PERSONAL RECOMMENDATIONS FOR SERVICE STATIONS

Abstract

A system comprises a monitoring module configured to determine at least one location and route of a vehicle, and a recommendation module configured to identify multiple service stations based on the vehicle's at least one location and route. The monitoring module compares a preference of a first user of the vehicle with preference data relating to a second user of a different vehicle, predicts a preferred service station from among the multiple service stations based on the comparison, and presents a recommendation to the first user, indicating the preferred service station.

Description

BACKGROUND

[0001] The subject matter of the disclosure relates to energy or power transfer and in particular to systems and methods for controlling power transfer between energy storage systems with different parameters.

[0002] Vehicles, including gasoline and diesel-powered vehicles as well as electric and hybrid electric vehicles, have battery storage systems to power electric motors, electronics, and other vehicle subsystems. Battery systems can be recharged via dedicated charging stations and other power sources such as homes and buildings connected to the electrical grid. When a vehicle is on the road, there may be multiple charging points available. It is desirable to provide a device or system that can identify a preferred or optimal charging point for the vehicle. SUMMARY

[0003] In an exemplary embodiment, a system comprises a monitoring module configured to determine at least one location and route of a vehicle, and a recommendation module configured to identify multiple service stations based on the vehicle's location and / or route. The monitoring module compares a preference of a first user of the vehicle with preference data relating to a second user of a different vehicle, predicts a preferred service station from among the multiple service stations based on the comparison, and presents a recommendation to the first user, specifying the preferred service station.

[0004] In addition to one or more of the features described here, the vehicle is an electric vehicle and the multiple service stations are multiple charging stations.

[0005] In addition to one or more of the characteristics described here, the first user and the second user belong to a group of multiple users, and the second user is selected from the group of multiple users based on a similarity between the second user and the first user.

[0006] In addition to one or more of the features described here, the similarity is determined based on comparing an attribute of the first user with an attribute of each of the multiple users.

[0007] In addition to one or more of the features described here, the prediction of the preferred service station includes assigning a predicted rating to at least one of the multiple service stations.

[0008] In addition to one or more of the features described here, the predicted score is determined based on a multi-user scoring matrix, where the multi-users include the first user and the second user, and the scoring matrix includes a score for each combination of a user and an identified service station.

[0009] In addition to one or more of the features described here, the similarity is determined based on a first set of latent factors for each user, the multiple users, and a second set of latent factors for the multiple service stations, with the first set of latent factors and the second set of latent factors being estimated based on a machine learning algorithm.

[0010] In addition to one or more of the features described here, predicting the preferred service station includes training the machine learning algorithm, generating a latent user vector for each user from the multiple users, generating a latent service station vector for each service station from the multiple service stations, and combining the latent user vectors and the latent service station vectors.

[0011] In addition to one or more of the features described here, predicting the preferred service station includes selecting the second user based on the combination and assigning a predicted rating to the first user based on a rating from the second user.

[0012] In another exemplary embodiment, a method comprises determining at least one location and route of a vehicle, identifying multiple service stations based on at least the location and route of the vehicle, comparing a preference of a first user of the vehicle with preference data relating to a second user of another vehicle, predicting a preferred service station from among the multiple service stations based on the comparison, and presenting a recommendation to the first user, wherein the recommendation specifies the preferred service station.

[0013] In addition to one or more of the features described here, the vehicle is an electric vehicle and the multiple service stations are multiple charging stations.

[0014] In addition to one or more of the characteristics described here, the first user and the second user belong to a group of multiple users, and the second user is selected from the group of multiple users based on a similarity between the second user and the first user.

[0015] In addition to one or more of the features described here, the prediction of the preferred service station includes assigning a predicted rating to at least one of the multiple service stations.

[0016] In addition to one or more of the features described here, similarity is determined on the basis of a multi-user rating matrix, where the multi-users include the first user and the multiple second users, and the rating matrix includes a rating for each combination of a user and an identified service station.

[0017] In addition to one or more of the features described here, the similarity is determined based on a first set of latent factors for each user, the multiple users, and a second set of latent factors for the multiple service stations, with the first set of latent factors and the second set of latent factors being estimated based on a machine learning algorithm.

