A method, system, device, and medium for vehicle speed prediction based on Gaussian processes.

Abstract

This application discloses a method, system, device, and medium for vehicle speed prediction based on Gaussian processes, relating to the field of intelligent driving. The method includes: constructing a Gaussian process prediction model using a kernel function; optimizing the hyperparameters of the kernel function by maximizing the logarithmic marginal likelihood function using a training set to obtain a trained Gaussian process prediction model; predicting the vehicle speed using the trained Gaussian process prediction model based on the vehicle speed over a preset time period to obtain an initial predicted speed; identifying the vehicle's current driving mode using a K-means clustering algorithm; the driving mode includes acceleration mode, constant speed mode, and deceleration mode; and correcting the initial predicted speed based on the identified driving mode and the vehicle's current acceleration value to obtain the final predicted speed. This application can adaptively identify driving modes, dynamically fuse acceleration information, and effectively reduce prediction uncertainty.

Description

Technical Field

[0001] This application relates to the field of intelligent driving, and in particular to a method, system, device and medium for predicting vehicle speed based on Gaussian processes. Background Technology

[0002] With the rapid development of intelligent driving technology and the widespread application of intelligent transportation systems, vehicle speed prediction, as a core component of vehicle control and decision-making, directly impacts driving safety, energy efficiency, and the driving experience through its accuracy and reliability. In autonomous driving systems, accurate vehicle speed prediction provides crucial input for key functions such as path planning, adaptive cruise control, and emergency braking, directly affecting the accuracy and timeliness of system decisions. However, existing vehicle speed prediction technologies still face numerous challenges, especially in complex and ever-changing driving scenarios where the uncertainty of prediction results increases significantly.

[0003] Current mainstream vehicle speed prediction methods mainly rely on historical data, sensor input, and simple models. However, in practical applications, due to the diversity of driving behaviors, the complexity of road conditions, and the dynamic changes in environmental factors, the prediction results often have significant deviations. This uncertainty not only affects the decision-making quality of autonomous driving systems but may also lead to energy waste, reduced driving efficiency, and even safety hazards.

[0004] While various prediction models have been proposed in the current technology, such as random forests and neural networks based on machine learning, these methods mostly use static parameters and lack the ability to adapt to dynamic driving modes. At the same time, existing systems often ignore the impact of acceleration changes on vehicle speed prediction and fail to effectively integrate driving behavior characteristics with speed prediction models, resulting in a significant decrease in prediction accuracy in scenarios such as rapid acceleration and deceleration.

[0005] Furthermore, traditional vehicle speed prediction methods struggle to cope with diverse driving environments, such as frequent starts and stops in urban areas, continuous lane changes on highways, and driving in adverse weather conditions. Vehicle speed changes in these scenarios are highly nonlinear and uncertain. Existing technologies often require significant manual intervention or frequent adjustments to model parameters when handling these complex scenarios, failing to meet the demands of real-time prediction. Summary of the Invention

[0006] The purpose of this application is to provide a vehicle speed prediction method, system, device and medium based on Gaussian process, which can adaptively identify driving modes, dynamically fuse acceleration information and effectively reduce prediction uncertainty.

[0007] To achieve the above objectives, this application provides the following solution:

[0008] Firstly, this application provides a vehicle speed prediction method based on a Gaussian process, including:

[0009] A Gaussian process prediction model is constructed using a kernel function, and a training set is used to optimize the hyperparameters of the kernel function by maximizing the logarithmic marginal likelihood function to obtain a trained Gaussian process prediction model; the training set includes vehicle speeds at historical moments.

[0010] Based on the vehicle speed over a preset time period, a trained Gaussian process prediction model is used to predict the vehicle speed, resulting in an initial predicted vehicle speed.

[0011] The K-means clustering algorithm is used to identify the current driving mode of the vehicle; the driving mode includes acceleration mode, constant speed mode and deceleration mode;

[0012] Based on the identified driving mode and the vehicle's current acceleration value, the initial predicted speed is corrected to obtain the final predicted speed.

