Energy consumption prediction model training method and device, energy consumption prediction method and device and electronic equipment
By supervising and fine-tuning the electric vehicle energy consumption prediction model, the problem of inaccurate energy consumption estimation in the existing technology is solved, and more accurate range prediction and energy consumption estimation of driving sections is achieved.
Patent Information
- Application Number
- CN202510115401.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-24
- Publication Date
- 2025-05-30
AI Technical Summary
The energy consumption estimates in the prior art are not accurate enough, resulting in large errors in the range prediction of electric vehicles.
The energy consumption prediction model is established by establishing a pre-trained model for energy consumption prediction and supervising and fine-tuning the pre-trained model based on the acquired data set to establish an energy consumption prediction model. This model is used to determine the energy consumption of segmented sections and determine the energy consumption of the travel sections based on the energy consumption of segmented sections.
It improves the accuracy of energy consumption estimation, makes the range prediction of electric vehicles more accurate, helps car owners reasonably arrange charging time and location to avoid battery life anxiety.
Smart Images
Figure CN120069188A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of energy consumption estimation, and more particularly, to a method for training an energy consumption prediction model, an energy consumption prediction method, a device, and an electronic device. Background Art
[0002] The driving range of an electric vehicle is directly related to its energy consumption. By estimating the energy consumption in real time, the vehicle owner can better plan the driving route, charging time, and charging location to ensure a smooth journey. Especially in the case of long-distance and high-speed driving, energy consumption estimation is even more necessary.
[0003] In the related art, usually, the geographical location information and the remaining driving range of the target vehicle are obtained, and then the energy consumption of a section of road is estimated by the average speed through the entire section of the road. However, this energy consumption is not accurate enough and there is a large error from the actual energy consumption. Summary of the Invention
[0004] The problem solved by the present invention is how to improve the accuracy of energy consumption estimation.
[0005] To solve the above problems, the present invention provides a method for training an energy consumption prediction model, an energy consumption prediction method, a device, and an electronic device.
[0006] In a first aspect, the present invention provides a method for training an energy consumption prediction model, including:
[0007] Establishing a pre-training model for energy consumption prediction;
[0008] Supervising and fine-tuning the pre-training model according to the obtained data set to establish an energy consumption prediction model, where the energy consumption prediction model is used to determine the segmented energy consumption of segmented road sections, so as to determine the energy consumption of the driving road section according to the segmented energy consumption; where the driving road section includes each of the segmented road sections.
[0009] Optionally, the data set includes a training set and a validation set, and the supervising and fine-tuning the pre-training model according to the obtained data set includes:
[0010] Setting a fine-tuning target and configuring hyperparameters;
[0011] Training the pre-training model according to the training set;
[0012] Evaluating the training result according to the validation set, and adjusting the hyperparameters according to the evaluation result to establish the energy consumption prediction model.
[0013] Optionally, before setting the fine-tuning target and configuring the training hyperparameters, it further includes:
[0014] A discrete distribution diagram is established based on the data set, where the discrete distribution diagram includes a plurality of data points, and each data point represents the energy consumption corresponding to a vehicle parameter group, and the vehicle parameter group includes vehicle speed and vehicle driving mileage;
[0015] Abnormal data points are deleted from the discrete distribution diagram according to a preset rule to obtain a data set for training the pre-trained model.
[0016] Optionally, the deleting abnormal data points from the discrete distribution diagram according to a preset rule includes:
[0017] Delete the data points corresponding to the vehicle speed less than the preset speed from the discrete distribution diagram;
[0018] Delete the data points corresponding to the intermittent stop distance less than the preset distance from the discrete distribution diagram, where the intermittent stop distance represents the distance between two consecutive positions where the vehicle speed is zero in the driving section;
[0019] Delete the data points corresponding to the estimated full-charge driving range exceeding the preset range of the rated driving range from the discrete distribution diagram, where the full-charge driving range is estimated according to the vehicle driving mileage and energy consumption corresponding to the data points.
