Vehicle driving analysis method, system, equipment and medium
By acquiring and processing vehicle driving data, using vehicle driving analysis models and flow analysis engines, combined with Flink and Kafka, the problem of low accuracy in vehicle driving abnormality analysis is solved, and real-time safety monitoring and analysis in automated driving is achieved.
Patent Information
- Application Number
- CN202411386499.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-30
- Publication Date
- 2025-01-07
AI Technical Summary
In the prior art, the accuracy of vehicle driving abnormalities analysis is low, and it is impossible to effectively deal with the uncertainty of driving scenarios and vehicle technical limitations, resulting in potential threats in automated driving.
By obtaining the current driving data sequence, converting it into a driving vector sequence, and using the trained vehicle driving analysis model, the target analysis tag set is determined, thereby determining the safety analysis results of the vehicle driving and improving the accuracy of the analysis. The method includes feature extraction, data acquisition and model training, using Flink and Kafka to process streaming data, and combining weak learning machine models for real-time analysis.
It improves the accuracy of vehicle driving analysis, can process large-scale data, monitors vehicle driving safety in real time, and enhances the safety of automated driving.
Smart Images
Figure CN119274339A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of intelligent driving technology, and in particular to a vehicle driving analysis method, system, device and medium. Background Art
[0002] Abnormal analysis and prediction of vehicle driving is one of the hot research directions in the field of traffic safety and intelligent transportation systems.
[0003] Due to the uncertainty of driving scenarios and the limitations of vehicle technology, there are still potential threats in the process of automated driving, so it is necessary to analyze the vehicle's driving. At present, statistical methods are often used to analyze vehicle driving anomalies, but statistical methods have low accuracy in some scenarios. Summary of the invention
[0004] In order to solve the problem of low accuracy in analyzing abnormal vehicle driving in the prior art, the present invention provides a vehicle driving analysis method, system, device and medium.
[0005] In a first aspect, an embodiment of the present disclosure provides a vehicle driving analysis method, the method comprising:
[0006] Get the current driving data sequence;
[0007] Determine a current driving vector sequence corresponding to the current driving data sequence;
[0008] Input the current driving vector sequence into the trained vehicle driving analysis model to obtain a target analysis label set corresponding to the current driving vector sequence; the target analysis labels in the target analysis label set correspond one-to-one to the current driving vector in the current driving vector sequence;
[0009] The safety analysis results of vehicle driving are determined based on the target analysis label set.
[0010] In one embodiment, determining a current driving vector sequence corresponding to a current driving data sequence includes:
[0011] Perform feature extraction on each current driving data in the current driving data sequence to obtain a current driving vector corresponding to each current driving data;
[0012] The current driving vector sequence is determined according to the current driving vector corresponding to each current driving data in the current driving data sequence.
[0013] In one embodiment, the current driving data sequence includes current first driving data and current second driving data;
[0014] Before obtaining the current driving data sequence, include:
[0015] Collecting initial driving data at first preset time intervals;
[0016] The initial driving data is updated to the current first driving data; and the first driving data corresponding to the previous first preset duration is updated to the current second driving data.
[0017] In one embodiment, the method further comprises the step of training to obtain a vehicle driving analysis model;
[0018] The trained vehicle analysis model includes:
[0019] Acquire a sample data set, where the sample data set includes a plurality of sample driving vectors and a preset analysis label corresponding to each of the plurality of sample driving vectors;
[0020] Build an initial model;
[0021] Based on the initial model, a vehicle driving trajectory analysis is performed on each sample driving vector among the multiple sample driving vectors to obtain a prediction analysis label corresponding to each sample driving vector;
[0022] Construct a loss function based on the predicted analysis label and preset analysis label corresponding to each sample driving vector;
[0023] Update the initial model according to the loss function;
[0024] When the loss function iterates to the minimum value, the initial model is determined as the vehicle analysis model.
[0025] In one embodiment, updating the initial model according to the loss function includes:
[0026] Determine multiple hyperparameters of the initial model and the value range of each hyperparameter in the multiple hyperparameters;
[0027] The value of each hyperparameter is selected from the range of values of each hyperparameter through a random search function.
[0028] In one embodiment, the initial model includes an iterative weak learning machine model.
[0029] In a second aspect, an exemplary embodiment of the present disclosure provides a vehicle driving analysis system, the system comprising:
[0030] An acquisition module, used to acquire the current driving data sequence;
[0031] A first determination module is used to determine a current driving vector sequence corresponding to a current driving data sequence; the current driving vectors in the current driving vector sequence correspond one-to-one to the current driving data in the current driving data sequence;
[0032] The label analysis module is used to input the current driving vector sequence into the trained vehicle driving analysis model to obtain a target analysis label set corresponding to the current driving vector sequence; the target analysis labels in the target analysis label set correspond one-to-one to the current driving vector in the current driving vector sequence;
[0033] The result analysis module is used to determine the safety analysis result of vehicle driving according to the target analysis tag set.
[0034] In one embodiment, the first determining module includes:
[0035] An extraction unit, used for performing feature extraction on each current driving data in the current driving data sequence to obtain a current driving vector corresponding to each current driving data;
[0036] The first determining unit is used to determine a current driving vector sequence according to a current driving vector corresponding to each current driving data.
[0037] In one embodiment, the current driving data sequence includes current first driving data and current second driving data;
[0038] Before executing the acquisition module, it also includes:
[0039] The collection module is used to collect initial driving data every first preset time period.
[0040] The updating module is used to update the initial driving data to the current first driving data; and to update the first driving data corresponding to the previous first preset duration to the current second driving data.
[0041] In one embodiment, the system further comprises a training module for training and obtaining a vehicle driving analysis model;
[0042] The training modules include:
[0043] An acquisition unit, used for acquiring a sample data set, the sample data set comprising a plurality of sample driving vectors and a preset analysis label corresponding to each of the plurality of sample driving vectors;
[0044] A first building unit, used to build an initial model;
[0045] An analysis unit, configured to perform a vehicle driving trajectory analysis on each sample driving vector among the plurality of sample driving vectors based on an initial model, and obtain a prediction analysis label corresponding to each sample driving vector;
[0046] A second construction unit is used to construct a loss function according to the prediction analysis label and the preset analysis label corresponding to each sample driving vector;
[0047] An updating unit, used to update the initial model according to the loss function;
[0048] The second determination unit is used to determine the initial model as the vehicle analysis model when the loss function iterates to a minimum value.
[0049] In one embodiment, the updating unit is further configured to:
[0050] Determine multiple hyperparameters of the initial model and the value range of each hyperparameter in the multiple hyperparameters; select the value of each hyperparameter from the value range of each hyperparameter through a random search function.
[0051] In one embodiment, the initial model includes an iterative weak learning machine model.
[0052] In a third aspect, an exemplary embodiment of the present disclosure provides an electronic device, the electronic device comprising a processor and a memory, the memory storing at least one instruction, at least one program, a code set or an instruction set, the at least one instruction, at least one program, a code set or an instruction set being loaded and executed by the processor to implement a vehicle driving analysis method as described in any one of the first aspects.
[0053] In a fourth aspect, an exemplary embodiment of the present disclosure provides a computer-readable storage medium, in which at least one instruction or at least one program is stored, and the at least one instruction or at least one program is loaded and executed by a processor to implement a vehicle driving analysis method as described in any one of the first aspects.
[0054] The vehicle driving analysis method, system, device and medium provided by the embodiments of the present disclosure have the following technical effects:
[0055] The vehicle's current driving data sequence is converted into a current driving vector sequence, and the current vector sequence target analysis label set is determined based on the vehicle driving analysis model to determine the vehicle's driving safety analysis result. The vehicle driving analysis model provided in this embodiment analyzes a current driving vector sequence that may contain multiple current driving vectors, so that the vehicle driving analysis model can process large-scale data. When analyzing the scenario in which the vehicle is driving, the accuracy of the vehicle driving analysis is improved.
[0056] 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
[0057] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present disclosure or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0058] Figure 1 is a schematic diagram of an application environment provided by an embodiment of the present disclosure;
[0059] Figure 2 is a first flow chart of a vehicle driving analysis method provided by an embodiment of the present disclosure;
[0060] Figure 3 is a flow chart of step S21 of a vehicle driving analysis method provided by an embodiment of the present disclosure;
[0061] Figure 4 is a flow chart of step S22 of a vehicle driving analysis method provided by an embodiment of the present disclosure;
[0062] Figure 5 is a second flow chart of a vehicle driving analysis method provided by an embodiment of the present disclosure;
[0063] Figure 6 is a structural diagram of a vehicle driving analysis system provided by an embodiment of the present disclosure;
[0064] Figure 7 It is a structural diagram of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION
[0065] The following will be combined with the drawings in the embodiments of the present disclosure to clearly and completely describe the technical solutions in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present disclosure.
[0066] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products, or devices.
[0067] See also Figure 1 , Figure 1 1 is a schematic diagram of an application environment provided by an embodiment of the present application, the schematic diagram includes a vehicle 101 and a server 102, wherein, in an optional implementation, the server 102 may be an onboard server disposed in the vehicle 101, the onboard server includes a vehicle driving analysis model, and can obtain desired data in real time, so as to prepare for subsequent analysis of the vehicle's driving safety. In another optional implementation, the vehicle 101 may be provided with its own onboard server, and the onboard server and Figure 1 The server 102 shown in the figure is not the same server. After the vehicle-mounted server transmits the acquired data to the server 102, the server can complete the subsequent steps and finally obtain the security analysis result. The vehicle-mounted server involved in the first case and the server involved in the second case are collectively referred to as servers.
[0068] Specifically, the server 102 obtains the current driving data sequence of the vehicle 101. At the same time, the server 102 can convert the current driving data sequence into a current driving vector sequence. Finally, the server 102 inputs the current driving vector sequence into the vehicle driving analysis model and determines the vehicle safety analysis result according to the output result of the vehicle driving analysis model.
[0069] Since the vehicle 101 continuously generates a large amount of current driving data during driving, the current driving data in this scenario can be regarded as stream data, which refers to a real-time data stream that is continuously generated and transmitted. This data is continuous, dynamic, and often highly time-sensitive. Compared with traditional batch data, stream data usually has a higher arrival speed and shorter processing time requirements.
[0070] In order to realize the processing capability of the server 102 for large-scale streaming data, the server 102 provided in this embodiment preferably adopts FLink (a stream processing framework), but is not limited thereto.
[0071] Flink is a distributed stream processing framework and engine that is specifically designed to process large-scale data streams. In practical applications, Flink can process both stream data (real-time data) and batch data (static data) and supports high-throughput, low-latency data processing. This makes it very suitable for scenarios that require real-time analysis, monitoring, and complex event processing.
[0072] The combination of Flink and Kafka can form a powerful tool for processing streaming data. The following is a brief description of the processing flow of Flink and Kafka.
[0073] Kafka is used as a data source. Kafka is a high-throughput distributed message queue system, which is often used as a source of streaming data. In this embodiment, the current driving data collected by vehicle 101 will be cached through Kafka, and Flink will be configured to read the current driving data from the Kafka message topic.
[0074] Flink processes streaming data. Flink is a stream processing framework with low latency and high throughput. It can process current driving data from Kafka, which will be described in detail in the following embodiments. In addition, the processing and training of the vehicle driving analysis model in this embodiment are all performed in Flink.
[0075] Kafka acts as a data receiver. The data processed by Flink, that is, the target analysis tags obtained in this embodiment, can be written back to the Kafka topic for use by other consumers.
[0076] The following describes a specific embodiment of a vehicle driving analysis method of the present application. Figure 2 It is a flowchart of a vehicle driving analysis method provided in an embodiment of the present application. This specification provides method operation steps such as the embodiment or flowchart, but may include more or fewer operation steps based on conventional or non-creative labor. The order of steps listed in the embodiment is only one way of executing the steps among many orders, and does not represent the only order of execution. When the actual system or server product is executed, it can be executed in the order shown in the embodiment or the accompanying drawings or in parallel (for example, in a parallel processor or multi-threaded processing environment). Specifically, Figure 2 As shown, the method may include:
[0077] S21. Obtain current driving data sequence.
[0078] Among them, a current driving data is the driving data generated by the vehicle during the driving process, including sub-data of at least one dimension collected at the same time node, such as trajectory sub-data, GPS (Global Positioning System) sub-data, acceleration sub-data, etc.
[0079] The current driving data sequence includes a plurality of current driving data, and specifically may include current driving data continuously generated within a second preset time period.
[0080] In one embodiment, the current driving data sequence includes current first driving data and current second driving data.
[0081] See also Figure 3 , before step S21, including:
[0082] S211. Collect initial driving data every first preset time period.
[0083] S212: Update the initial driving data to the current first driving data.
[0084] S213: Update the first driving data corresponding to the previous first preset duration to the current second driving data.
[0085] For the current driving data of the above-mentioned stream data type, a corresponding current driving data sequence is generated through a sliding time window, and the sliding time window is set to a second preset time length. The length of the second preset time length can be set according to actual conditions and is not specifically limited here.
[0086] For the current driving data sequence, the current first driving data is at the head of the sequence, and the current second driving data is at the tail of the sequence. After every first preset time, the server will obtain new initial driving data and put it into the head of the current driving data sequence. Since the length of the sliding time window is fixed, in this embodiment, the current first driving data will be updated to the current second driving data, and the second driving data corresponding to the previous first preset time will be discarded.
[0087] In one embodiment, the current driving data is collected based on sensors installed on the vehicle 101. The current driving data can be reported to the server 102 via a Tbox (TelematicsBOX, a remote information processor) terminal or a message middleware, such as Kafka, MQTT (Message Queuing Telemetry Transport, a lightweight message transmission protocol).
[0088] For the reported current driving data, the server 102 can collect the generated current driving data through Logstash (data processing pipeline tool) and perform preliminary processing, such as cleaning or removing illegal data, etc. After processing the current driving data, Logstash sends the current driving data to Kafka under FLink for caching.
[0089] S22. Determine a current driving vector sequence corresponding to the current driving data sequence.
[0090] See also Figure 4 , step S22 specifically includes:
[0091] S221. Perform feature extraction on each current driving data in the current driving data sequence to obtain a current driving vector corresponding to each current driving data.
[0092] For example, a current driving data includes sub-data of the following four dimensions: longitude and latitude value t1 (Geo function converts longitude and latitude), speed value t2, acceleration value t3, and steering angle value t4. Feature extraction is performed on the above current driving data to obtain the corresponding current driving vector X=[x1,x2,x3,x4].
[0093] S222. Determine a current driving vector sequence according to a current driving vector corresponding to each current driving data in the current driving data sequence.
[0094] Among them, the current driving vector sequence can be expressed by C i Characterization, C i ={X i,k |k=t0...t n}, C i Represents the current driving vector X of the i-th current driving vector sequence at time k.
[0095] In addition, during the vehicle driving analysis process, different brands and / or different models will have a certain impact on the output target analysis labels. In order to fully consider the association between different brands and / or different models and the target analysis labels and reduce the impact of such factors on the results, in this embodiment, vectors representing brands and models are introduced into the current driving vector sequence to facilitate the processing of the vehicle analysis model, which can be specifically processed through unique hot encoding.
[0096] One-Hot Encoding is a technique for converting categorical data into numerical data. It is widely used in machine learning and data preprocessing. The basic idea is to represent each category as a unique binary vector. i =one_hot(C i), each different brand and / or different model will be represented as a unique binary vector with only one position set to 1 and the rest set to 0.
[0097] In one embodiment, before step S22, the method further includes: preprocessing the current driving data sequence. The preprocessing includes but is not limited to data deletion, data filling and other operations on the current driving data in the current driving data sequence.
[0098] Specifically, since the current driving data uploaded by the vehicle 101 is time series data and has a correlation before and after, in this embodiment, the missing current driving data can be filled by selecting a linear difference. The missing current driving data can be filled with reference to the following formula:
[0099]
[0100] Among them, the start and end time of the sliding window are set to (x1, x2), the current driving data corresponding to the start and end time is (y1, y2), x is the time point corresponding to the missing current driving data, and y is the missing current driving data.
[0101] At the same time, data that does not affect the prediction results of the target analysis label can be deleted directly.
[0102] S23. Input the current driving vector sequence into the trained vehicle driving analysis model to obtain a target analysis label set corresponding to the current driving vector sequence.
[0103] Among them, the target analysis tags in the target analysis tag set correspond one-to-one to the current driving vector in the current driving vector sequence.
[0104] The target analysis tag is used to characterize whether the corresponding current driving vector has an abnormality. Since the current driving vector corresponds to the current driving data one by one, the target analysis tag can also characterize whether the corresponding current driving data has an abnormality.
[0105] The target analysis label can indicate that the current driving data has an anomaly, or that a dimension in the current driving data has an anomaly. Specifically, it can be performed in binary, where 0 indicates no anomaly and 1 indicates an anomaly. For example, if the current driving data has three dimensions, 001 indicates that the first dimension of the current driving data has an anomaly, and so on. I will not repeat it here.
[0106] In addition, step S22-step S23 is performed in Flink, and the target analysis tag will be output as the output result of the model and will be output to Kafka for caching for use by other consumers or by the vehicle analysis model provided in this embodiment.
[0107] S24. Determine a safety analysis result of vehicle driving according to the target analysis tag set.
[0108] Among them, other consumers, such as prediction models, can obtain the target analysis tag set from Kafka and determine the security analysis results based on the target analysis tag set.
[0109] The safety analysis results can indicate whether the vehicle's driving process is safe or whether there is a possibility of a traffic accident. The specific selection can be made based on the actual situation.
[0110] In this embodiment, a vehicle driving analysis model and a streaming analysis engine, namely Flink, are combined to collect and analyze the current driving data generated during the vehicle driving process in real time. At the same time, the middleware Kafka is used to improve the ability to process data. It is possible to perform streaming analysis on a large amount of current driving data generated during the vehicle driving process, improve the processing capability of large-scale data, and monitor the vehicle driving safety in real time.
[0111] In one embodiment, the method further comprises the step of training a vehicle driving analysis model, see Figure 5 , specifically including:
[0112] S51. Obtain a sample data set.
[0113] The sample data set includes a plurality of sample driving vectors and a preset analysis label corresponding to each of the plurality of sample driving vectors.
[0114] Regarding the sample driving vector, the sample driving vector is determined according to the sample driving data. The sample driving data may be the historical driving data of the vehicle during driving, or the current driving data of the vehicle after the vehicle driving analysis model is loaded. The server 102 obtains the real-time data of the vehicle to perform real-time training on the vehicle driving analysis model. In addition, after the vehicle driving analysis model is trained, the server 102 may also update and retrain it.
[0115] The process of converting the sample driving data into the sample driving vector can refer to the process of converting the current driving data into the current driving vector, which will not be repeated here.
[0116] Regarding the preset analysis label, the preset analysis label is used to characterize whether the corresponding sample driving vector is abnormal. Since the sample driving vector corresponds to the sample driving data one by one, the preset analysis label can also characterize whether the corresponding sample driving data is abnormal. The preset analysis label is the actual analysis result corresponding to the sample driving vector, while the predicted analysis label and the target analysis label are the analysis results output by the model.
[0117] The preset analysis label can be determined based on any one of an expert knowledge base, a threshold determination method, and the like.
[0118] S52. Construct an initial model.
[0119] Among them, the initial model preferably adopts a continuously iterative weak learning machine model (XGboost), but is not limited to this.
[0120] XGBoost is an improvement based on Gradient Boosting Trees, and its performance is greatly improved compared to other models. