Vehicle refueling behavior recognition methods and systems, servers, and storage media

Abstract

This invention relates to a vehicle refueling behavior recognition method and system, server, and storage medium, comprising: receiving vehicle status data uploaded by the vehicle in real time; the vehicle status data including at least a user identifier, fuel tank level signal, and power-on status signal; inputting the fuel tank level signal and power-on status signal into a pre-trained machine learning model for processing and outputting the probability that the vehicle is in a refueling state and its corresponding first fuel tank level change rate; determining whether the vehicle is refueling based on the probability of the vehicle being in a refueling state; when the vehicle is refueling, generating and outputting a refueling behavior profile of the vehicle user based on the user identifier, the initial fuel tank level signal, the final fuel tank level signal during the refueling process, and the first fuel tank level change rate. This invention can identify vehicle user refueling behavior, providing objective basic data input for customer-oriented automotive control development.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vehicle user behavior analysis, in particular to a vehicle refueling behavior identification method and system, a server and a computer readable storage medium. BACKGROUND

[0002] With the vigorous development of the new four modernizations of vehicles, vehicle development needs to closely integrate actual user needs to develop innovative and practical system products to improve user perception value and realize digital upgrading of the vehicle industry in the new era. The vehicle energy saving and emission reduction technology route is not only the requirement of national regulations, but also an important direction of vehicle development. Fuel consumption, as an intuitive feature of this technology, is also an important link that users experience and particularly care about. How to identify vehicle refueling actions through existing sensor data, understand the actual user's refueling habits and in-use refueling station fuel quality is a technical problem that needs to be solved in the industry. SUMMARY

[0003] The purpose of the present application is to provide a vehicle refueling behavior identification method and system, and a computer readable storage medium, which can identify the refueling behavior of vehicle users in order to understand the refueling habits of vehicle users.

[0004] To achieve the above purpose, the first aspect of the present application provides a vehicle refueling behavior identification method, comprising:

[0005] Step S1, real-time receiving vehicle state data uploaded by a vehicle; the vehicle state data at least includes a user identifier, a fuel tank liquid level signal and a power-on state signal;

[0006] Step S2, inputting the fuel tank liquid level signal and the power-on state signal into a pre-trained machine learning model for processing to output a probability that the vehicle is in a refueling state and a corresponding first fuel tank liquid level change rate;

[0007] Step S3, judging whether the vehicle has a refueling behavior according to the probability that the vehicle is in a refueling state;

[0008] Step S4, when the vehicle has a refueling behavior, generating a refueling behavior portrait of a vehicle user according to the user identifier and the initial fuel tank liquid level signal, the terminal fuel tank liquid level signal and the first fuel tank liquid level change rate in the refueling process, and outputting.

[0009] Optionally, the step S2 comprises:

[0010] The machine learning model determines whether the vehicle is in a stationary state according to the power-on state signal, and the power-on state signal has a step state from 1 to 0. If yes, the machine learning model determines that the probability that the vehicle is in a refueling state is a preset initial probability X% and starts to decay. If not, the machine learning model determines that the vehicle is not in a refueling state.

[0011] The machine learning model determines whether the oil tank level of the vehicle rises within the preset time according to the oil tank level signal, if yes, the probability of the vehicle being in the refueling state corresponds to an increase, and when the oil tank level stops rising, the probability of the vehicle being in the refueling state and the corresponding first oil tank level change rate are output, if no, the probability of the vehicle being in the refueling state within the preset time decays to 0 with a preset initial probability X% as the initial value.

[0012] Optionally, the step S2 comprises:

[0013] If the oil tank level rises from an initial oil tank level Y% to a terminal oil tank level Z% within the preset time, and (Z%-Y%)>F%, the machine learning model determines that the probability of the vehicle being in the refueling state is 100%; wherein F% is a preset threshold, and Y% and Z% are the oil tank level signals uploaded by the vehicle.

[0014] Optionally, the vehicle state data further comprises driving mileage data, average fuel consumption per 100 kilometers data and oil tank volume data;

[0015] The step S3 comprises:

[0016] The driving mileage data when the last refueling behavior of the vehicle occurs is obtained, and a second oil tank level change rate is calculated according to the driving mileage data when the last refueling behavior of the vehicle occurs, the current driving mileage data, the average fuel consumption per 100 kilometers data and the oil tank volume data;

[0017] According to the comparison result of the first oil tank level change rate and the second oil tank level change rate, it is determined whether the output result of the machine learning model is reliable, if yes, the probability of the vehicle being in the refueling state output by the machine learning model is multiplied by a coefficient k to obtain a corrected probability of the vehicle being in the refueling state; wherein the coefficient k is a value greater than 0;

[0018] If the corrected probability of the vehicle being in the refueling state is greater than a preset probability threshold, it is determined that the vehicle has a refueling behavior.

