Vehicle road gradient real-time estimation method based on deep learning
The estimation model built through deep learning, utilizing vehicle CAN bus data and tire pressure dynamics data, solves the problems of high cost of gyroscopes and low accuracy of existing models, and achieves low-cost, real-time vehicle gradient and mass estimation.
Patent Information
- Application Number
- CN202510885410.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-30
- Publication Date
- 2025-10-17
AI Technical Summary
In existing technologies, using gyroscopes to estimate vehicle road slope is costly and has limited applicability, while Kalman filtering and vehicle dynamics models are not accurate enough to calculate vehicle slope in real time.
A deep learning-based approach is used to construct estimation models using feedforward neural networks and convolutional neural networks. Real-time estimation of vehicle mass and gradient is performed using vehicle CAN bus data, and training and prediction are conducted using tire pressure data and vehicle power data.
It achieves low-cost, real-time vehicle gradient and mass estimation, reduces hardware costs, improves estimation accuracy, reduces calculation cycles, and ignores the impact of vehicle configuration differences and gradient changes.
Smart Images
Figure CN120792833A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of vehicles, in particular to a vehicle road slope real-time estimation method based on deep learning. BACKGROUND
[0002] At present, the road slope data of commercial vehicles is mainly obtained by adding a gyroscope to the vehicle. However, using a gyroscope for slope estimation not only has high cost, but also has limited application scenarios. Specifically, the sensor has a high unit price and needs to be customized and installed for different vehicle models, which greatly increases the hardware and maintenance costs. The sensor is sensitive to road conditions and is easily disturbed by road vibration, resulting in a decrease in precision and poor stability in complex road conditions.
[0003] The existing Kalman filter and vehicle dynamics model can only calculate the vehicle mass, and the precision is not high, and the road slope of the vehicle cannot be calculated in real time.
[0004] Therefore, there is an urgent need for a low-cost and real-time vehicle road slope estimation method. SUMMARY
[0005] The present application provides a vehicle road slope real-time estimation method based on deep learning, which can estimate the road slope of the vehicle at low cost and in real time.
[0006] In order to achieve the above purpose, the present application adopts the following technical solutions:
[0007] In a first aspect, the present application provides a vehicle road slope real-time estimation method based on deep learning, comprising:
[0008] Obtaining historical running data of the vehicle, including tire pressure data and vehicle power data;
[0009] Preprocessing the historical running data to obtain training data;
[0010] Constructing an estimation model, the estimation model comprising a feedforward neural network and a convolutional neural network, the output layer of the feedforward neural network being connected to the input layer of the convolutional neural network, the feedforward neural network being configured to receive the tire pressure data and output the vehicle mass, and the convolutional neural network being configured to receive the vehicle mass and the vehicle power data and output the slope of the vehicle;
[0011] Training the estimation model using the training data to obtain a pre-trained estimation model;
[0012] Determining whether the vehicle has started, and if the vehicle enters the starting stage, collecting real-time tire pressure data and vehicle power data from the CAN bus of the vehicle to form a real-time data set;
[0013] After preprocessing the real-time data set, the real-time data set is input into the pre-trained estimation model to obtain a real-time slope of the vehicle.
[0014] In a preferred example of the present application, the tire pressure data can further include ambient temperature, tire temperature, tire unloaded pressure, real-time tire pressure, and tire cavity pressure.
[0015] In a preferred example of the present application, the vehicle power data can further include vehicle mass, vehicle acceleration, vehicle deceleration, engine torque, vehicle speed, and brake pedal stroke.
[0016] In a preferred example of the present application, the vehicle power data can further include vehicle mass, vehicle acceleration, vehicle deceleration, engine torque, vehicle speed, and brake pedal stroke.
[0017] In a preferred example of the present application, the step of preprocessing the historical operation data can include:
[0018] The fields of the historical operation data are filtered by a theoretical formula of a sliding mean filter.
[0019] In a preferred example of the present application, the feedforward neural network can include an input layer, a hidden layer, and an output layer.
[0020] In a preferred example of the present application, the convolutional neural network can include an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer. During training of the estimation model, a gradient optimizer Adam is used to update parameters of the convolutional neural network.
[0021] In a preferred example of the present application, the feedforward neural network can include an input layer, a hidden layer, and an output layer.
[0022] The training data acquisition module is configured to acquire historical operation data of a vehicle, including tire pressure data and vehicle power data.
[0023] The preprocessing module is configured to preprocess the historical operation data to obtain training data.
