Vehicle speed control method, vehicle, and computer-readable storage medium

Abstract

The embodiment of the application provides a vehicle speed control method, a vehicle and a computer readable storage medium. The method comprises the following steps: acquiring a first rainfall collected by a rainfall sensor on the vehicle during vehicle driving; predicting a first water film thickness of a road where the vehicle is located based on the first rainfall; determining a target vehicle speed of the vehicle based on the first water film thickness, a braking parameter of the vehicle and a system parameter of a driving system on the vehicle; and controlling the vehicle to drive at a limited speed based on the target vehicle speed. The application solves the technical problem of poor control accuracy of an intelligent driving system for vehicle speed in the related art.

Description

Technical Field

[0001] This application relates to the field of intelligent driving, and more specifically, to a vehicle speed control method, a vehicle, and a computer-readable storage medium. Background Technology

[0002] With the rapid development of intelligent driving technology, the vehicle's adaptive control capability in complex environments has become crucial for improving safety and user experience. However, current intelligent driving systems exhibit poor accuracy in controlling vehicle speed during rainy weather, thus impacting traffic efficiency and driving safety.

[0003] There is currently no good solution to the above problems. Summary of the Invention

[0004] This application provides a vehicle speed control method, a vehicle, and a computer-readable storage medium to at least solve the technical problem of poor accuracy in vehicle speed control by intelligent driving systems in related technologies.

[0005] According to one aspect of the embodiments of this application, a vehicle speed control method is provided, comprising: acquiring a first rainfall amount collected by a rain sensor on the vehicle during vehicle operation; predicting a first water film thickness on the road where the vehicle is located based on the first rainfall amount; determining a target vehicle speed based on the first water film thickness, vehicle braking parameters, and system parameters of the vehicle's driving system; and controlling the vehicle to drive at a speed limit based on the target vehicle speed.

[0006] Further, based on the first rainfall, the first water film thickness of the road where the vehicle is located is predicted, including: inputting the first rainfall into the water film thickness prediction model, and using the water film thickness prediction model to predict the first water film thickness, wherein the water film thickness prediction model is trained using the second water film thickness and the second rainfall measured during the process of the vehicle driving on the preset road under rainfall conditions.

[0007] Furthermore, the method also includes: obtaining the second water film thickness and the second rainfall; constructing an initial prediction model; inputting the second rainfall into the initial prediction model and using the initial prediction model to predict the third water film thickness; and adjusting the model parameters of the initial prediction model based on the second water film thickness and the third water film thickness to obtain a water film thickness prediction model.

[0008] Furthermore, obtaining the second water film thickness and the second rainfall includes: constructing a target road based on the preset construction depth, preset road slope, and preset road width of the road surface material; controlling the target road to be under rainfall conditions based on the preset rainfall intensity, and controlling vehicles to drive on the target road; collecting the water film thickness of the target road using data acquisition equipment to obtain the first water film thickness; and collecting rainfall using a rain sensor on the vehicle to obtain the second rainfall.

[0009] Furthermore, an initial prediction model is constructed, including: determining the number of first neurons in the input layer, the number of second neurons in the output layer, and the number of hidden layers; determining the number of third neurons in the hidden layer based on the number of first and second neurons; and constructing the initial prediction model based on the number of first neurons, the number of second neurons, the number of third neurons, and the number of layers.

[0010] Further, based on the number of the first and second neurons, the number of the third neurons in the hidden layer is determined, including: determining a range of the number of neurons based on the number of the first and second neurons, wherein the range of the number of neurons includes multiple candidate neuron numbers; determining the prediction error of the initial model corresponding to the multiple candidate neuron numbers; and determining the number of the third neurons from the multiple candidate neuron numbers based on the prediction error.

[0011] Furthermore, based on the first water film thickness, the vehicle's braking parameters, and the system parameters of the vehicle's driving system, the target vehicle speed is determined, including: inputting the braking parameters, system parameters, and the first water film thickness into the vehicle speed prediction model, and using the vehicle speed prediction model to predict the target vehicle speed. The vehicle speed prediction model is constructed based on the vehicle's braking distance model and a preset relationship model. The preset relationship model is used to characterize the correlation between the first water film thickness, the road surface adhesion coefficient of the target road, and the target vehicle speed.

[0012] Furthermore, the target vehicle speed is predicted using a vehicle speed prediction model, including: inputting braking parameters and system parameters into a braking distance model to obtain a first correlation between the target vehicle speed and the road surface adhesion coefficient of the target road, wherein the braking distance model is used to characterize the correlation between braking parameters, system parameters, target vehicle speed, and road surface adhesion coefficient; inputting a first water film thickness into a preset relationship model to obtain a second correlation between the target vehicle speed and the road surface adhesion coefficient of the target road, wherein the preset relationship model is used to characterize the correlation between the first water film thickness, road surface adhesion coefficient, and target vehicle speed; and determining the target vehicle speed based on the first and second correlations.

[0013] According to another aspect of the embodiments of this application, a vehicle speed control device is also provided, comprising: a first acquisition module, configured to acquire a first rainfall amount collected by a rain sensor on the vehicle during vehicle operation; a first prediction module, configured to predict a first water film thickness on the road where the vehicle is located based on the first rainfall amount; a vehicle speed determination module, configured to determine a target vehicle speed based on the first water film thickness, the vehicle's braking parameters and the system parameters of the vehicle's driving system; and a vehicle control module, configured to control the vehicle to drive at a speed limit based on the target vehicle speed.

[0014] According to another aspect of the embodiments of this application, a vehicle is also provided, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods in various embodiments of this application when it runs.

[0015] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0016] According to another aspect of the embodiments of this application, a computer program product is also provided, including a computer program that, when executed by a processor, implements the methods of various embodiments of this application.

[0017] According to another aspect of the embodiments of this application, a computer program product is also provided, including a non-volatile computer-readable storage medium storing a computer program that, when executed by a processor, implements the methods in various embodiments of this application.

[0018] According to another aspect of the embodiments of this application, a computer program is also provided, which, when executed by a processor, implements the methods of the various embodiments of this application.

