Range extender control method, device, electronic device and storage medium based on federated learning

Abstract

The present application relates to the field of vehicle control technology and provides a method, device, electronic device, and storage medium for controlling a range extender based on federated learning. The method obtains global federated learning model parameters from a server to obtain a local federated learning model. The local federated learning model is trained based on the vehicle's historical usage data and the historical usage data of other vehicles of the same vehicle type as the vehicle to obtain a model gradient value. The model gradient value is sent to the server so that the server updates the global federated learning model based on the model gradient values ​​received from each vehicle. The above steps are repeated until the global federated learning model converges. The model parameters of the converged global federated learning model are used to update a target local federated learning model. The target local federated learning model is then used to determine the power allocation of the range extender of a first target vehicle. This method can improve the accuracy of determining the range extender power allocation using a prediction model and enhance the user experience.

Description

Technical Field

[0001] The present application relates to the field of vehicle control technology, and in particular to a range extender control method, device, electronic device and storage medium based on federated learning. Background Art

[0002] Plug-in hybrid electric vehicles (PHEVs), such as range-extended hybrid electric vehicles (REPs), are a key component of the current new energy vehicle market. These REPs utilize a range extender to charge the vehicle's battery, which then powers the motor and other onboard devices, addressing the range concerns of pure electric vehicles.

[0003] Range-extended hybrid vehicles can operate in both pure electric and extended-range modes. Furthermore, when the vehicle is in extended-range mode, the range extender can operate at varying power levels. To balance vehicle performance and cost, related technologies employ machine learning models to automatically control the range extender's activation timing and adjust its power levels in different situations.

[0004] However, the various models used in related technologies fail to differentiate between different vehicle types, resulting in insufficient control accuracy. Developing appropriate range extender management strategies for different vehicle types to balance performance and cost remains a technical challenge. Summary of the Invention

[0005] In view of this, the embodiments of the present application provide a range extender control method, device, electronic device and storage medium based on federated learning to solve the problem in the prior art that the control accuracy of the range extender control model obtained through training is low due to insufficient data from a single supplier.

[0006] A first aspect of an embodiment of the present application provides a range extender control method based on federated learning, including:

[0007] Receive the global federated learning model parameters sent by the server, and initialize the local federated learning model based on the global federated learning model parameters;

[0008] Obtaining historical usage data of the first target vehicle;

[0009] Obtaining a vehicle identification of a first target vehicle, and determining a vehicle type based on the vehicle identification;

[0010] Obtaining historical usage data of a second target vehicle, where the second target vehicle is of the same vehicle type as the first target vehicle;

[0011] Training a local federated learning model based on historical usage data of the first target vehicle and historical usage data of the second target vehicle to obtain a model gradient value;

[0012] Send the model gradient value to the server;

[0013] Receiving updated global federated learning model parameters sent by the server, where the updated global federated learning model is determined by the server after updating the global federated learning model based on the received multiple model gradient values;

[0014] Repeating the steps of receiving the global federated learning model parameters sent by the server, training the local federated learning model based on the historical usage data of the first target vehicle and the historical usage data of the second target vehicle to obtain a model gradient value, and receiving the updated global federated learning model parameters determined by the server according to the received model gradient value until target global federated learning model parameters are received, where the target global federated learning model parameters are the model parameters when the global federated learning model converges;

[0015] Update the local federated learning model based on the target global federated learning model parameters to obtain the target local federated learning model;

[0016] The operating condition information, vehicle status information, and vehicle type of the first target vehicle during this driving are obtained, and the power distribution of the range extender of the first target vehicle is determined using the target local federated learning model.

[0017] A second aspect of an embodiment of the present application provides a range extender control device based on federated learning, including:

[0018] a receiving module configured to receive global federated learning model parameters sent by the server and initialize the local federated learning model based on the global federated learning model parameters;

[0019] an acquisition module configured to acquire historical usage data of a first target vehicle;

[0020] The acquisition module is further configured to acquire a vehicle identification of the first target vehicle and determine the vehicle type based on the vehicle identification;

[0021] The acquisition module is further configured to acquire historical usage data of a second target vehicle, the second target vehicle being of the same vehicle type as the first target vehicle;

[0022] A training module is configured to train a local federated learning model based on historical usage data of the first target vehicle and historical usage data of the second target vehicle to obtain a model gradient value;

[0023] A communication module is configured to send the model gradient value to the server;

[0024] The communication module is further configured to receive updated global federated learning model parameters sent by the server, where the updated global federated learning model is determined by the server after updating the global federated learning model based on the received multiple model gradient values;

[0025] The training module is further configured to repeatedly perform the steps of receiving global federated learning model parameters sent by the server, training the local federated learning model based on the historical usage data of the first target vehicle and the historical usage data of the second target vehicle to obtain a model gradient value, and receiving updated global federated learning model parameters determined by the server according to the received model gradient value, until target global federated learning model parameters are received, where the target global federated learning model parameters are model parameters when the global federated learning model converges;

[0026] The training module is further configured to update the local federated learning model based on the target global federated learning model parameters to obtain a target local federated learning model;

[0027] The control module is configured to obtain operating condition information, vehicle status information, and vehicle type of the first target vehicle during this driving, and determine the power distribution of the range extender of the first target vehicle using the target local federated learning model.

