A vehicle load estimation method and device, electronic equipment and readable storage medium

Abstract

This application relates to the field of automotive technology and provides a vehicle load estimation method, apparatus, electronic device, and readable storage medium. The method includes: collecting raw load data of a vehicle while it is driving on a road; preprocessing the raw load data to obtain preprocessed load data; inputting the preprocessed load data into a preset vehicle load estimation model to output a first vehicle load estimation value; collecting in-vehicle image information and vehicle model information, and determining a second vehicle load estimation value based on the in-vehicle image information and vehicle model information; determining a third vehicle load estimation value based on the second vehicle load estimation value, road feature information, and driving road map information; and determining a final vehicle load estimation value based on the first, second, and third vehicle load estimation values. This application can estimate the vehicle load under different road conditions and driving states with high accuracy.

Description

Technical Field

[0001] This application relates to the field of automotive technology, and in particular to a vehicle load estimation method, apparatus, electronic device, and readable storage medium. Background Technology

[0002] Vehicle load refers to the weight of goods or passengers carried by a vehicle during operation. Every vehicle has a maximum load capacity, and overloading will cause significant damage to both the vehicle and road safety. For vehicles, overloading accelerates fatigue damage to vehicle components, reduces their safety performance, shortens their lifespan, and increases the risk of loss of control, directly impacting driving safety. For roads, overloading can damage the road surface and subgrade, increasing road maintenance costs and even threatening road safety.

[0003] Therefore, in road traffic, strictly controlling vehicle load is of great significance for reducing fatigue damage to vehicle parts, extending the service life of parts, ensuring vehicle driving safety, and avoiding damage to the road surface and roadbed, thus ensuring road safety.

[0004] In related technologies, the weighing method for vehicle load weight mainly involves collecting weight signals from weighing sensors deployed in the lane weighing area, and then using a microcontroller or host computer to analyze the data to determine the vehicle load weight. However, while existing vehicle load weight weighing methods are relatively accurate in measuring the vehicle load of a static vehicle, their accuracy in estimating the vehicle load under different road conditions and driving states is poor. Summary of the Invention

[0005] In view of this, embodiments of this application provide a vehicle load estimation method, device, electronic device, and readable storage medium to solve the problem that existing vehicle load weight weighing methods are relatively accurate in measuring the vehicle load of a static vehicle, but have poor accuracy in estimating the vehicle load under different road conditions and driving states.

[0006] A first aspect of this application provides a vehicle load estimation method, including:

[0007] The raw load data of the vehicle while it is driving on the road is collected, and the raw load data is preprocessed to obtain the preprocessed load data. The raw load data includes the six components of the wheel center force, seat acceleration, wheel axle head acceleration, body acceleration, wheel center runout displacement and vehicle chassis stress.

[0008] The preprocessed load data is input into a preset vehicle load estimation model, and the first vehicle load estimation value is output.

[0009] Collect in-vehicle image information and vehicle model information, and determine the estimated value of the second vehicle load based on the in-vehicle image information and vehicle model information;

[0010] Collect road feature information and driving road map information, and determine the third vehicle load estimate based on the first vehicle load estimate, road feature information, and driving road map information;

[0011] The final vehicle load estimate is determined based on the first vehicle load estimate, the second vehicle load estimate, and the third vehicle load estimate.

[0012] A second aspect of this application provides a vehicle load estimation device, comprising:

[0013] The first acquisition module is configured to acquire raw load data of the vehicle while it is driving on the road, preprocess the raw load data to obtain preprocessed load data. The raw load data includes wheel center six-component force, seat acceleration, wheel axle head acceleration, vehicle body acceleration, wheel center runout displacement and vehicle chassis stress.

[0014] The output module is configured to input the preprocessed load data into a preset vehicle load estimation model and output the first vehicle load estimation value.

[0015] The second acquisition module is configured to acquire in-vehicle image information and vehicle model information, and determine the estimated value of the second vehicle load based on the in-vehicle image information and vehicle model information.

[0016] The third acquisition module is configured to acquire road feature information and driving road map information, and determine the third vehicle load estimate based on the first vehicle load estimate, road feature information and driving road map information;

[0017] The determination module is configured to determine the final vehicle load estimate based on the first vehicle load estimate, the second vehicle load estimate, and the third vehicle load estimate.

[0018] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0019] A fourth aspect of this application provides a readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described method.

[0020] Compared with the prior art, the beneficial effects of this application embodiment include at least the following: preprocessing the collected raw load data, then inputting the preprocessed load data into a preset vehicle load estimation model to output a first vehicle load estimation value; collecting in-vehicle image information and vehicle model information, and determining a second vehicle load estimation value based on the in-vehicle image information and vehicle model information; collecting road feature information and driving road map information, and determining a third vehicle load estimation value based on the first vehicle load estimation value, road feature information, and driving road map information; and determining a final vehicle load estimation value based on the first vehicle load estimation value, the second vehicle load estimation value, and the third vehicle load estimation value. This enables the estimation of vehicle load under different road conditions and driving states with high accuracy. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram illustrating one application scenario of this application.