[0018] In addition to one or more of the features described here, predicting the preferred service station includes training the machine learning algorithm, generating a latent user vector for each user from the multiple users, generating a latent service station vector for each service station from the multiple service stations, and combining the latent user vectors and the latent service station vectors.

[0019] In addition to one or more of the features described here, predicting the preferred service station includes selecting the second user based on the combination and assigning a predicted rating to the first user based on a rating from the selected second user.

[0020] In another exemplary embodiment, a vehicle system comprises a memory containing computer-readable instructions and a processing device for executing the computer-readable instructions, wherein the computer-readable instructions control the processing device to perform a method. The method comprises determining at least one location and route of a vehicle, identifying multiple gas stations based on at least the location and route of the vehicle, and comparing a preference of a first user of the vehicle with preference data relating to a second user of another vehicle, wherein the first user and the second user are part of multiple users.The procedure also includes predicting a preferred service station from among the multiple service stations based on the comparison and presenting a recommendation to the first user, with the recommendation indicating the preferred service station.

[0021] In addition to one or more of the characteristics described here, the second user is selected from the multiple users based on a similarity between the second user and the first user.

[0022] In addition to one or more of the features described here, the similarity is determined based on a rating matrix for the multiple users, wherein the rating matrix includes a rating for each combination of a user and an identified service station, and the similarity is determined based on a first set of latent factors for each user of the multitude of users and a second set of latent factors for the multiple service stations, wherein the first set of latent factors and the second set of latent factors are estimated based on a machine learning algorithm.

[0023] The aforementioned features and advantages, as well as other features and advantages of the disclosure, are readily apparent from the following detailed description in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Further features, advantages and details are only examples included in the following detailed description, which refers to the drawings in which they are illustrated: Fig. Figure 1 is a schematic top view of a motor vehicle with a battery system according to an exemplary embodiment; Fig. Figure 2 schematically shows a series of vehicle users and charging stations and illustrates aspects of a method for predicting a user preference and presenting a service station recommendation according to an exemplary embodiment; Fig. Figure 3 is a flowchart illustrating aspects of a procedure for recommending a service station to a vehicle user according to an exemplary embodiment. Fig. Section 4 shows aspects of determining a service station recommendation based on an evaluation matrix and a collaborative filtering technique according to an exemplary embodiment; Fig. Figure 5 shows aspects of determining a service station recommendation based on latent factors learned via the machine learning model, according to an exemplary embodiment; Fig. Figure 6 shows aspects of determining a service station recommendation based on latent factors learned via the machine learning model, according to an exemplary embodiment; Fig. Figure 7 shows an example of an evaluation matrix according to an exemplary embodiment; Fig. Figure 8 shows an example of a customer filter according to an exemplary embodiment; Fig. Figure 9 shows an example of a multidimensional embedding that includes clusters associated with users with similar preferences, in accordance with an exemplary embodiment; and Fig. Figure 10 shows a computer system according to an exemplary embodiment. DETAILED DESCRIPTION

[0025] The following description is merely exemplary and is not intended to limit the present disclosure, its application, or use. It is understood that identical or corresponding parts and features in the drawings are identified by appropriate reference numerals.

[0026] In accordance with one or more exemplary embodiments, methods, devices, and systems are provided to offer a user recommendations regarding available charging stations or other service points (e.g., gas stations and / or diesel stations for combustion engine and hybrid vehicles, hydrogen stations for fuel cell vehicles, etc.). One embodiment of a recommendation system is designed to provide a vehicle user with personalized recommendations based on user preferences, vehicle type, and behavior. The recommendation may be based on a rating assigned to each charging / service point for the user.

[0027] In one embodiment, the rating assigned to a charging station for a user (also referred to as the "first user") is estimated based on determining preferences and other information for several other users (also referred to as "second users") and determining a similarity between the first user and each of the several second users. The similarity can be determined by a collaborative filtering technique that uses matrix factorization to identify latent factors associated with each user and each charging station. The latent factor(s) and / or other information associated with the other users are compared to identify another user or users with the greatest similarity. One or more charging stations (e.g.,Charging stations that have not yet been assigned a rating will be assigned a rating based on an existing rating assigned to one or more charging stations by a similar user. The points assigned to each charging station can then be used to recommend an optimal or desired charging station to the user.