[0013] Secondly, this application provides a vehicle speed prediction system based on a Gaussian process, comprising:

[0014] The model building and training module is used to build a Gaussian process prediction model using a kernel function, and to optimize the hyperparameters of the kernel function by maximizing the log marginal likelihood function using a training set to obtain a trained Gaussian process prediction model; the training set includes vehicle speeds at historical times.

[0015] The initial vehicle speed prediction module is used to predict the vehicle speed based on the vehicle speed over a preset time period using a trained Gaussian process prediction model, and obtain the initial predicted vehicle speed.

[0016] The driving mode recognition module is used to identify the current driving mode of the vehicle using the K-means clustering algorithm; the driving mode includes acceleration mode, constant speed mode and deceleration mode;

[0017] The correction module is used to correct the initial predicted vehicle speed based on the identified driving mode and the vehicle's current acceleration value to obtain the final predicted vehicle speed.

[0018] Thirdly, this application provides a computer 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 above-described vehicle speed prediction method based on Gaussian processes.

[0019] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described vehicle speed prediction method based on a Gaussian process.

[0020] According to the specific embodiments provided in this application, this application has the following technical effects:

[0021] (1) Effectively improve the accuracy of vehicle speed prediction under non-steady-state conditions: By introducing the K-means clustering algorithm to identify the current driving mode (acceleration / uniform speed / deceleration), and combining the current acceleration value to make targeted corrections to the initial prediction results, active compensation for the prediction deviation direction of the Gaussian process is realized.

[0022] (2) Achieve the adaptive capability of Gaussian process prediction model to driving behavior: Couple driving mode recognition with Gaussian process prediction to form a "perception-recognition-correction" closed loop. The prediction no longer depends solely on historical vehicle speed, but integrates semantic information of driving mode, so that the model output can be dynamically adjusted according to the driving mode without manual intervention or offline retraining, thus meeting the dual requirements of real-time performance and generalization of intelligent driving.

[0023] (3) Make full use of the uncertainty quantification capability of Gaussian process to avoid blindly relying on point prediction and improve the safety of system decision-making: Gaussian process naturally outputs mean and variance. This application further adjusts the mean based on acceleration, so that the prediction results reflect both the trend and the risk boundary, which helps the system to adopt conservative strategies (such as deceleration in advance) when uncertainty is high, thereby improving driving safety and energy utilization efficiency. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments 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.

[0025] Figure 1 A flowchart illustrating a vehicle speed prediction method based on a Gaussian process, provided as an embodiment of this application;

[0026] Figure 2 This is a flowchart illustrating the K-means clustering algorithm. Detailed Implementation

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

[0028] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0029] In one exemplary embodiment, such as Figure 1 As shown, a vehicle speed prediction method based on a Gaussian process is provided. This method is executed by a computer device, specifically a terminal or server, or both. In this embodiment, the method is described using a server as an example, and includes the following steps S1 to S4. Wherein:

[0030] S1: A Gaussian process prediction model is constructed using a kernel function, and a training set is used to optimize the hyperparameters of the kernel function by maximizing the logarithmic marginal likelihood function to obtain a trained Gaussian process prediction model; the training set includes vehicle speeds at historical moments.

[0031] S2: Based on the vehicle speed over a preset time period, a trained Gaussian process prediction model is used to predict the vehicle speed, resulting in an initial predicted speed.

[0032] S3: The K-means clustering algorithm is used to identify the current driving mode of the vehicle; the driving mode includes acceleration mode, constant speed mode and deceleration mode.

[0033] S4: Based on the identified driving mode and the vehicle's current acceleration value, the initial predicted speed is corrected to obtain the final predicted speed.

[0034] This application constructs a Gaussian process prediction model and combines driving mode recognition and acceleration features to achieve adaptive optimization of vehicle speed prediction, significantly reducing prediction uncertainty and providing more accurate and reliable vehicle speed information support for intelligent driving systems, thereby improving the overall system safety and efficiency.

