Data processing method and device, vehicle, storage medium and program product
By using a pre-trained random forest model to identify abnormal nodes in the data link in the autonomous driving system, the problem of identifying abnormal nodes in the prior art is solved, and the stability and safety of the vehicle are improved.
Patent Information
- Application Number
- CN202411911347.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-23
- Publication Date
- 2025-03-14
AI Technical Summary
In autonomous driving systems, it is difficult for the prior art to quickly and accurately identify abnormal nodes in the data link, resulting in the stability and safety of vehicle operation being affected.
Through the pre-trained random forest model, the performance characteristic data of nodes in the vehicle data link is obtained, the status of nodes is determined, abnormal nodes are identified, and the target sublinks are dynamically split for monitoring.
It realizes the rapid and accurate identification of abnormal nodes in the data link, improves the stability and security of vehicle operation, and reduces false alarms and missed reports of abnormal nodes.
Smart Images
Figure CN119622537A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of autonomous driving, and in particular to a data processing method, device, vehicle, storage medium, and program product. Background Art
[0002] In the development of autonomous driving systems, the real-time and accuracy of data processing are crucial to ensuring vehicle safety. By implementing end-to-end link monitoring, every key node from sensor data acquisition, data processing, decision making to execution control can be monitored in real time, so as to promptly identify and resolve potential computing delays and effectively prevent safety risks. In addition, the data link monitoring mechanism helps to optimize the performance of the autonomous driving system and improve the reliability and efficiency of autonomous driving vehicles. Summary of the invention
[0003] In order to overcome the problems existing in the related art, the present disclosure provides a data processing method, device, vehicle, storage medium and program product.
[0004] According to a first aspect of an embodiment of the present disclosure, a data processing method is provided, including: Acquire performance characteristic data of at least one node in a data link of the vehicle, wherein the performance characteristic data represents the operating performance of the node in the data link; According to the performance characteristic data, the node state of each node is obtained through a pre-trained state determination model; the node state represents whether the node is normal or abnormal, and the state determination model is obtained by training a preset random forest model.
[0005] Optionally, the method further comprises: According to the node status, a target sub-link to be monitored is determined from the data link, and the target sub-link includes an abnormal node whose node status is abnormal.
[0006] Optionally, determining a target sub-link to be monitored from the data link according to the node state includes: In the data link where the abnormal node is located, a link formed by a preset number of target nodes and the abnormal node is used as the target sub-link; the preset number of target nodes are nodes adjacent to the abnormal node in the at least one node in series in the data link.
[0007] Optionally, the state determination model is trained in the following manner: Acquire a training sample set, the training sample set comprising first sample performance characteristic data of at least one node of the data link and a first sample node state corresponding to each first sample performance characteristic data; The preset random forest model is trained using the training sample set to obtain the state determination model.
[0008] Optionally, the training the preset random forest model by using the training sample set to obtain the state determination model includes: Dividing the training sample set into a plurality of training sample subsets, each of the training sample subsets corresponds to a preset decision tree in the preset random forest model; The preset decision trees are trained by using the training sample subsets corresponding to each of the preset decision trees to obtain multiple target decision trees; The random forest models corresponding to the multiple target decision trees are used as the state determination model.
[0009] Optionally, the training the preset decision tree by using the training sample subset corresponding to each preset decision tree to obtain multiple target decision trees comprises: By using the training sample subset corresponding to each of the preset decision trees, the splitting point of the preset decision tree is determined according to the Gini impurity to obtain the multiple target decision trees; or, By using the training sample subset corresponding to each of the preset decision trees, the splitting point of the preset decision tree is determined according to the information gain to obtain the multiple target decision trees.
[0010] Optionally, the state determination model is verified by: Acquire a verification sample set, the verification sample set comprising second sample performance characteristic data of at least one node of the data link and a second sample node state corresponding to each second sample performance characteristic data; The state determination model is cross-validated using the validation sample set.
[0011] Optionally, the performance characteristic data includes computing delay and / or CPU occupancy.
[0012] According to a second aspect of an embodiment of the present disclosure, there is provided a data processing device, including: an acquisition module, configured to acquire performance characteristic data of at least one node in a data link of the vehicle, wherein the performance characteristic data represents an operating performance of the node in the data link; The first determination module is configured to obtain the node state of each node through a pre-trained state determination model according to the performance characteristic data; the node state represents whether the node is normal or abnormal, and the state determination model is obtained by training a preset random forest model.