[0028] The embodiments described here offer numerous advantages and technical benefits. For example, they improve navigation, user experience, the charging process, and vehicle performance by providing personalized recommendations for charging stations that are most advantageous for a user. These embodiments offer benefits such as reduced or minimized travel and / or charging times, increased user satisfaction, and more. They enable the user to utilize a charging station that best suits their behavior, route, and preferences, thereby saving time and ensuring that the most suitable charging station is used.

[0029] Existing recommendation systems suggest public charging stations along a driving route; however, these recommendations are based solely on distance from the route. The implementations described here adapt the recommendations for specific users based on their preferences and behavior, thereby optimizing user satisfaction and the overall experience with public chargers. Furthermore, user preferences and ratings for a particular charging station can be derived without directly surveying the user.

[0030] The embodiments are not limited to use with a specific vehicle, device, or system that uses battery assemblies and can be applicable in various contexts. For example, the embodiments can be used in cars, trucks, aircraft, construction equipment, agricultural equipment, automated factory plants, and / or other devices or systems that can use charging stations.

[0031] Fig. Figure 1 shows an embodiment of a motor vehicle 10 comprising a vehicle body 12 that at least partially forms a passenger compartment 14. The vehicle body 12 also carries various subsystems of the vehicle, including a drive system 16 and other subsystems for supporting the functions of the drive system 16 and other vehicle components, such as a brake subsystem, a suspension system, a steering subsystem, a fuel injection subsystem, an exhaust subsystem, and the like.

[0032] Vehicle 10 can be a vehicle with an internal combustion engine, an electric vehicle (EV), or a hybrid electric vehicle (HEV). In one example, vehicle 10 is a hybrid vehicle comprising an internal combustion engine 18 and an electric motor 20.

[0033] The vehicle 10 comprises a battery system 22, which may be electrically connected to the motor 20 and / or other components, such as the vehicle electronics. In one embodiment, the battery system 22 comprises a battery assembly, e.g., a high-voltage battery pack 24 with multiple battery modules 26. Each of the battery modules 26 comprises an array of individual cells (not shown). The battery system 22 may also include a monitoring unit 28 configured to receive measurements from sensors 30. Each sensor 30 may be an assembly or a system with one or more sensors for measuring various battery and environmental parameters, such as temperature, current, and voltages. The monitoring unit 28 comprises components such as a processor, memory, an interface, a bus, and / or other suitable components.

[0034] The battery system 22 comprises various conversion devices for controlling the power supply from the battery pack 24 to the motor 20 and / or to electronic components. The conversion devices include a direct current (DC)-DC converter module 32 with a DC-DC converter 34. The conversion devices also include an inverter module 36 with an inverter 38, which receives direct current (DC) from the DC-DC converter 34 and converts the direct current into alternating current (AC), which is supplied to the electric motor 20.

[0035] The vehicle 10 also includes a charging system that can be used to charge the battery system 22 and / or to supply power from the battery system 22 to charge another energy storage system (e.g., vehicle-to-vehicle (V2V) and / or vehicle-to-everything (V2X) charging). The vehicle's charging system includes a charging control device 40, e.g., an onboard charging module (OBCM), which is connected to a charging port 42.

[0036] The charging control device 40 can be designed to perform other functions, such as monitoring battery parameters (e.g., temperature, voltage, current, and impedance) during a charging process, controlling aspects of a charging process, and / or providing recommendations for charging stations, as described here.

[0037] The vehicle 10 comprises at least one processor or processing unit for controlling aspects of the identification and recommendation of charging stations, which is referred to as processor 44. Processor 44 can be, as shown, a separate device or part of the vehicle's monitoring and / or navigation system. It should be noted that embodiments are not limited to a specific control or processing unit and may include multiple processors or control units.

[0038] The vehicle 10 also includes a computer system 48, which comprises one or more processing units 50 and a user interface 52. The computer system 48 can, for example, communicate with a control unit or a vehicle system to provide it with commands in response to user input. The various processing units, modules, and units can communicate with each other via a communication device or system, such as a Controller Area Network (CAN) or Transmission Control Protocol (TCP) bus.