[0035] In a specific embodiment, step S1 specifically includes:

[0036] (1) Determine the training set.

[0037] 1) The training set consists of two parts: input and output. The feature vector of the input is expressed as follows:

[0038]

[0039] in, The training samples represent time t. It includes multiple element features, namely the historical vehicle speed from time t to time t+k, i.e. arrive ; The dimension of the input feature vector; This indicates the matrix transpose.

[0040] 2) The output format is as follows:

[0041]

[0042] in, The output at time t, i.e., the predicted velocity at time t. , This represents the dimension of the output.

[0043] The training set constructed in this embodiment is:

[0044]

[0045] in, (i=1,2,...,t) represents the training samples at time i. (i=1,2,...,t) represents the training samples at time i. The corresponding output, This is the output corresponding to the training set X.

[0046] (2) Kernel function training process.

[0047] The kernel function has the following form:

[0048]

[0049] in, Here, a kernel function is used to measure the training samples at time i. Training samples at time j The similarity between them For signal variance parameters, For signal length scale parameters, Training samples and radial distance, (i=1,2,...,t) are the training samples at time j. . For training samples The p-th element feature, For training samples The p-th element feature.

[0050] Hyperparameters to be optimized for:

[0051] .

[0052] First, construct the kernel matrix. , The element in the i-th row and j-th column is to The calculated value is obtained after substituting it into the kernel function. To avoid matrix singularity, a noise term is introduced:

[0053]

[0054] in, For a noisy kernel matrix, For noise variance, It is an identity matrix.

[0055] The hyperparameters are trained by maximizing the logarithmic marginal likelihood function. The purpose is to measure, during the training phase, the degree of matching between the training set data and the Gaussian process prior under the current hyperparameters.

[0056] The expression for the logarithmic marginal likelihood function is:

[0057]

[0058] in, In hyperparameters Under the corresponding Gaussian process prior, the training set Corresponding output The probability of occurrence. It is a data fitting term, and its function is to ensure that the model fits the speed trend of the training set. It is a complexity penalty term, and its purpose is to avoid overfitting. is the determinant value of the kernel matrix.

[0059] In this embodiment, the gradient ascent method is used to optimize the hyperparameter values ​​during the training process.

[0060] In a specific embodiment, step S2 specifically includes: inputting the vehicle speed for a preset time period into the trained Gaussian process prediction model, calculating the cross kernel matrix between the vehicle speed for the preset time period and the vehicle speed at historical moments in the training set; calculating the mean vehicle speed based on the cross kernel matrix; and using the mean vehicle speed as the initial predicted vehicle speed.

[0061] (1) In this embodiment, the vehicle speed during the preset time period is the predicted time. Vehicle speed data in the last 20 seconds ,Right now: . arrive for The speed of the car at that moment The speed of the vehicle at any given moment.

[0062] (2) Calculate the cross kernel matrix , It is all samples in the training set and The set of similarities:

[0063]

[0064] in, Cross kernel matrix The One element, Indicates training samples and The similarity.

[0065] (3) Predict vehicle speed.

[0066] Calculate the average vehicle speed : .

[0067] Posterior covariance : .

[0068] in, , for The value of its own kernel function.

[0069] In a specific embodiment, step S3 specifically includes: using the K-means clustering algorithm to cluster the historical acceleration data of the vehicle to obtain multiple clusters; each cluster represents a driving mode; calculating the similarity distance between the current acceleration value of the vehicle and the cluster center of each cluster; and taking the cluster corresponding to the cluster center with the smallest similarity distance as the current driving mode of the vehicle.

[0070] K-means is a distance-based unsupervised clustering algorithm. Its core idea is to divide n samples into w clusters, maximizing the similarity of samples within the same cluster and minimizing the similarity between samples in different clusters. For driving mode recognition, this embodiment sets the number of clusters w=3, corresponding to acceleration, constant speed, and deceleration driving modes. The overall process is as follows: Figure 2 As shown, categories 1 through 3 correspond to acceleration mode, deceleration mode, and constant speed mode, respectively.