[0020] In a second aspect, the present invention provides an energy consumption prediction method, including:
[0021] The driving section is divided into multiple segmented sections according to the historical vehicle speed of the vehicle passing through the driving section, where the difference between the maximum vehicle speed and the minimum vehicle speed in each segmented section is less than a preset threshold;
[0022] Determine the segmented energy consumption of each segmented section according to the energy consumption prediction model established by the energy consumption prediction model training method;
[0023] Determine the energy consumption of the driving section according to the segmented energy consumption of all the segmented sections.
[0024] Optionally, the dividing the driving section into multiple segmented sections according to the historical vehicle speed of the vehicle passing through the driving section includes:
[0025] Determine the driving section according to the starting point, passing points and ending point;
[0026] Determine the road condition of the driving section according to the road signs and road forms;
[0027] After estimating the change of the vehicle speed according to the road condition, divide the driving section according to the vehicle speed.
[0028] In a third aspect, the present invention provides an energy consumption prediction model training device, including:
[0029] A first module, configured to establish a pre-training model for energy consumption prediction;
[0030] A second module, configured to perform supervised fine-tuning on the pre-training model according to the obtained data set to establish an energy consumption prediction model, wherein the energy consumption prediction model is used to determine the segmented energy consumption of segmented road sections, so as to determine the energy consumption of the driving road section according to the segmented energy consumption; wherein, the driving road section includes each of the segmented road sections.
[0031] In a fourth aspect, the present invention provides an energy consumption prediction device, including:
[0032] A third module, configured to divide the driving road section into multiple segmented road sections according to the historical vehicle speeds of vehicles passing through the driving road section, wherein the difference between the maximum vehicle speed and the minimum vehicle speed in each of the segmented road sections is less than a preset threshold;
[0033] A fourth module, configured to determine the segmented energy consumption of each of the segmented road sections according to the energy consumption prediction model established by the energy consumption prediction model training method;
[0034] A fifth module, configured to determine the energy consumption of the driving road section according to the segmented energy consumption of all the segmented road sections.
[0035] In a fifth aspect, the present invention provides an electronic device, including a memory and a processor;
[0036] The memory is configured to store a computer program;
[0037] The processor is configured to, when executing the computer program, implement the energy consumption prediction model training method as described in the first aspect or the energy consumption prediction method as described in the second aspect.
[0038] In a sixth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the energy consumption prediction model training method as described in the first aspect or the energy consumption prediction method as described in the second aspect is implemented.
[0039] The beneficial effect of the energy consumption prediction model training method of the present invention is that: by establishing a pre-training model for energy consumption prediction and performing supervised fine-tuning on the pre-training model according to the obtained data set, the parameters of the pre-training model are adjusted, thereby establishing an energy consumption prediction model. The energy consumption prediction model can be used to determine the segmented energy consumption of segmented road sections in the driving road section, and adding the segmented energy consumption of each segmented road section can obtain the energy consumption of the driving road section. Compared with the solution of estimating energy consumption according to the average speed of the entire driving road section in the related art, more accurate energy consumption can be obtained. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] Figure 1 Schematic flowchart of the method for training an energy consumption prediction model according to an embodiment of the present invention;
[0041] Figure 2 Schematic flowchart of the supervised fine-tuning according to an embodiment of the present invention;
[0042] Figure 3 Schematic flowchart of the data preprocessing according to an embodiment of the present invention;
[0043] Figure 4 Schematic flowchart of the process of deleting abnormal data points according to an embodiment of the present invention;
[0044] Figure 5 Schematic flowchart of the energy consumption prediction method according to an embodiment of the present invention;
[0045] Figure 6 Schematic flowchart of the process of dividing a road section into segments according to an embodiment of the present invention;
[0046] Figure 7 Schematic diagram of the principle of dividing a road section into segments according to an embodiment of the present invention;
[0047] Figure 8 System architecture diagram of the energy consumption prediction model training device according to an embodiment of the present invention;
[0048] Figure 9 System architecture diagram of the energy consumption prediction device according to an embodiment of the present invention;
[0049] Figure 10 System architecture diagram of the electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0050] To make the above objects, features, and advantages of the present invention more apparent and understandable, the following detailed description of the specific embodiments of the present invention is provided in conjunction with the accompanying drawings. Although some embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. On the contrary, these embodiments are provided to more thoroughly and completely understand the present invention. It should be understood that the drawings and embodiments of the present invention are only for exemplary purposes and are not used to limit the protection scope of the present invention.