[0019] In this embodiment, the method involves acquiring the first rainfall amount collected by a rain sensor on the vehicle during vehicle operation; predicting the first water film thickness on the road where the vehicle is located based on the first rainfall amount; determining the target vehicle speed based on the first water film thickness, the vehicle's braking parameters, and the system parameters of the vehicle's driving system; and controlling the vehicle's speed limit based on the target vehicle speed. By accurately predicting the first water film thickness of the current road segment based on the first rainfall amount collected by the rain sensor, and then combining the vehicle's inherent braking parameters and the system parameters of the driving system, the target vehicle speed that can ensure safe braking under the first water film thickness is derived, and the vehicle speed limit is controlled based on the target vehicle speed, rather than using a fixed threshold or a simple on / off strategy. This effectively avoids the problem of overly conservative or overly aggressive speed control due to inaccurate quantification of rainfall changes, and significantly improves the accuracy and robustness of the driving system's speed control in rainy conditions. This solves the technical problem of poor speed control accuracy in related technologies. Attached Figure Description

[0020] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0021] Figure 1 This is a flowchart of a vehicle speed control method according to an embodiment of this application;

[0022] Figure 2 This is a schematic diagram of an optional water film thickness prediction process according to an embodiment of this application;

[0023] Figure 3 This is a schematic diagram of the architecture of an optional initial prediction model according to an embodiment of this application;

[0024] Figure 4 This is a schematic diagram of the prediction process of an optional water film thickness prediction model according to an embodiment of this application;

[0025] Figure 5 This is a schematic diagram illustrating the construction process of an optional water film thickness prediction model according to an embodiment of this application;

[0026] Figure 6 This is a schematic diagram of a vehicle speed control device according to an embodiment of this application. Detailed Implementation

[0027] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0028] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0029] According to an embodiment of this application, an embodiment of a vehicle speed control method is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0030] This embodiment provides a vehicle speed control method. Figure 1 This is a flowchart of a vehicle speed control method according to an embodiment of this application, such as... Figure 1 As shown, the process includes the following steps:

[0031] Step S102: During the vehicle's operation, acquire the first rainfall data collected by the rain sensor on the vehicle.

[0032] The aforementioned rain sensor can be an optical detection device installed on the inside or outside of a vehicle's windshield. It can emit infrared light pulses and receive changes in the intensity of reflected light. Based on the scattering effect of raindrops on the light path, which causes the received signal to attenuate, it outputs an electrical signal related to the density and size of the raindrops. This signal is used to quantify the precipitation intensity per unit time on the windshield surface. The signal output by the rain sensor can be converted into a digital quantity by a signal processing circuit to reflect the instantaneous rainfall intensity at the location where the rain sensor is installed.

[0033] The aforementioned first rainfall can be a value representing the intensity of precipitation on the windshield surface, which is collected and output in real time by the aforementioned rain sensor under the current driving state of the vehicle. This value can be an estimated value of rainfall intensity calculated by the rain sensor based on the light intensity attenuation within the sampling period, and is used as an input parameter for the vehicle speed control strategy in the intelligent driving system. This value can change dynamically with the rainfall to reflect the direct impact of the current local environmental precipitation on the sensing area of ​​the rain sensor.

[0034] In one alternative embodiment, considering that the rain sensor can reflect the rainfall intensity on the windshield surface in real time, the vehicle speed control system (hereinafter referred to as the control system) can collect the first rainfall amount through the rain sensor on the vehicle during vehicle operation.

[0035] Specifically, the rain sensor's internal transmitter emits infrared light, which is directed at the vehicle's windshield at a preset angle. When the windshield surface is dry, most of the infrared light is reflected back and received by the rain sensor's receiver.

[0036] If raindrops fall on the windshield, they scatter the incident light, reducing the amount of light reflected back to the receiver. The greater the rainfall, the more numerous or larger the raindrops, the more the light is scattered, and the weaker the reflected light signal. Therefore, this rain sensor can determine the initial rainfall amount based on the intensity of the infrared light received by the receiver. This initial rainfall amount can then be used as an input variable to characterize the direct impact of the current rainfall environment on the rate of water film formation on the road surface, providing raw observational data for establishing a mapping relationship between the initial rainfall amount and the initial water film thickness.

[0037] For example, during vehicle operation, the onboard rain sensor can continuously output analog signals based on changes in optical reflection intensity. These signals are then converted from analog to digital and transmitted to the intelligent driving system. The intelligent driving system can read the raw digital data at a fixed sampling period and perform filtering to extract the value of the first rainfall at the current moment. Subsequently, the intelligent driving system can send the value of the first rainfall to the control system.

[0038] For example, during vehicle operation, the output signal of the rain sensor is sent to the control system in the form of periodic messages via the CAN (Controller Area Network) bus. The control system can parse the rain data field in the message and extract the preprocessed standardized rain value as the first rainfall.

[0039] For example, during vehicle operation, the raw light intensity differential signal from the rain sensor can be input to the edge computing module. This edge computing module can directly convert the light intensity change rate into a rainfall value through a pre-trained linear regression model. This rainfall value can be written into the shared memory area as the first rainfall amount for the control system to read and use in real time.

[0040] Step S104: Based on the first rainfall, predict the first water film thickness on the road where the vehicle is located.

[0041] The aforementioned first water film thickness can refer to the average vertical thickness of a continuous water layer formed by rainfall on the surface of the road where vehicles travel. The value of the first water film thickness can represent the height of the liquid water layer formed when the water accumulated on a unit area of ​​the road surface is evenly distributed under the action of gravity. This value can be determined by the first rainfall recorded by the rain sensor, as well as the rainfall intensity, rainfall duration, road material characteristics, road longitudinal slope and lateral drainage capacity, and can be obtained through actual measurement or model calculation, so as to accurately reflect the degree of road surface slipperiness.

[0042] In one optional embodiment, considering the nonlinear mapping relationship between the first rainfall collected by the rain sensor and the thickness of the water film on the road surface, the control system can use the first rainfall as input to a trained BP (Backpropagation) neural network model. The model then outputs the first water film thickness corresponding to the first rainfall to quantify the direct impact of rainfall on road surface slippage, thereby providing accurate physical state parameters for subsequent calculations of the adhesion coefficient and safe speed.

[0043] In another alternative embodiment, the control system can standardize the first rainfall output from the rain sensor and feed it into a linear regression model. The coefficients of the linear regression model can be obtained by least squares fitting of historical measured rainfall and road water film thickness dataset. The linear regression model can directly output the water film thickness value as the first water film thickness of the road where the vehicle is located.

[0044] In another alternative embodiment, the control system can send the first rainfall collected by the rain sensor to a lookup module. This lookup module can store an offline calibrated comparison table of rainfall and water film thickness. The data in this table can be constructed by collecting water film thickness and corresponding rainfall data from the actual vehicle under different rainfall intensities. The control system can use an interpolation algorithm to match the water film thickness value corresponding to the current first rainfall in the table and output the first water film thickness of the road where the vehicle is located.

[0045] For ease of understanding, Figure 2 This is a schematic diagram of an optional water film thickness prediction process according to an embodiment of this application, as shown below. Figure 2 As shown, firstly, the control system can input data into the prediction module, which can be the aforementioned first rainfall.

[0046] Subsequently, the prediction module can normalize the input data. After normalization, the prediction module can determine the BP neural network structure and initialize the neural network weights and thresholds.

[0047] The prediction module then iteratively obtains optimal weights and thresholds and determines whether the termination condition is met. If the termination condition is not met, the prediction module continues to obtain optimal weights and thresholds until the termination condition is met.

[0048] Finally, the prediction module can use a BP neural network structure containing the optimal weights and thresholds to make predictions and output the prediction results, thereby obtaining the aforementioned first water film thickness.

[0049] Step S106: Based on the first water film thickness, the vehicle's braking parameters, and the system parameters of the vehicle's driving system, determine the vehicle's target speed.