[0024] The model construction module is configured to construct an estimation model, the estimation model including a feedforward neural network and a convolutional neural network, an output layer of the feedforward neural network being connected to an input layer of the convolutional neural network, the feedforward neural network being configured to receive the tire pressure data and output vehicle mass, the convolutional neural network being configured to receive the vehicle mass and the vehicle power data and output a slope of the vehicle.
[0025] The training module is configured to train the estimation model using the training data to obtain a pre-trained estimation model.
[0026] An estimation module is configured to determine whether the vehicle starts, and if the vehicle enters a starting stage, start collecting real-time tire pressure data and vehicle power data from a CAN bus of the vehicle to form a real-time data set; after preprocessing the real-time data set, input the real-time data set into the pre-trained estimation model to obtain real-time slope of the vehicle.
[0027] In a third aspect, the present application provides a computer device, 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 deep learning based real-time vehicle road slope estimation method according to any one of the above aspects when executing the computer program.
[0028] In a fourth aspect, the present application provides a computer readable storage medium, wherein the computer readable storage medium stores a program, and wherein the program is executed by a processor to implement the deep learning based real-time vehicle road slope estimation method according to any one of the above aspects.
[0029] In a fifth aspect, the present application provides a computer program product, comprising computer instructions, wherein the computer instructions implement the steps of the deep learning based real-time vehicle road slope estimation method according to any one of the above aspects when executed by a processor.
[0030] In summary, compared with the prior art, the technical scheme provided by the embodiments of the present application has at least the following beneficial effects:
[0031] The deep learning model is used to construct an estimation model capable of accurately estimating the vehicle mass and real-time calculating the vehicle road slope. The estimation model is used for slope estimation, without the need for multiple labeling of the vehicle, the configuration differences of different vehicles can be ignored, and any slope sensor and gyroscope does not need to be installed to real-time calculate the operating road slope of the vehicle and the vehicle mass. Moreover, through continuous iterative learning of the deep learning model, the error can be controlled within a very small range. Thus, the estimation cost of the slope is greatly reduced.
[0032] Based on the data uploaded on the CAN bus of the whole vehicle, the tire pressure data is used to calculate the mass of the vehicle, and after obtaining the mass, the vehicle operating data is used to predict the slope of the vehicle operating road, thereby reducing the calculation period and improving the real-time estimation capability. Specifically, the mass prediction is completed within a very short time in the starting stage of the vehicle, and in this stage, the slope change is very small and is almost zero, so the influence of the slope does not need to be considered. BRIEF DESCRIPTION OF DRAWINGS
[0033] Figure 1 A flowchart of a deep learning based real-time vehicle road slope estimation method according to an embodiment of the present application is provided.
[0034] Figure 2A structural diagram of an estimation model of a vehicle road slope real-time estimation method based on deep learning is provided for an embodiment of the present application.
[0035] Figure 3 A module diagram of a vehicle road slope real-time estimation device based on deep learning is provided for an embodiment of the present application. DETAILED DESCRIPTION
[0036] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.
[0037] In an embodiment of the present application, a vehicle road slope real-time estimation method based on deep learning is provided, please refer to Figure 1 The method comprises the following steps:
[0038] S100: Obtain historical running data of a vehicle, including tire pressure data and vehicle power data.
[0039] Specifically, the tire pressure data and the vehicle data are obtained from historical data on the vehicle CAN bus, and the tire pressure data and the vehicle power data are corresponding.
[0040] The tire pressure data includes ambient temperature, tire temperature, tire unloaded pressure, real-time tire pressure, and tire cavity pressure.
[0041] The vehicle power data includes vehicle mass, vehicle acceleration, vehicle deceleration, engine torque, vehicle speed, and brake pedal stroke.
[0042] Since the tire-ground contact area and the tire cavity volume change little under different vehicle loads, the change amounts of these two parameters can be ignored.
[0043] S200: Preprocess the historical running data to obtain training data.
[0044] Specifically, the data obtained by the vehicle CAN bus may have random noise and oscillation, so the uploaded fields are filtered.
[0045] S300: Construct an estimation model, the estimation model comprising a feedforward neural network and a convolutional neural network, the output layer of the feedforward neural network being connected to the input layer of the convolutional neural network, the feedforward neural network being configured to receive the tire pressure data and output vehicle mass, and the convolutional neural network being configured to receive the vehicle mass and the vehicle power data and output the slope of the vehicle.
[0046] Specifically, the structure of the estimation model please refer to Figure 2 As shown in the application of the model, the tire pressure data is first input into the feedforward neural network, and the feedforward neural network outputs the mass of the vehicle, i.e. the vehicle load, after processing the data. Then, the feedforward neural network outputs the vehicle mass as one of the input data to the input layer of the convolutional neural network connected to its output layer.