[0051] It should be understood that the various steps recited in the method embodiments of the present invention can be executed in a different order and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this regard.
[0052] As used herein, the term "including" and its variations are open-ended, i.e., "including but not limited to"; the term "based on" means "at least partially based on"; the term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments"; the term "optionally" means "optional embodiments". The relevant definitions of other terms will be given in the following description. It should be noted that the concepts such as "first", "second", etc. mentioned in the present invention are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.
[0053] It should be noted that the modification of "one" and "multiple" mentioned in the present invention is illustrative rather than restrictive. Those skilled in the art should understand that unless otherwise clearly specified in the context, it should be understood as "one or more".
[0054] The names of the messages or information exchanged between multiple devices in the embodiments of the present invention are only for illustrative purposes and are not used to limit the scope of these messages or information.
[0055] As Figure 1 shown, a method for training an energy consumption prediction model provided by an embodiment of the present invention includes:
[0056] S100: Establish a pre-training model for energy consumption prediction.
[0057] Specifically, first establish a pre-training model for energy consumption prediction. For example, call a pre-trained BERT model (Bidirectional Encoder Representations from Transformers) from the model library. After the model is trained, it can output the predicted energy consumption by inputting the running parameters, environmental parameters, and other text description information of the vehicle.
[0058] S200: Perform supervised fine-tuning on the pre-training model according to the obtained data set to establish an energy consumption prediction model, where the energy consumption prediction model is used to determine the segmented energy consumption of segmented road sections, so as to determine the energy consumption of the driving road section according to the segmented energy consumption; where the driving road section includes each of the segmented road sections.
[0059] Specifically, the pre-trained model is a preliminary and unoptimized model. For example, when establishing the pre-trained model, it may be based on default hyperparameters or initial parameters, but these are usually not optimal, and the performance of the model may not be good enough. To improve its accuracy and performance, the model needs to be adjusted. After obtaining the dataset, the pre-trained model is supervised and fine-tuned according to the dataset, keeping most of the weights of the pre-trained model unchanged and only adjusting the part related to the target task to achieve the adjustment of the parameters of the pre-trained model. After the training is completed, the obtained energy consumption prediction model is saved. The energy consumption prediction model can be used to determine the segmented energy consumption of the segmented sections in the driving section. Adding up the segmented energy consumption of each segmented section can obtain the energy consumption of the driving section. By accurately estimating the energy consumption, it can help the vehicle owner understand the endurance ability of the vehicle under different working conditions, reasonably arrange the charging time and location, avoid range anxiety, and support the optimization of the driving strategy. For example, the energy utilization rate can be maximized through a reasonable driving method (constant speed driving, avoiding frequent hard braking).
[0060] In this embodiment, by establishing a pre-trained model for energy consumption prediction and supervising and fine-tuning the pre-trained model according to the obtained dataset, the parameters of the pre-trained model are adjusted, thereby establishing an energy consumption prediction model. The energy consumption prediction model can be used to determine the segmented energy consumption of the segmented sections in the driving section. Adding up the segmented energy consumption of each segmented section can obtain the energy consumption of the driving section. Compared with the related technology that estimates the energy consumption according to the average speed of the entire driving section, a more accurate energy consumption can be obtained.
[0061] Optionally, the dataset includes a training set and a validation set. The supervision and fine-tuning of the pre-trained model according to the obtained dataset includes:
[0062] S210: Set the fine-tuning target and configure the hyperparameters.
[0063] Specifically, in combination with Figure 2As shown, energy consumption estimation is usually regarded as a regression problem. Therefore, the corresponding fine-tuning objective can be set to "regression task", and a linear regression layer can be added to the output layer of the BERT model to output a continuous value. For hyperparameters, it is usually necessary to configure the learning rate, batch size, number of training epochs, optimizer, weight decay, and gradient accumulation, etc.; usually a smaller learning rate is used during fine-tuning to avoid unstable training caused by too large a learning rate; the batch size determines the number of samples used for each update. For a large model like BERT, the batch size is usually set to 8, 16, or 32, which can be adjusted according to computing resources; BERT usually requires fewer training epochs (such as 3 to 4 epochs) to avoid overfitting; the fine-tuning of the BERT model usually uses the AdamW optimizer, which is more adaptable to the pre-trained model; the BERT model usually uses weight decay to prevent overfitting, which can be set to 0.01 or 0.1, and the specific value can be adjusted according to the performance of the validation set; if the video memory is insufficient, or a larger batch size is desired, gradient accumulation can be used.