[0050] The aforementioned braking parameters can refer to a set of physical and engineering parameters that affect the braking performance of a vehicle. These parameters may include, but are not limited to, the efficiency coefficient of the braking system, the coefficient of friction of the brake, the braking response delay time, the maximum static friction coefficient between the tire and the road surface, the vehicle mass, the thermal fade characteristics of the brake, the pressure transmission characteristics of the brake lines, and the maximum braking force output capacity of the brake. These parameters can collectively determine the distance and time required for the vehicle to decelerate from its current speed to a standstill under the first water film thickness.

[0051] The aforementioned driving system can refer to the integrated hardware and software system in a vehicle used to perform autonomous driving functions. It may include, but does not require, environmental perception sensors (such as rain sensors, cameras, millimeter-wave radar), decision control units, actuators (such as electronic stability programs, electronic power steering, and electronic braking systems), high-precision positioning modules, path planning algorithm modules, and speed control modules. This driving system can adjust the longitudinal and lateral movements of the vehicle to achieve functions such as adaptive cruise control and lane keeping.

[0052] The aforementioned system parameters can refer to control constants and algorithm parameters that are pre-calibrated or configured in real time in the driving system. These parameters may include, but are not limited to, target braking distance safety margin, system response time threshold, maximum permissible deceleration limit, speed adjustment step size, rain and water film conversion coefficient, adhesion coefficient and speed mapping curve slope, filter time constant, sensor noise tolerance, dynamic weighting coefficient, etc. These parameters can be used to constrain the driving system's response behavior to environmental changes, thereby enhancing the driving system's ability to respond to environmental changes.

[0053] The aforementioned target speed can refer to the longitudinal speed value calculated by the control system under the conditions of the first water film thickness, the vehicle's braking parameters, and system parameters, which allows the vehicle to continue driving while maintaining a safe braking distance. This speed value can ensure that the vehicle can still complete emergency braking without losing control or colliding under rain conditions.

[0054] In one alternative embodiment, the thickness of the first water film is considered to significantly affect the road surface adhesion coefficient of the road on which the vehicle travels, thereby changing the vehicle's braking distance.

[0055] Based on this, the control system can determine the vehicle speed at which a safe stop can be guaranteed under the first water film thickness by combining the vehicle's braking parameters such as braking coefficient and braking response time, as well as the driving system's system parameters such as safe distance setting and system response delay, through the pre-built correlation between the first water film thickness, braking parameters, and system parameters, thereby determining the vehicle's target speed.

[0056] For example, the control system can input the thickness of the first water film into a pre-built braking distance model. This braking distance model can combine system parameters such as the vehicle's inherent braking coefficient, the response time of the driving system, and the safety distance to solve for the upper limit of vehicle speed that meets the braking distance constraint, and can use this upper limit of vehicle speed as the target vehicle speed.

[0057] For example, the control system can substitute the thickness of the first water film, the upper limit of the deceleration capability of the vehicle braking system, the response delay time of the driving system, and the preset safety margin into the dynamic braking distance equation to solve for the upper limit of vehicle speed, and can use the solved upper limit of vehicle speed as the target vehicle speed.

[0058] For example, the control system can combine the first water film thickness, vehicle mass, tire contact area, ABS (Antilock Braking System) control threshold and the longitudinal control bandwidth of the intelligent driving system, and use linear programming to solve the upper limit of cruise speed that will not trigger brake failure under the condition that the current road surface adhesion coefficient decreases, and can use the upper limit of cruise speed as the target speed.

[0059] Step S108: Control the vehicle to travel at a speed limit based on the target vehicle speed.

[0060] In one alternative embodiment, the target vehicle speed is considered to be a safe speed constructed by the control system based on the first water film thickness of the road where the vehicle is currently traveling, as well as the vehicle's own braking capability and the responsiveness of the driving system. This speed can ensure that the vehicle can still drive safely in the current rainy environment.

[0061] Based on this, the control system can perform vehicle speed limit control based on the calculated target vehicle speed, so that the vehicle can drive at a limited speed, thereby ensuring that the driving system can still maintain braking safety and driving continuity under dynamic rain conditions.

[0062] In this embodiment, the method involves acquiring the first rainfall amount collected by a rain sensor on the vehicle during vehicle operation; predicting the first water film thickness on the road where the vehicle is located based on the first rainfall amount; determining the target vehicle speed based on the first water film thickness, the vehicle's braking parameters, and the system parameters of the vehicle's driving system; and controlling the vehicle's speed limit based on the target vehicle speed. By accurately predicting the first water film thickness of the current road segment based on the first rainfall amount collected by the rain sensor, and then combining the vehicle's inherent braking parameters and the system parameters of the driving system, the target vehicle speed that can ensure safe braking under the first water film thickness is derived, and the vehicle speed limit is controlled based on the target vehicle speed, rather than using a fixed threshold or a simple on / off strategy. This effectively avoids the problem of overly conservative or overly aggressive speed control due to inaccurate quantification of rainfall changes, and significantly improves the accuracy and robustness of the driving system's speed control in rainy conditions. This solves the technical problem of poor speed control accuracy in related technologies.

[0063] Further, based on the first rainfall, the first water film thickness of the road where the vehicle is located is predicted, including: inputting the first rainfall into the water film thickness prediction model, and using the water film thickness prediction model to predict the first water film thickness, wherein the water film thickness prediction model is trained using the second water film thickness and the second rainfall measured during the process of the vehicle driving on the preset road under rainfall conditions.

[0064] The aforementioned water film thickness prediction model can be a model built based on machine learning methods. The input of the model can be real-time rainfall intensity data collected by a rain gauge, and the output of the model can be an estimated value of the water film thickness formed on the corresponding road surface. The model can be trained using historical data. To ensure the accuracy and reliability of the training process, the training data can include the second water film thickness and the second rainfall output by the rain gauge, which are recorded synchronously during the driving of a vehicle on a preset road under real rainfall conditions. The structure of the model can adopt a BP neural network.

[0065] The aforementioned second water film thickness can be the actual thickness of the water film on the surface of the vehicle's driving path, obtained directly from on-site measurement equipment during the training phase of the water film thickness prediction model. The measurement method can be to quantify and record the vertical thickness of the road surface water layer at different sampling points using non-contact optical sensors or laser rangefinders. This value can be used as one of the training data for the water film thickness prediction model and can reflect the water accumulation situation under real road conditions.

[0066] The aforementioned second rainfall can be rainfall intensity data collected synchronously with the second water film thickness during the training phase of the water film thickness prediction model. This data is output by a rain sensor installed on the vehicle's windshield and can be calculated based on the change in the reflectivity of infrared light on the glass surface. It is used to describe the amount of precipitation per unit area per unit time. This data can be used as one of the training data for the water film thickness prediction model and combined with the second water film thickness to establish a mapping relationship between rainfall and water film thickness.

[0067] In one alternative embodiment, considering that traditional water film thickness measurement methods directly estimate water film thickness using fixed thresholds or empirical formulas without taking into account the nonlinear dynamic relationship between actual rainfall intensity and road surface water film thickness, resulting in significant prediction errors under different rainfall intensities, road surface materials, and temperature conditions, which in turn affects the accuracy of subsequent adhesion coefficient and safe vehicle speed calculations.