[0028] According to a third aspect of an embodiment of the present application, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0029] According to a fourth aspect of the embodiments of the present application, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, the steps of the above method are implemented.

[0030] Compared with the prior art, the embodiments of the present application have the following beneficial effects: the embodiments of the present application obtain global federated learning model parameters from a server to obtain a local federated learning model, train the local federated learning model based on the historical usage data of the vehicle and the historical usage data of other vehicles of the same vehicle type as the vehicle, obtain a model gradient value, send the model gradient value to the server so that the server updates the global federated learning model based on the model gradient values ​​received from each vehicle, repeat the above steps until the global federated learning model converges, use the model parameters of the converged global federated learning model to update the target local federated learning model, and then use the target local federated learning model to determine the power distribution of the range extender of the first target vehicle. When the historical data held by the vehicle supplier is limited, a prediction model with higher prediction accuracy can be trained, thereby improving the accuracy of using the prediction model to determine the power distribution of the range extender, better balancing the performance and cost of the vehicle during operation, and improving the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments or descriptions of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0032] Figure 1 It is a scenario diagram of the application scenario of the embodiment of the present application.

[0033] Figure 2 This is a flow chart of a range extender control method based on federated learning provided in an embodiment of the present application.

[0034] Figure 3 This is a flow chart of a method provided in an embodiment of the present application for training a local federated learning model based on historical usage data of a first target vehicle and historical usage data of a second target vehicle to obtain a model gradient value.

[0035] Figure 4 It is a flowchart of a method for a server to determine updated global federated learning model parameters provided in an embodiment of the present application.

[0036] Figure 5 1 is a flow chart of a method for determining power allocation of a first target vehicle range extender using a target local federated learning model provided in an embodiment of the present application.

[0037] Figure 6 is a schematic diagram of a range extender control device based on federated learning provided in an embodiment of the present application.

[0038] Figure 7 Schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0039] In the following description, specific details such as specific system structures and techniques are provided for purposes of illustration rather than limitation to facilitate a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obscuring the description of the present application with unnecessary detail.

[0040] A range extender control method and device based on federated learning according to an embodiment of the present application will be described in detail below with reference to the accompanying drawings.

[0041] Figure 1 1 is a schematic diagram of an application scenario of an embodiment of the present application. The application scenario may include terminal devices 1, 2 and 3, a server 4 and a network 5.

[0042] The terminal devices 1, 2, and 3 can be hardware or software. When the terminal devices 1, 2, and 3 are hardware, they can be various electronic devices with display screens that support communication with the server 4, including but not limited to smartphones, tablet computers, laptop computers, and desktop computers; when the terminal devices 1, 2, and 3 are software, they can be installed in the electronic devices described above. The terminal devices 1, 2, and 3 can be implemented as multiple software or software modules, or as a single software or software module, and the embodiments of the present application are not limited to this. Furthermore, various applications can be installed on the terminal devices 1, 2, and 3, such as data processing applications, instant messaging tools, social platform software, search applications, shopping applications, etc.

[0043] Server 4 can be a server that provides various services, for example, a backend server that receives requests sent by terminal devices that establish communication connections with it. The backend server can receive and analyze the requests sent by the terminal devices, and generate processing results. Server 4 can be a single server, a server cluster consisting of multiple servers, or a cloud computing service center, and the embodiments of the present application are not limited thereto.

[0044] It should be noted that the server 4 can be either hardware or software. When the server 4 is hardware, it can be various electronic devices that provide various services to the terminal devices 1, 2, and 3. When the server 4 is software, it can be multiple software or software modules that provide various services to the terminal devices 1, 2, and 3, or it can be a single software or software module that provides various services to the terminal devices 1, 2, and 3, and this embodiment of the application does not limit this.

[0045] The network 5 can be a wired network connected by coaxial cable, twisted pair and optical fiber, or it can be a wireless network that can interconnect various communication devices without wiring, such as Bluetooth, Near Field Communication (NFC), infrared, etc., which is not limited in the embodiment of the present application.