[0064] S220: Train the pre-trained model according to the training set.
[0065] Specifically, in combination with Figure 2 As shown, when training the training model according to the training set, calculate the loss and perform backpropagation to update the model parameters.
[0066] S230: Evaluate the training results according to the validation set, and adjust the hyperparameters according to the evaluation results to establish the energy consumption prediction model.
[0067] Specifically, in combination with Figure 2 As shown, after each round of training, use the validation set to evaluate the model, calculate the performance of the model in the energy consumption estimation task. For example, use the mean squared error or mean absolute error as the loss function to evaluate the training results. If the model performance is not good, the hyperparameters can be adjusted, such as the learning rate, batch size, or add more training data, etc. After training, establish the energy consumption prediction model.
[0068] In this optional embodiment, by setting the fine-tuning objective and configuring the hyperparameters, then training the training model according to the training set and using the validation set to evaluate the model, the energy consumption prediction model can effectively adapt to the energy consumption prediction task of electric vehicles and achieve a good estimation effect.
[0069] Optionally, before setting the fine-tuning objective and configuring the training hyperparameters, the energy consumption prediction model training method further includes:
[0070] S201: Establish a discrete distribution graph based on the dataset, where the discrete distribution graph includes multiple data points, and each data point represents the energy consumption corresponding to a vehicle parameter group, and the vehicle parameter group includes vehicle speed and vehicle driving mileage.
[0071] Specifically, as shown in Figure 3 , the dataset usually includes driving data, battery data, vehicle status, location data, environmental data, driving behavior data, etc. Common data includes vehicle driving mileage, vehicle speed, energy consumption, vehicle weight, and vehicle temperature difference. Among them, vehicle driving mileage, vehicle speed (collectively referred to as vehicle parameter group) and energy consumption can generate a discrete distribution graph, and the discrete distribution graph can represent the relationship between vehicle driving mileage and energy consumption at different vehicle speeds, or can represent the relationship between vehicle speed and energy consumption at different vehicle driving mileages, and be converted into a fuel consumption per 100 kilometers graph.
[0072] S202: Delete abnormal data points from the discrete distribution graph according to a preset rule to obtain a dataset for training the pre-trained model.
[0073] Specifically, as shown in Figure 3 , since the discrete distribution graph contains some abnormal data points, these abnormal data points will affect model training, so they need to be deleted. By checking and removing abnormal data points, the validity of the data is ensured, and thus a dataset for training the pre-trained model is obtained.
[0074] In this optional embodiment, by deleting abnormal data points from the discrete distribution graph, the quality and consistency of the training dataset are effectively improved, thereby improving the accuracy of the model.
[0075] Optionally, the deleting abnormal data points from the discrete distribution graph according to a preset rule includes:
[0076] S203: Delete the data points corresponding to vehicle speeds less than the preset speed from the discrete distribution graph.
[0077] Specifically, as shown in Figure 4 , in the discrete distribution graph, there are large noise points at low speeds, so data with vehicle speeds less than the preset speed (such as 10 km / h) can be deleted, that is, the corresponding data points are deleted from the discrete distribution graph.
[0078] S204: Delete the data points corresponding to the intermittent stop distance less than the preset distance from the discrete distribution graph, where the intermittent stop distance represents the distance between two consecutive positions where the vehicle speed is zero in the driving section.
[0079] Specifically, as shown in Figure 4As shown, similarly, data with an intermittent stop distance (the distance between two consecutive positions where the vehicle speed is zero on a driving section) less than a preset distance (e.g., 1 km) is deleted, that is, the corresponding data points are deleted from the discrete distribution diagram.
[0080] S205: Delete the data points corresponding to the estimated full - charge driving range exceeding the preset range of the rated driving range from the discrete distribution diagram, where the full - charge driving range is estimated based on the vehicle driving mileage and energy consumption corresponding to the data points.