[0047] The convolutional neural network receives the vehicle mass and the vehicle power data, and finally outputs the slope of the vehicle.
[0048] S400: Train the estimation model using the training data to obtain a pre-trained estimation model.
[0049] Specifically, the model parameters are adjusted by the gradient optimizer and the loss function during training to obtain the pre-trained estimation model.
[0050] S500: Determine whether the vehicle starts, if the vehicle enters the starting stage, start collecting real-time tire pressure data and vehicle power data from the CAN bus of the vehicle to form a real-time data set;
[0051] S600: After preprocessing the real-time data set, input the real-time data set into the pre-trained estimation model to obtain the real-time slope of the vehicle.
[0052] Specifically, the load of the vehicle is usually constant, so only the mass estimation needs to be performed during the starting stage of the vehicle, which is also a factor considered during the design of the model structure. During the starting stage of the vehicle, the slope changes very little and is almost zero, so the influence of the slope is not considered.
[0053] Through the estimation model, the real-time slope of the vehicle is finally obtained.
[0054] In this embodiment, a deep learning model is used to construct an estimation model that can accurately estimate the mass of the vehicle and calculate the real-time slope of the vehicle. The estimation model is used for slope estimation, without the need for multiple labeling of the vehicle, and the configuration differences of different vehicles can be ignored. Without the need for adding any slope sensor and gyroscope, the operating road slope of the vehicle and the mass of the vehicle can be calculated in real time. Moreover, through the continuous iterative learning of the deep learning model, the error can be controlled within a very small range. Therefore, the estimation cost of the slope is greatly reduced.
[0055] Based on the data uploaded on the vehicle CAN bus, the mass of the vehicle is calculated by using the tire pressure data, and after obtaining the mass, the slope of the vehicle operating road is predicted by using the vehicle operation data, which reduces the calculation period and improves the real-time estimation capability. Specifically, the mass prediction is completed in a very short time during the starting stage of the vehicle, and in this stage, the slope change is very small and almost zero, so the influence of the slope does not need to be considered.
[0056] In some embodiments, the step of preprocessing the historical operation data comprises:
[0057] The fields of the historical operation data are filtered by a sliding mean filter theoretical formula.
[0058] In a specific implementation, the sliding mean filter theoretical formula is:
[0059]
[0060] Wherein, x[n] is an input signal, and N is the size of the sliding window.
[0061] In this embodiment, the random noise and oscillation of the vehicle data collected from the vehicle CAN bus can be reduced.
[0062] In some embodiments, the convolutional neural network comprises an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer, and when the estimation model is trained, the parameters of the convolutional neural network are updated using a gradient optimizer Adam.
[0063] In this embodiment, the training effect of the estimation model is improved.
[0064] The application also provides a vehicle road slope real-time estimation device based on deep learning, as shown in Figure 3 The device comprises:
[0065] The training data acquisition module 100 is configured to obtain historical operation data of a vehicle, including tire pressure data and vehicle power data.
[0066] The preprocessing module 200 is configured to preprocess the historical operation data to obtain training data.
[0067] The model construction module 300 is configured to construct an estimation model, wherein the estimation model comprises a feedforward neural network and a convolutional neural network, the output layer of the feedforward neural network is connected to the input layer of the convolutional neural network, the feedforward neural network is configured to receive the tire pressure data and output the mass of the vehicle, and the convolutional neural network is configured to receive the mass of the vehicle and the vehicle power data and output the slope of the vehicle.
[0068] The training module 400 is configured to train the estimation model using the training data to obtain a pre-trained estimation model.
[0069] The estimation module 500 is configured to determine whether the vehicle starts, and if the vehicle enters a starting stage, start collecting real-time tire pressure data and vehicle power data from a CAN bus of the vehicle to form a real-time data set, pre-process the real-time data set, input the real-time data set into the pre-trained estimation model, and obtain real-time slope of the vehicle.
[0070] The functions of the modules in the above-mentioned deep learning-based real-time vehicle road slope estimation device correspond to the steps in the above-mentioned deep learning-based real-time vehicle road slope estimation method embodiment, and the functions and implementation processes will not be repeated here.
[0071] The application also provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the deep learning-based real-time vehicle road slope estimation method according to any one of the above-mentioned embodiments when executing the computer program.
[0072] The application also provides a computer readable storage medium, which stores a program, wherein the computer readable storage medium refers to a carrier for storing data, which can include, but is not limited to, floppy disks, optical disks, hard disks, flash memories, USB flash disks, Memory Sticks, and the like, and the computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The working process, working details, and technical effects of the computer readable storage medium provided in the embodiment can be referred to the above-mentioned embodiments of the deep learning-based real-time vehicle road slope estimation method, and will not be repeated here.