[0039] The processor 44, the computer system 48, and / or other processing components in the vehicle 10 can be configured to communicate with various remote devices and systems, such as charging stations and other vehicles. Such communication can take place, for example, via a network 54 (e.g., a mobile network, a cloud, etc.) and / or wirelessly. Thus, the vehicle 10 can communicate with various charging stations 56, a remote unit 58 (e.g., a workstation, a fleet management system, a computer, a server, a mapping system, etc.), and / or a database 60. The database 60 can store information about the locations and parameters of charging stations (e.g., conventional or DC fast charging stations) as well as user information. A rating matrix, as described above, can also be stored in the database 60.

[0040] Embodiments comprise one or more methods for identifying and recommending one or more charging stations to a user. Generally, the method involves identifying potential charging stations that the user could use based on their route and / or location. The method also includes providing a recommendation to a user based on ratings or preferences of one or more other users who are sufficiently similar to the user.

[0041] Fig. Figure 2 schematically shows a series of vehicle users and charging stations and illustrates aspects of an example of a recommendation procedure described here. In this example, a first user 70 drives an electric vehicle, and it is determined that the vehicle 10 should visit a charging station. Based on the vehicle's location and / or route, four charging stations 72a, 72b, 72c, and 72d are considered available (e.g., within a selected distance from vehicle 10 and / or at a location along a route).

[0042] The recommendation process involves identifying one or more similar users. A "similar" user refers to another user of a different vehicle, where the other user and / or vehicle share at least one attribute (or at least one attribute that is sufficiently similar). In this example, user information such as vehicle type and demographic data is used to identify a similar user, referred to as Second User 74.

[0043] A processing device accesses the charging station and user data (e.g., in database 60), which includes information about the first user 70 and the second user 74 (and possibly one or more additional users / drivers). The charging station and user data also include information about the available charging stations.

[0044] The second user, 74, is associated with a rating or ranking for each charging station 72a, 72b, 72c, and 72d. It should be noted that a rating or ranking is specific to a particular user and charging station.

[0045] The ratings of the charging stations for the second user 74 include, for example, a rating R2a for charging station 72a, a rating R2b for charging station 72b, a rating R2c for charging station 72c, and a rating R2d for charging station 72d. Each rating in this example is indicated by a "thumbs up" symbol, representing a positive (or relatively high) rating, or by a "thumbs down" symbol, representing a negative (or relatively low) rating.

[0046] The first user, 70, is associated with ratings for each of the charging stations 72a, 72b, and 72c, but has no rating for charging station 72d. The ratings for the charging stations for the first user, 70, include, for example, a rating of R1a for charging station 72a, a rating of R1b for charging station 72b, and a rating of R1c for charging station 72c.

[0047] The procedure involves predicting a rating that the first user 70 would assign based on the preferences of the second user 74. The procedure then involves assigning an initial user rating (thumbs up) to charging station 72d, which is the same as the rating for charging station 72d associated with the second user 74 (thumbs up). Since all charging stations now have points assigned to user 70, a recommendation can be presented to user 70.

[0048] Fig. Figure 3 shows an embodiment of a method 80 for recommending a charging station or other service station or location to a user. The method 80 comprises a series of steps or stages represented by blocks 81-88. The method 80 is not limited in the number or sequence of steps, since some of the steps represented by blocks 81-88 may be performed in a different order than that described below, or fewer than all of the steps may be performed.

[0049] Method 80 is described for illustrative purposes in connection with vehicle 10 and processor 44. It is understood that method 80 can be carried out with any type of vehicle and any suitable processing equipment or combination of processing equipment.

[0050] Although the embodiments are described in connection with electric vehicles and charging stations, this does not limit their applicability. The embodiments can, for example, apply to combustion engine and hybrid vehicles, as well as to other types of service stations (e.g., gas stations, mechanics, dealers, etc.).

[0051] In block 81, processor 44 determines that it is desirable for vehicle 10 to visit a public charging station. Processor 44 can make this determination based on a user request or a signal indicating that the battery system has a low charge level. A driver or user of vehicle 10 is referred to here as the "first user".

[0052] In block 82, the processor collects or accesses information 44 describing characteristics of the first user and the vehicle 10. This information may include user preferences (e.g., the charging station should be located near a restaurant or other point of interest), user demographics (e.g., age), any limitations of the first user (e.g., mobility issues that may affect the type of charging station the first user can comfortably use), and other information relevant to identifying similarities between the first user and other users. This information may also include the vehicle type and charging capabilities.