[0068] Based on this, the control system can collect data on the second water film thickness and the second rainfall while the vehicle is driving on a preset road under real rainfall conditions, and construct a water film thickness prediction model with rainfall as input and water film thickness as output. The measured second water film thickness and the second rainfall can be used as training datasets to supervise the training of the water film thickness prediction model, so that the model can learn the complex mapping relationship between rainfall and water film thickness.

[0069] In practical applications, the control system can directly input the first rainfall collected by the rain sensor into the water film thickness prediction model. The water film thickness prediction model can then output the first water film thickness that matches the actual rainfall conditions, improving the adaptability and accuracy of the prediction, and thus enhancing the accuracy of the control decisions of the subsequent control system.

[0070] For example, the control system can input the first rainfall amount into a pre-trained water film thickness prediction model. This model can map the nonlinear relationship between the first rainfall amount and the first water film thickness through its internal model parameters, thereby outputting the first water film thickness corresponding to the first rainfall amount.

[0071] The model can be trained using a backpropagation neural network structure based on the second rainfall data collected when the vehicle is driving on a preset road and the second water film thickness data measured simultaneously. The input of the model is the normalized rainfall value output by the rain sensor, and the output is the measured water film thickness value. During the training process, the network weights of the model can be adjusted by minimizing the mean squared error (MSE) until the prediction performance of the model reaches a better state for actual water film thickness prediction.

[0072] For example, the control system can use the first rainfall as an input feature to feed into the water film thickness prediction model. Based on the learned statistical correlation between rainfall and water film thickness, the model can perform forward propagation calculation and output the predicted value of the road water film thickness corresponding to the first rainfall as the first water film thickness. The weights and bias parameters of the model are determined during the training phase by multiple sets of rainfall and measured water film thickness data to reduce the prediction error.

[0073] For example, the control system can input the first rainfall amount into a water film thickness prediction model built using machine learning methods. This model can employ support vector regression or random forest regression structures, and be trained by fitting the model with historically collected second rainfall amounts and corresponding second water film thickness data. After training, the model can directly output continuous predicted values ​​of road water film thickness based on the input first rainfall amount, thus obtaining the first water film thickness. The entire prediction process does not rely on the derivation of physical formulas; it establishes the mapping relationship between rainfall and water film thickness solely through a data-driven approach.

[0074] Furthermore, the method also includes: obtaining the second water film thickness and the second rainfall; constructing an initial prediction model; inputting the second rainfall into the initial prediction model and using the initial prediction model to predict the third water film thickness; and adjusting the model parameters of the initial prediction model based on the second water film thickness and the third water film thickness to obtain a water film thickness prediction model.

[0075] The aforementioned initial prediction model can be a model built based on the output signal of the rain sensor and historical water film thickness data. It is used to preliminarily estimate the water film thickness on the road surface based on the current rainfall input when there is no measured water film thickness. The structure of the model can be linear regression, multinomial fitting or neural network. The parameters of the model have not been calibrated with measured data and only have preliminary mapping ability.

[0076] The aforementioned third water film thickness can be a predicted value output by the initial prediction model based on the input second rainfall. It represents the initial prediction model's estimate of the road surface water film thickness under the current rainfall conditions. This predicted value has not been verified by actual measurements and is only used as an intermediate output within the initial prediction model for subsequent parameter adjustments.

[0077] The aforementioned model parameters can be adjustable variables used in the initial prediction model to define the relationship between input and output. These can include, but are not limited to, weight coefficients, bias terms, neural network connection weights, activation function thresholds, or regression equation coefficients. The values ​​of these model parameters can determine the prediction accuracy of the initial prediction model. The model can be iteratively adjusted by comparing the error between the predicted output of the initial prediction model and the measured thickness of the second water film, so that the output of the initial prediction model approaches the true value.

[0078] In one alternative embodiment, considering that directly using the first rainfall and the first water film thickness data to train the water film thickness prediction model without introducing independent validation samples, the water film thickness prediction model may overfit on the training set, resulting in a decrease in generalization ability and difficulty in accurately reflecting the nonlinear mapping relationship between rainfall and water film thickness under actual road conditions, thereby reducing the reliability of water film thickness prediction.

[0079] Based on this, the control system can construct an initial prediction model by acquiring independently collected data on the second water film thickness and the second rainfall, and use the second water film thickness and the second rainfall to make forward predictions to obtain the third water film thickness.

[0080] Subsequently, the control system can use the second water film thickness as the true value to calculate the error between the second water film thickness and the third water film thickness. Then, it can adjust the model parameters such as the weights and bias parameters of the initial prediction model through the backpropagation algorithm, thereby minimizing the loss function value of the initial prediction model and completing the parameter adjustment of the initial prediction model to obtain a water film thickness prediction model with actual generalization ability.

[0081] For example, the control system can first obtain the second water film thickness and the second rainfall. The second water film thickness can be obtained by road surface sampling equipment at different rainfall periods, while the second rainfall can be collected synchronously by the vehicle-mounted rain sensor.

[0082] After obtaining the second water film thickness and the second rainfall, the control system can construct an initial prediction model based on these data. This model can be a BP neural network containing an input layer, a single hidden layer, and an output layer.

[0083] Subsequently, the control system can use the second rainfall as input data to the initial prediction model, which can then calculate and output the third water film thickness through forward propagation.

[0084] Then, the control system can calculate the mean square error based on the second and third water film thicknesses, use the backpropagation algorithm to update the weights and bias parameters of the hidden layer, and iteratively train the initial prediction model until the error converges, thus obtaining the final water film thickness prediction model.

[0085] For example, after the control system obtains the second water film thickness and the second rainfall, the initial prediction model can predict the third water film thickness based on the second rainfall. At this point, the control system can minimize the absolute error function between the third and second water film thicknesses, and use the Adam (Adaptive Moment Estimation) algorithm to update all trainable parameters of the initial prediction model in batches. The update frequency is p iterations, and training stops after the cumulative number of updates reaches q, thus forming a water film thickness prediction model.

[0086] For example, the control system can input the second rainfall as data into the initial prediction model, which then outputs the third water film thickness using a pre-defined parameter structure. The control system can then compare the difference between the third and second water film thicknesses and update the weights and bias parameters of the initial prediction model using gradients based on this difference. This gradient update process can be repeated until the initial prediction model converges, resulting in the final water film thickness prediction model.

[0087] For ease of understanding, Figure 3 This is a schematic diagram of the architecture of an optional initial prediction model according to an embodiment of this application, such as... Figure 3 As shown, the initial prediction model consists of an input layer, a hidden layer, and an output layer.

[0088] During training, the control system can perform a forward propagation process from the input layer to the output layer to obtain the third water film thickness. After comparing the second and third water film thicknesses and obtaining the prediction error, the control system can perform a backward propagation process from the output layer to the input layer based on the prediction error to iteratively adjust the model parameters of the initial prediction model, thereby obtaining the water film thickness prediction model.

[0089] Figure 4 This is a schematic diagram of the prediction process of an optional water film thickness prediction model according to an embodiment of this application, as shown below.Figure 4 As shown, each neuron in the input layer of the water film thickness prediction model corresponds to an input feature. For example, these input features may include construction depth, slope, rainfall intensity, and drainage length. After these input features are processed by the hidden layer and the output layer, the water film thickness can be output.