[0019] Optionally, the vehicle state data further comprises a positioning signal;

[0020] The step S3 further comprises:

[0021] When it is determined that the vehicle has a refueling behavior according to the probability of the vehicle being in the refueling state, the credibility of the refueling behavior is further verified according to the positioning signal when the refueling behavior occurs, and it is determined whether there is a gas station within a preset distance range centered on the position of the positioning signal according to map data; if yes, the refueling behavior is credible, and it is determined that there is a refueling behavior, if no, the refueling behavior is not credible, and it is determined that there is no refueling behavior.

[0022] Optionally, the step S3 further comprises: when the refueling behavior is credible, determining that the refueling behavior exists, and acquiring the gas station information within the preset distance range;

[0023] The step S4 further comprises: updating the gas station heat data according to the gas station information, and outputting.

[0024] The second aspect of the present application provides a vehicle refueling behavior recognition system, comprising:

[0025] A data receiving unit is configured to receive vehicle state data uploaded by a vehicle in real time, wherein the vehicle state data at least comprises a user identifier, a tank level signal, and a power-on state signal;

[0026] A refueling behavior recognition unit is configured to input the tank level signal and the power-on state signal into a pre-trained machine learning model to process and output a probability that the vehicle is in a refueling state and a corresponding first tank level change rate.

[0027] A credibility judgment unit is configured to determine whether the vehicle has a refueling behavior according to the probability that the vehicle is in a refueling state; and

[0028] A portrait generation unit is configured to generate a refueling behavior portrait of a vehicle user according to the user identifier and the initial tank level signal, the terminal tank level signal, and the first tank level change rate during the refueling process when the vehicle has a refueling behavior, and output the refueling behavior portrait.

[0029] Optionally, the machine learning model determines whether the vehicle is in a stationary state according to the power-on state signal, and the power-on state signal has a step state from 1 to 0, if yes, the machine learning model determines that the probability that the vehicle is in a refueling state is a preset initial probability X% and starts to decay, if no, the machine learning model determines that the vehicle is not in a refueling state.

[0030] The machine learning model determines whether the tank level of the vehicle rises within the preset time according to the tank level signal, if yes, the probability that the vehicle is in a refueling state corresponds to an increase, and when the tank level stops rising, the probability that the vehicle is in a refueling state and the corresponding first tank level change rate are output, if no, the probability that the vehicle is in a refueling state within the preset time decays to 0 with the preset initial probability X% as the initial value.

[0031] If the tank level rises from an initial tank level Y% to a terminal tank level Z% within the preset time, and (Z%-Y%)>F%, the machine learning model determines that the probability that the vehicle is in a refueling state is 100%, wherein F% is a parameter threshold value determined by the machine learning model, and Y% and Z% are tank level signals uploaded by the vehicle.

[0032] Optionally, the vehicle state data further comprises driving mileage data, average fuel consumption per 100 kilometers data, fuel tank volume data, positioning signal;

[0033] The credibility judging unit comprises:

[0034] The first verification subunit is configured to acquire driving mileage data when the last refueling behavior of the vehicle occurs, calculate a second fuel tank level change rate based on the driving mileage data when the last refueling behavior of the vehicle occurs, the current driving mileage data, the average fuel consumption per 100 kilometers data and the fuel tank volume data, and determine whether the output result of the machine learning model is credible based on a comparison result of the first fuel tank level change rate and the second fuel tank level change rate. If the output result of the machine learning model is credible, the probability of the vehicle being in the refueling state output by the machine learning model is multiplied by a coefficient k to obtain a corrected probability of the vehicle being in the refueling state. The coefficient k is a value greater than 0. If the corrected probability of the vehicle being in the refueling state is greater than a preset probability threshold, it is determined that the refueling behavior of the vehicle exists.

[0035] The second verification subunit is configured to, when it is determined that the refueling behavior of the vehicle exists based on the probability of the vehicle being in the refueling state, further verify the credibility of the refueling behavior based on the positioning signal when the refueling behavior occurs, and determine whether a gas station exists within a preset distance range centered on the position of the positioning signal based on map data. If the gas station exists, the refueling behavior is credible, and it is determined that the refueling behavior exists. If the gas station does not exist, the refueling behavior is not credible, and it is determined that the refueling behavior does not exist.

[0036] The third aspect of the present application provides a server, comprising:

[0037] A memory for storing a computer program;

[0038] A processor for executing the computer program to implement the steps of the vehicle refueling behavior identification method according to the first aspect.

[0039] The fourth aspect of the present application provides a computer readable storage medium having computer program instructions stored thereon, wherein the computer program instructions are executed by a processor to implement the steps of the vehicle refueling behavior identification method according to the first aspect.