[0046] Terminal devices 1, 2 and 3 can establish a communication connection with server 4 via network 5 to receive or send information, etc. Specifically, terminal devices 1, 2 and 3 can receive the global federated learning model from server 4, and use the data that can be obtained by themselves to perform model training locally on terminal devices 1, 2 and 3, respectively, and return the model gradient values ​​obtained by training to server 4. Server 4 updates the global federated learning model based on all received model gradient values. Repeat the steps of the terminal receiving the global federated learning model, local training returning the model gradient value, and the server updating the global federated learning model based on the model gradient value until the model converges. Server 4 sends the converged global federated learning model to terminal devices 1, 2 and 3, so that terminal devices 1, 2 and 3 can perform range extender power distribution control based on the model.

[0047] It should be noted that the specific types, quantities and combinations of the terminal devices 1, 2 and 3, the server 4 and the network 5 can be adjusted according to the actual needs of the application scenario, and the embodiments of the present application do not limit this.

[0048] As mentioned above, to balance vehicle performance and cost, related technologies can employ machine learning models to automatically control the range extender's activation timing and its operation at different power levels under different circumstances. However, these models fail to differentiate between different vehicle types, resulting in insufficient control accuracy. Developing appropriate range extender management strategies for different vehicle types to balance performance and cost remains a technical challenge.

[0049] Furthermore, due to the current proliferation of vehicle suppliers, data from different suppliers is not shared. Therefore, each supplier can only use its own data for model training. However, the data held by a single supplier is limited, and the distribution of vehicle types varies across suppliers. Training a model for range extender power allocation by each supplier individually can be ineffective.

[0050] In view of this, an embodiment of the present application provides a range extender control method based on federated learning, which obtains global federated learning model parameters from a server to obtain a local federated learning model, trains the local federated learning model based on the vehicle's historical usage data and the historical usage data of other vehicles of the same vehicle type as the vehicle, obtains a model gradient value, and sends the model gradient value to the server so that the server updates the global federated learning model based on the model gradient values ​​received from each vehicle. The above steps are repeated until the global federated learning model converges, and the model parameters of the converged global federated learning model are used to update the target local federated learning model. The target local federated learning model is then used to determine the power distribution of the range extender of the first target vehicle. When the historical data held by the vehicle supplier is limited, a prediction model with higher prediction accuracy can be trained, thereby improving the accuracy of determining the power distribution of the range extender using the prediction model, better balancing the performance and cost of the vehicle during operation, and improving the user experience.

[0051] Figure 2 This is a flow chart of a range extender control method based on federated learning provided in an embodiment of the present application. Figure 2 The range extender control method based on federated learning can be obtained by Figure 1 The terminal device executes. Figure 2 As shown, the range extender control method based on federated learning includes the following steps:

[0052] In step S201, global federated learning model parameters sent by the server are received, and a local federated learning model is initialized based on the global federated learning model parameters.

[0053] In step S202 , historical usage data of the first target vehicle is obtained.

[0054] In step S203, the vehicle identification of the first target vehicle is obtained, and the vehicle type is determined based on the vehicle identification.

[0055] In step S204 , historical usage data of the second target vehicle is obtained.

[0056] The second target vehicle and the first target vehicle have the same vehicle type.

[0057] In step S205 , a local federated learning model is trained based on the historical usage data of the first target vehicle and the historical usage data of the second target vehicle to obtain a model gradient value.

[0058] In step S206, the model gradient value is sent to the server.

[0059] In step S207, the updated global federated learning model parameters sent by the server are received.

[0060] Among them, the updated global federated learning model is determined by the server after updating the global federated learning model based on the received multiple model gradient values.

[0061] In step S208, the steps of receiving the global federated learning model parameters sent by the server, training the local federated learning model based on the historical usage data of the first target vehicle and the historical usage data of the second target vehicle to obtain the model gradient value, and receiving the updated global federated learning model parameters determined by the server based on the received model gradient value are repeated until the target global federated learning model parameters are received.

[0062] Among them, the target global federated learning model parameters are the model parameters when the global federated learning model converges

[0063] In step S209, the local federated learning model is updated based on the target global federated learning model parameters to obtain a target local federated learning model.

[0064] In step S210, the operating condition information, vehicle status information and vehicle type of the first target vehicle during this driving are obtained, and the power distribution of the range extender of the first target vehicle is determined using the target local federated learning model.

[0065] In the embodiments of the present application, the range extender control method based on federated learning can be executed by the vehicle. Furthermore, the method can be executed by an onboard unit (OBU) in the vehicle, which is a processing unit with certain computing capabilities. The OBU also has communication capabilities and can communicate with a server to receive or send data, and can also communicate with the range extender to control the power distribution of the range extender.

[0066] In the embodiment of the present application, the range extender control method based on federated learning can also be executed by a local server of a vehicle supplier. The local server of the vehicle supplier can store data of all vehicles managed by the vehicle supplier.