[0081] Specifically, as shown in Figure 4 Based on the vehicle driving mileage and energy consumption corresponding to the data points, the full - charge driving range can be estimated. If the full - charge driving range exceeds the preset range of the rated driving range (e.g., exceeds 1.2 times the rated driving range or is less than 0.8 times the rated driving range), it indicates that the corresponding data points are abnormal and need to be deleted.
[0082] In this optional embodiment, by deleting abnormal data points corresponding to low speed, intermittent stops, and non - compliant full - charge driving ranges from the discrete distribution diagram, the quality and consistency of the training dataset are effectively improved, thereby improving the accuracy of the model.
[0083] As shown in Figure 5 An energy consumption prediction method provided by an embodiment of the present invention includes:
[0084] S300: Divide the driving section into multiple segmented sections according to the historical vehicle speeds of vehicles passing through the driving section, where the difference between the maximum vehicle speed and the minimum vehicle speed in each segmented section is less than a preset threshold.
[0085] Specifically, before energy consumption prediction, the driving section is divided into multiple segmented sections according to the historical vehicle speeds of vehicles passing through the driving section. The difference between the maximum vehicle speed and the minimum vehicle speed in each segmented section is less than a preset threshold (e.g., 20 km / h). For example, 0 to 20 km / h is divided into one section, 20 km / h to 30 km / h is divided into one section, and so on.
[0086] S400: Determine the segmented energy consumption of each segmented section according to the energy consumption prediction model established by the energy consumption prediction model training method.
[0087] Specifically, for each segmented section, the above - mentioned energy consumption prediction model is used to determine the segmented energy consumption of each segmented section. For example, data such as the historical vehicle speed and vehicle driving mileage corresponding to the segmented section are input into the energy consumption prediction model, and the energy consumption prediction model outputs the segmented energy consumption of each segmented section.
[0088] S500: Determine the energy consumption of the driving section according to the segmented energy consumption of all the segmented sections.
[0089] Specifically, by adding the sectional energy consumption of each sectional section of the driving section, the energy consumption of the driving section can be obtained.
[0090] Optionally, the dividing the driving section into multiple sectional sections according to the historical vehicle speed passing through the driving section includes:
[0091] S310: Determine the driving section according to the starting point, waypoints and ending point.
[0092] Specifically, as combined with Figure 6 shown, when determining the driving section, it can be determined according to the starting point longitude and latitude, waypoint longitude and latitude (or key point longitude and latitude) and ending point longitude and latitude.
[0093] S320: Determine the road condition of the driving section according to the road signs and road forms.
[0094] Specifically, as combined with Figure 6 and Figure 7 shown, A001 represents the sectional section of the first segment, A002 represents the road sign (such as traffic lights), A003 represents the dividing line. For A001, the speeds are similar. When encountering traffic lights, the speed will drop suddenly. When the speed drops to a certain value, it will be segmented. In addition, when the vehicle passes through a curve, the speed will also decrease, so it will also be segmented.
[0095] S330: After predicting the change of the vehicle speed according to the road condition, divide the driving section according to the vehicle speed.
[0096] Specifically, according to the road condition, the change of the vehicle speed can be predicted, so that the driving section can be accurately divided according to the vehicle speed, and the division result can be verified according to the road condition. For example, if the vehicle speed drops significantly at a certain curve, but no separate sectional section is divided, it means that there is a problem with the division result.
[0097] In this optional embodiment, after determining the road condition of the driving section according to the road signs and road forms, predicting the change of the vehicle speed according to the road condition, and then dividing the driving section according to the vehicle speed, the sectional energy consumption of each sectional section can be calculated respectively, improving the accuracy of the energy consumption estimation of the driving section.
[0098] As Figure 8 shown, an energy consumption prediction model training device 800 provided by an embodiment of the present invention includes:
[0099] The first module 810 is used to establish a pre-training model for energy consumption prediction;
[0100] A second module 820, configured to perform supervised fine-tuning on the pre-trained model according to the obtained data set to establish an energy consumption prediction model, where the energy consumption prediction model is used to determine the segment energy consumption of segmented road sections, so as to determine the energy consumption of a driving road section according to the segment energy consumption; where the driving road section includes each of the segmented road sections.