[0023] Figure 2 This is a schematic flowchart of a vehicle load estimation method provided in an embodiment of this application;

[0024] Figure 3 This is a schematic diagram of the architecture of a vehicle load estimation model training system provided in an embodiment of this application;

[0025] Figure 4 This is a schematic diagram of a working condition distribution provided in an embodiment of this application;

[0026] Figure 5 This is a schematic diagram of a vehicle load estimation device provided in an embodiment of this application;

[0027] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0028] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0029] A vehicle load estimation method and apparatus according to embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0030] Figure 1 This is a schematic diagram illustrating an application scenario according to an embodiment of this application. The application scenario may include a vehicle 101 and a server 102. The vehicle 101 and the server 102 can be connected via a network (e.g., a wired network connected by coaxial cable, twisted pair, and fiber optic cable, or a wireless network that enables interconnection of various communication devices without wiring, such as Bluetooth, Near Field Communication (NFC), Infrared, etc.).

[0031] Vehicle 101 includes multiple sensors, such as a six-component force sensor and a displacement sensor installed on the vehicle's wheels, an acceleration sensor installed on key components such as the vehicle's seats and axle heads, a Global Positioning System (GPS), strain gauges installed on key areas of interest on the vehicle, and communication devices.

[0032] Accelerometers can directly reflect the severity of vehicle vibrations, and axle acceleration signals can effectively estimate wheel loads. Accelerometers can be categorized into unidirectional and tridirectional sensors based on their measurement range; the appropriate sensor can be selected according to the required measurement points. Displacement sensors are commonly used to measure changes in the travel of the damping system or the degree of frame deformation. GPS sensors are primarily used to record vehicle routes and speed information.

[0033] Server 102 can be a server that provides various services. Specifically, it can be a single server, a server cluster consisting of several servers, or a cloud computing service center. This application embodiment does not limit this.

[0034] Figure 2 This is a schematic flowchart of a vehicle load estimation method provided in an embodiment of this application. Figure 2 Vehicle load estimation methods can be derived from Figure 1 Server 102 executes. For example... Figure 2As shown, the vehicle load estimation method includes:

[0035] Step S201: Collect the original load data of the vehicle while it is driving on the road, preprocess the original load data to obtain the preprocessed load data. The original load data includes the six components of the wheel center force, seat acceleration, wheel axle head acceleration, vehicle body acceleration, wheel center jump displacement and vehicle chassis stress.

[0036] In one embodiment, a six-component force sensor installed on the vehicle's wheels can be used to collect the six-component force at the wheel center, namely the three-directional force and torque of the wheel; acceleration sensors installed at key components such as the vehicle's seats and axle heads can be used to collect seat acceleration, axle head acceleration, and vehicle body acceleration; displacement sensors installed at the vehicle's wheels can be used to collect wheel center jump displacement; and strain gauges installed at key areas of the vehicle can be used to collect vehicle chassis stress. Then, the collected raw load data is uploaded to server 102 for subsequent data analysis and processing.

[0037] In practical applications, various sensors related to vehicle load estimation can be integrated. When the vehicle is in motion, data from each sensor can be collected synchronously and uploaded to server 102.

[0038] Raw load data refers to the unprocessed sensor data collected by various sensors on the vehicle, namely the "load-time" history data model, which is the signal of the vehicle components changing over time during the driving process.

[0039] Since raw load data cannot be directly applied to vehicle load estimation, it is necessary to preprocess and compile a large amount of disordered raw load data into stable and reliable analytical data that can be directly applied to vehicle load estimation, thereby improving the accuracy of subsequent vehicle load estimation.

[0040] Step S202: Input the preprocessed load data into the preset vehicle load estimation model and output the first vehicle load estimation value.

[0041] Vehicle load estimation models can be neural network models (such as LSTM (Long Short-Term Memory) neural network models) or joint learning models.

[0042] The first vehicle load estimate refers to the vehicle load estimate (the maximum load capacity that the vehicle can withstand) obtained by collecting sensor data related to vehicle load estimation using sensor technology and processing the sensor data based on a preset vehicle load estimation model.

[0043] Step S203: Collect in-vehicle image information and vehicle model information, and determine the estimated value of the second vehicle load based on the in-vehicle image information and vehicle model information.

[0044] In one embodiment, the in-vehicle image information includes cabin interior image information. The cabin interior image information can be acquired by a vehicle-mounted camera device (such as a vehicle-mounted camera) to capture images of the occupants and cargo inside the cabin.

[0045] In another embodiment, the in-vehicle image information may also include images of the vehicle's trunk. People typically place large or heavy items (such as suitcases) in the trunk, so a camera can be installed in the trunk to capture images of the trunk's contents.

[0046] Vehicle information includes the vehicle manufacturer, brand, size, and model. For example, car models mainly include small cars, microcars, compact cars, mid-size cars, high-end cars, luxury cars, sedans, CDVs, MPVs, and SUVs.