[0040] The vehicle refueling behavior identification method and system, the server and the computer readable storage medium have at least the following beneficial effects:

[0041] By pre-training a machine learning model for identifying whether a vehicle is in a refueling state according to a vehicle power-on state signal and a fuel tank level signal, in use, the vehicle terminal device is in communication connection with the cloud server through the 4G / 5G network, the vehicle terminal device sends the vehicle state data to the cloud server in real time, the vehicle state data at least includes a user identifier (the user identifier can be used as a unique identifier to distinguish different vehicle users), a power-on state signal and a fuel tank level signal, the cloud server uses the machine learning model to identify the vehicle refueling state of the data uploaded by the vehicle terminal, and outputs the probability that the vehicle is in a refueling state and the corresponding first fuel tank level change rate; further, whether the vehicle has a refueling behavior can be judged according to the probability that the vehicle is in a refueling state; if there is a refueling behavior, a portrait of the vehicle user's this time refueling behavior is generated according to the user identifier and the initial fuel tank level signal, the termination fuel tank level signal and the first fuel tank level change rate in the refueling process, for each refueling behavior of the vehicle user, the embodiment of the application outputs a refueling behavior portrait, so as to understand the refueling habits of the vehicle user and provide objective basic data input for customer-oriented automobile control development.

[0042] Other features and advantages of the vehicle refueling behavior identification method and system, server and computer readable storage medium described above will be described in the subsequent specification. BRIEF DESCRIPTION OF DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0044] Figure 1 A flow chart of a vehicle refueling behavior identification method in an embodiment of the present application.

[0045] Figure 2 An application scenario diagram of a vehicle refueling behavior identification method in an embodiment of the present application.

[0046] Figure 3 A structural framework diagram of a vehicle refueling behavior identification system in an embodiment of the present application.

[0047] Figure 4 A structural framework diagram of a server in an embodiment of the present application. DETAILED DESCRIPTION

[0048] The various exemplary embodiments, features, and aspects of this disclosure will now be described in detail with reference to the accompanying drawings. Furthermore, numerous specific details are set forth in the following detailed embodiments to better illustrate the invention. Those skilled in the art will understand that the invention can be practiced without certain specific details. In some instances, means well known to those skilled in the art have not been described in detail in order to highlight the spirit of the invention.

[0049] See Figure 1 One embodiment of the present invention proposes a method for recognizing vehicle refueling behavior. The application scenarios of this method include, for example... Figure 2 As shown, Figure 2 This example only shows four vehicles; in practice, more vehicles can be included. The vehicle-side terminal devices connect to the cloud server via network communication, such as 4G / 5G / 6G networks, or based on existing vehicle-to-everything (V2X) networks. The cloud server maintains communication connections with the terminal devices of multiple vehicles. The cloud server interacts with any vehicle, executing the following steps S1-S4:

[0050] Step S1: Receive vehicle status data uploaded by the vehicle in real time; the vehicle status data includes at least the user identifier, fuel tank level signal, and power-on status signal.

[0051] Specifically, the vehicle is equipped with a terminal device, such as a T-BOX, for collecting vehicle status data. This terminal device interacts with a cloud server through a communication network and uploads the collected vehicle status data to the cloud server. The cloud server receives the vehicle status data and stores it in a database. The cloud server distinguishes and identifies data uploaded by different vehicles based on the user identifier in the vehicle status data.

[0052] It should be noted that the form of the user identifier is not limited. For example, it can be a T-BOX user account or other user information; it can also be a vehicle identifier, such as a vehicle VIN code, as long as it can be used to distinguish different vehicles.

[0053] Step S2: Input the fuel tank level signal and power-on status signal into a pre-trained machine learning model for processing and output the probability that the vehicle is in refueling state and its corresponding first fuel tank level change rate.

[0054] Specifically, the first fuel tank level change rate refers to the change in fuel tank level during the vehicle's current refueling process, which can be calculated based on the initial and final fuel tank levels. The vehicle's power-on status signal determines whether the vehicle is ready to refuel, since the vehicle is always stationary during refueling. Furthermore, when the vehicle is ready to refuel, the fuel tank level signal received over a continuous period determines whether the vehicle is ready to refuel, thus outputting the probability of the vehicle being in a refueling state and its corresponding first fuel tank level change rate.

[0055] It should be noted that the machine learning model adopts the black box theory. The black box theory refers to treating a specific system as an opaque black box when studying it. The research does not involve the internal structure and interrelationships of the system, but only understands the system's patterns from its input and output characteristics. This black box approach yields insights into the patterns of a system. Instead of analyzing the internal structure and interrelationships of an ecosystem, it derives the structural and functional patterns of the ecosystem based on the overall input and output relationships of matter and energy and their influencing factors. There are many possible model structures for the machine learning model; this embodiment does not specifically limit itself to any particular model structure. Based on a basic model structure, after training with a given large dataset (real historical data of vehicles), any model can achieve step S2 of this embodiment.