[0067] In an embodiment of the present application, an onboard unit (OBU) or a local server can receive global federated learning model parameters sent by a server and initialize a local federated learning model based on the global federated learning model parameters. The server can be a cloud server or a server cluster with strong computing power, and the server can communicate with multiple OBUs managed by different vehicle suppliers, or with local servers of multiple different vehicle suppliers. Furthermore, the inputs to the local federated learning model can include vehicle operating condition information, vehicle status information, and vehicle type, and the outputs can include information about when to activate the vehicle's range extender and the range extender's output power corresponding to each operating condition.

[0068] In an embodiment of the present application, the on-board unit or local server can use the data it can obtain to train the initialized local federated learning model. The data that the on-board unit or local server can obtain include the historical usage data of the vehicle, as well as the usage data of other vehicles owned by the vehicle supplier and managed by the supplier. Furthermore, since different vehicles correspond to different vehicle types, and different vehicle types may correspond to different range extender power allocation strategies due to their different daily driving routes, driving environments, and user driving needs, in order to improve control accuracy, the local federated learning model can be trained to provide different range extender power allocation strategies for different types of vehicles.

[0069] That is, the on-board unit or local server can first obtain the historical usage data of the first target vehicle, and at the same time obtain the vehicle identification of the first target vehicle, determine the vehicle type of the first target vehicle based on the vehicle identification, and then obtain the historical usage data of the second target vehicle of the same vehicle type as the first target vehicle.

[0070] In this embodiment of the present application, a local federated learning model can be trained based on the historical usage data of the first target vehicle and the historical usage data of the second target vehicle to obtain a model gradient value. The specific training method is basically the same as the machine learning model training method in the related art and will not be repeated here.

[0071] In an embodiment of the present application, the model gradient value obtained through training can be sent to a server. The server receives the model gradient values ​​sent by multiple on-board units or local servers, updates the global federated learning model based on the multiple model gradient values, and sends the updated global federated learning model parameters to the on-board unit or local server again, so that the on-board unit or local server updates the local federated learning model based on the updated global federated learning model parameters. The on-board unit or local server repeats the steps of receiving the global federated learning model parameters sent by the server, training the updated local federated learning model based on the historical usage data of the first target vehicle and the historical usage data of the second target vehicle to obtain the model gradient value, and receiving the updated global federated learning model parameters determined by the server based on the received model gradient value, until the target global federated learning model parameters are received. The target global federated learning model parameters are the model parameters when the global federated learning model converges.

[0072] In an embodiment of the present application, the onboard unit or local server can update the local federated learning model based on the received target global federated learning model parameters to obtain a target local federated learning model. Furthermore, the onboard unit or local server can also obtain the operating condition information, vehicle status information, and vehicle type of the first target vehicle during the current trip, and use the target local federated learning model to determine the power allocation for the range extender of the first target vehicle based on the operating condition information, vehicle status information, and vehicle type of the first target vehicle during the current trip.

[0073] According to the technical solution provided in the embodiments of the present application, a local federated learning model is obtained by obtaining global federated learning model parameters from a server, and the local federated learning model is trained based on the historical usage data of the vehicle and the historical usage data of other vehicles of the same vehicle type as the vehicle to obtain a model gradient value. The model gradient value is sent to the server so that the server updates the global federated learning model based on the model gradient values ​​received from each vehicle. The above steps are repeated until the global federated learning model converges, and the model parameters of the converged global federated learning model are used to update the target local federated learning model. The target local federated learning model is then used to determine the power distribution of the range extender of the first target vehicle. In this way, a prediction model with higher prediction accuracy can be trained when the historical data held by the vehicle supplier is limited, thereby improving the accuracy of using the prediction model to determine the power distribution of the range extender, better balancing the performance and cost of the vehicle during operation, and improving the user experience.

[0074] In the embodiment of the present application, the vehicle types may include private car types, logistics vehicle types, and online car-hailing types. Different vehicle types may correspond to different range extender power allocation strategies. For example, for private cars, since their driving time periods and driving routes on weekdays are usually relatively fixed, and their charging locations are usually relatively fixed, when formulating their range extender power allocation strategies, priority may be given to the impact of factors such as the congestion situation and weather conditions of each road section corresponding to their driving routes on the range extender power allocation. For logistics vehicles, since their driving routes are usually long and there are many highway sections passing through, when formulating their range extender power allocation strategies, priority may be given to the impact of factors such as their charging habits during driving, the power allocation switching method for highway sections and ordinary sections on the range extender power allocation. For online car-hailing, since their driving routes are usually irregular, when formulating their range extender power allocation strategies, priority may be given to the impact of factors such as their charging habits during driving on the range extender power allocation.

[0075] By adopting the technical solution of the embodiment of the present application, by classifying vehicles into different types and obtaining corresponding historical data based on different vehicle types to train a global federated learning model, a more accurate prediction model can be obtained, thereby improving the accuracy of range extender power distribution and enhancing user experience.