[0053] In block 83, the processor identifies 44 available charging stations that are located within a selected distance to vehicle 10 and / or are easily accessible from a planned route.

[0054] In block 84, processor 44 accesses user data for one or more other users (secondary users) who have used the available charging stations and / or provided a ranking or other preference information regarding the available charging stations. The user data for these other users may include ratings or preferences for each other user's charging stations, as well as demographic information.

[0055] In block 85, the user data is compared with the information about the first user, and processor 44 determines a degree of similarity between the first user and each of the other users. The degree of similarity can be determined, at least in part, by finding matching or similar characteristics among the users. Examples of such characteristics include age (and / or other demographic characteristics), preferences, vehicle type, loading capabilities, and more.

[0056] In one embodiment, the degree of similarity is determined at least partially by estimating latent factors for each user (the first user and the other users). A “latent factor” is any feature or attribute of a user that is determined by machine learning, as further explained here. Latent factors can be discovered without having to question or prompt the user of vehicle 10, thus enabling a similarity determination without user input.

[0057] In block 86, the processor identifies which other user or group of users is most similar (the "similar user" or "similar users") and assigns a predicted rating to each available charging station for the first user (or any available charging station that does not have a predefined rating for the first user). The predicted rating is based on the scores or preferences of similar users.

[0058] If a rating is predefined or already assigned to one or more charging stations (for the first user), the ratings for similar users are used to predict ratings, and each of the remaining charging stations is assigned a predicted rating. If multiple similar users have assigned different ratings to a charging station, the predicted rating can be based on an average of the ratings or another value derived from the ratings.

[0059] In block 87, processor 44 determines which charging station has the highest rating (a predefined rating or a predicted rating) or which group of charging stations has the highest rating, and makes a recommendation as to which charging station the first user prefers. The recommendation can be in the form of a single charging station or several charging stations that correspond to the first user's preferences. For example, a list of charging stations can be displayed graphically or textually, along with their respective rankings or ratings (e.g., numerical rankings, colors, thumbs up / down icons, or other symbols, etc.). A recommendation can be presented via any suitable modality (e.g., graphically via a touchscreen or heads-up display, audibly, etc.).

[0060] The predicted rating and recommendation of charging stations can take into account additional factors beyond those used to determine similarity. For example, the machine learning model can also consider the expected availability along a planned route, the distance from the planned route, reliability issues, and other charging-related aspects. This can be achieved using a weighted rating formula.

[0061] In Block 88, various actions can be performed based on the recommendation. For example, directions to a recommended charging station can be provided, or if the vehicle 10 has an autonomous control function, it can be autonomously controlled to drive to the recommended charging station. In another example, the vehicle 10 can communicate with a network and / or the recommended charging station.

[0062] Fig. Figure 4 schematically shows an embodiment of method 80 of Fig. 3, where similarity determination is based on the factorization of a matrix of user ratings. Similarities are identified through collaborative filtering, with the matrix factorization used to learn latent factors. The latent factors are used to determine which other user(s) are similar to an initial user. A rating assigned to a charging station for a similar user can then be used to predict a rating and assign the predicted rating to the charging station for the initial user.

[0063] With reference to Fig. 4. Charging station data 90 and user data 92 are accessed and used to create or update a user-station matrix 94 of user ratings or scores for a variety of users (including the first user of the vehicle 10) and charging stations. The matrix 94 is referred to as the "rating matrix".

[0064] The charging station data 90 comprises various types of information for each of the multiple charging stations. Examples include an identifier (e.g., a numerical ID) and a location (e.g., from GPS communication) for each charging station. The charging station data 90 can also include other characteristics of each charging station, such as the charging stage (e.g., DC fast charging (DCFC)), the capability for automatic charging, the connector type, and more.

[0065] The user data 92 comprises various types of information for each of the multiple users, which can be used to determine similarities between users and user preferences. Examples include an identifier (e.g., a numerical identifier) ​​and demographic information for each user.

[0066] The user data 92 can also include information about users' experiences with different charging stations and their ratings. Such information can include actual user ratings, the number of visits to a particular charging station with successful charging sessions, the number of visits with unsuccessful attempts, the charging speed, and more.