[0071] In a specific embodiment, step S4 specifically includes: when the identified driving mode is an acceleration mode or a deceleration mode, the average vehicle speed is corrected based on the current acceleration value, and the corrected average vehicle speed is used as the final predicted vehicle speed; when the identified driving mode is a constant speed mode, the initial predicted vehicle speed is used as the final predicted vehicle speed.

[0072] When a vehicle is in acceleration mode, the predicted speed is often higher than the actual speed. In this case, the average predicted speed needs to be reduced appropriately based on the acceleration value. Similarly, when a vehicle is in deceleration mode, the predicted speed is often lower than the actual speed. In this case, the average predicted speed needs to be increased appropriately based on the acceleration value.

[0073] In this embodiment, the formula is used. The average vehicle speed is corrected.

[0074] in, This is the corrected average vehicle speed. For sensitivity parameters, This is the current acceleration value in acceleration mode. In deceleration mode .

[0075] The posterior covariance also needs to be adjusted, and its expression is as follows:

[0076]

[0077] It is the adjusted posterior covariance. It is a sensitivity parameter related to acceleration, and its expression is: , represent After gain coefficient Weighted quantization value This is a correction factor.

[0078] This application can effectively improve prediction accuracy by taking into account the inaccuracies in predicting vehicle speed.

[0079] Based on the same inventive concept, this application also provides a system for implementing the vehicle speed prediction method based on Gaussian processes described above. The solution provided by this system is similar to the solution described in the above method; therefore, the specific limitations of one or more vehicle speed prediction system embodiments based on Gaussian processes provided below can be found in the limitations of the vehicle speed prediction method based on Gaussian processes described above, and will not be repeated here.

[0080] In one exemplary embodiment, a vehicle speed prediction system based on a Gaussian process is provided, including the following modules.

[0081] The model building and training module is used to build a Gaussian process prediction model using a kernel function, and to optimize the hyperparameters of the kernel function by maximizing the logarithmic marginal likelihood function using a training set to obtain a trained Gaussian process prediction model; the training set includes vehicle speeds at historical times.

[0082] The prediction module is used to predict the vehicle speed based on the vehicle's speed over a preset time period using a trained Gaussian process prediction model, and obtain the initial predicted vehicle speed.

[0083] The driving mode recognition module is used to identify the current driving mode of the vehicle using the K-means clustering algorithm; the driving modes include acceleration mode, constant speed mode and deceleration mode.

[0084] The correction module is used to correct the initial predicted vehicle speed based on the identified driving mode and the vehicle's current acceleration value to obtain the final predicted vehicle speed.

[0085] In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments. The computer device may be a server or a terminal. The computer device includes a processor, a memory, an input / output interface (I / O), and a communication interface. The processor, memory, and I / O interface are connected via a system bus, and the communication interface is connected to the system bus via the I / O interface. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device stores data to be processed. The I / O interface of the computer device is used for exchanging information between the processor and external devices. The communication interface of the computer device is used for communicating with an external terminal via a network connection. When the computer program is executed by the processor, it implements the steps in the above-described method embodiments.

[0086] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0087] 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, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0088] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0089] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0090] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0091] This application uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. In summary, the content of this specification should not be construed as a limitation of this application.