[0090] Figure 5 This is a schematic diagram illustrating the construction process of an optional water film thickness prediction model according to an embodiment of this application, as shown below. Figure 5 As shown, the process mainly consists of input and output layer design, hidden layer design, and model parameter setting. In the input and output layer design process, the input and output layers are constructed separately, with the input layer containing four neurons and the output layer containing one neuron. Subsequently, the control system can normalize these neurons.

[0091] The hidden layer design process includes determining the number of hidden layers and the range of the number of neurons. Subsequently, the control system can select different candidate neuron numbers from this range based on empirical formulas to form different initial models for testing. For each initial model, R (correlation coefficient) and MSE are calculated, and finally, the optimal number of hidden layer neurons is determined.

[0092] After completing the design of the input layer, output layer, and hidden layer, the control system can construct an initial prediction model based on the input, output, and hidden layers, and then set the model parameters. This process can be completed through multiple iterative training iterations. During iterative training, the control system can determine the number of iterations, the learning rate, and the error threshold to keep the entire iterative training process under control. Finally, after completing the iterative training, a water film thickness prediction model can be obtained.

[0093] Furthermore, in the process of constructing the water film thickness prediction model, a suitable activation function needs to be selected in the BP neural network. This activation function can transform the linear function of the input signal into a nonlinear function, thereby enabling error backpropagation and thus adjusting the parameters of the model.

[0094] The three commonly used activation functions in BP neural networks are log sig (Logistic Sigmoid Function), tan sig (Tangent Sigmoid Function), and purelin (Linear Pure Function). The purelin function can be used in the output layer. The output value of this function can be any value, which is suitable for regression problems and can provide predictions for continuous values.

[0095] Furthermore, obtaining the second water film thickness and the second rainfall includes: constructing a target road based on the preset construction depth, preset road slope, and preset road width of the road surface material; controlling the target road to be under rainfall conditions based on the preset rainfall intensity, and controlling vehicles to drive on the target road; collecting the water film thickness of the target road using data acquisition equipment to obtain the second water film thickness; and collecting rainfall using a rain sensor on the vehicle to obtain the second rainfall.

[0096] The aforementioned pavement material refers to engineering materials used for road surface paving. It can be composed of asphalt binder, graded crushed stone, mineral powder and additives, and has anti-skid, anti-rutting and drainage properties. This pavement material can directly affect the retention and diffusion characteristics of rainwater on the road surface.

[0097] The aforementioned target road can refer to a standardized road segment selected to meet the requirements of the rainfall and water film thickness data collection experiment. The characteristics of the road may include, but are not limited to, the width of the road on one side, the longitudinal slope, and the pavement material. The road can be without intersections, traffic signals, or frequent passage of large vehicles to ensure that the water film formation process under rainfall conditions is not disturbed by external factors and has repeatability and spatial homogeneity.

[0098] The aforementioned data acquisition equipment can refer to non-contact or contact measuring instruments used to directly measure the thickness of water film on the road surface. These instruments may include, but are not limited to, laser rangefinders, optical reflective water film thickness gauges, or ultrasonic water film sensors. The data acquisition equipment can be installed on a fixed bracket or mobile measuring vehicle beside the road, vertically aligned with the road surface, and calculates the water film thickness based on the optical path difference or sound wave propagation time difference by transmitting and receiving reflected signals.

[0099] In one alternative embodiment, if the output signal of the rain sensor is directly used as the input feature for the water film thickness without physical consistency control between the sensor's environment and the actual road surface water film formation conditions, the rainfall in the training dataset and the actual road surface water film thickness will lack spatial and temporal synchronization. Furthermore, this process does not limit key parameters such as road surface construction depth, slope, and road width, causing the collected data to be affected by factors such as local water accumulation, uneven drainage, and differences in lane cross slope, introducing non-rainfall-related noise and reducing the generalization ability of the initial prediction model.

[0100] Based on this, the control system can determine the standardized target road for training data collection by setting the construction depth, road slope and single-side road width that conform to the material standards of highways, respectively. This ensures that the pavement geometry and material properties of the target road are consistent with the actual working conditions, and that the water film formation process is stable and repeatable under controllable rainfall intensity.

[0101] After the target road is determined, the control system can use a high-precision water film thickness sensor deployed on the target road to synchronously collect the actual water film thickness as the aforementioned second water film thickness.

[0102] Meanwhile, the control system can record optical rainfall signals at the same time and on the same road segment using an onboard rain sensor, which can be used as the second rainfall data. This enables a strict correspondence between rainfall input and water film output in terms of spatial location, timestamp, and environmental conditions. Consequently, a training dataset with physical consistency and environmental controllability can be constructed, eliminating the interference of non-rainfall factors on the input-output relationship of the initial prediction model and improving the accuracy and transferability of the final water film thickness prediction model trained by the initial prediction model.

[0103] For example, the control system can lay the target road based on a preset construction depth, preset road slope, and preset road width. Subsequently, a rainfall simulation system can be activated to continuously induce rainfall on the target road according to a preset rainfall intensity, maintaining uniform rainfall.

[0104] During rainfall, vehicles can be controlled to travel at a constant speed on the target road, ensuring that there is no braking or acceleration during data collection. While the vehicle is in motion, the water film thickness on the target road surface can be acquired in real time using water film thickness data acquisition equipment installed on the target road, thus obtaining a second water film thickness.

[0105] At the same time, it can also read the infrared light intensity change signal of the vehicle's windshield rain sensor. After processing, this signal can be output as the corresponding rainfall value, which can be used as a second rainfall value.

[0106] For example, a sprinkler system can be used to simulate natural rainfall. The vehicle travels at a constant speed on the target road, and the control system can collect water film thickness data through a capacitive water film thickness sensor array installed on the road surface. After filtering and interpolation, the data is output as a second water film thickness.

[0107] In addition, the control system can also simultaneously collect the light intensity attenuation rate output by the vehicle's rain sensor within the same sampling period, and convert it into a rainfall value after calibration to obtain a second rainfall value.

[0108] Furthermore, an initial prediction model is constructed, including: determining the number of first neurons in the input layer, the number of second neurons in the output layer, and the number of hidden layers; determining the number of third neurons in the hidden layer based on the number of first and second neurons; and constructing the initial prediction model based on the number of first neurons, the number of second neurons, the number of third neurons, and the number of layers.

[0109] The number of the first neurons mentioned above can be the number of neurons in the input layer. Each neuron can correspond to an input feature variable, which is used to receive and transmit the raw rainfall data collected by the rain sensor or other preprocessed feature values. The number of the first neurons can be equal to the dimension of the input data.

[0110] The number of the second neurons mentioned above can be the number of neurons in the output layer. Each neuron can correspond to a predicted output result, that is, the value of the thickness of the water film on the road surface. The number of the second neurons can be equal to the dimension of the target variable to be predicted.