[0056] Step S3: Determine whether the vehicle is refueling based on the probability that the vehicle is in a refueling state.

[0057] Specifically, as an example, the presence of refueling activity can be determined by comparing the probability of the vehicle being in a refueling state with a preset probability threshold. For instance, if the probability of the vehicle being in a refueling state is greater than or equal to the preset probability threshold of 95%, the vehicle is determined to be refueling; if the probability of the vehicle being in a refueling state is less than the preset probability threshold of 95%, the vehicle is determined not to be refueling. Of course, this is only a very simple specific embodiment for illustration. Other factors can also be considered in combination to determine whether the vehicle is refueling. More specific embodiments will be described in detail below.

[0058] Step S4: When the vehicle is refueling, generate a refueling behavior profile of the vehicle user based on the user identifier, the initial fuel tank level signal, the final fuel tank level signal, and the first fuel tank level change rate during the refueling process, and output the profile.

[0059] Specifically, in this embodiment, when a vehicle is detected refueling, a refueling behavior profile of the vehicle user is generated. This refueling behavior profile mainly includes a user identifier representing the vehicle user, as well as the user's usual refueling times and amounts. The initial fuel tank level signal can determine when the user usually refuels, while the termination fuel tank level signal and the first fuel tank level change rate can determine the amount of fuel added each time. The generated refueling behavior profile is stored in a database on a cloud server. Based on multiple refueling behavior profiles of vehicle users, their refueling behavior characteristics can be statistically analyzed to understand the user's refueling habits.

[0060] As described in the above embodiments, the method of this invention pre-trains a machine learning model capable of identifying whether a vehicle is in a refueling state based on the vehicle's power-on status signal and fuel tank level signal. In use, the vehicle terminal device communicates with a cloud server via a 4G / 5G network. The vehicle terminal device sends vehicle status data to the cloud server in real time. The cloud server uses the machine learning model to identify the vehicle's refueling status based on the data uploaded by the vehicle terminal, outputting the probability that the vehicle is in a refueling state and its corresponding first fuel tank level change rate. Furthermore, the probability of the vehicle being in a refueling state can be used to determine whether the vehicle has engaged in refueling behavior. If refueling behavior has occurred, a profile of the vehicle user's current refueling behavior is generated based on the user identifier, the initial fuel tank level signal, the final fuel tank level signal, and the first fuel tank level change rate during the refueling process. For each refueling behavior of the vehicle user, this embodiment of the invention outputs a refueling behavior profile to understand the vehicle user's refueling habits and provide objective data input for customer-oriented automotive control development.

[0061] Preferably, the machine learning model described in this embodiment is implemented using a classification supervision algorithm, such as a logistic regression model, a decision tree model, or an SVM model.

[0062] In some embodiments, step S2 includes:

[0063] The machine learning model determines whether the vehicle is stationary based on the power-on status signal ICM_KeySt and whether the power-on status signal ICM_KeySt has a step state from 1 to 0. If so, the machine learning model determines that the probability of the vehicle being in refueling state is a preset initial probability X% and begins to decay. If not, the machine learning model determines that the vehicle is not in refueling state.

[0064] Specifically, the machine learning model determines whether the power-on status signal ICM_KeySt is 0. If the power-on status signal ICM_KeySt = 0, the vehicle is determined to be stationary. Simultaneously, it detects whether the power-on status signal ICM_KeySt has a step state from 1 to 0. If both conditions are met, the vehicle is considered ready for refueling, and an initial probability X% is assigned as the probability that the vehicle is in refueling mode. A time decay coefficient K is set, and the real-time value of the probability that the vehicle is in refueling mode is the result of decaying based on this coefficient K starting from the initial probability X%. If neither condition is met simultaneously, the machine learning model determines that the vehicle is not ready for refueling, and the machine learning model identifies that the vehicle is not in refueling mode.

[0065] In addition, as another method, whether the vehicle is stationary can also be determined by the engine speed signal EMS_EngSpd. When EMS_EngSpd = 0, the vehicle is considered to be stationary.

[0066] The machine learning model determines whether the vehicle's fuel tank level rises within a preset time period based on the fuel tank level signal ICM_FuelLevel. If so, the probability that the vehicle is in refueling mode increases accordingly. When the fuel tank level stops rising, the model outputs the probability that the vehicle is in refueling mode and its corresponding first fuel tank level change rate. If not, the probability that the vehicle is in refueling mode within the preset time period is decayed to 0 with a preset initial probability X%.

[0067] Specifically, the preset time is preferably, but not limited to, 5 minutes. It is understood that the fuel level in the tank will rise during refueling. During refueling, the vehicle sends its fuel level signal to the cloud server in real time. The cloud server inputs the received fuel level signal into the machine learning model. When the machine learning model recognizes that the fuel level in the vehicle's tank has risen within the preset time, it increases the real-time probability that the vehicle is in refueling mode. While increasing the probability, it continues to decrease the probability using the coefficient K. If the fuel level does not rise within the preset time, the probability that the vehicle is in refueling mode within the preset time decreases from an initial preset probability X% to 0.