[0013] Optionally, the device further comprises: The second determination module is configured to determine a target sub-link to be monitored from the data link according to the node status, wherein the target sub-link includes an abnormal node whose node status is abnormal.
[0014] Optionally, the second determining module is configured to: In the data link where the abnormal node is located, a link formed by a preset number of target nodes and the abnormal node is used as the target sub-link; the preset number of target nodes are nodes adjacent to the abnormal node in the at least one node in series in the data link.
[0015] Optionally, the state determination model is trained in the following manner: Acquire a training sample set, the training sample set comprising first sample performance characteristic data of at least one node of the data link and a first sample node state corresponding to each first sample performance characteristic data; The preset random forest model is trained using the training sample set to obtain the state determination model.
[0016] Optionally, the training the preset random forest model by using the training sample set to obtain the state determination model includes: Dividing the training sample set into a plurality of training sample subsets, each of the training sample subsets corresponds to a preset decision tree in the preset random forest model; The preset decision trees are trained by using the training sample subsets corresponding to each of the preset decision trees to obtain multiple target decision trees; The random forest models corresponding to the multiple target decision trees are used as the state determination model.
[0017] Optionally, the training the preset decision tree by using the training sample subset corresponding to each preset decision tree to obtain multiple target decision trees comprises: By using the training sample subset corresponding to each of the preset decision trees, the splitting point of the preset decision tree is determined according to the Gini impurity to obtain the multiple target decision trees; or, By using the training sample subset corresponding to each of the preset decision trees, the splitting point of the preset decision tree is determined according to the information gain to obtain the multiple target decision trees.
[0018] Optionally, the state determination model is verified by: Acquire a verification sample set, the verification sample set comprising second sample performance characteristic data of at least one node of the data link and a second sample node state corresponding to each second sample performance characteristic data; The state determination model is cross-validated using the validation sample set.
[0019] Optionally, the performance characteristic data includes computing delay and / or CPU occupancy.
[0020] According to a third aspect of an embodiment of the present disclosure, a vehicle is provided, comprising: processor; a memory for storing processor-executable instructions; Wherein, the processor is configured to: Acquire performance characteristic data of at least one node in a data link of the vehicle, wherein the performance characteristic data represents the operating performance of the target node in the data link; According to the performance characteristic data, the node state of each node is obtained through a pre-trained state determination model; the node state represents whether the node is normal or abnormal, and the state determination model is obtained by training a preset random forest model.
[0021] According to a fourth aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described in the first aspect of the embodiment of the present disclosure are implemented.
[0022] According to a fifth aspect of an embodiment of the present disclosure, a computer program product is provided, including a computer program, which, when executed by a processor, implements the steps of the method described in the first aspect of the embodiment of the present disclosure.
[0023] The technical solution provided by the embodiments of the present disclosure may have the following beneficial effects: The present disclosure pre-trains a preset random forest model to obtain a state determination model. During the operation of the vehicle, the performance characteristic data of at least one node in the data link of the vehicle is input into the state determination model to obtain the node state of each node, wherein the node state represents whether the node is normal or abnormal. In this way, the pre-trained state determination model can quickly and accurately identify abnormal nodes in the data link, thereby improving the stability and safety of vehicle operation.
[0024] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.
[0026] Figure 1 The present invention is a flowchart of a method for processing data according to an exemplary embodiment.
[0027] Figure 2 The present invention is a flowchart of another data processing method according to an exemplary embodiment.
[0028] Figure 3 A method for determining a state determination model is shown according to an exemplary embodiment.
[0029] Figure 4 It is an application scenario diagram of a state determination model according to an exemplary embodiment.
[0030] Figure 5 The invention is a block diagram of a data processing device according to an exemplary embodiment.
[0031] Figure 6 It is a block diagram of another data processing device according to an exemplary embodiment.
[0032] Figure 7 is a block diagram of a vehicle according to an exemplary embodiment. DETAILED DESCRIPTION
[0033] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.
[0034] It should be noted that all actions of acquiring signals, information or data in the present disclosure are carried out in compliance with the relevant data protection laws and policies of the country where the device is located and with the authorization given by the owner of the corresponding device.
[0035] Before introducing a data processing method, device, vehicle, storage medium, and program product shown in an embodiment of the present disclosure, an application scenario involved in the embodiment of the present disclosure is first introduced.