[0111] The number of the third neuron mentioned above can be the number of neurons in the hidden layer, that is, the total number of neurons contained in each hidden layer in the non-output layer between the input layer and the output layer. It can be used to perform nonlinear transformation and feature abstraction on the input features. The number of the third neuron can be determined by empirical formulas or experimental tuning. It can affect the expressive power and generalization performance of the initial prediction model. The number of the third neuron can be an integer greater than the input and output dimensions. The value range can be calculated based on the number of input and output nodes and empirical coefficients.

[0112] In an alternative embodiment, it is considered that if the number of layers and nodes of the neural network is set directly using a fixed structure or empirical rules without systematically configuring it based on the mathematical relationship between the input and output dimensions, the capacity of the initial prediction model will not match the complexity of the problem, resulting in the output of the model being underfitted or overfitted, thereby reducing the model's generalization ability from the mapping of rainfall reduction to water film thickness.

[0113] Based on this, the control system can construct an initial prediction model by setting the number of the first neurons in the input layer as the feature dimension of the rain sensor output, setting the number of the second neurons in the output layer as the prediction target dimension of the water film thickness, determining the number of the third neurons in the hidden layer according to empirical formulas, and then combining the single hidden layer structure. This ensures that the structure of the initial prediction model strictly corresponds to the data dimension, guaranteeing that the parameter space of the initial prediction model is within the trainable range and has sufficient expressive power.

[0114] For example, we can assume that the number of neurons in the first input layer is 'a', corresponding to the dimension of the original signal features output by the rain sensor. We can assume that the number of neurons in the second output layer is 'b', corresponding to the dimension of the predicted value of the road surface water film thickness. Meanwhile, we can set the number of hidden layers to 'c'.

[0115] Based on empirical formulas, the number of neurons in the first layer (a) and the number of neurons in the second layer (b) can be used to calculate the number of neurons in the third layer (d). The value of d can be an integer within the interval [a+b+1, a+b+10]. The control system can then select a value of d that minimizes the mean square error of the training set as the final number of neurons within this range.

[0116] Finally, the control system can construct a feedforward BP neural network structure with a hidden layer based on the parameter configuration of a, b, d, and c, thus forming an initial prediction model.

[0117] For example, when constructing the initial prediction model, the control system can set the number of neurons in the first input layer to 'a', the number of neurons in the second output layer to 'b', and the number of hidden layers to 'c'. Subsequently, the control system can set the number of neurons in the third hidden layer to 'd', which is the floor value of (a+b) / 2, based on the arithmetic mean of a and b.

[0118] The control system can construct a deep feedforward network with a hidden layer c based on a, b, d, and c. Each hidden layer is fully connected to the previous and next layers to form an initial prediction model.

[0119] Further, based on the number of the first and second neurons, the number of the third neurons in the hidden layer is determined, including: determining a range of the number of neurons based on the number of the first and second neurons, wherein the range of the number of neurons includes multiple candidate neuron numbers; determining the prediction error of the initial model corresponding to the multiple candidate neuron numbers; and determining the number of the third neurons from the multiple candidate neuron numbers based on the prediction error.

[0120] The aforementioned range of the number of neurons can refer to the closed interval or discrete set of values ​​that allow the number of neurons in the hidden layer to take the form of the first neuron in the input layer and the second neuron in the output layer when constructing the hidden layer of the initial prediction model, based on empirical formulas or engineering experience. This range can define the upper and lower limits of the number of selectable neurons in the hidden layer, ensuring that the network structure of the hidden layer maintains a reasonable balance between computational complexity and expressive power.

[0121] The aforementioned number of candidate neurons can refer to a number of discrete integer values ​​that are set manually or generated by the system within the range of the number of neurons. Each value can represent a hidden layer neuron configuration scheme to be tested, which is used to compare the impact of different hidden layer structures on model performance through training and validation, so as to accurately determine the number of third neurons.

[0122] The aforementioned prediction error can refer to the numerical deviation between the predicted value output by the initial model and the corresponding true label value under given input data. It can be expressed by quantitative indicators such as mean squared error or mean absolute error, reflecting the fitting accuracy of the initial model to the training set or validation set data. The smaller the error, the stronger the prediction ability of the initial model.

[0123] In one alternative embodiment, considering that the number of third neurons in the hidden layer directly affects the expressive power and generalization performance of the initial prediction model, too many third neurons can lead to overfitting, while too few can not fit complex nonlinear relationships.

[0124] Based on this, the control system can calculate a reasonable range of the number of neurons by using empirical formulas, based on the number of the first neurons in the input layer and the number of the second neurons in the output layer. This range can include multiple candidate neuron numbers to cover possible optimal hidden layer structures.

[0125] Subsequently, the control system can construct an initial model corresponding to each number of candidate neurons, and train the initial model using the same training set and validation set. The mean square error of the initial model on the validation set is calculated as the prediction error. The smaller the prediction error, the stronger the modeling ability of the initial model to model the input-output mapping.

[0126] Therefore, the control system can compare the prediction errors of the initial model corresponding to the number of candidate neurons and select the number of candidate neurons with smaller errors as the number of third neurons, thereby ensuring that the hidden layer structure achieves better fitting performance under the current data distribution.

[0127] For example, the control system can calculate a range of neuron counts based on the number of the first neuron in the input layer and the number of the second neuron in the output layer, using an empirical formula. This range includes multiple candidate neuron counts. For each candidate neuron count, the control system can construct a corresponding initial backpropagation (BP) neural network model and train it using a normalized dataset of rainfall and water film thickness to obtain the predicted outputs of each model.

[0128] Then, the control system can calculate the mean square error between the predicted output of each model and the measured water film thickness, as the prediction error. By comparing the prediction errors corresponding to the number of candidate neurons, the extended system can select the number of candidate neurons with smaller prediction errors as the number of third neurons in the hidden layer.

[0129] For example, an empirical formula can be shown as follows:

[0130] ;

[0131] ;

[0132] in, The number of hidden layer nodes. The number of nodes in the input layer. This represents the number of nodes in the output layer. The value is an integer from 1 to 10. Based on this formula, the control system can calculate the mean square error of the neural network with the number of hidden layer nodes within a range of the number of neurons, and can use the mean square error of the neural network as a performance evaluation parameter for selecting the number of nodes.

[0133] Furthermore, based on the first water film thickness, the vehicle's braking parameters, and the system parameters of the vehicle's driving system, the target vehicle speed is determined, including: inputting the braking parameters, system parameters, and the first water film thickness into the vehicle speed prediction model, and using the vehicle speed prediction model to predict the target vehicle speed. The vehicle speed prediction model is constructed based on the vehicle's braking distance model and a preset relationship model. The preset relationship model is used to characterize the correlation between the first water film thickness, the road surface adhesion coefficient of the target road, and the target vehicle speed.

[0134] The aforementioned vehicle speed prediction model can be a computational model built based on mathematical modeling and data-driven methods. The inputs to the model can be the first water film thickness of the current road, the vehicle's braking parameters, and the system parameters of the driving system. The output of the model can be the cruise speed that can be safely executed under the current environmental conditions.

[0135] This model can be trained by fusing physical equations and measured data. Its internal structure can include a braking distance model and a preset relationship model, which are used to calculate the target vehicle speed allowed by the control system in real time under a given environmental disturbance, so as to ensure that the braking distance under the current environmental disturbance does not exceed the safety threshold.