[0068] In some embodiments, step S2 further includes:

[0069] If, within the preset time period, the fuel tank level rises from the initial fuel tank level Y% to the final fuel tank level Z%, and (Z% - Y%) > F%, then the machine learning model determines that the probability of the vehicle being in a refueling state is 100%; where F% is a preset threshold.

[0070] Specifically, in this embodiment, a threshold F% is set to eliminate some misjudgments of liquid level changes caused by vehicle vibration.

[0071] The relevant parameters of the machine learning model can be selected within the following range, and can be fine-tuned in combination with the actual fuel tank volume, refueling rate, etc.

[0072] The initial refueling probability X% is preferably, but not limited to, 50%.

[0073] The preferred, but not limited, time decay coefficient K is 0.1, representing a 10% decay per minute;

[0074] The initial oil tank level Y% is preferably, but not limited to, greater than 10%, to filter out drastic changes in the low oil level that could lead to misjudgment.

[0075] Among them, the preferred, but not limited, value of the termination tank level Z% is less than 100%, in order to filter out the error parameter of the level sensor exceeding the upper limit.

[0076] The preferred, but not limited, value of the fuel tank level change rate F% is greater than 15%, in order to filter out misjudgments of level changes caused by vehicle vibration.

[0077] During the training of a machine learning model, parameters X, Y, Z, K, and F are continuously adjusted. When the recognition accuracy of the machine learning model reaches 95% or higher, the parameters corresponding to X, Y, Z, K, and F can be used as the final model parameters.

[0078] In some embodiments, the vehicle status data further includes mileage data, average fuel consumption data per 100 kilometers, and fuel tank volume data; specifically, the mileage data and the average fuel consumption data per 100 kilometers are the vehicle's displayed fuel consumption data and displayed mileage data; the fuel tank volume data is a fixed value, that is, the volume of the fuel tank is constant.

[0079] Step S3 includes:

[0080] Step S31: Obtain the mileage data when the vehicle last refueling occurred, and calculate the second fuel tank level change rate based on the mileage data when the vehicle last refueling occurred, the currently received mileage data, average fuel consumption data per 100 kilometers, and fuel tank volume data.

[0081] Specifically, the cloud server's database stores the mileage data corresponding to each refueling operation; the specific calculation formula for the second fuel tank level change rate is as follows:

[0082] The vehicle mileage corresponding to the last vehicle refueling behavior is named mileage A; the vehicle mileage corresponding to the current vehicle refueling behavior is named mileage B; the average fuel consumption per 100 kilometers between the two refueling times is C, and the fuel tank volume is D. The change rate of the second fuel tank level (actual refueling amount) is calculated using the following formula and compared with the change rate of the first fuel tank level under the refueling behavior detected by the machine learning model to cross-validate the credibility of the refueling behavior.

[0083] The rate of change of the liquid level in the second oil tank = (BA)*C / (100*D)

[0084] Where A is in km, B is in km, C is in L / 100km, and D is in L;

[0085] Step S32: Determine whether the output result of the machine learning model is reliable based on the comparison result of the first fuel tank level change rate and the second fuel tank level change rate. If so, multiply the probability of the vehicle being in refueling mode output by the machine learning model by the coefficient k to obtain the corrected probability of the vehicle being in refueling mode; where the coefficient k is a value greater than 0.

[0086] Step S33: If the probability that the corrected vehicle is in a refueling state is greater than the preset probability threshold, then it is determined that the vehicle is refueling.

[0087] For example, if the monitored distance for two refueling events is 500 kilometers, and the average fuel consumption per kilometer is 10L / 100km, then the total fuel consumption is 50L. If the fuel tank capacity is 75L, then the fuel level change should be 67%. Compare this to the actual monitored fuel level change rate F%. If F% is within the range of 67% ± 5%, then the refueling event can be considered reliable. The probability of recognizing the refueling event can be increased by multiplying the output of the machine recognition model by a coefficient of 1.05. If the result after multiplying by the coefficient is greater than 100%, then the probability is represented as 100%.

[0088] It should be noted that steps S31-S33 are further verifications of the output results of step S2.

[0089] In some embodiments, the vehicle status data further includes a positioning signal;

[0090] Step S3 further includes:

[0091] Step S34: When it is determined that the vehicle is refueling based on the probability that the vehicle is in a refueling state, the credibility of the refueling behavior is further verified based on the positioning signal when the refueling behavior occurs. Based on the map data, it is determined whether there is a gas station within a preset distance range centered on the location of the positioning signal. If yes, the refueling behavior is credible and the existence of refueling behavior is confirmed. If no, the refueling behavior is not credible and the absence of refueling behavior is confirmed.