[0036] In an autonomous driving system, ensuring the real-time and accuracy of data processing is crucial to ensuring driving safety. By implementing end-to-end link monitoring, every key node from sensor data acquisition, data processing, decision making to execution control can be monitored in real time. Specifically, the data link of an autonomous driving system may include: data transmission paths from sensors (such as cameras, radars, and lidars) to data processing units (such as on-board computers); data transmission paths from data processing units to decision modules (such as path planning algorithms, obstacle avoidance algorithms); and control signal transmission paths from decision modules to actuators (such as steering systems, braking systems, and acceleration systems). For example, in the automatic parking APA (English: Automated Parking Assist, Chinese: Automatic Parking Assist System) scenario, the data link may include: 1. Data transmission path from sensor to data processing unit: Participating nodes: camera, ultrasonic sensor, radar → data acquisition module → on-board computer.
[0037] Monitoring goal: Ensure that sensor data can be transmitted to the data processing unit in real time and accurately to perceive the vehicle's surrounding environment.
[0038] 2. Data transmission path from the data processing unit to the decision module: Participating nodes: on-board computer (data processing unit) → middleware (such as ROS2 (English: Robot Operating System, Chinese: Robot Operating System)) → path planning algorithm → obstacle avoidance algorithm.
[0039] Monitoring objective: Ensure that the delay in data processing is within an acceptable range to avoid decision-making errors caused by data processing delays.
[0040] 3. Control signal transmission path from decision module to actuator: Participating nodes: path planning algorithm → obstacle avoidance algorithm and other decision modules → control signal processor → execution node (such as steering system, braking system, acceleration system).
[0041] Monitoring objective: Ensure that control signals can be transmitted to the actuator in a timely and accurate manner to perform correct operations.
[0042] The autonomous driving system includes a lot of links. How to quickly and accurately identify abnormal links with a high probability is a task with a huge workload. The disclosed embodiment pre-trains the preset random forest model to obtain a state determination model. During the operation of the vehicle, the performance characteristic data of at least one node in the vehicle's data link is input into the state determination model to obtain the node state of each node, where the node state represents whether the node is normal or abnormal. In this way, abnormal nodes in the data link of the autonomous driving system can be quickly and accurately identified, thereby improving the stability and safety of vehicle operation.
[0043] Figure 1 is a flow chart of a data processing method according to an exemplary embodiment. Figure 1 As shown, the method may include the following steps.
[0044] Step S101, obtaining performance characteristic data of at least one node in a data link of a vehicle.
[0045] For example, the vehicle in the embodiment of the present disclosure may include one or more data links, and different data links may correspond to the vehicle running in different modes or in different scenarios. Each data link may include at least one node. When the node includes multiple nodes, data may be transmitted between multiple nodes. During the operation of the vehicle, performance characteristic data of at least one node in the data link may be obtained, wherein the performance characteristic data may characterize the operating performance of the target node in the data link, and the performance characteristic data may include, for example, the computing delay, CPU occupancy, bit error rate, etc. of the node. Taking the data link including three nodes of radar, data acquisition module and on-board computer as an example, the performance characteristic data of radar, data acquisition module and on-board computer may be obtained respectively.
[0046] Step S102, according to the performance characteristic data, the node state of each node is obtained through a pre-trained state determination model, wherein the node state can represent whether the node is normal or abnormal.
[0047] For example, a preset random forest model can be trained in advance to obtain a state determination model for determining the node state. In some practical examples, multiple decision trees can be constructed based on different training data sets, and each decision tree independently learns how to distinguish between normal nodes and abnormal nodes to obtain the node state of the node. Through the majority voting mechanism, the trained state determination model can combine the prediction results of all decision trees to determine whether a node is abnormal.
[0048] During the operation of the vehicle, the performance characteristic data of each node can be input into a pre-trained state determination model. The state determination model can process the input performance characteristic data and output the node status of the node, so that abnormal nodes in the data link can be identified quickly and accurately, reducing false positives and false negatives of abnormal nodes and improving the stability and safety of vehicle operation.
[0049] In summary, the present disclosure pre-trains a preset random forest model to obtain a state determination model. During the operation of the vehicle, the performance characteristic data of at least one node in the data link of the vehicle is input into the state determination model to obtain the node state of each node, wherein the node state represents whether the node is normal or abnormal. In this way, through the pre-trained state determination model, abnormal nodes in the data link can be quickly and accurately identified, thereby improving the stability and safety of vehicle operation.