[0073] The application also provides a computer program product, which includes computer instructions executable by a processor to implement the steps of the deep learning-based real-time vehicle road slope estimation method according to any one of the above-mentioned embodiments.
[0074] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiment methods. Any reference to memory, storage, database or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0075] The technical features of the above embodiments can be combined in any manner. In order to make the description simple, not all possible combinations of the technical features in the above embodiments are described, however, as long as the combinations of the technical features do not contradict, they should be considered as within the scope of the present disclosure. The above embodiments only express several implementation manners of the present application, and the description is specific and detailed, but it should not be considered as a limitation on the scope of the patent. It should be pointed out that for those skilled in the art, without departing from the concept of the present application, some modifications and improvements can be made, which are within the scope of the present application. Therefore, the scope of the patent of the present application should be subject to the appended claims.
Claims
1. A real-time estimation method for vehicle road slope based on deep learning, characterized in that: include: Obtain historical vehicle operating data, including tire pressure data and vehicle power data; Preprocessing the historical operation data to obtain training data; Constructing an estimation model, the estimation model including a feedforward neural network and a convolutional neural network, wherein the output layer of the feedforward neural network is connected to the input layer of the convolutional neural network, the feedforward neural network is used to receive the tire pressure data and output the vehicle mass, and the convolutional neural network is used to receive the vehicle mass and the vehicle power data and output the vehicle slope; Using the training data to train the estimation model to obtain a pre-trained estimation model; Determine whether the vehicle is starting. If the vehicle enters the starting phase, start collecting real-time tire pressure data and vehicle power data from the vehicle's CAN bus to form a real-time data set; After preprocessing the real-time data set, the real-time data set is input into the pre-trained estimation model to obtain the real-time slope of the vehicle.
2. The method for real-time estimation of vehicle road slope based on deep learning according to claim 1, characterized in that: The tire pressure data includes ambient temperature, tire temperature, tire no-load pressure, tire real-time pressure and tire cavity pressure.
3. The method for real-time estimation of vehicle road slope based on deep learning according to claim 2, characterized in that: The vehicle dynamics data includes vehicle mass, vehicle acceleration, vehicle deceleration, engine torque, vehicle speed and brake pedal travel.
4. The method for real-time estimation of vehicle road slope based on deep learning according to claim 3, characterized in that: The step of preprocessing the historical operation data includes: Each field of the historical operation data is filtered using a sliding mean filter theoretical formula.
5. The method for real-time estimation of vehicle road slope based on deep learning according to claim 1, characterized in that: The feedforward neural network includes an input layer, a hidden layer and an output layer.
6. The method for real-time estimation of vehicle road slope based on deep learning according to claim 5, characterized in that: The convolutional neural network includes an input layer, a convolution layer, a pooling layer, a fully connected layer and an output layer. When the estimation model is trained, the gradient optimizer Adam is used to update the parameters of the convolutional neural network.
7. A real-time estimation device for vehicle road slope based on deep learning, characterized in that: include: The training data acquisition module is used to obtain the historical operating data of the vehicle, including tire pressure data and vehicle power data; A preprocessing module, used for preprocessing the historical operation data to obtain training data; a model construction module, configured to construct an estimation model, the estimation model comprising a feedforward neural network and a convolutional neural network, wherein the output layer of the feedforward neural network is connected to the input layer of the convolutional neural network, the feedforward neural network is configured to receive the tire pressure data and output the vehicle mass, and the convolutional neural network is configured to receive the vehicle mass and the vehicle power data and output the vehicle slope; A training module, configured to train the estimation model using the training data to obtain a pre-trained estimation model; The estimation module is used to determine whether the vehicle has started. If the vehicle enters the starting phase, it starts to collect real-time tire pressure data and vehicle power data from the vehicle's CAN bus to form a real-time data set; After preprocessing the real-time data set, the real-time data set is input into the pre-trained estimation model to obtain the real-time slope of the vehicle.
8. A computer 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 vehicle road slope real-time estimation method based on deep learning are implemented as described in any one of claims 1 to 6.
9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a program, wherein when the program is executed by the processor, the method for real-time estimation of vehicle road slope based on deep learning as described in any one of claims 1 to 6 is implemented.
10. A computer program product comprising computer instructions, characterized in that When executed by a processor, the computer instructions implement the steps of the vehicle road slope real-time estimation method based on deep learning according to claims 1 to 6.