[0047] Vehicle load capacity is related to factors such as vehicle structure and materials, engine output power, suspension system, and tire size. Different vehicle models often have different structures and materials, engine output power, suspension systems, and tire sizes, sometimes with significant differences. Therefore, the maximum vehicle load capacity of different vehicle models is often different.

[0048] By collecting in-vehicle image information and vehicle model information for data analysis, and combining the passenger and cargo situation inside the vehicle with the maximum vehicle load corresponding to the vehicle model, the accuracy of vehicle load estimation can be improved.

[0049] Step S204: Collect road feature information and driving road map information, and determine the third vehicle load estimate based on the first vehicle load estimate, road feature information, and driving road map information.

[0050] Road feature information is information used to characterize the road surface features of the road on which a vehicle is currently traveling. For example, information used to characterize features such as road bumps (e.g., speed bumps), road depressions (e.g., potholes), icy or snowy roads, and slippery roads.

[0051] Step S205: Determine the final vehicle load estimate based on the first vehicle load estimate, the second vehicle load estimate, and the third vehicle load estimate.

[0052] The technical solution provided in this application preprocesses the collected raw load data, then inputs the preprocessed load data into a preset vehicle load estimation model to output a first vehicle load estimate; it collects in-vehicle image information and vehicle model information, and determines a second vehicle load estimate based on the in-vehicle image information and vehicle model information; it collects road feature information and driving road map information, and determines a third vehicle load estimate based on the first vehicle load estimate, road feature information, and driving road map information; and it determines a final vehicle load estimate based on the first vehicle load estimate, the second vehicle load estimate, and the third vehicle load estimate. This method can estimate the vehicle load under different road conditions and driving states with high accuracy.

[0053] In other embodiments, vehicle 101 may also include a vehicle load estimation device, which can be used to perform the steps of the above-described vehicle load estimation method.

[0054] In some embodiments, the raw load data includes N sampling points, each sampling point corresponding to a set of sampling signals; a set of sampling signals includes wheel center force, seat acceleration, wheel axle head acceleration, vehicle body acceleration, wheel center jump displacement and vehicle chassis stress, where N is a positive integer.

[0055] The steps for preprocessing the raw load data to obtain preprocessed load data include: calculating the mean and standard deviation of the raw load data; calculating a threshold value based on the standard deviation and a preset adjustment coefficient, wherein the adjustment coefficient is ≥4; calculating the absolute value of the difference between each sampling point and the mean; if the absolute value is greater than the threshold value, the sampling point is identified as an abnormal sampling point; removing abnormal sampling points from the raw load data and extracting the signal peak and signal valley values ​​from the raw load data to obtain the preprocessed load data.

[0056] The original load data is a series of load-time history data consisting of N sets of sampled signals corresponding to N sampling points.

[0057] A sampling point can be understood as a sampling time point; one sampling time point corresponds to a set of sampled signals.

[0058] As an example, for the raw load data , i =1,2,3,...N, calculate the original load data according to the following formula (1). mean The original load data are calculated according to the following formula (2). Standard deviation .

[0059] (1);

[0060] (2).

[0061] Next, the threshold value is calculated according to the following formula (3).

[0062] (3).

[0063] In equation (3), Indicates the threshold value. This represents the adjustment factor.

[0064] If it is the original load data The first in i The sampled signal and mean of each sampling point The absolute value of the difference, i.e. ,like If so, the sampling point is identified as an abnormal sampling point.

[0065] When the adjustment coefficient When the adjustment coefficient is less than 4, too many sampling points will be removed, resulting in a reduction in data volume, compromising the integrity and accuracy of the original load data, and thus reducing the accuracy of vehicle load estimation. When the value is ≥4, significant outlier sampling points can be removed while ensuring the integrity and authenticity of the original load data, which helps to improve the accuracy of vehicle load estimation.

[0066] Even after removing outlier sampling points, the original load data is still quite large, which places high demands on computing power. This embodiment of the application extracts the signal peaks and valleys from the original load data after removing outlier sampling points. This effectively compresses the size of the original load data while maintaining its waveform essentially unchanged, reducing the amount of data input to the subsequent vehicle load estimation model. This reduces the computing power requirements and improves the model's computational efficiency for the preprocessed load data.

[0067] In some embodiments, the above-mentioned preset vehicle load estimation model is trained by the following steps:

[0068] Start training of the first joint learning model to obtain the first load estimation guidance model. Based on the first load estimation guidance model, perform knowledge transfer learning on the initial load estimation model to obtain the first load estimation learning model, and upload the first model parameters of the first load estimation learning model to the server.

[0069] Upon receiving the first aggregated parameters returned by the server after aggregating multiple first model parameters, the first load estimation learning model is iteratively updated based on the first aggregated parameters to obtain the second load estimation learning model.