[0050] Figure 2 is a flowchart of another data processing method according to an exemplary embodiment. Figure 2 As shown, the method also includes: Step S103: determining a target sub-link to be monitored from the data link according to the node status, wherein the target sub-link may include an abnormal node whose node status is abnormal.
[0051] For example, after obtaining the node status of each node in the data link, an abnormal node with an abnormal node status may be determined, and a target sub-link to be monitored may be determined from the data link according to the position of the abnormal node in the data link.
[0052] In some embodiments, a link consisting of a preset number of target nodes and the abnormal node in the data link where the abnormal node is located can be used as a target sub-link. The preset number of target nodes can be nodes adjacent to the abnormal node in at least one node in series in the data link, and the preset number of target nodes can be located downstream of the abnormal node in the data transmission direction, or can be located upstream of the abnormal node, or can be evenly distributed upstream and downstream of the abnormal node, and the present disclosure does not make specific limitations on this.
[0053] For example, taking the preset number as 2, the data link includes seven nodes in series, A, B, C, D, E, F, and G, and the abnormal node is C. When the preset number of target nodes are all located upstream in the data transmission direction, A, B, and C can be used as the target sub-link to be monitored. When the preset number of target nodes are all located downstream in the data transmission direction, the short path composed of nodes C, D, and E can be used as the target sub-link to be monitored. When the preset number of target nodes are located upstream and downstream in the data transmission direction, the short path composed of nodes B, C, and D can be used as the target sub-link to be monitored.
[0054] In this way, abnormal nodes are identified through a pre-trained state determination model, and the data links to be monitored are determined based on the abnormal nodes. This not only improves the accuracy of locating abnormal nodes and reduces false positives and negatives of abnormal nodes, but also dynamically splits the target sub-links to be monitored to enhance fine-grained monitoring and optimization, ensuring that abnormal problems at nodes can be discovered and handled in a timely manner, thereby improving the stability and reliability of the autonomous driving system.
[0055] Figure 3 A method for determining a state determination model is shown according to an exemplary embodiment. Figure 3 As shown, the state determination model is trained in the following way: Step S201 : acquiring a training sample set, where the training sample set includes first sample performance characteristic data of at least one node of a data link and a first sample node state corresponding to each first sample performance characteristic data.
[0056] For example, taking the first sample performance characteristic data including computing delay and CPU occupancy as an example, the training sample set D can be expressed as: D={(x 1 ,x 2 ,y)}, where x 1 represents the computation delay, x 2 Indicates the CPU usage, y is the node status. For example, y=0 indicates that the node status is normal, and y=1 indicates that the node status is abnormal. 1 , N 2 , N 3 For example, each node may correspond to multiple first sample performance characteristic data, each of which may include computing latency, CPU occupancy, and node status. The training sample set D may be expressed as: D={(x 11 ,x 12 ,y 1 ), (x 21 ,x 22 ,y 2 ), (x 31 ,x 32 ,y3 ), ...}, where (x 11 ,x 12 ,y 1 ) is node N 1 A first sample performance characteristic data, (x 21 ,x 22 ,y 2 ) is node N 2 A first sample performance characteristic data, (x 31 ,x 32 ,y 3 ) is node N 3 A first sample performance characteristic data.
[0057] Step S202, training a preset random forest model using a training sample set to obtain a state determination model.
[0058] In some embodiments, the first sample performance characteristic data in the training sample set may be normalized. The purpose of normalization is to make all the first sample performance characteristic data at the same level to avoid some data dominating the learning process of the model due to their large magnitude. The first sample performance characteristic data in the training sample set may be normalized using the normalization formula shown below.
[0059]
[0060] in, is the normalized performance characteristic data of the first sample, is the original first sample performance characteristic data, and are the minimum and maximum values of the first sample performance characteristic data in the training sample set, respectively.
[0061] To calculate the delay x 1 For example, assume that the minimum value of the computational delay is 10ms and the maximum value is 100ms. The computational delay x of a node 1 is 50ms, and the normalized calculation delay is: ’ 1 = (50-10) / (100-10) = 40 / 90 = 4 / 9. Taking CPU usage as an example, assume that the minimum CPU usage is 10% and the maximum is 90%. The CPU usage of a node is 50%. The normalized CPU usage is: x ’ 2 =(50%-10%) / (90%-10%)=40 / 80=0.5.
[0062] For example, the definition of a pre-set random forest model can be as follows:
[0063] in, It can represent the performance characteristic data of the first sample by the preset random forest model The prediction results, It can represent the number of decision trees in the preset random forest model. Can indicate the The performance characteristic data of the first sample is prediction results.