[0101] As Figure 9 shown, an energy consumption prediction device 900 provided by an embodiment of the present invention includes:
[0102] A third module 930, configured to divide the driving road section into multiple segmented road sections according to the historical vehicle speed of the driving road section, where the difference between the maximum vehicle speed and the minimum vehicle speed in each segmented road section is less than a preset threshold;
[0103] A fourth module 940, configured to determine the segment energy consumption of each of the segmented road sections according to the energy consumption prediction model established by the energy consumption prediction model training method;
[0104] A fifth module 950, configured to determine the energy consumption of the driving road section according to the segment energy consumption of all the segmented road sections.
[0105] As Figure 10 shown, an electronic device 1000 provided by an embodiment of the present invention includes a memory 1020 and a processor 1010; the memory 1020 is used to store a computer program; the processor 1010 is used to implement the above-mentioned energy consumption prediction model training method or energy consumption prediction method when executing the computer program.
[0106] Or, an electronic device 1000 includes a memory 1020 and a processor 1010 coupled to the memory 1020; the memory 1020 is configured to store a computer program; the processor 1010 is configured to perform the following operations when executing the computer program:
[0107] Establish a pre-trained model for energy consumption prediction;
[0108] Perform supervised fine-tuning on the pre-trained model according to the obtained data set to establish an energy consumption prediction model, where the energy consumption prediction model is used to determine the segment energy consumption of segmented road sections, so as to determine the energy consumption of a driving road section according to the segment energy consumption; where the driving road section includes each of the segmented road sections.
[0109] Or divide the driving road section into multiple segmented road sections according to the historical vehicle speed of the driving road section, where the difference between the maximum vehicle speed and the minimum vehicle speed in each segmented road section is less than a preset threshold;
[0110] Determine the segment energy consumption of each of the segmented road sections according to the energy consumption prediction model established by the energy consumption prediction model training method;
[0111] Determine the energy consumption of the driving road section according to the segment energy consumption of all the segmented road sections.
[0112] A computer-readable storage medium provided by an embodiment of the present invention, on which a computer program is stored. When the computer program is executed by a processor, the energy consumption prediction model training method or the energy consumption prediction method described above is implemented.
[0113] Or, a non-volatile computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the processor performs the following operations:
[0114] Establish a pre-training model for energy consumption prediction;
[0115] Supervise and fine-tune the pre-training model according to the obtained data set to establish an energy consumption prediction model, where the energy consumption prediction model is used to determine the segment energy consumption of the segmented road sections to determine the energy consumption of the driving road section according to the segment energy consumption; where the driving road section includes each of the segmented road sections.
[0116] Or divide the driving road section into multiple segmented road sections according to the historical vehicle speeds of the vehicles passing through the driving road section, where the difference between the maximum vehicle speed and the minimum vehicle speed in each of the segmented road sections is less than a preset threshold;
[0117] Determine the segment energy consumption of each of the segmented road sections according to the energy consumption prediction model established by the energy consumption prediction model training method;
[0118] Determine the energy consumption of the driving road section according to the segment energy consumption of all the segmented road sections.
[0119] Now, an electronic device 1000 that can be used as a server or a client of the present invention will be described. It is an example of a hardware device that can be applied to various aspects of the present invention. The electronic device 1000 is intended to represent various forms of digital electronic computer devices, such as, a laptop computer, a desktop computer, a workbench, a personal digital assistant, a server, a blade server, a mainframe computer, and other suitable computers. The electronic device 1000 can also represent various forms of mobile devices, such as, a personal digital processor, a cellular phone, a smart phone, a wearable device, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described herein and / or claimed.
[0120] The electronic device 1000 includes a computing unit that can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) or a computer program loaded from a storage unit into a random access memory (RAM). In the RAM, various programs and data required for device operation can also be stored. The computing unit, ROM, and RAM are connected to each other via a bus. An input / output (I / O) interface is also connected to the bus.
[0121] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), etc. In this application, the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiments of the present invention. In addition, in each embodiment of the present invention, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.
[0122] Although the present invention is disclosed as above, the protection scope of the present invention is not limited thereto. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention, and these changes and modifications will all fall within the protection scope of the present invention.