[0070] Start training the second joint learning model, use the second load estimation learning model as the second load estimation guidance model, perform knowledge transfer learning on the first load estimation guidance model to obtain the third load estimation learning model, and upload the second model parameters of the third load estimation learning model to the server.

[0071] Upon receiving the second aggregated parameters returned by the server after aggregating multiple second model parameters, the second model parameters of the third load estimation learning model are iteratively updated according to the second aggregated parameters to obtain the third load estimation guidance model.

[0072] When the training of the first joint learning model and the training of the second joint learning model reach the preset convergence condition, the vehicle load estimation model is obtained.

[0073] Combination Figure 3 As an example, server 102 can select at least two participants to train the first joint learning model according to actual needs. For example, server 102 selects participants 301, 302, and 303 to participate in the training of the first joint learning model. Among them, participants 301, 302, and 303 can be vehicle manufacturers with a large amount of vehicle load data, or they can be vehicle manufacturers with a small amount of vehicle load data or no vehicle load data, but who want to optimize the vehicle load estimation model.

[0074] The first load estimation guidance model is typically a vehicle load estimation model pre-trained by vehicle manufacturers with abundant vehicle load data as training data. This model usually has a large number of weight parameters, placing high demands on the device's memory and computing power.

[0075] The initial load estimation model can be a basic model pre-trained by each participant using their local vehicle load data, or it can be a basic model established by server 102 and distributed to each participant.

[0076] Based on the first load estimation guidance model, knowledge transfer learning is performed on the initial load estimation model to obtain the first load estimation learning model. That is, each participant 301, 302, and 303 transfers the knowledge of the first load estimation guidance model to their initial load estimation model, thereby optimizing and adjusting the model parameters of their respective initial load estimation models to obtain the first load estimation learning model.

[0077] Each participant 301, 302, and 303 uploads the first model parameters (including weights and biases) of their respective first load estimation learning models to server 102. Upon receiving the first model parameters uploaded by each participant 301, 302, and 303, server 102 aggregates the first model parameters from each participant to obtain first aggregated parameters. After receiving the first aggregated model parameters, each participant 301, 302, and 303 uses these first aggregated parameters to iteratively update the first model parameters of the first load estimation learning model to obtain the second load estimation learning model.

[0078] Next, the training of the second joint learning model is initiated. The second load estimation learning models of each participant (301, 302, and 303) are identified as the second load estimation guidance models. Knowledge transfer is performed from the second load estimation guidance model to the first load estimation guidance model. That is, the first load estimation guidance model adjusts its model parameters based on the knowledge of the second load estimation guidance models of each participant, resulting in the third load estimation learning model. Afterward, the second model parameters of each of the obtained third load estimation learning models are uploaded to server 102. Upon receiving the second model parameters of the third load estimation learning models uploaded by each participant (301, 302, and 303), server 102 aggregates these second model parameters to obtain second aggregated parameters. These second aggregated parameters are then returned to each participant. Each participant (301, 302, and 303) uses these second aggregated parameters to iteratively update the second model parameters of the third load estimation learning model, obtaining the third load estimation guidance model, thus completing one round of bidirectional knowledge transfer learning (i.e., including one training cycle of the first joint learning model and one training cycle of the second joint learning model).

[0079] Next, following the steps above, start the next round of bidirectional knowledge transfer learning until the training of the first joint learning model and the training of the second joint learning model reach the preset convergence conditions (e.g., the training rounds reach the preset number of bidirectional knowledge transfer learning rounds, the model accuracy reaches the preset accuracy, etc.), and then obtain the vehicle load estimation model.

[0080] Training the vehicle load estimation model through bidirectional knowledge transfer learning not only makes the model lighter but also improves its performance, thereby enhancing the accuracy of vehicle load estimation.

[0081] In some embodiments, the step of determining the second vehicle load estimate based on in-vehicle image information and vehicle model information includes:

[0082] Identify the target region in the in-vehicle image information and extract the feature information of the target object in the target region;

[0083] Based on the target object's characteristic information, the estimated value of the vehicle's second vehicle load is determined.

[0084] In one embodiment, the driver and passenger area inside the vehicle cabin, i.e., the target area, can be identified based on a target detection algorithm (e.g., the YOLO series of algorithms). Then, the target object feature information in this target area is extracted. The target object feature information indicates whether there are drivers or passengers in the driver and passenger area, and whether other items (e.g., backpacks or other miscellaneous items) are loaded in the driver and passenger area.

[0085] The second vehicle load estimate refers to the estimated load (weight) of the occupants and other items loaded on the seats inside the vehicle's cabin.

[0086] In one embodiment, the load corresponding to the target object feature information can be obtained by querying a pre-set correspondence table between target object feature information and load amount, and then the various load amounts can be superimposed to obtain the second vehicle load estimate value of the vehicle.

[0087] The correspondence between target feature information and load amount can be set as shown in Table 1 below.