[0076] In this embodiment of the present application, historical usage data for the second target vehicle can be obtained by uploading the vehicle type of the first target vehicle to the vehicle supplier's server; and receiving historical usage data for the second target vehicle from the supplier server. The second target vehicle and the first target vehicle have the same supplier and the same vehicle type. In other words, historical usage data for the second target vehicle of the same vehicle type as the first target vehicle can be obtained from the vehicle supplier. In this case, the second target vehicle and the first target vehicle also belong to the same vehicle supplier.

[0077] In an embodiment of the present application, the historical usage data includes: historical operating condition information of the vehicle, historical start-up timing information of the range extender, historical output power of the range extender under different operating conditions, and information on the remaining battery power of the vehicle under different operating conditions. Furthermore, the operating conditions include starting conditions, slow acceleration conditions, medium and low speed cruising, rapid acceleration conditions, high speed cruising conditions, and deceleration conditions. Furthermore, the step of training a local federated learning model based on the historical usage data of the first target vehicle and the historical usage data of the second target vehicle may be to use the historical operating condition information of the first target vehicle and the second target vehicle and the remaining battery power information of the vehicle under different operating conditions as input parameters of the local federated learning model, and use the historical start-up timing information of the range extender of the first target vehicle and the second target vehicle and the historical output power of the range extender under different operating conditions as verification outputs of the local federated learning model to train the local federated learning model.

[0078] Furthermore, the operating condition information includes the operating condition duration information and the adjacent operating condition information. In other words, the range extender power allocation strategy can be determined based on the range extender's historical output power training under different vehicle operating conditions and different vehicle remaining power information.

[0079] Figure 3 This is a flow chart of a method for training a local federated learning model based on the historical usage data of the first target vehicle and the historical usage data of the second target vehicle to obtain a model gradient value provided by an embodiment of the present application. Figure 3 As shown, the method includes the following steps:

[0080] In step S301, the historical operating condition information of the first and second target vehicles, the remaining power information of the vehicles under different operating conditions, and the vehicle type are used as model inputs, and a local federated learning model is used to predict the predicted range extender start-up timing information and the predicted range extender output power under different operating conditions for the first and second target vehicles.

[0081] In step S302, the difference between the historical range extender start-up timing information and the historical range extender output power under different operating conditions of the first and second target vehicles and the predicted range extender start-up timing information and the predicted range extender output power under different operating conditions of the first and second target vehicles is determined as a model gradient value.

[0082] In this embodiment of the present application, a vehicle's historical operating condition information and information about the vehicle's remaining battery power under different operating conditions can be used as input parameters for a local federated learning model, and a type parameter of the local federated learning model is determined based on the vehicle type. Furthermore, a global federated learning output is obtained based on the input parameters and the type parameter. The global federated learning output includes a predicted activation time for the vehicle's range extender and / or a predicted output power of the range extender under specific operating conditions.

[0083] In embodiments of the present application, the predicted vehicle range extender activation timing can be compared with the actual range extender activation timing in historical data to obtain a model gradient value. Furthermore, the predicted range extender output power under specific operating conditions can be compared with the actual range extender output power under the specific operating conditions in historical data to obtain a model gradient value.

[0084] Figure 4 Schematic diagram of the process of determining updated global federated learning model parameters by the server according to the embodiment of the present application. Figure 4 As shown, the method includes the following steps:

[0085] In step S401 , a plurality of model gradient values ​​and vehicle types corresponding to different model gradient values ​​are received.

[0086] In step S402 , the global federated learning model is updated based on the model gradient value and the vehicle type to obtain updated global federated learning model parameters.

[0087] The global federated learning model parameters include network parameters and / or type parameters of the global federated learning model.

[0088] In an embodiment of the present application, a server may receive multiple model gradient values ​​and the vehicle types corresponding to the different model gradient values. The multiple vehicles may be managed by different vehicle suppliers. Each vehicle receives global federated learning model parameters from the server, trains its own local federated learning model locally, obtains its own model gradient value, and sends the model gradient value to the server. The server updates the global federated learning model based on the received model gradient value and vehicle type, obtaining updated global federated learning model parameters. The global federated learning model parameters may include network parameters and / or type parameters.

[0089] Figure 51 is a flow chart of a method for determining the power allocation of a first target vehicle range extender using a target local federated learning model provided in an embodiment of the present application. Figure 5 As shown, the method includes the following steps:

[0090] In step S501, the route information of the first target vehicle during this travel is obtained, and the operating condition information of the first target vehicle during this travel is estimated based on the route information.

[0091] In step S502, vehicle status information of the first target vehicle is acquired, where the vehicle status information at least includes remaining power information of the first target vehicle.

[0092] In step S503, the operating condition information, vehicle status information, and vehicle type of the first target vehicle during this driving are input into the target local federated learning model to predict the range extender start timing information and / or the range extender output power corresponding to each operating condition contained in the operating condition information of the first target vehicle.