Claims
1. A method for training an energy consumption prediction model, characterized in that: include: Establish a pre-trained model for energy consumption prediction; The pre-trained model is supervised and fine-tuned according to the acquired data set to establish an energy consumption prediction model, wherein the energy consumption prediction model is used to determine the segmented energy consumption of the segmented road section, so as to determine the energy consumption of the driving section according to the segmented energy consumption; wherein the driving section includes each segmented road section.
2. The energy consumption prediction model training method according to claim 1, characterized in that: The data set includes a training set and a validation set, and the supervised fine-tuning of the pre-trained model according to the acquired data set includes: Set fine-tuning targets and configure hyperparameters; Training the pre-trained model according to the training set; The training results are evaluated according to the validation set, and the hyperparameters are adjusted according to the evaluation results to establish the energy consumption prediction model.
3. The energy consumption prediction model training method according to claim 2 is characterized in that: Before setting the fine-tuning target and configuring the training hyperparameters, also include: Establishing a discrete distribution graph according to the data set, wherein the discrete distribution graph includes a plurality of data points, each of the data points represents energy consumption corresponding to a vehicle parameter group, and the vehicle parameter group includes a vehicle speed and a vehicle mileage; Abnormal data points are deleted from the discrete distribution graph according to preset rules to obtain a data set for training the pre-trained model.
4. The energy consumption prediction model training method according to claim 3 is characterized in that: Deleting abnormal data points from the discrete distribution graph according to a preset rule includes: Deleting data points corresponding to the vehicle speed being less than a preset speed from the discrete distribution graph; Deleting data points corresponding to the interval-stop distance being less than a preset distance from the discrete distribution graph, wherein the interval-stop distance represents the distance between two consecutive positions where the vehicle speed is zero in the driving section; Data points corresponding to estimated full-charge ranges exceeding a preset range of rated ranges are deleted from the discrete distribution diagram, wherein the full-charge range is estimated based on the vehicle mileage and energy consumption corresponding to the data points.
5. A method for predicting energy consumption, characterized in that: include: Dividing the driving section into a plurality of segmented sections according to the historical vehicle speeds passing through the driving section, wherein the difference between the maximum vehicle speed and the minimum vehicle speed in each segmented section is less than a preset threshold; Determine the segmented energy consumption of each segmented road section according to the energy consumption prediction model established by the energy consumption prediction model training method according to any one of claims 1 to 4; The energy consumption of the driving section is determined according to the segmented energy consumptions of all the segmented sections.
6. The energy consumption prediction method according to claim 5, characterized in that: The step of dividing the driving section into a plurality of segmented sections according to the historical speed of vehicles passing through the driving section comprises: Determine the driving section according to the starting point, the waypoints and the end point; Determine the road condition of the driving section according to road signs and road forms; After estimating the change of the vehicle speed according to the road condition, the driving section is divided according to the vehicle speed.
7. An energy consumption prediction model training device, characterized in that: include: The first module is used to establish a pre-trained model for energy consumption prediction; The second module is used to supervise and fine-tune the pre-trained model according to the acquired data set to establish an energy consumption prediction model, wherein the energy consumption prediction model is used to determine the segmented energy consumption of the segmented road section, so as to determine the energy consumption of the driving section according to the segmented energy consumption; wherein the driving section includes each segmented road section.
8. An energy consumption prediction device, characterized in that: include: The third module is used to divide the driving section into a plurality of segmented sections according to the historical vehicle speeds passing through the driving section, wherein the difference between the maximum vehicle speed and the minimum vehicle speed in each segmented section is less than a preset threshold; The fourth module is used to determine the segmented energy consumption of each segmented road section according to the energy consumption prediction model established by the energy consumption prediction model training method; The fifth module is used to determine the energy consumption of the driving section according to the segmented energy consumption of all the segmented sections.
9. An electronic device, characterized in that: including memory and processor; The memory is used to store computer programs; The processor is used to implement the energy consumption prediction model training method as described in any one of claims 1 to 5, or the energy consumption prediction method as described in claim 6 or 7 when executing the computer program.
10. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by the processor, it implements the energy consumption prediction model training method as described in any one of claims 1 to 5, or the energy consumption prediction method as described in claim 6 or 7.