[0092] Specifically, the positioning signal is GPS coordinates, represented as (longitude GPS_Longitude, latitude GPS_Latitude). Through the standardized interface defined by the map provider, it is possible to query gas stations within a specified location and range. For example, the POI interface for gas station query on Gaode Map is 010100. If there are no gas stations within 3km of the potential refueling location, it can be considered that the refueling behavior has been identified incorrectly, and a probability correction can be performed.

[0093] It should be noted that step S34 is a further verification of the output result of step S33.

[0094] In some embodiments, step S3 further includes: when the refueling behavior is reliable and it is determined that refueling behavior exists, obtaining information on gas stations within the preset distance range.

[0095] Specifically, through the standardized interface defined by the map provider, the information of the gas station where the refueling behavior was carried out can be obtained by querying gas stations within a specified location and range. Under normal circumstances, if a refueling behavior actually exists, there should only be one gas station within the preset distance range, and that gas station is the gas station corresponding to the current refueling behavior. If there are multiple gas stations within the preset distance range, the gas station closest to the vehicle is selected as the gas station corresponding to the current refueling behavior.

[0096] Step S4 further includes: updating the gas station popularity data according to the gas station information and outputting it.

[0097] Specifically, by statistically analyzing the gas stations corresponding to the refueling behavior of each vehicle user, it is possible to rank the gas stations by popularity and refueling frequency. This data can provide reference for the design of fuel system features and for fuel product analysis and gas station recommendations for user-oriented services.

[0098] See Figure 3 Another embodiment of the present invention provides a vehicle refueling behavior recognition system, comprising:

[0099] The data receiving unit is used to receive vehicle status data uploaded by the vehicle in real time; the vehicle status data includes at least the user identifier, fuel tank level signal, and power-on status signal.

[0100] The refueling behavior recognition unit is used to input the fuel tank level signal and the power-on status signal into a pre-trained machine learning model for processing and outputting the probability that the vehicle is in a refueling state and its corresponding first fuel tank level change rate.

[0101] A credibility determination unit is used to determine whether the vehicle has engaged in refueling activity based on the probability that the vehicle is in a refueling state; and

[0102] The profile generation unit is used to generate and output a profile of the vehicle user's refueling behavior based on the user identifier, the initial fuel tank level signal, the final fuel tank level signal, and the first fuel tank level change rate during the refueling process when the vehicle is refueling.

[0103] In some embodiments, the machine learning model determines whether the vehicle is stationary based on the power-on status signal and the power-on status signal has a step state from 1 to 0. If yes, the machine learning model determines that the probability of the vehicle being in refueling state is a preset initial probability X% and begins to decay. If no, the machine learning model determines that the vehicle is not in refueling state.

[0104] The machine learning model determines whether the vehicle's fuel tank level rises within a preset time period based on the fuel tank level signal. If so, the probability that the vehicle is in refueling mode increases accordingly. When the fuel tank level stops rising, the model outputs the probability that the vehicle is in refueling mode and its corresponding first fuel tank level change rate. If not, the probability that the vehicle is in refueling mode within the preset time period is decayed to 0 with a preset initial probability of X%.

[0105] If, within the preset time period, the fuel tank level rises from the initial fuel tank level Y% to the final fuel tank level Z%, and (Z%-Y%)>F%, then the probability that the machine learning model determines the vehicle is in a refueling state is 100%; where F% is a parameter threshold determined by the machine learning model in pre-training, and Y% and Z% are the fuel tank level signals uploaded by the vehicle.

[0106] In some embodiments, the vehicle status data further includes mileage data, average fuel consumption per 100 kilometers data, fuel tank volume data, and positioning signal;

[0107] The credibility determination unit includes:

[0108] The first verification subunit is used to acquire the mileage data of the vehicle at the time of the last refueling, calculate the second fuel tank level change rate based on the mileage data of the last refueling, the current mileage data, the average fuel consumption data per 100 kilometers, and the fuel tank volume data; and determine whether the output result of the machine learning model is reliable based on the comparison result of the first fuel tank level change rate and the second fuel tank level change rate. If so, the probability of the vehicle being in a refueling state output by the machine learning model is multiplied by a coefficient k to obtain the corrected probability of the vehicle being in a refueling state; where the coefficient k is a value greater than 0; if the corrected probability of the vehicle being in a refueling state is greater than a preset probability threshold, it is determined that the vehicle has refueled.

[0109] The second verification subunit is used to further verify the credibility of the refueling behavior based on the location signal at the time of the refueling behavior when it is determined that the vehicle is in a refueling state based on the probability that the vehicle is in a refueling state. It also uses map data to determine whether there is a gas station within a preset distance range centered on the location of the location signal. If so, the refueling behavior is credible and the existence of refueling behavior is confirmed. If not, the refueling behavior is unreliable and the absence of refueling behavior is confirmed.