[0064] In other embodiments, the parameters of the preset random forest model can be set, such as the number of trees (n_estimators), maximum depth (max_depth), minimum sample split (min_samples_split), etc. The number of trees in the random forest model determines how many decision trees will independently evaluate the node state of each node. More decision trees can improve the stability and accuracy of the model, but also increase the computational cost. Shallower trees may not be able to fully capture complex patterns, while deeper trees may lead to overfitting. For data sets with more features, deeper trees may be more suitable for capturing complex relationships, but for data sets with fewer features, shallower trees may be more appropriate. By reasonably setting the parameters of the preset random forest model, it can be ensured that the model detects node anomalies more accurately and stably.
[0065] In other embodiments, since the random forest model can be composed of multiple decision trees, the training sample set can be divided into multiple training sample subsets, wherein each training sample subset corresponds to a preset decision tree in the preset random forest model. Then, the preset decision tree can be trained by the training sample subset corresponding to each preset decision tree to obtain multiple target decision trees, and the random forest models corresponding to the multiple target decision trees are used as the state determination model.
[0066] In some other embodiments, step S202 may be implemented in the following manner: Through the training sample subset corresponding to each preset decision tree, the split point of the preset decision tree is determined according to the Gini impurity to obtain multiple target decision trees. Or, Through the training sample subset corresponding to each preset decision tree, the split point of the preset decision tree is determined according to the information gain to obtain multiple target decision trees.
[0067] For example, since each preset decision tree can be a binary tree structure, it is necessary to select the best split feature and split point when constructing the decision tree. The best split feature in the disclosed embodiment may include calculation delay and CPU occupancy rate, and the split point of the prediction decision tree may be determined by Gini impurity or information gain.
[0068] Gini impurity can be used to measure the probability of a sample in a node being misclassified. Taking the optimal splitting features including computational latency and CPU occupancy as an example, if a node has a high computational latency or a high CPU occupancy, then the node may be abnormal. Therefore, Gini impurity can be used to determine which split point can better separate normal nodes from abnormal nodes. The lower the Gini impurity, the purer the node and the better the classification effect. By using the training sample subset corresponding to each preset decision tree and determining the split point of the preset decision tree according to the Gini impurity, multiple target decision trees can be trained to build an efficient classification model. Among them, Gini impurity is defined as follows:
[0069] Where G(D) is the Gini impurity, p 2 k is the probability of belonging to the kth class. For the binary classification problem, K=2, p 1 and p 2 are the ratios of positive and negative classes, respectively.
[0070] Information gain can be used to measure the amount of uncertainty reduction brought about by feature splitting. Taking the best splitting features including computing delay and CPU occupancy as an example, if the computing delay or CPU occupancy of a node deviates significantly from the normal range, then the information gain of this feature is relatively high, because they can better distinguish normal nodes from abnormal nodes. Through the training sample subset corresponding to each preset decision tree, the split point of the preset decision tree is determined according to the information gain, and multiple target decision trees can be trained. Among them, the definition of information gain is as follows:
[0071] Among them, IG(D,A) is the information gain, H(D) is the entropy of the training sample set D, and H(D|A) is the conditional entropy of the training sample set D given the optimal split feature A.
[0072] In some other embodiments, the trained state determination model may be verified in the following manner: A verification sample set is obtained, wherein the verification sample set may include second sample performance characteristic data of at least one node of the data link and a second sample node state corresponding to each second sample performance characteristic data.
[0073] The state determination model was cross-validated using the validation sample set.
[0074] For example, the state determination model can be validated by using K-fold cross validation on the validation sample set, where K can be 5.
[0075] In one possible implementation, the state determination model can be cross-validated at the node level through the validation sample set. Through the cross-validation at the node level, the performance of the state determination model on different nodes can be evaluated. The computing latency and CPU occupancy of different nodes may be different. Cross-validation at the node level helps ensure the generalization ability of the state determination model on various nodes.
[0076] In another possible implementation, each link may consist of at least one node, and the state determination model may be cross-validated at the link level through the validation sample set. Through the cross-validation at the link level, the performance of the state determination model on the entire data link may be evaluated. The data distribution of different data links may be different, and the cross-validation at the link level helps ensure the generalization ability of the model on different links.
[0077] In other embodiments, the performance of the state determination model can be evaluated by analyzing the confusion matrix based on the validation sample set, and the accuracy, recall rate, F1 score and other indicators of the state determination model can be calculated to evaluate the state determination model.