[0136] The aforementioned braking distance model can be a model describing the distance a vehicle travels from the time the braking command takes effect until it comes to a complete stop. This model can be constructed based on Newton's laws of motion and the principle of energy conservation to determine the safe braking distance required for a vehicle to travel at different speeds under different road surface adhesion coefficients.

[0137] The aforementioned pre-defined relationship model can refer to a model established through fitting experimental data or theoretical derivation, which can be used to characterize the deterministic mathematical dependency relationship between road water film thickness, road surface adhesion coefficient, and vehicle safe driving speed.

[0138] In one alternative embodiment, it is considered that if a fixed speed limit is used, or if the speed is limited only based on the discrete levels output by the rain sensor without quantifying the continuous influence of the road surface water film thickness on the road surface adhesion coefficient, the speed setting will be too conservative under moderate rainfall conditions or the speed will still be high under heavy rainfall, thereby reducing the accuracy of the target speed determination, and thus reducing driving efficiency or increasing the risk of skidding.

[0139] Based on this, the control system can input the first water film thickness, the vehicle's braking parameters, and the driving system's system parameters into the vehicle speed prediction model, and use the vehicle speed prediction model to predict the target vehicle speed. This model can be constructed based on the braking distance model and a preset relationship model.

[0140] The braking distance model can be defined by the braking coefficient, gradient, driving system response time, and safe distance. The preset relationship model can be constructed from a nonlinear function expression between the water film thickness, road surface adhesion coefficient, and target vehicle speed, obtained by fitting experimental data. This allows the upper limit of the safe vehicle speed under a given braking distance constraint to be derived by using the road surface adhesion coefficient as an input parameter. Furthermore, the target vehicle speed can be continuously and dynamically predicted based on the actual water film thickness, ensuring that the speed limit matches real road conditions and avoiding braking performance evaluation deviations caused by neglecting the attenuation effect of water film thickness on the road surface adhesion coefficient.

[0141] Furthermore, the target vehicle speed is predicted using a vehicle speed prediction model, including: inputting braking parameters and system parameters into a braking distance model to obtain a first correlation between the target vehicle speed and the road surface adhesion coefficient of the target road, wherein the braking distance model is used to characterize the correlation between braking parameters, system parameters, target vehicle speed, and road surface adhesion coefficient; inputting a first water film thickness into a preset relationship model to obtain a second correlation between the target vehicle speed and the road surface adhesion coefficient of the target road, wherein the preset relationship model is used to characterize the correlation between the first water film thickness, road surface adhesion coefficient, and target vehicle speed; and determining the target vehicle speed based on the first and second correlations.

[0142] The aforementioned first correlation can be a mathematical mapping relationship between the target vehicle speed and the road surface adhesion coefficient established through a braking distance model. It is used to describe the constraint law of the target vehicle speed changing with the road surface adhesion coefficient under given braking parameters and system parameters.

[0143] The aforementioned second correlation can be a nonlinear functional relationship between the road surface adhesion coefficient and the target vehicle speed established through a preset relationship model. This relationship is used to characterize the regulatory effect of water film thickness changes on the road surface adhesion coefficient and the limiting effect of water film thickness on safe vehicle speed.

[0144] In one alternative embodiment, it is considered that if a fixed speed limit or a simple linear mapping based on the output of a rain sensor is used to determine the target speed without considering the nonlinear effect of the actual road surface water film thickness on the adhesion coefficient, the speed setting will be out of sync with the actual road safety conditions when the rainfall changes but the water film thickness is not updated synchronously. This may result in the braking distance exceeding the acceptable range or excessive speed restriction reducing traffic efficiency.

[0145] Based on this, the control system can establish the first correlation between the target vehicle speed and the road adhesion coefficient by inputting braking parameters and system parameters into the braking distance model. This model can accurately characterize the physical relationship that the vehicle speed is limited as the adhesion coefficient decreases under a given braking distance constraint, based on the vehicle dynamics equations and braking performance parameters.

[0146] Meanwhile, the control system can input the first water film thickness, which is trained by the rain sensor and measured data, into a preset relationship model. This model can be obtained by fitting experimental data, characterizing the empirical functional relationship between the water film thickness and the road surface adhesion coefficient. Furthermore, it can derive the second correlation between the target vehicle speed and the road surface adhesion coefficient through the braking constraint relationship between the road surface adhesion coefficient and the vehicle speed.

[0147] Ultimately, the control system can jointly solve the upper limit of safe vehicle speed set by the braking system in the first correlation relationship and the vehicle speed corresponding to the available adhesion coefficient determined by the actual road water film thickness in the second correlation relationship. By calculating the intersection, the system determines the target vehicle speed that meets the braking safety constraints and is adapted to the current road conditions, thereby achieving physical consistency between dynamic vehicle speed adjustment and actual road adhesion capability.

[0148] For example, under the control of an intelligent driving system, the vehicle's stopping distance S can be composed of three parts: the distance traveled S1 within the response time of the intelligent driving linkage, the braking distance S2, and the safety distance S3. The calculation process for S1 can be shown in the following formula:

[0149] ;

[0150] in, Indicates the target speed. S2 represents the response time of the intelligent driving system, and 3.6 represents the unit conversion factor. The calculation process of S2 can be shown in the following formula:

[0151] ;

[0152] in, Indicates the braking coefficient. This indicates the gradient of a road section. Generally, the gradient of highways is relatively small and can be ignored. This represents the road surface adhesion coefficient, with 254 being the unit conversion constant. The calculation process of S2 can serve as the braking distance model described above. Furthermore, since there is a certain mathematical relationship between the water film thickness, the road surface adhesion coefficient, and the target speed, this mathematical relationship is the aforementioned preset correlation.

[0153] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0154] According to an embodiment of this application, a vehicle speed control device is provided. It should be noted that this device can be used to execute the aforementioned vehicle speed control method. The specific implementation process and application scenarios are the same as those in the above embodiments, and will not be repeated here. Figure 6 This is a schematic diagram of a vehicle speed control device according to an embodiment of this application, such as... Figure 6 As shown, the device includes:

[0155] The first acquisition module 602 is used to acquire the first rainfall collected by the rain sensor on the vehicle during the vehicle's operation.

[0156] The first prediction module 604 is used to predict the first water film thickness on the road where the vehicle is located based on the first rainfall.

[0157] The vehicle speed determination module 606 is used to determine the target vehicle speed based on the first water film thickness, the vehicle's braking parameters, and the system parameters of the vehicle's driving system.

[0158] The vehicle control module 608 is used to control the vehicle to travel at a speed limit based on the target vehicle speed.

[0159] Furthermore, the first prediction module is also used to: input the first rainfall into the water film thickness prediction model, and use the water film thickness prediction model to predict the first water film thickness, wherein the water film thickness prediction model is trained using the second water film thickness and the second rainfall measured during the process of a vehicle driving on a preset road under rainfall conditions.