[0088] Table 1. Correspondence between target object feature information and load amount

[0089]

[0090] Generally, if the target object is a driver or passenger, the target object's characteristic information can include the driver's or passenger's age, gender, and ethnicity. People of different ages, genders, and ethnicities often have significantly different weights (load capacities); for example, the weight of an infant differs considerably from that of an adult.

[0091] In practical applications, big data platforms can be used to collect data on people of different ages, genders, and races and their weight. Through data analysis, a correlation between people of different ages, genders, and races and their weight (load capacity) can be established.

[0092] Similarly, for other objects, the same method of establishing the correspondence between people of different ages, genders, and races and their weight (load capacity) can be used to establish the correspondence between different objects and their load capacity.

[0093] In some embodiments, the step of determining a third vehicle load estimate based on a first vehicle load estimate, road feature information, and driving road map information includes:

[0094] Based on road feature information and driving route map information, determine the working condition distribution map of the vehicle when driving on the road;

[0095] Determine the load distribution diagram of the vehicle when it is driving on the road based on the working condition distribution diagram;

[0096] The estimated load of the first vehicle is corrected based on the load distribution diagram to obtain the estimated load of the third vehicle.

[0097] A road condition distribution map refers to the distribution of road conditions corresponding to various driving segments on the road where the vehicle is currently traveling. These road conditions include, but are not limited to, straight road conditions, transition curve conditions, sharp bends, uphill conditions, downhill conditions, potholed road conditions, icy and snowy road conditions, and slippery road conditions.

[0098] Figure 4 This is a schematic diagram of a working condition distribution provided in an embodiment of this application.

[0099] Combination Figure 4 Assuming that based on the road feature information and road map information of the current road segment, the driving road is divided into driving segments ①, ②, ③, and ④. Among them, driving segment ① corresponds to a straight road condition; driving segment ② corresponds to a transition curve condition; driving segment ③ corresponds to a sharp bend condition; and driving segment ④ corresponds to a slippery road condition.

[0100] A load distribution diagram refers to the load distribution of a vehicle under different operating conditions on different road sections. Generally, different road conditions have varying impacts on vehicle load. For example, on straight roads, the overall vehicle load is relatively stable; on sharp bends, the load varies significantly in certain localized areas.

[0101] In addition, the vehicle's operating status (e.g., traction, coasting, braking) and the driver's driving behavior also have a significant impact on the vehicle load. Therefore, the road load corresponding to a certain driving segment can be determined by combining the vehicle's operating status, the driver's driving behavior, and the operating conditions of the driving segment.

[0102] Based on the above examples, and according to the working conditions of each road segment ①, ②, ③, and ④, combined with the vehicle's operating status and the driver's driving habits when driving on each road segment, the corresponding road segment loads a (corresponding to road segment ①), b (corresponding to road segment ②), c (corresponding to road segment ③), and d (corresponding to road segment ④) are obtained when the vehicle is driving on each road segment. Thus, the load distribution map of the vehicle when driving on the road (including road segments ①, ②, ③, and ④) is obtained.

[0103] Next, based on the load distribution diagram above, the estimated load value of the first vehicle is corrected to obtain the estimated load value of the third vehicle.

[0104] The step of correcting the estimated load of the first vehicle based on the load distribution diagram to obtain the estimated load of the third vehicle includes:

[0105] Based on the load distribution map, the road currently being traveled by the vehicle is divided into multiple travel segments;

[0106] A road segment load correction factor is set for each driving segment based on the load distribution map;

[0107] The first vehicle load estimate for each driving segment is corrected based on the load correction coefficient for each segment to obtain the third vehicle load estimate.

[0108] To facilitate understanding, we will continue with the above example. Assume that the road the vehicle is currently traveling on is divided into segments based on the load distribution map obtained above, resulting in driving segments ①, ②, ③, and ④. Furthermore, based on the road conditions of each driving segment ①, ②, ③, and ④, the corresponding road segment loads are determined to be a (corresponding to driving segment ①), b (corresponding to driving segment ②), c (corresponding to driving segment ③), and d (corresponding to driving segment ④), respectively. Then, a road segment load correction coefficient θ1, θ2, θ3, and θ4 are set for each driving segment ①, ②, ③, and ④. Next, the estimated first vehicle load ① corresponding to driving segment ① is corrected according to the segment load correction factor θ1 to obtain the corrected load value ①; the estimated first vehicle load ② corresponding to driving segment ② is corrected according to the segment load correction factor θ2 to obtain the corrected load value ②; the estimated first vehicle load ③ corresponding to driving segment ③ is corrected according to the segment load correction factor θ3 to obtain the corrected load value ③; and the estimated first vehicle load ④ corresponding to driving segment ④ is corrected according to the segment load correction factor θ4 to obtain the corrected load value ④. Finally, the corrected load values ​​①, ②, ③, and ④ are summed to obtain the estimated third vehicle load value.

[0109] By combining the working condition distribution map and load distribution map of each driving segment on the road, setting the corresponding road segment load correction coefficient, and correcting the first vehicle load estimate value corresponding to each driving segment according to the road segment load correction coefficient, the accuracy of vehicle load estimation can be improved.