Claims

1. A vehicle speed prediction method based on Gaussian processes, characterized in that, include: A Gaussian process prediction model is constructed using a kernel function, and a training set is used to optimize the hyperparameters of the kernel function by maximizing the logarithmic marginal likelihood function to obtain a trained Gaussian process prediction model; the training set includes vehicle speeds at historical moments. Based on the vehicle speed over a preset time period, a trained Gaussian process prediction model is used to predict the vehicle speed to obtain an initial predicted vehicle speed. Specifically, this includes: inputting the vehicle speed over the preset time period into the trained Gaussian process prediction model, calculating the cross-kernel matrix between the vehicle speed over the preset time period and the vehicle speed at historical moments in the training set; calculating the mean vehicle speed based on the cross-kernel matrix; and using the mean vehicle speed as the initial predicted vehicle speed. The K-means clustering algorithm is used to identify the current driving mode of the vehicle; the driving mode includes acceleration mode, constant speed mode and deceleration mode; Based on the identified driving mode and the vehicle's current acceleration value, the initial predicted vehicle speed is corrected to obtain the final predicted vehicle speed; specifically, when the identified driving mode is an acceleration mode or a deceleration mode, the average vehicle speed is corrected based on the current acceleration value, and the corrected average vehicle speed is used as the final predicted vehicle speed; when the identified driving mode is a constant speed mode, the initial predicted vehicle speed is used as the final predicted vehicle speed. Using formula The average vehicle speed is corrected, among which, This is the corrected average vehicle speed. The average vehicle speed For sensitivity parameters, This is the current acceleration value in acceleration mode. In deceleration mode .

2. The vehicle speed prediction method based on Gaussian process according to claim 1, characterized in that, The expression for the kernel function is: in, Here, a kernel function is used to measure the training samples at time i. Training samples at time j The similarity between them For signal variance parameters, For signal length scale parameters, yes and The radial distance.

3. The vehicle speed prediction method based on Gaussian process according to claim 1, characterized in that, The expression for the logarithmic marginal likelihood function is: in, In hyperparameters Under the corresponding Gaussian process prior, the training set Corresponding output The probability of occurrence This is a kernel matrix with noise; is the determinant value of the kernel matrix.

4. The vehicle speed prediction method based on Gaussian process according to claim 1, characterized in that, The K-means clustering algorithm is used to identify the vehicle's current driving mode, specifically including: The historical acceleration data of vehicles were clustered using the K-means clustering algorithm to obtain multiple clusters; each cluster represents a driving mode. Calculate the similarity distance between the vehicle's current acceleration value and the cluster centers of each cluster; The cluster corresponding to the cluster center with the smallest similarity distance is taken as the current driving mode of the vehicle.

5. A vehicle speed prediction system based on Gaussian processes, characterized in that, include: The model building and training module is used to build a Gaussian process prediction model using a kernel function, and to optimize the hyperparameters of the kernel function by maximizing the log marginal likelihood function using a training set to obtain a trained Gaussian process prediction model; the training set includes vehicle speeds at historical times. The prediction module is used to predict the vehicle speed based on the vehicle speed over a preset time period using a trained Gaussian process prediction model, and obtain an initial predicted vehicle speed. Specifically, it includes: inputting the vehicle speed over the preset time period into the trained Gaussian process prediction model; calculating the cross kernel matrix between the vehicle speed over the preset time period and the vehicle speed at historical moments in the training set; calculating the mean vehicle speed based on the cross kernel matrix; and using the mean vehicle speed as the initial predicted vehicle speed. The driving mode recognition module is used to identify the current driving mode of the vehicle using the K-means clustering algorithm; the driving mode includes acceleration mode, constant speed mode and deceleration mode; The correction module is used to correct the initial predicted vehicle speed based on the identified driving mode and the vehicle's current acceleration value to obtain the final predicted vehicle speed. Specifically, it includes: when the identified driving mode is an acceleration mode or a deceleration mode, correcting the average vehicle speed based on the current acceleration value and using the corrected average vehicle speed as the final predicted vehicle speed; when the identified driving mode is a constant speed mode, using the initial predicted vehicle speed as the final predicted vehicle speed. Using formula The average vehicle speed is corrected, among which, This is the corrected average vehicle speed. The average vehicle speed For sensitivity parameters, This is the current acceleration value in acceleration mode. In deceleration mode .

6. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the vehicle speed prediction method based on a Gaussian process as described in any one of claims 1-4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the vehicle speed prediction method based on the Gaussian process as described in any one of claims 1-4.

Images