[0110] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0111] It should be noted that the system described in the above embodiments corresponds to the method described in the above embodiments. Therefore, the parts of the system described in the above embodiments that are not described in detail can be obtained by referring to the content of the method described in the above embodiments. That is, the specific steps recorded in the method described in the above embodiments can be understood as the functions that the system of this embodiment can achieve, and will not be repeated here.

[0112] Furthermore, if the vehicle refueling behavior recognition system described in the above embodiments is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0113] Another embodiment of the present invention provides a server, comprising:

[0114] Memory, used to store computer programs;

[0115] A processor is configured to execute the computer program to implement the steps of the vehicle refueling behavior recognition method as described in the above embodiments.

[0116] For example, the server may include a memory 11, a processor 12, and a bus 13;

[0117] The memory 21 includes at least one type of readable storage medium, such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 21 can be an internal storage unit of the server. In other embodiments, the memory 21 can be an external storage device of the server, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., all equipped on the server. Furthermore, the memory 21 can include both internal and external storage devices. The memory 21 can be used not only to store application software and various types of data installed on the server, such as the code of a program executing a vehicle refueling behavior recognition method, but also to temporarily store data that has been output or will be output.

[0118] In some embodiments, processor 22 may be a central processing unit (CPU), controller, microcontroller, microprocessor or other data processing chip, used to run program code stored in memory 21 or process data, such as the code of a program for executing a vehicle refueling behavior recognition method.

[0119] This bus 23 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 4 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0120] Furthermore, the server may also include a network interface 24, which may optionally include a wired interface and / or a wireless interface (such as a Wi-Fi interface, a Bluetooth interface, etc.), typically used to establish communication connections between the server and other electronic devices.

[0121] Optionally, the server may further include a user interface 25, which may include a display, input units such as buttons, and optionally, a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the server and to display a visual user interface.

[0122] Figure 4 Only servers with components 21-25 are shown; those skilled in the art will understand that... Figure 4 The structure shown does not constitute a limitation on the server and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0123] Another embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the vehicle refueling behavior recognition method described in the above embodiments.

[0124] Specifically, the computer-readable storage medium may include any entity or device capable of carrying the computer program instructions, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0125] The various embodiments of the present invention have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or technical improvements to the embodiments in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A vehicle refueling behavior recognition method, characterized by, The method comprises the following steps: Step S1, receiving vehicle state data uploaded by a vehicle in real time; The vehicle state data at least includes user identification, tank level signal, power-on state signal, driving mileage data, average fuel consumption data and tank volume data; Step S2, inputting the tank level signal and the power-on state signal into a pre-trained machine learning model to process and output the probability of the vehicle being in a refueling state and the corresponding first tank level change rate; Step S3, judging whether the vehicle has a refueling behavior according to the probability of the vehicle being in a refueling state; Step S4, when the vehicle has a refueling behavior, generating a refueling behavior portrait of the vehicle user according to the user identification and the initial tank level signal, the terminal tank level signal and the first tank level change rate in the refueling process, and outputting the refueling behavior portrait; The step S3 comprises: Obtaining the driving mileage data when the last refueling behavior of the vehicle occurs, and calculating a second tank level change rate according to the driving mileage data when the last refueling behavior of the vehicle occurs, the current driving mileage data, the average fuel consumption data and the tank volume data; According to the comparison result of the first tank level change rate and the second tank level change rate, it is determined whether the output result of the machine learning model is reliable, if yes, the probability of the vehicle being in a refueling state output by the machine learning model is multiplied by a coefficient k to obtain a modified probability of the vehicle being in a refueling state; Wherein the coefficient k is a value greater than 0; If the modified probability of the vehicle being in a refueling state is greater than a preset probability threshold, it is determined that the vehicle has a refueling behavior.

2. The vehicle refueling behavior recognition method according to claim 1, characterized in that, The step S2 comprises: The machine learning model determines whether the vehicle is in a stationary state according to the power-on state signal, and the power-on state signal has a step state from 1 to 0, if yes, the machine learning model determines that the probability of the vehicle being in a refueling state is a preset initial probability X% and starts to decay, if not, the machine learning model determines that the vehicle is not in a refueling state; The machine learning model determines whether the tank level of the vehicle rises within the preset time according to the tank level signal, if yes, the probability of the vehicle being in a refueling state corresponds to an increase, and when the tank level stops rising, the probability of the vehicle being in a refueling state and the corresponding first tank level change rate are output, if not, the probability of the vehicle being in a refueling state decays to 0 with the preset initial probability X% as the initial value within the preset time.