[0067] The user and charging station data are used to create the rating matrix 94. A rating is calculated for each user and each charging station (provided sufficient information is available for the calculation).

[0068] In Fig. The rating matrix 94 comprises one row for m users (U1 ... U4). i ... U m ), and a column for n charging stations (CS1 ... CS1) j ... CS nThe rating matrix is ​​populated with a rating (e.g., 1-5) in one or more entries, indicating or calculating the preference or ranking. The rating can be taken directly from a known ranking or derived from other information. For example, the rating can be based on the number of successful charging attempts by a user at a specific charging station, the number of kilowatt-hours charged during a session, and / or the charging speed. A number of entries may be empty if the preference or ranking of a charging station for a particular user is unknown.

[0069] Processor 44 uses a collaborative filtering technique (represented by element 96) that can first find similar characteristics between users and vehicles (represented by element 98) and similarities between charging stations. These relationships can be used to identify one or more users who are most similar to the first user and to identify similar charging stations.

[0070] Processor 44 collects data for multiple users and charging stations from the rating matrix 94. Data collection can be performed for all users and charging stations or for a subset based on similar characteristics. For example, if a prediction and recommendation is performed for User1, data is collected for a group of other users who exhibit a sufficient degree of similarity.

[0071] In one embodiment, the collaborative filtering 96 includes learning latent factors of the detected users and charging stations (element 100). The latent factors are learned using machine learning techniques, as discussed below. The semantic relationships and / or latent factors are then used to predict the values ​​for U1 (represented by element 102).

[0072] The Fig. 5 and Fig. 6 are aspects of embodiments of method 80. In these embodiments, a neural network or other machine learning model is trained to detect latent factors of the users and the charging stations, which are used to determine similarities and predict ratings.

[0073] In this embodiment, user and charging station rating data 110 from the rating matrix 94 (e.g., a list of user and charging station identifiers, ratings, or combinations thereof) are entered into a latent feature space or embedding space (embedding). The user and charging station data 110 include, for example, a column for the user identifier (UID) and a column for the charging station identifier (CSID). A rating column (S) includes a numerical rating for each combination of a user and a charging station.

[0074] In one embodiment, the user data describing the characteristics of each user is input into an embedding layer 112. The embedding layer 112 is trained to generate clusters 114 of similar users. The clusters provide a set of latent user vectors (ULVs) 116 for each user identification.

[0075] Similarly, charging station data describing the characteristics of each available charging station is fed into an embedding layer 120, which is trained to generate clusters 122 of similar charging stations. The clusters provide a set of latent vectors (SCLV) 124 for each charging station identifier.

[0076] In one embodiment, which is in Fig. As shown in Figure 5, the latent vectors 116 are combined to form a dense layer 118, and the latent vectors 124 are combined to form a dense layer 126. The dense layers 118 and 126 are combined by calculating a cross product of the dense layers (represented by element 119). The result is a set of predicted ratings 128 for each combination of user and charging station. Losses (differences between predicted and actual ratings) can be used to refine the rating predictions.

[0077] Fig. 6 presents an alternative to the embodiment of Fig. 5. In this embodiment, the matrix information and the latent vectors are applied to another machine learning model 130 that learns the dot product via a deep neural network.

[0078] The training process for predicting results can be repeated as often as desired. For example, the training can be repeated at specific intervals (e.g., daily, weekly, etc.).

[0079] Fig. Figure 7 shows an example of the rating matrix 94 and examples of existing or predefined ratings. In this example, the rating matrix represents four users (U1 to U4) and four charging stations (CS1 to CS4). Points are missing in this example (represented by "?"), which can be predicted based on the similarities between the users. By applying this data to the machine learning model, including embedding layers 112 and 120, a missing rating can be added based on a rating from a similar driver. For example, if U1 and U3 are found to be similar (i.e., they have similar preferences), the rating "2" can be assigned to the combination of U1 and CS4.

[0080] In one embodiment, 80 user-defined filters can be added to the method to accommodate specific requirements or characteristics, or to further analyze the similarity between users and charging stations. A user-defined filtering procedure can be implemented to provide additional filtering. This procedure includes selecting or creating a category, such as the connector type, or indicating whether a charging mode, such as automatic charging, is available.