[0078] Figure 4 It is a scenario diagram of the application of the state determination model in the production environment. Figure 4 , the trained state determination model can be deployed on the R&D vehicle, and the performance characteristic data of the R&D vehicle can be input into the state determination model in real time to continuously monitor the node status of the nodes of the data link in the R&D vehicle, and then determine the target sub-link to be monitored based on the location of the identified abnormal node in the data link. After multiple iterations, the state determination model can be deployed on mass-produced vehicles.
[0079] In summary, the present disclosure pre-trains a preset random forest model to obtain a state determination model. During the operation of the vehicle, the performance characteristic data of at least one node in the data link of the vehicle is input into the state determination model to obtain the node state of each node, wherein the node state represents whether the node is normal or abnormal. In this way, through the pre-trained state determination model, abnormal nodes in the data link can be quickly and accurately identified, thereby improving the stability and safety of vehicle operation.
[0080] Figure 5 is a block diagram of a data processing device according to an exemplary embodiment. Figure 5 As shown, the device 300 may include: The acquisition module 301 is configured to acquire performance characteristic data of at least one node in a data link of the vehicle, where the performance characteristic data represents the operating performance of the node in the data link.
[0081] The first determination module 302 is configured to obtain the node state of each node through a pre-trained state determination model according to the performance characteristic data. The node state represents whether the node is normal or abnormal, and the state determination model is obtained by training a preset random forest model.
[0082] Figure 6 is a block diagram of another data processing device according to an exemplary embodiment. Figure 6 As shown, the device 300 may also include: The second determination module 303 is configured to determine a target sub-link to be monitored from the data link according to the node status, and the target sub-link includes an abnormal node whose node status is abnormal.
[0083] In some embodiments, the second determination module 303 may be configured to: In the data link where the abnormal node is located, a link formed by a preset number of target nodes and the abnormal node is used as a target sub-link. The preset number of target nodes are nodes adjacent to the abnormal node in at least one node in series in the data link.
[0084] In some other embodiments, the state determination model is trained in the following manner: A training sample set is obtained, where the training sample set includes first sample performance characteristic data of at least one node of the data link and a first sample node state corresponding to each first sample performance characteristic data.
[0085] The preset random forest model is trained through the training sample set to obtain the state determination model.
[0086] In some other embodiments, the state determination model is obtained by training a preset random forest model through a training sample set, including: The training sample set is divided into multiple training sample subsets, and each training sample subset corresponds to a preset decision tree in the preset random forest model.
[0087] The preset decision trees are trained using the training sample subsets corresponding to each preset decision tree to obtain multiple target decision trees.
[0088] The random forest model corresponding to multiple target decision trees is used as the state determination model.
[0089] In other embodiments, the preset decision trees are trained by using the training sample subsets corresponding to each preset decision tree to obtain multiple target decision trees including: Through the training sample subset corresponding to each preset decision tree, the split point of the preset decision tree is determined according to the Gini impurity to obtain multiple target decision trees. Or, Through the training sample subset corresponding to each preset decision tree, the split point of the preset decision tree is determined according to the information gain to obtain multiple target decision trees.
[0090] In some other embodiments, the state determination model is verified by: A verification sample set is obtained, where the verification sample set includes second sample performance characteristic data of at least one node of the data link and a second sample node state corresponding to each second sample performance characteristic data.
[0091] The state determination model was cross-validated using the validation sample set.
[0092] In other embodiments, the performance characteristic data includes computing latency and / or CPU occupancy.
[0093] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.
[0094] In summary, the present disclosure pre-trains a preset random forest model to obtain a state determination model. During the operation of the vehicle, the performance characteristic data of at least one node in the data link of the vehicle is input into the state determination model to obtain the node state of each node, wherein the node state represents whether the node is normal or abnormal. In this way, abnormal nodes in the data link can be quickly and accurately identified, thereby improving the stability and safety of vehicle operation.
[0095] The present disclosure also provides a computer-readable storage medium having computer program instructions stored thereon, and when the program instructions are executed by a processor, the steps of the data processing method provided by the present disclosure are implemented.
[0096] Figure 7 4 is a block diagram of a vehicle 400 according to an exemplary embodiment. For example, the vehicle 400 may be a hybrid vehicle, a non-hybrid vehicle, an electric vehicle, a fuel cell vehicle, or other types of vehicles. The vehicle 400 may be an autonomous vehicle, a semi-autonomous vehicle, or a non-autonomous vehicle.