[0160] Furthermore, the device also includes: a second acquisition module for acquiring the second water film thickness and the second rainfall; a model building module for building an initial prediction model; a second prediction module for inputting the second rainfall into the initial prediction model and using the initial prediction model to predict the third water film thickness; and a parameter adjustment module for adjusting the model parameters of the initial prediction model based on the second water film thickness and the third water film thickness to obtain a water film thickness prediction model.

[0161] Furthermore, the second acquisition module is also used to: construct a target road based on the preset construction depth, preset road slope and preset road width of the road surface material; control the target road to be under rainfall conditions based on the preset rainfall intensity and control the vehicle to drive on the target road; collect the water film thickness of the target road using data acquisition equipment to obtain the first water film thickness; and collect the rainfall using the rain sensor on the vehicle to obtain the second rainfall.

[0162] Furthermore, the model building module is also used to: determine the number of first neurons in the input layer, the number of second neurons in the output layer, and the number of hidden layers; determine the number of third neurons in the hidden layer based on the number of first neurons and second neurons; and build an initial prediction model based on the number of first neurons, second neurons, third neurons, and the number of layers.

[0163] Furthermore, the model building module is also used to: determine the range of the number of neurons based on the number of the first and second neurons, wherein the range of the number of neurons includes multiple candidate neuron numbers; determine the prediction error of the initial model corresponding to the multiple candidate neuron numbers; and determine the number of the third neuron from the multiple candidate neuron numbers based on the prediction error.

[0164] Furthermore, the vehicle speed determination module is also used to: input braking parameters, system parameters and the first water film thickness into the vehicle speed prediction model, and use the vehicle speed prediction model to predict the target vehicle speed. The vehicle speed prediction model is constructed based on the vehicle's braking distance model and a preset relationship model. The preset relationship model is used to characterize the correlation between the first water film thickness, the road surface adhesion coefficient of the target road and the target vehicle speed.

[0165] Furthermore, the vehicle speed determination module is also used to: input braking parameters and system parameters into a braking distance model to obtain a first correlation between the target vehicle speed and the road surface adhesion coefficient of the target road, wherein the braking distance model is used to characterize the correlation between braking parameters, system parameters, target vehicle speed and road surface adhesion coefficient; input a first water film thickness into a preset relationship model to obtain a second correlation between the target vehicle speed and the road surface adhesion coefficient of the target road, wherein the preset relationship model is used to characterize the correlation between the first water film thickness, road surface adhesion coefficient and target vehicle speed; and determine the target vehicle speed based on the first correlation and the second correlation.

[0166] Embodiments of this application also provide a vehicle, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods described in various embodiments of this application when it runs.

[0167] Embodiments of this application also provide a computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0168] Embodiments of this application also provide a computer program product, including a computer program that, when executed by a processor, implements the methods of various embodiments of this application.

[0169] Embodiments of this application also provide a computer program product, including a non-volatile computer-readable storage medium for storing a computer program that, when executed by a processor, implements the methods in various embodiments of this application.

[0170] Embodiments of this application also provide a computer program that, when executed by a processor, implements the methods described in the various embodiments of this application.

[0171] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0172] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0173] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0174] Furthermore, the functional units in the various embodiments of this application 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.

[0175] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0176] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A vehicle speed control method, characterized in that, include: During vehicle operation, the first rainfall data collected by the rain sensor on the vehicle is obtained. Based on the first rainfall, the first water film thickness on the road where the vehicle is located is predicted; Based on the thickness of the first water film, the braking parameters of the vehicle, and the system parameters of the driving system on the vehicle, the target speed of the vehicle is determined. The vehicle is controlled to travel at a limited speed based on the target vehicle speed.

2. The method according to claim 1, characterized in that, Based on the first rainfall, the first water film thickness on the road where the vehicle is located is predicted, including: The first rainfall is input into the water film thickness prediction model, and the first water film thickness is predicted using the water film thickness prediction model. The water film thickness prediction model is trained using the second water film thickness and the second rainfall measured during the vehicle's journey on a preset road under rainfall conditions.

3. The method according to claim 2, characterized in that, The method further includes: Obtain the thickness of the second water film and the second rainfall; Construct an initial prediction model; The second rainfall is input into the initial prediction model, and the third water film thickness is predicted using the initial prediction model. Based on the second water film thickness and the third water film thickness, the model parameters of the initial prediction model are adjusted to obtain the water film thickness prediction model.

4. The method according to claim 3, characterized in that, Obtaining the second water film thickness and the second rainfall includes: The target road is constructed based on the preset construction depth, preset road slope, and preset road width of the road surface material. The target road is controlled to be in a rainy condition based on a preset rainfall intensity, and the vehicle is controlled to travel on the target road. The water film thickness of the target road is collected using data acquisition equipment to obtain the second water film thickness; The second rainfall is obtained by collecting rainfall data using the rain sensor on the vehicle.

5. The method according to claim 3, characterized in that, Constructing an initial prediction model includes: Determine the number of neurons in the first input layer, the number of neurons in the second output layer, and the number of hidden layers; The number of third neurons in the hidden layer is determined based on the number of the first neurons and the number of the second neurons. The initial prediction model is constructed based on the number of the first neuron, the number of the second neuron, the number of the third neuron, and the number of layers.

6. The method according to claim 5, characterized in that, Determining the number of third neurons in the hidden layer based on the number of the first and second neurons includes: Based on the number of the first neuron and the number of the second neuron, a range of neuron numbers is determined, wherein the range of neuron numbers includes multiple candidate neuron numbers; Determine the prediction error of the initial model corresponding to the number of the plurality of candidate neurons; Based on the prediction error, the number of the third neuron is determined from the number of candidate neurons.

7. The method according to any one of claims 1 to 6, characterized in that, Based on the thickness of the first water film, the vehicle's braking parameters, and the system parameters of the vehicle's driving system, the target vehicle speed is determined, including: The braking parameters, the system parameters, and the first water film thickness are input into the vehicle speed prediction model, and the target vehicle speed is predicted using the vehicle speed prediction model. The vehicle speed prediction model is constructed based on the vehicle's braking distance model and a preset relationship model. The preset relationship model is used to characterize the correlation between the first water film thickness, the road surface adhesion coefficient of the target road, and the target vehicle speed.

8. The method according to claim 7, characterized in that, The target vehicle speed is predicted using the vehicle speed prediction model, including: The braking parameters and the system parameters are input into the braking distance model to obtain the first correlation between the target vehicle speed and the road surface adhesion coefficient of the target road. The braking distance model is used to characterize the correlation between the braking parameters, the system parameters, the target vehicle speed and the road surface adhesion coefficient. The first water film thickness is input into a preset relationship model to obtain a second correlation between the target vehicle speed and the road surface adhesion coefficient of the target road. The preset relationship model is used to characterize the correlation between the first water film thickness, the road surface adhesion coefficient and the target vehicle speed. The target vehicle speed is determined based on the first association relationship and the second association relationship.

9. A vehicle, characterized in that, include: Memory, which stores executable programs; A processor for running the program, wherein the program, when running, performs the method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored executable program, wherein, when the executable program is executed, it controls the device on which the storage medium is located to perform the method according to any one of claims 1 to 8.

Images