[0110] In some embodiments, the step of determining a final vehicle load estimate based on a first vehicle load estimate, a second vehicle load estimate, and a third vehicle load estimate includes:

[0111] The first vehicle load value is calculated based on the estimated first vehicle load value and the first weighting coefficient.

[0112] The second vehicle load value is calculated based on the estimated second vehicle load value and the second weighting coefficient.

[0113] The third vehicle load value is calculated based on the estimated third vehicle load value and the third weighting coefficient.

[0114] The first vehicle load value, the second vehicle load value, and the third vehicle load value are superimposed to obtain the final estimated vehicle load value.

[0115] The first, second, and third weighting coefficients can be flexibly set according to the actual situation, as long as their sum is 1. For example, the first, second, and third weighting coefficients can be set to 0.4, 0.1, and 0.5, respectively.

[0116] The final estimated vehicle load value is calculated according to the following formula (4).

[0117] (4).

[0118] In equation (4), L represents the final estimated vehicle load. This represents the estimated load of the first vehicle. This represents the first weighting coefficient. This indicates the estimated load of the second vehicle. This represents the second weighting coefficient. This indicates the estimated load of the third vehicle. This represents the third weighting coefficient.

[0119] By calculating a weighted average of the first, second, and third vehicle load estimates, the impact of different road conditions and driving states on vehicle load can be comprehensively considered, thereby improving the accuracy of vehicle load estimation.

[0120] All of the above-mentioned optional technical solutions can be combined in any way to form the optional embodiments of this application, and will not be described in detail here.

[0121] The following are embodiments of the apparatus described in this application, which can be used to execute the embodiments of the method described in this application. For details not disclosed in the apparatus embodiments of this application, please refer to the embodiments of the method described in this application.

[0122] Figure 5 This is a schematic diagram of a vehicle load estimation device provided in an embodiment of this application. Figure 5 As shown, the vehicle load estimation device includes:

[0123] The first acquisition module 501 is configured to acquire raw load data of the vehicle while it is driving on the road, preprocess the raw load data to obtain preprocessed load data, and the raw load data includes wheel center six-component force, seat acceleration, wheel axle head acceleration, vehicle body acceleration, wheel center jump displacement and vehicle chassis stress.

[0124] The output module 502 is configured to input the preprocessed load data into a preset vehicle load estimation model and output a first vehicle load estimation value.

[0125] The second acquisition module 503 is configured to acquire in-vehicle image information and vehicle model information, and determine the estimated value of the second vehicle load based on the in-vehicle image information and vehicle model information.

[0126] The third acquisition module 504 is configured to acquire road feature information and driving road map information, and determine the third vehicle load estimate based on the first vehicle load estimate, road feature information and driving road map information;

[0127] The determination module 505 is configured to determine the final vehicle load estimate based on the first vehicle load estimate, the second vehicle load estimate, and the third vehicle load estimate.

[0128] The technical solution provided in this application preprocesses the collected raw load data, then inputs the preprocessed load data into a preset vehicle load estimation model to output a first vehicle load estimate; it collects in-vehicle image information and vehicle model information, and determines a second vehicle load estimate based on the in-vehicle image information and vehicle model information; it collects road feature information and driving road map information, and determines a third vehicle load estimate based on the first vehicle load estimate, road feature information, and driving road map information; and it determines a final vehicle load estimate based on the first vehicle load estimate, the second vehicle load estimate, and the third vehicle load estimate. This method can estimate the vehicle load under different road conditions and driving states with high accuracy.

[0129] In some embodiments, the raw load data includes N sampling points, each sampling point corresponding to a set of sampling signals; a set of sampling signals includes wheel center force, seat acceleration, wheel axle head acceleration, vehicle body acceleration, wheel center jump displacement and vehicle chassis stress, where N is a positive integer.

[0130] The first acquisition module 501 mentioned above includes a data preprocessing unit, which is configured to preprocess the original load data to obtain preprocessed load data.

[0131] This data preprocessing unit includes:

[0132] The first calculation component is configured to calculate the mean and standard deviation of the raw load data;

[0133] The second calculation component is configured to calculate the threshold value based on the standard deviation and a preset adjustment factor, wherein the adjustment factor is ≥4;

[0134] The third calculation component is configured to calculate the absolute value of the difference between the sampled signal at each sampling point and the mean.

[0135] The anomaly determination component is configured to determine a sampling point as an anomaly sampling point if the absolute value is greater than a threshold value.

[0136] The removal component is configured to remove abnormal sampling points from the original load data and extract signal peaks and valleys from the original load data to obtain preprocessed load data.

[0137] In some embodiments, the preset vehicle load estimation model is trained by the following steps:

[0138] Start training of the first joint learning model to obtain the first load estimation guidance model. Based on the first load estimation guidance model, perform knowledge transfer learning on the initial load estimation model to obtain the first load estimation learning model, and upload the first model parameters of the first load estimation learning model to the server.