[0081] The user-defined filtering procedure involves creating a vector representation that is a concatenation of two vector types. The first vector is a learned embedding, denoted as "e". The norm of these vectors is significantly less than one (||e|| = ε << 1).

[0082] A second vector (scalar) represents a desired category "c", which serves as an indicator function and has a value of one or zero.

[0083] The concatenated vector is denoted by x. For a specific user with an embedding x1 and a specific charging station with an embedding x2, the concatenated vector is represented by: x2Tx1=x1⋅x2=e1⋅e2+c1⋅c2.

[0084] If x1 and x2 belong to the same category, the concatenated vector is represented by: x2Tx1=x1⋅x2=e1⋅e2+c1⋅c2=‖e1‖‖e2‖cos θ+1≈ε+1.

[0085] If x1 and x2 do not belong to the same category, the concatenated vector is represented by: x2Tx1=x1⋅x2=e1⋅e2+c1⋅c2=‖e1‖‖e2‖cos θ+0≈ε.

[0086] Adding the indicator function results in vectors of the same categories having significantly higher point products.

[0087] Fig. Figure 8 shows an example of user rating information 132 for user U1, which includes a predicted rating S assigned to each charging station within a group of charging stations, and also an indicator function for each of two categories. The first category (denoted UA) is whether the user's vehicle has auto-charging capability. The indicator function value is one if the user's vehicle has auto-charging capability and zero if the user's vehicle does not. The second category (denoted CSA) is whether a charging station has auto-charging capability. An indicator function value of one is provided if the charging station has auto-charging capability, and a value of zero is provided if the charging station does not.

[0088] Fig. Figure 8 also shows the category score for user U1 and each charging station. As shown, using an indicator function results in a higher point product and, consequently, a higher score. For example, the scores for a user and a charging station in the same category are significantly higher (4.8 and 4.5) than the scores for charging stations in a different category than the user (2, 2.3, and 1.5).

[0089] It should be noted that the embedding layers can be two-dimensional, three-dimensional, or in any number of dimensions. Fig. Figure 9 shows an example of a 15-dimensional user embedding layer 140, which represents multiple users. The embedding layer comprises different clusters, such as cluster 142, which represent users who are assumed to have similar preferences (i.e., similar users).

[0090] Fig.Figure 10 illustrates aspects of an embodiment of a computer system 240 that can perform various aspects of the embodiments described herein. The computer system 240 comprises at least one processing device 242, which generally includes one or more processors for performing aspects of the image acquisition and analysis procedures described herein.

[0091] Components of the computer system 240 include the processing device 242 (such as one or more processors or processing units), a memory 244, and a bus 246 that connects various system components, including the system memory 244, to the processing device 242. The system memory 244 can be a non-transient, computer-readable medium and can comprise a variety of computer-readable media. These media can be any available media accessible to the processing device 242 and can include both volatile and non-volatile media, as well as removable and non-removable media.

[0092] System memory 244 includes, for example, non-volatile memory 248, such as a hard disk, and may also include volatile memory 250, such as random access memory (RAM) and / or cache memory. The computer system 240 may also include other removable / non-removable, volatile / non-volatile computer system storage media.

[0093] The system memory 244 can comprise at least one program product with a set (i.e., at least one) of program modules configured to perform the functions of the embodiments described herein. For example, the system memory 244 stores various program modules that generally perform the functions and / or methodologies of the embodiments described herein. A module 252 can be included for performing functions related to conducting impedance measurements, and a module 254 can be included for performing functions related to controlling charging processes.

[0094] System 240 is not limited by this, as other modules can also be included. As used herein, the term "module" refers to processing circuits that may include an application-specific integrated circuit (ASIC), an electronic circuit, a processor (shared, dedicated, or as a group), memory that executes one or more software or firmware programs, a combinational logic circuit, and / or other suitable components that provide the described functionality.

[0095] The processing device 242 can also communicate with one or more external devices 256, such as a keyboard, a pointing device, and / or other devices (e.g., a network card, a modem, etc.), which enable the processing device 242 to communicate with one or more other computer devices. Communication with various devices can take place via the input / output (I / O) interfaces 264 and 265.

[0096] The processing device 242 can also communicate with one or more networks 266, such as a local area network (LAN), a wide area network (WAN), a bus network, and / or a public network (e.g., the Internet), via a network adapter 268. It is understood that other hardware and / or software components can also be used in conjunction with the computer system 40, even if they are not shown. Examples include, but are not limited to: microcode, device drivers, redundant processing units, external hard disk arrays, RAID systems, data archiving systems, etc.

[0097] The terms "a" and "a / an / an" do not imply a limitation of quantity, but rather denote the presence of at least one of the mentioned items. The term "or" means "and / or" unless the context clearly indicates otherwise. When the entire description refers to "an aspect," this means that a particular element (e.g., a feature, a structure, a step, or a property) described in connection with the aspect is encompassed by at least one of the aspects described herein and may or may not be present in other aspects. It is understood that the described elements can be combined in any suitable way across the various aspects.

[0098] When an element, such as a layer, film, area, or substrate, is described as lying "on" another element, it may lie directly on top of the other element, or there may be intermediate elements. Conversely, when an element is described as lying "directly on" another element, there are no intermediate elements.

[0099] Unless otherwise stated herein, all testing standards are the latest standard in force on the filing date of this application or, if priority is claimed, the filing date of the earliest priority application in which the testing standard appears.

[0100] Unless otherwise defined, the technical and scientific terms used herein have the same meaning as they are generally understood by a person skilled in the art in the field of the present invention.

[0101] Although the above disclosure has been described with reference to exemplary embodiments, the person skilled in the art understands that various modifications can be made and equivalent elements substituted without departing from the scope of application. Furthermore, many modifications can be made to adapt a particular situation or material to the teachings of the disclosure without deviating from its essential scope. Therefore, the present disclosure is not intended to be limited to the particular embodiments, but rather to encompass all embodiments that fall within its scope.

Claims

[1] System, encompassing: a monitoring module designed to determine at least one location and one route of a vehicle; and a recommendation module designed for implementation: Identifying multiple service stations based on at least the location and route of the vehicle; Comparing a preference of a first user of the vehicle with preference data relating to a second user of a different vehicle; Predictions of a preferred service station from among the multiple service stations based on comparison; and Presenting a recommendation to the first user, where the recommendation specifies the preferred service station. [2] System according to claim 1, wherein the vehicle is an electric vehicle and the multiple service stations are multiple charging stations. [3] System according to claim 1, wherein the first user and the second user are part of several users and the second user is selected from the several users on the basis of a similarity between the second user and the first user. [4] System according to claim 3, wherein the similarity is determined on the basis of comparing an attribute of the first user with an attribute of each user from the multiple users. [5] System according to claim 1, wherein predicting the preferred service station comprises assigning a predicted rating to at least one of the multiple service stations. [6] System according to claim 5, wherein the predicted rating is determined on the basis of a rating matrix for multiple users, wherein the multiple users include the first user and the second user, and wherein the rating matrix includes a rating for each combination of a user and an identified service station. [7] System according to claim 6, wherein the similarity is determined on the basis of a first set of latent factors for each user from the multiple users and a second set of latent factors for the multiple service stations, wherein the first set of latent factors and the second set of latent factors are estimated on the basis of a machine learning algorithm. [8] System according to claim 7, wherein predicting the preferred service station comprises training the machine learning algorithm, generating a latent user vector for each user from the multiple users, generating a latent service station vector for each service station from the multiple service stations, combining the latent user vectors and the latent service station vectors, selecting the second user based on the combination, and assigning a predicted rating to the first user based on a rating of the second user. [9] Procedures, comprehensive: Determine at least one location and route of a vehicle; Identifying multiple service stations based on at least the location and route of the vehicle; Comparing a preference of a first user of the vehicle with preference data relating to a second user of a different vehicle; Predictions of a preferred service station from among the multiple service stations based on comparison; and Presenting a recommendation to the first user, where the recommendation specifies the preferred service station. [10] Method according to claim 9, wherein the first user and the second user are part of several users, the second user is selected from several users on the basis of a similarity between the second user and the first user, and the similarity is determined on the basis of a rating matrix for several users, wherein the several users include the first user and the several second users, and wherein the rating matrix includes a rating for each combination of a user and an identified service station.

Images