[0097] Reference Figure 7 , the vehicle 400 may include various subsystems, for example, an infotainment system 410, a perception system 420, a decision control system 430, a drive system 440, and a computing platform 450. The vehicle 400 may also include more or fewer subsystems, and each subsystem may include multiple components. In addition, each subsystem and each component of the vehicle 400 may be interconnected by wire or wireless means.
[0098] In some embodiments, the infotainment system 410 may include a communication system, an entertainment system, and a navigation system, among others.
[0099] The perception system 420 may include several sensors for sensing information about the environment around the vehicle 400. For example, the perception system 420 may include a global positioning system (the global positioning system may be a GPS system, or a Beidou system or other positioning systems), an inertial measurement unit (IMU), a laser radar, a millimeter wave radar, an ultrasonic radar, and a camera.
[0100] The decision control system 430 may include a computing system, a vehicle controller, a steering system, a throttle, and a braking system.
[0101] The drive system 440 may include components that provide power movement for the vehicle 400. In one embodiment, the drive system 440 may include an engine, an energy source, a transmission system, and wheels. The engine may be one or a combination of an internal combustion engine, an electric motor, and an air compression engine. The engine is capable of converting energy provided by the energy source into mechanical energy.
[0102] Some or all functions of the vehicle 400 are controlled by a computing platform 450. The computing platform 450 may include at least one processor 451 and a memory 452, and the processor 451 may execute instructions 453 stored in the memory 452.
[0103] The processor 451 may be any conventional processor, such as a commercially available CPU. The processor may also include a graphics processor (Graphic Process Unit, GPU), a field programmable gate array (Field Programmable Gate Array, FPGA), a system on chip (System on Chip, SOC), an application specific integrated circuit (Application Specific Integrated Circuit, ASIC) or a combination thereof.
[0104] The memory 452 may be implemented by any type of volatile or nonvolatile memory device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.
[0105] In addition to instructions 453 , memory 452 may also store data, such as road maps, route information, and data such as the location, direction, and speed of the vehicle. The data stored in memory 452 may be used by computing platform 450 .
[0106] In the embodiment of the present disclosure, the processor 451 may execute the instruction 453 to complete all or part of the steps of the above-mentioned data processing method.
[0107] In another exemplary embodiment, a computer program product is also provided. The computer program product includes a computer program executable by a programmable device, and the computer program has a code portion for executing the above-mentioned data processing method when executed by the programmable device.
[0108] Those skilled in the art may also understand that the various illustrative logical blocks and steps listed in the embodiments of the present application may be implemented by electronic hardware, computer software, or a combination of the two. Whether such functions are implemented by hardware or software depends on the specific application and the design requirements of the entire system. Those skilled in the art may use various methods to implement the functions described for each specific application, but such implementation should not be understood as exceeding the scope of protection of the embodiments of the present application.
[0109] In addition, the word "exemplary" is used herein to indicate serving as an example, instance, or diagram. Any aspect or design described as "exemplary" in this article is not necessarily understood to be advantageous compared to other aspects or designs. On the contrary, the use of the word exemplary is intended to present concepts in a specific way. As used herein, the term "or" is intended to represent an inclusive "or" rather than an exclusive "or". That is, unless otherwise specified or clear from the context, "X applies A or B" is intended to represent any one of the natural inclusive arrangements. That is, if X applies A; X applies B; or X applies both A and B, "X applies A or B" is satisfied under any of the aforementioned examples. In addition, unless otherwise specified or clearly pointed to a singular form from the context, the articles "one" and "an" as used in this application and the appended claims are generally understood to mean "one or more".
[0110] Likewise, although the present disclosure has been shown and described with respect to one or more implementations, equivalent variations and modifications will occur to those skilled in the art after reading and understanding the specification and drawings. The present disclosure includes all such modifications and variations and is limited only by the scope of the claims. In particular, with respect to the various functions performed by the components (e.g., elements, resources, etc.) described above, unless otherwise indicated, the terms used to describe such components are intended to correspond to any component (functionally equivalent) that performs the specific functions of the described components, even if the structure is not equivalent to the disclosed structure. In addition, although specific features of the present disclosure may have been disclosed with respect to only one of several implementations, such features may be combined with one or more other features of other implementations as may be desired and beneficial to any given or specific application. In addition, with respect to "including", "having", "having", "having", or variations thereof used in a specific embodiment or claim, such terms are intended to be inclusive in a manner similar to the term "comprising".