[0139] Upon receiving the first aggregated parameters returned by the server after aggregating multiple first model parameters, the first load estimation learning model is iteratively updated based on the first aggregated parameters to obtain the second load estimation learning model.

[0140] Start training the second joint learning model, use the second load estimation learning model as the second load estimation guidance model, perform knowledge transfer learning on the first load estimation guidance model to obtain the third load estimation learning model, and upload the second model parameters of the third load estimation learning model to the server.

[0141] Upon receiving the second aggregated parameters returned by the server after aggregating multiple second model parameters, the second model parameters of the third load estimation learning model are iteratively updated according to the second aggregated parameters to obtain the third load estimation guidance model.

[0142] When the training of the first joint learning model and the training of the second joint learning model reach the preset convergence condition, the vehicle load estimation model is obtained.

[0143] In some embodiments, the second acquisition module 503 includes a first load determination unit, configured to determine a second vehicle load estimate based on in-vehicle image information and vehicle model information.

[0144] The first load determination unit includes:

[0145] The region determination component is configured to determine the target region in the in-vehicle image information and extract the feature information of the target object in the target region;

[0146] The estimation component is configured to estimate the second vehicle load estimate of the vehicle based on target feature information.

[0147] In some embodiments, the third acquisition module 504 includes a second load determination unit, configured to determine a third vehicle load estimate based on a first vehicle load estimate, road feature information, and driving road map information.

[0148] The second load determination unit includes:

[0149] The load condition distribution determination component is configured to determine the load condition distribution map of a vehicle while it is driving on a road based on road feature information and driving road map information.

[0150] The load distribution determination component is configured to determine the load distribution map of the vehicle when it is traveling on the road based on the load condition distribution map.

[0151] The correction component is configured to correct the first vehicle load estimate based on the load distribution map to obtain the third vehicle load estimate.

[0152] In some embodiments, the above-mentioned corrective component includes:

[0153] The segmentation device is configured to divide the road currently being traveled by the vehicle into multiple travel segments based on the load distribution map.

[0154] The device is configured to set a segment load correction factor for each travel segment based on the load distribution map;

[0155] The correction device is configured to correct the first vehicle load estimate corresponding to each driving segment according to the load correction coefficient of each segment, so as to obtain the third vehicle load estimate.

[0156] In some embodiments, the determining module 505 includes:

[0157] The first calculation unit is configured to calculate the first vehicle load value based on the first vehicle load estimate and the first weighting coefficient.

[0158] The second calculation unit is configured to calculate the second vehicle load value based on the second vehicle load estimate and the second weighting coefficient.

[0159] The third calculation unit is configured to calculate the third vehicle load value based on the third vehicle load estimate and the third weighting coefficient.

[0160] The fourth calculation unit is configured to superimpose the first vehicle load value, the second vehicle load value, and the third vehicle load value to obtain the final estimated vehicle load value.

[0161] It should be understood that the sequence number of each step in the above embodiments does not imply 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.

[0162] Figure 6 This is a schematic diagram of the electronic device 6 provided in an embodiment of this application. Figure 6 As shown, the electronic device 6 of this embodiment includes a processor 601, a memory 602, and a computer program 603 stored in the memory 602 and executable on the processor 601. When the processor 601 executes the computer program 603, it implements the steps in the various method embodiments described above. Alternatively, when the processor 601 executes the computer program 603, it implements the functions of each module / unit in the various device embodiments described above.

[0163] Electronic device 6 can be a desktop computer, laptop, handheld computer, cloud server, or other electronic device. Electronic device 6 may include, but is not limited to, processor 601 and memory 602. Those skilled in the art will understand that... Figure 6 This is merely an example of electronic device 6 and does not constitute a limitation on electronic device 6. It may include more or fewer components than shown, or different components.

[0164] The processor 601 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0165] The memory 602 can be an internal storage unit of the electronic device 6, such as a hard disk or RAM of the electronic device 6. The memory 602 can also be an external storage device of the electronic device 6, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc., equipped on the electronic device 6. The memory 602 can also include both internal and external storage units of the electronic device 6. The memory 602 is used to store computer programs and other programs and data required by the electronic device.

[0166] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to 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 integrated unit can be implemented in hardware or as a software functional unit.