3. The vehicle refueling behavior recognition method according to claim 2, characterized in that, The step S2 comprises: If the tank level rises from the initial tank level Y% to the terminal tank level Z% within the preset time, and (Z%-Y%)>F%, the machine learning model determines that the probability of the vehicle being in a refueling state is 100%; wherein F% is a preset threshold, Y% and Z% are tank level signals uploaded by the vehicle.

4. The vehicle refueling behavior recognition method according to claim 1, characterized in that, The vehicle state data further includes a positioning signal; The step S3 further comprises: When it is determined that the vehicle has the refueling behavior according to the probability that the vehicle is in the refueling state, the credibility of the refueling behavior is further verified according to the positioning signal at the time when the refueling behavior occurs, and whether there is a gas station in a preset distance range centered on the position of the positioning signal is determined according to the map data; if yes, the refueling behavior is credible, and it is determined that the refueling behavior exists; if no, the refueling behavior is not credible, and it is determined that the refueling behavior does not exist.

5. The vehicle refueling behavior recognition method according to claim 4, characterized in that, The step S3 further includes: when the refueling behavior is credible and it is determined that the refueling behavior exists, obtaining gas station information in the preset distance range; The step S4 further includes: updating the gas station heat data according to the gas station information, and outputting.

6. A vehicle refueling behavior recognition system characterized by, Comprise: A data receiving unit is configured to receive vehicle state data uploaded by a vehicle in real time; The vehicle state data at least includes a user identifier, a tank level signal, a power-on state signal, driving mileage data, average fuel consumption data per 100 kilometers, and tank volume data; A refueling behavior identification unit is configured to input the tank level signal and the power-on state signal into a pre-trained machine learning model to process and output a probability that the vehicle is in a refueling state and a corresponding first tank level change rate; A credibility determination unit is configured to determine whether the vehicle has a refueling behavior according to the probability that the vehicle is in the refueling state; And An image generation unit is configured to generate a refueling behavior image of a vehicle user according to the user identifier and initial and terminal tank level signals during the refueling process and the first tank level change rate when the vehicle has the refueling behavior, and output the refueling behavior image. The credibility determination unit comprises: A first verification subunit is configured to obtain driving mileage data when a previous refueling behavior of the vehicle occurs, calculate a second tank level change rate according to the driving mileage data when the previous refueling behavior of the vehicle occurs, current driving mileage data, average fuel consumption data per 100 kilometers, and tank volume data, and determine whether the output result of the machine learning model is credible according to a comparison result of the first tank level change rate and the second tank level change rate; if yes, multiply the probability that the vehicle is in the refueling state output by the machine learning model by a coefficient k to obtain a modified probability that the vehicle is in the refueling state; wherein the coefficient k is a value greater than 0; if the modified probability that the vehicle is in the refueling state is greater than a preset probability threshold, it is determined that the vehicle has the refueling behavior.

7. The vehicle refueling behavior recognition system of claim 6, wherein, The machine learning model determines whether the vehicle is in a stationary state according to the power-on state signal, and the power-on state signal has a step state from 1 to 0; if yes, the machine learning model determines that the probability that the vehicle is in the refueling state is a preset initial probability X% and starts to decay; if no, the machine learning model determines that the vehicle is not in the refueling state. The machine learning model determines whether the oil tank level of the vehicle rises within the preset time according to the oil tank level signal. If yes, the probability of the vehicle being in the refueling state rises, and when the oil tank level stops rising, the probability of the vehicle being in the refueling state and the corresponding first oil tank level change rate are output. If no, the probability of the vehicle being in the refueling state within the preset time decays to 0 with a preset initial probability X% as an initial value; If the oil tank level rises from an initial oil tank level Y% to a terminal oil tank level Z% within the preset time, and (Z%-Y%)>F%, the machine learning model determines that the probability of the vehicle being in the refueling state is 100%. F% is a parameter threshold value determined by pre-training of the machine learning model, and Y% and Z% are the oil tank level signals uploaded by the vehicle.

8. The vehicle refueling behavior recognition system of claim 6, wherein, The vehicle state data further includes a positioning signal; The credibility judgment unit comprises: The second inspection subunit is configured to, when determining that the vehicle has the refueling behavior according to the probability of the vehicle being in the refueling state, further determine the credibility of the refueling behavior according to the positioning signal at the time of the refueling behavior, and determine whether there is a gas station within a preset distance range centered on the position of the positioning signal according to map data. If yes, the refueling behavior is credible, and it is determined that the refueling behavior exists. If no, the refueling behavior is not credible, and it is determined that the refueling behavior does not exist.

9. A server, characterized by It comprises: A memory for storing a computer program; A processor for executing the computer program to implement the steps of the vehicle refueling behavior identification method according to any one of claims 1 to 5.

10. A computer-readable storage medium having stored thereon computer program instructions, wherein, The computer program instructions are executed by the processor to implement the steps of the vehicle refueling behavior identification method according to any one of claims 1 to 5.

Images