[0111] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any modification, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present disclosure are indicated by the appended claims.
[0112] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.
[0113] It should be understood that, unless otherwise specifically noted, the features of some embodiments of the various present disclosures described herein may be combined with each other. As used herein, the term "and / or" includes any one of the related listed items and any combination of any two or more; similarly, "at least one of . . . " includes any one of the related listed items and any combination of any two or more.
[0114] Although terms such as "first", "second" and "third" can be used herein to describe various components, parts, regions, layers or sections, these components, parts, regions, layers or sections are not limited to these terms. On the contrary, these terms are only used to distinguish one component, component, region, layer or section from another component, component, region, layer or section. Therefore, without departing from the teachings of each example, the first component, component, region, layer or section mentioned in the examples described herein may also be referred to as the second component, component, region, layer or section. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Thus, the features defined as "first" and "second" may expressly or implicitly include at least one of the features. In the description herein, the meaning of "multiple" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.
Claims
1. A data processing method, characterized in that: include: Acquire performance characteristic data of at least one node in a data link of the vehicle, wherein the performance characteristic data represents the operating performance of the node in the data link; According to the performance characteristic data, the node state of each node is obtained through a pre-trained state determination model; the node state represents whether the node is normal or abnormal, and the state determination model is obtained by training a preset random forest model.
2. The method according to claim 1, characterized in that The method further comprises: According to the node status, a target sub-link to be monitored is determined from the data link, and the target sub-link includes an abnormal node whose node status is abnormal.
3. The method according to claim 2, characterized in that Determining the target sub-link to be monitored from the data link according to the node state includes: In the data link where the abnormal node is located, a link formed by a preset number of target nodes and the abnormal node is used as the target sub-link; the preset number of target nodes are nodes adjacent to the abnormal node in the at least one node in series in the data link.
4. The method according to claim 1, characterized in that The state determination model is trained in the following way: Acquire a training sample set, the training sample set comprising first sample performance characteristic data of at least one node of the data link and a first sample node state corresponding to each first sample performance characteristic data; The preset random forest model is trained using the training sample set to obtain the state determination model.
5. The method according to claim 4, characterized in that The training of the preset random forest model by the training sample set to obtain the state determination model comprises: Dividing the training sample set into a plurality of training sample subsets, each of the training sample subsets corresponds to a preset decision tree in the preset random forest model; The preset decision trees are trained by using the training sample subsets corresponding to each of the preset decision trees to obtain multiple target decision trees; The random forest models corresponding to the multiple target decision trees are used as the state determination model.
6. The method according to claim 5, characterized in that The training of the preset decision trees by the training sample subsets corresponding to each preset decision tree to obtain multiple target decision trees comprises: By using the training sample subset corresponding to each of the preset decision trees, the splitting point of the preset decision tree is determined according to the Gini impurity to obtain the multiple target decision trees; or, By using the training sample subset corresponding to each of the preset decision trees, the splitting point of the preset decision tree is determined according to the information gain to obtain the multiple target decision trees.
7. The method according to claim 1, characterized in that The state determination model is verified by: Acquire a verification sample set, the verification sample set comprising second sample performance characteristic data of at least one node of the data link and a second sample node state corresponding to each second sample performance characteristic data; The state determination model is cross-validated using the validation sample set.
8. The method according to any one of claims 1 to 7, characterized in that The performance characteristic data includes computing delay and / or CPU occupancy rate.
9. A data processing device, characterized in that: include: an acquisition module, configured to acquire performance characteristic data of at least one node in a data link of the vehicle, wherein the performance characteristic data represents an operating performance of the node in the data link; The first determination module is configured to obtain the node state of each node through a pre-trained state determination model according to the performance characteristic data; the node state represents whether the node is normal or abnormal, and the state determination model is obtained by training a preset random forest model.
10. A vehicle, characterized in that: include: processor; a memory for storing processor-executable instructions; Wherein, the processor is configured to: Acquire performance characteristic data of at least one node in a data link of the vehicle, wherein the performance characteristic data represents the operating performance of the target node in the data link; According to the performance characteristic data, the node state of each node is obtained through a pre-trained state determination model; the node state represents whether the node is normal or abnormal, and the state determination model is obtained by training a preset random forest model.
11. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.
12. A computer program product, characterized in that The invention comprises a computer program, which, when executed by a processor, implements the steps of the method according to any one of claims 1 to 8.