[0093] In step S504 , the power distribution of the range extender of the first target vehicle is determined based on the start-up timing information of the range extender and / or the output power of the range extender corresponding to each operating condition.

[0094] In an embodiment of the present application, after determining the target local federated learning model, the on-board unit or the local server obtains the route information of the first target vehicle's current travel, and estimates the operating condition information of the first target vehicle's current travel based on the route information. At the same time, the vehicle status information of the first target vehicle can also be obtained, wherein the vehicle status information at least includes the vehicle's remaining power information. Next, the operating condition information and vehicle status information of the first target vehicle's current travel are used as input parameters of the updated target local federated learning model, and the vehicle type of the first target vehicle is input into the target local federated learning model as the type parameter of the target local federated learning model, so that the start-up timing information of the range extender and / or the estimated output power of the range extender under various operating conditions of the first target vehicle's current travel can be predicted. Finally, the on-board unit or the local server determines the power distribution of the range extender of the first target vehicle based on the start-up timing information of the range extender and / or the output power of the range extender corresponding to each operating condition.

[0095] All of the above optional technical solutions can be combined in any way to form optional embodiments of the present application, and will not be described in detail here.

[0096] The following are device embodiments of the present application, which can be used to implement the method embodiments of the present application. For details not disclosed in the device embodiments of the present application, please refer to the method embodiments of the present application.

[0097] Figure 6Schematic diagram of a range extender control device based on federated learning provided in an embodiment of the present application. Figure 6 As shown, the range extender control device based on federated learning includes:

[0098] The receiving module 601 is configured to receive the global federated learning model parameters sent by the server and initialize the local federated learning model based on the global federated learning model parameters.

[0099] The acquisition module 602 is configured to acquire historical usage data of the first target vehicle.

[0100] The acquisition module 602 is further configured to acquire a vehicle identification of the first target vehicle and determine the vehicle type based on the vehicle identification.

[0101] The acquisition module 602 is further configured to acquire historical usage data of a second target vehicle, where the second target vehicle has the same vehicle type as the first target vehicle.

[0102] The training module 603 is configured to train a local federated learning model based on the historical usage data of the first target vehicle and the historical usage data of the second target vehicle to obtain a model gradient value.

[0103] The communication module 604 is configured to send the model gradient value to the server;

[0104] The communication module 604 is further configured to receive updated global federated learning model parameters sent by the server. The updated global federated learning model is determined by the server after updating the global federated learning model based on the received multiple model gradient values.

[0105] The training module 603 is also configured to repeatedly execute the steps of receiving the global federated learning model parameters sent by the server, training the local federated learning model based on the historical usage data of the first target vehicle and the historical usage data of the second target vehicle to obtain the model gradient value, and receiving the updated global federated learning model parameters determined by the server according to the received model gradient value, until the target global federated learning model parameters are received, and the target global federated learning model parameters are the model parameters when the global federated learning model converges.

[0106] The training module 603 is further configured to update the local federated learning model based on the target global federated learning model parameters to obtain a target local federated learning model.

[0107] The control module 605 is configured to obtain the operating condition information, vehicle status information and vehicle type of the first target vehicle during this driving, and determine the power distribution of the range extender of the first target vehicle using the target local federated learning model.

[0108] According to the technical solution provided in the embodiments of the present application, a local federated learning model is obtained by obtaining global federated learning model parameters from a server, and the local federated learning model is trained based on the historical usage data of the vehicle and the historical usage data of other vehicles of the same vehicle type as the vehicle to obtain a model gradient value. The model gradient value is sent to the server so that the server updates the global federated learning model based on the model gradient values ​​received from each vehicle. The above steps are repeated until the global federated learning model converges, and the model parameters of the converged global federated learning model are used to update the target local federated learning model. The target local federated learning model is then used to determine the power distribution of the range extender of the first target vehicle. In this way, a prediction model with higher prediction accuracy can be trained when the historical data held by the vehicle supplier is limited, thereby improving the accuracy of using the prediction model to determine the power distribution of the range extender, better balancing the performance and cost of the vehicle during operation, and improving the user experience.

[0109] In an embodiment of the present application, vehicle types include: private car type, logistics vehicle type, and online car-hailing type; obtaining historical usage data of the second target vehicle includes: uploading the vehicle type of the first target vehicle to the vehicle supplier server; receiving the historical usage data of the second target vehicle issued by the supplier server, and the second target vehicle has the same supplier and the same vehicle type as the first target vehicle.

[0110] In the embodiment of the present application, the historical usage data includes: historical operating condition information of the vehicle, historical start-up timing information of the range extender, historical output power of the range extender under different operating conditions, and vehicle remaining power information under different operating conditions.

[0111] In the embodiment of the present application, the operating conditions include starting conditions, slow acceleration conditions, medium and low speed cruising conditions, rapid acceleration conditions, high speed cruising conditions and deceleration conditions; the operating condition information includes the operating condition maintenance time information and adjacent operating condition information of the operating condition.

