parked. The determining unit 511 is specifically used to: determine the single electricity consumption duration of the target electrical equipment based on the off-time and on-time; determine the abnormal electricity consumption frequency of the vehicle based on the comparison between the single electricity consumption duration and a first preset duration; and calculate the electricity consumption risk coefficient based on the abnormal electricity consumption frequency.

[0151] In one embodiment, the start-stop timing feature includes the vehicle's engine shutdown time within a preset time period. The determining unit 511 is further configured to: select a target time from the end time of the preset time period or the vehicle's ignition time within the preset time period; determine the vehicle's engine shutdown duration within the preset time period based on the engine shutdown time and the target time; determine the frequency of abnormal parking of the vehicle within the preset time period based on a comparison between the engine shutdown duration and a second preset duration; and calculate a parking risk coefficient based on the frequency of abnormal parking.

[0152] In one embodiment, the prediction unit 512 is specifically used to: use vehicle information, environmental characteristics, electricity risk coefficient, parking risk coefficient, health status and health status decay slope as input features of the prediction model, and use the prediction model to determine the complaint probability value.

[0153] In one embodiment, the calculation unit 513 is used to calculate the importance score of each input feature based on the Gini coefficient of each input feature; the determination unit 511 is also used to determine the target input feature from multiple input features based on the importance score; and the output unit 514 is used to output risk attribution information for battery depletion based on the feature information of the vehicle on the target input feature.

[0154] In one embodiment, the acquisition unit 510 is further configured to acquire a historical training dataset, wherein each training sample in the historical training dataset includes: vehicle information, environmental features, electricity risk coefficient, parking risk coefficient, battery health and health decay slope, and a label corresponding to the training sample. The label is used to indicate whether the user corresponding to the training sample has complained about battery depletion. The setting unit 515 is configured to set category weights for multiple training samples according to the labels, such that the weight of complaint samples is greater than the weight of non-complaint samples. The weight of complaint samples indicates that the user corresponding to the complaint samples has complained about battery depletion, and the weight of non-complaint samples indicates that the user corresponding to the non-complaint samples has not complained about battery depletion. The training unit 516 is configured to train the initial prediction model using the historical training dataset with set category weights, with the optimization objective of maximizing recall, to obtain a trained prediction model.

[0155] In one embodiment, the determining unit 511 is further configured to determine the risk level of a user's complaint regarding battery depletion based on multiple complaint probability values; the determining unit 511 is further configured to determine the user's influence score based on the user's user characteristics; and the determining unit 511 is further configured to determine a response strategy for battery depletion based on the complaint risk level and the influence score.

[0156] In one embodiment, the user characteristics include the number of followers of the user on the information publishing platform, the amount of interaction of the user with the target content, and the amount of dissemination of the information posted by the user regarding the battery being low on power. The determining unit 511 is further configured to: determine the influence score based on the ratio of the number of followers to a first preset value, the ratio of the amount of interaction to a second preset value, and the ratio of the amount of dissemination to a third preset value, wherein the first preset value is greater than the second preset value, and the second preset value is greater than the third preset value.

[0157] In several embodiments of this application, the vehicle's electricity consumption risk coefficient can be quantified in a fine-grained manner by the user's electricity consumption behavior characteristics after the vehicle is parked; the vehicle's parking risk coefficient can be quantified in a fine-grained manner by the vehicle's start-stop timing characteristics; and then, by combining fine-grained dynamic characteristics such as vehicle information, environmental characteristics, electricity consumption risk coefficient, parking risk coefficient, health status and health status decay slope, the probability value of user complaints about battery depletion can be accurately predicted.

[0158] like Figure 6 The diagram shown is a schematic representation of the structure of an electronic device for implementing a complaint prediction method according to an embodiment of this application.

[0159] In one embodiment of this application, the electronic device 100 includes, but is not limited to, a memory 1102, a processor 1101, and a computer program, such as a complaint prediction program, stored in the memory 1102 and executable on the processor 1101.

[0160] Those skilled in the art will understand that the schematic diagram is merely an example of the electronic device 100 and does not constitute a limitation on the electronic device 100. It may include more or fewer components than shown, or combine certain components, or different components. For example, the electronic device 100 may also include input / output devices, network access devices, buses, etc.

[0161] Processor 1101 can 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. The general-purpose processor can be a microprocessor or any conventional processor. Processor 1101 is the computing core and control center of electronic device 100, connecting various parts of electronic device 100 through various interfaces and lines, and acquiring the operating system of electronic device 100 and various installed application programs and program code.

[0162] Processor 1101 acquires the operating system and various installed applications of electronic device 100. Processor 1101 acquires these applications to implement the steps in the various complaint prediction method embodiments described above, for example... Figures 2 to 4 The steps are shown.

[0163] The memory 1102 can be used to store computer programs and / or modules. The processor 1101 implements various functions of the electronic device 100 by running or retrieving the computer programs and / or modules stored in the memory 1102, and by calling the data stored in the memory 1102. The memory 1102 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 1102 may include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other non-volatile solid-state storage device.

[0164] The memory 1102 can be the external memory and / or internal memory of the electronic device 100. Furthermore, the memory 1102 can be a memory in physical form, such as a memory stick, a TF card (Trans-flash Card), etc.

[0165] If the modules / units integrated in the electronic device 100 are implemented as software functional units and sold or used as independent workpieces, they can be stored in 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.

[0166] Computer programs include computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. Computer-readable media can 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), and random access memory (RAM).

[0167] For example, a computer program can be divided into one or more modules / units, one or more of which are stored in memory 1102 and executed by processor 1101 to complete this application. One or more modules / units can be a series of computer program segments capable of performing specific functions, which describe the execution process of the computer program in electronic device 100. For example, the computer program can be divided into an acquisition unit 510, a determination unit 511, a prediction unit 512, a calculation unit 513, an output unit 514, a setting unit 515, and a training unit 516.

[0168] For detailed information on the functions of each module / unit, please refer to the above text. Figures 2 to 4 The detailed description will not be repeated here.

[0169] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0170] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0171] Furthermore, the functional modules in the various embodiments of this application 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 in the form of hardware plus software functional modules.

[0172] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of this application is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within this application. No appended diagram markings in the claims should be construed as limiting the scope of the claims.

[0173] Furthermore, it is clear that the word "including" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices can also be implemented by a single unit or device through software or hardware. Terms such as "first," "second," etc., are used to indicate names and do not indicate any specific order.

[0174] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application and are not intended to limit it. Although this application has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of this application without departing from the spirit and scope of the technical solutions of this application.

Claims

1. A method for predicting complaints, characterized in that, The method includes: Obtain vehicle information and environmental characteristics of the environment in which the vehicle is located; The power consumption risk coefficient of the vehicle is determined based on the user's power consumption behavior characteristics after the vehicle is parked. Based on the start-stop timing characteristics of the vehicle, determine the parking risk coefficient of the vehicle; Based on the health status of the battery in the vehicle, determine the health degradation slope of the battery; Based on the vehicle information, environmental characteristics, electricity risk coefficient, parking risk coefficient, health status, and health status decay slope, the probability value of the user's complaint regarding the battery being depleted is predicted.

2. The complaint prediction method according to claim 1, characterized in that, The electricity consumption behavior characteristics include the start-up and shutdown times of the target electrical equipment in the vehicle after the vehicle is parked. Determining the vehicle's electricity consumption risk coefficient based on the user's electricity consumption behavior characteristics after the vehicle is parked includes: Based on the shutdown time and the opening time, determine the single power consumption duration of the target electrical equipment; Based on the comparison between the duration of a single power consumption and a first preset duration, the frequency of abnormal power consumption of the vehicle is determined; The electricity consumption risk coefficient is calculated based on the frequency of abnormal electricity consumption.

3. The complaint prediction method according to claim 1, characterized in that, The start-stop timing characteristics include the vehicle's engine shutdown time within a preset time period. Determining the vehicle's parking risk coefficient based on these start-stop timing characteristics includes: Select the target time from the end time of the preset time period or the ignition time of the vehicle within the preset time period; Based on the engine shutdown time and the target time, the engine shutdown duration of the vehicle within the preset time period is determined; Based on the comparison between the engine shutdown duration and the second preset duration, the frequency of abnormal parking of the vehicle within the preset time period is determined. The parking risk coefficient is calculated based on the frequency of abnormal parking.

4. The complaint prediction method according to claim 1, characterized in that, The method of predicting the probability of a user's complaint regarding a depleted battery based on the vehicle information, environmental characteristics, electricity risk coefficient, parking risk coefficient, battery health, and the rate of health decay includes: The vehicle information, environmental characteristics, electricity risk coefficient, parking risk coefficient, health status, and health status decay slope are used as input features of the prediction model, and the prediction model is used to determine the complaint probability value.

5. The complaint prediction method according to claim 4, characterized in that, The method further includes: Based on the Gini coefficient of each input feature, calculate the importance score of each input feature; Based on the importance score, the target input feature is determined from multiple input features; Based on the vehicle's feature information on the target input features, output risk attribution information for the battery depletion.

6. The complaint prediction method according to claim 4, characterized in that, The training method for the prediction model includes: Obtain a historical training dataset, wherein each training sample in the historical training dataset includes: vehicle information, environmental characteristics, electricity risk coefficient, parking risk coefficient, battery health and health decay slope, and a label corresponding to the training sample, wherein the label is used to indicate whether the user corresponding to the training sample has complained about battery depletion. Based on the labels, class weights are set for multiple training samples, such that the weight of complaint samples is greater than the weight of non-complaint samples. The weight of the complaint samples indicates that the user corresponding to the complaint sample has complained about battery depletion, and the weight of the non-complaint samples indicates that the user corresponding to the non-complaint samples has not complained about battery depletion. With the goal of maximizing recall, the initial prediction model is trained using the historical training dataset with class weights set, resulting in a trained prediction model.

7. The complaint prediction method according to any one of claims 1 to 6, characterized in that, The method further includes: Based on multiple complaint probability values, determine the risk level of the user's complaint regarding the battery's depletion; Based on the user's user characteristics, determine the user's influence score; Based on the complaint risk level and the influence score, a response strategy for the battery depletion is determined.

8. The complaint prediction method according to claim 7, characterized in that, The user characteristics include the number of followers the user has on the information publishing platform, the amount of interaction the user has with the target content, and the dissemination of the information the user posted about the battery being low on power. Determining the user's influence score based on these user characteristics includes: The influence score is determined based on the ratio of the number of fans to a first preset value, the ratio of the interaction volume to a second preset value, and the ratio of the dissemination volume to a third preset value, wherein the first preset value is greater than the second preset value, and the second preset value is greater than the third preset value.

9. An electronic device, characterized in that, include: A memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the complaint prediction method as described in any one of claims 1 to 8.

10. A vehicle, characterized in that, The vehicle is equipped with the electronic equipment as described in claim 9.

Images