[0167] If integrated modules / units are implemented as software functional units and sold or used as independent products, they can be stored in a readable storage medium (e.g., a computer-readable storage medium). Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program may include computer program code, which may be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable storage medium may include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0168] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A vehicle load estimation method characterized by comprising: include: The original load data of the vehicle while it is driving on the road is collected, and the original load data is preprocessed to obtain preprocessed load data. The original load data includes wheel center six-component force, seat acceleration, wheel axle head acceleration, body acceleration, wheel center runout displacement and vehicle chassis stress. The preprocessed load data is input into a preset vehicle load estimation model, and the first vehicle load estimation value is output. The vehicle interior image information and vehicle model information are collected, and the estimated value of the second vehicle load is determined based on the vehicle interior image information and vehicle model information. Collect road feature information and driving road map information of the road, and determine the third vehicle load estimate based on the first vehicle load estimate, road feature information and driving road map information; The final vehicle load estimate is determined based on the first vehicle load estimate, the second vehicle load estimate, and the third vehicle load estimate. Based on the first vehicle load estimate, road feature information, and driving route map information, a third vehicle load estimate is determined, including: Based on the road feature information and the driving route map information, a working condition distribution map of the vehicle when driving on the road is determined; Determine the load distribution diagram of the vehicle when it is driving on the road based on the working condition distribution diagram; The method of correcting the first vehicle load estimate based on the load distribution map to obtain a third vehicle load estimate includes: dividing the road currently being traveled by the vehicle into multiple travel segments based on the load distribution map; setting a segment load correction coefficient for each travel segment based on the load distribution map; and correcting the first vehicle load estimate corresponding to each travel segment based on each segment load correction coefficient to obtain a third vehicle load estimate.

2. The method of claim 1, wherein, The original load data includes N sampling points, each sampling point corresponding to a set of sampling signals; a set of sampling signals includes six components of wheel center force, seat acceleration, wheel axle head acceleration, vehicle body acceleration, wheel center jump displacement and vehicle chassis stress, where N is a positive integer; The original load data is preprocessed to obtain preprocessed load data, including: Calculate the mean and standard deviation of the original load data; The threshold value is calculated based on the standard deviation and the preset adjustment coefficient, wherein the adjustment coefficient is ≥4; Calculate the absolute value of the difference between the sampled signal at each of the sampling points and the mean; If the absolute value is greater than the threshold value, the sampling point is determined to be an abnormal sampling point; The abnormal sampling points in the original load data are removed, and the signal peaks and valleys in the original load data are extracted to obtain the preprocessed load data.

3. The method according to claim 1, characterized in that, The preset vehicle load estimation model is trained by the following steps: Start training of the first joint learning model to obtain the first load estimation guidance model. Based on the first load estimation guidance model, perform knowledge transfer learning on the initial load estimation model to obtain the first load estimation learning model, and upload the first model parameters of the first load estimation learning model to the server. Upon receiving the first aggregated parameters returned by the server after aggregating multiple first model parameters, the first load estimation learning model is iteratively updated according to the first aggregated parameters to obtain the second load estimation learning model. Start training of the second joint learning model, use the second load estimation learning model as the second load estimation guidance model, perform knowledge transfer learning on the first load estimation guidance model to obtain the third load estimation learning model, and upload the second model parameters of the third load estimation learning model to the server; Upon receiving the second aggregated parameters returned by the server, which are obtained by aggregating multiple second model parameters, the second model parameters of the third load estimation learning model are iteratively updated according to the second aggregated parameters to obtain the third load estimation guidance model. When the training of the first joint learning model and the training of the second joint learning model reach the preset convergence condition, the vehicle load estimation model is obtained.

4. The method according to claim 1, characterized in that, Based on the in-vehicle image information and vehicle model information, the estimated value of the second vehicle load is determined, including: Identify the target region in the in-vehicle image information and extract the feature information of the target object in the target region; Based on the target object's characteristic information, the estimated value of the vehicle's second vehicle load is determined.

5. The method according to claim 1, characterized in that, Based on the first vehicle load estimate, the second vehicle load estimate, and the third vehicle load estimate, the final vehicle load estimate is determined, including: The first vehicle load value is calculated based on the first estimated vehicle load value and the first weighting coefficient. The second vehicle load value is calculated based on the second vehicle load estimate and the second weighting coefficient. The third vehicle load value is calculated based on the estimated third vehicle load value and the third weighting coefficient. The first vehicle load value, the second vehicle load value, and the third vehicle load value are superimposed to obtain the final estimated vehicle load value.

6. A vehicle load estimation device, characterized in that, The apparatus is used to implement the method as described in any one of claims 1 to 5, the apparatus comprising: The first acquisition module is configured to acquire raw load data of the vehicle while it is driving on the road, and to preprocess the raw load data to obtain preprocessed load data. The raw load data includes wheel center six-component force, seat acceleration, wheel axle head acceleration, vehicle body acceleration, wheel center jump displacement and vehicle chassis stress. The output module is configured to input the preprocessed load data into a preset vehicle load estimation model and output a first vehicle load estimation value. The second acquisition module is configured to acquire in-vehicle image information and vehicle model information of the vehicle, and determine the estimated value of the second vehicle load based on the in-vehicle image information and vehicle model information. The third acquisition module is configured to acquire road feature information and driving road map information of the road, and determine the third vehicle load estimate based on the first vehicle load estimate, road feature information and driving road map information; The determination module is configured to determine the final vehicle load estimate based on the first vehicle load estimate, the second vehicle load estimate, and the third vehicle load estimate.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 5.

8. A readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 5.

Images