[0112] In an embodiment of the present application, a local federated learning model is trained based on historical usage data of a first target vehicle and historical usage data of a second target vehicle to obtain a model gradient value, including: using historical operating condition information of the first and second target vehicles, vehicle remaining power information under different operating conditions, and vehicle type as model inputs, respectively, and using the local federated learning model to predict predicted range extender start-up timing information and predicted range extender output power under different operating conditions for the first and second target vehicles; and determining the difference between the historical range extender start-up timing information and the historical output power of the range extender under different operating conditions of the first and second target vehicles and the predicted predicted range extender start-up timing information and the predicted range extender output power under different operating conditions for the first and second target vehicles as the model gradient value.

[0113] In an embodiment of the present application, the updated global federated learning model parameters are determined by the server in the following manner: receiving multiple model gradient values ​​and vehicle types corresponding to different model gradient values; updating the global federated learning model based on the model gradient values ​​and vehicle types to obtain updated global federated learning model parameters; wherein the global federated learning model parameters include network parameters and / or type parameters of the global federated learning model.

[0114] In an embodiment of the present application, a target local federated learning model is used to determine the power allocation of a range extender of a first target vehicle, including: obtaining route information of the first target vehicle's current travel, and estimating operating condition information of the first target vehicle's current travel based on the route information; obtaining vehicle status information of the first target vehicle, the vehicle status information at least including remaining power information of the first target vehicle; inputting the operating condition information of the first target vehicle's current travel, the vehicle status information, and the vehicle type of the first target vehicle into the target local federated learning model, and predicting the range extender start-up timing information of the first target vehicle and / or the range extender output power corresponding to each operating condition included in the operating condition information; and determining the power allocation of the range extender of the first target vehicle based on the range extender start-up timing information and / or the range extender output power corresponding to each operating condition.

[0115] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0116] Figure 7 Schematic diagram of an electronic device provided in an embodiment of the present application. Figure 7 As shown, the electronic device 7 of this embodiment includes: a processor 701, a memory 702, and a computer program 703 stored in the memory 702 and executable by the processor 701. When the processor 701 executes the computer program 703, the steps of the above-mentioned method embodiments are implemented. Alternatively, when the processor 701 executes the computer program 703, the functions of the modules / units in the above-mentioned device embodiments are implemented.

[0117] The electronic device 7 may be a desktop computer, a notebook, a PDA, a cloud server or other electronic device. The electronic device 7 may include but is not limited to a processor 701 and a memory 702. Those skilled in the art will understand that Figure 7 This is merely an example of the electronic device 7 and does not limit the electronic device 7 . The electronic device 7 may include more or fewer components than shown in the figure, or different components.

[0118] The processor 701 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0119] The memory 702 may be an internal storage unit of the electronic device, such as a hard disk or memory of the electronic device 7. The memory 702 may also be an external storage device of the electronic device 7, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash memory card, etc. equipped on the electronic device 7. The memory 702 may also include both an internal storage unit of the electronic device 7 and an external storage device. The memory 702 is used to store computer programs and other programs and data required by the electronic device.

[0120] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

[0121] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. The computer program may include computer program code, which may be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0122] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A range extender control method based on federated learning, characterized in that: include: receiving global federated learning model parameters sent by the server, and initializing a local federated learning model based on the global federated learning model parameters; Obtaining historical usage data of the first target vehicle; Obtaining a vehicle identification of a first target vehicle, and determining a vehicle type based on the vehicle identification; Acquiring historical usage data of a second target vehicle, where the second target vehicle is of the same vehicle type as the first target vehicle; Training the local federated learning model based on the historical usage data of the first target vehicle and the historical usage data of the second target vehicle to obtain a model gradient value, specifically comprising taking the historical operating condition information of the first and second target vehicles, the vehicle remaining power information under different operating conditions, and the vehicle type as model inputs, respectively, and using the local federated learning model to predict the first and second target vehicles' predicted range extender start-up timing information and the predicted range extender output power under different operating conditions; determining the difference between the historical range extender start-up timing information and the historical range extender output power under different operating conditions of the first and second target vehicles and the predicted predicted range extender start-up timing information and the predicted range extender output power under different operating conditions as the model gradient value; Sending the model gradient value to a server; receiving updated global federated learning model parameters sent by the server, where the updated global federated learning model is determined by the server after updating the global federated learning model according to the received multiple model gradient values; Repeating the steps of receiving global federated learning model parameters sent by the server, training the local federated learning model based on the historical usage data of the first target vehicle and the historical usage data of the second target vehicle to obtain a model gradient value, and receiving updated global federated learning model parameters determined by the server according to the received model gradient value, until target global federated learning model parameters are received, where the target global federated learning model parameters are the model parameters of the global federated learning model when convergence occurs; Updating the local federated learning model based on the target global federated learning model parameters to obtain a target local federated learning model; The operating condition information, vehicle status information, and vehicle type of the first target vehicle during this driving are obtained, and the power distribution of the range extender of the first target vehicle is determined using the target local federated learning model.

2. The method according to claim 1, characterized in that The vehicle types include: private car type, logistics vehicle type, and online car-hailing type; The obtaining of historical usage data of the second target vehicle includes: Uploading the vehicle type of the first target vehicle to the supplier server of the vehicle; Receive historical usage data of a second target vehicle sent by the supplier server, where the second target vehicle has the same supplier and the same vehicle type as the first target vehicle.

3. The method according to claim 1, characterized in that The historical usage data includes: historical operating condition information of the vehicle, historical start-up timing information of the range extender, historical output power of the range extender under different operating conditions, and vehicle remaining power information under different operating conditions; The step of training the local federated learning model based on the historical usage data of the first target vehicle and the historical usage data of the second target vehicle is to use the historical operating condition information of the first target vehicle and the second target vehicle and the vehicle remaining power information under different operating conditions as input parameters of the local federated learning model, and use the historical start-up timing information of the range extender of the first target vehicle and the second target vehicle and the historical output power of the range extender under different operating conditions as verification outputs of the local federated learning model to train the local federated learning model.

4. The method according to claim 3, characterized in that The operating conditions include starting conditions, slow acceleration conditions, medium and low speed cruising conditions, rapid acceleration conditions, high speed cruising conditions and deceleration conditions; The working condition information includes working condition maintenance time information and adjacent working condition information of the working condition.

5. The method according to claim 1, characterized in that The updated global federated learning model parameters are determined by the server in the following manner: Receive multiple model gradient values ​​and vehicle types corresponding to different model gradient values; updating the global federated learning model based on the model gradient value and the vehicle type to obtain the updated global federated learning model parameters; The global federated learning model parameters include network parameters and / or type parameters of the global federated learning model.

6. The method according to claim 1, characterized in that The determining the power allocation of the first target vehicle range extender using the target local federated learning model includes: Acquiring route information of the first target vehicle during this travel, and estimating operating condition information of the first target vehicle during this travel based on the route information; Acquire vehicle status information of the first target vehicle, the vehicle status information including at least remaining power information of the first target vehicle; Inputting the operating condition information, vehicle status information, and vehicle type of the first target vehicle during this driving of the first target vehicle into the target local federated learning model, and predicting the range extender start timing information of the first target vehicle and / or the range extender output power corresponding to each operating condition included in the operating condition information; The power distribution of the range extender of the first target vehicle is determined based on the start-up timing information of the range extender and / or the output power of the range extender corresponding to each operating condition.

7. A range extender control device based on federated learning, characterized in that: include: a receiving module configured to receive global federated learning model parameters sent by the server, and initialize the local federated learning model based on the global federated learning model parameters; an acquisition module configured to acquire historical usage data of a first target vehicle; The acquisition module is further configured to acquire a vehicle identification of the first target vehicle and determine the vehicle type based on the vehicle identification; The acquisition module is further configured to acquire historical usage data of a second target vehicle, the second target vehicle being of the same vehicle type as the first target vehicle; a training module configured to train the local federated learning model based on historical usage data of the first target vehicle and the second target vehicle to obtain a model gradient value, and specifically configured to use historical operating condition information, vehicle remaining power information under different operating conditions, and vehicle type of the first and second target vehicles as model inputs, respectively, and use the local federated learning model to predict predicted range extender start-up timing information and predicted range extender output power under different operating conditions for the first and second target vehicles, respectively; and determine the difference between the historical range extender start-up timing information and the historical range extender output power under different operating conditions of the first and second target vehicles and the predicted predicted range extender start-up timing information and the predicted range extender output power under different operating conditions for the first and second target vehicles as the model gradient value; a communication module, configured to send the model gradient value to a server; The communication module is further configured to receive updated global federated learning model parameters sent by the server, where the updated global federated learning model is determined by the server after updating the global federated learning model according to the received multiple model gradient values; The training module is further configured to repeatedly perform the steps of receiving global federated learning model parameters sent by the server, training the local federated learning model based on the historical usage data of the first target vehicle and the historical usage data of the second target vehicle to obtain a model gradient value, and receiving updated global federated learning model parameters determined by the server according to the received model gradient value, until target global federated learning model parameters are received, wherein the target global federated learning model parameters are model parameters when the global federated learning model converges; The training module is further configured to update the local federated learning model based on the target global federated learning model parameters to obtain a target local federated learning model; The control module is configured to obtain operating condition information, vehicle status information, and vehicle type of the first target vehicle during this driving, and determine the power distribution of the range extender of the first target vehicle using the target local federated learning model.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images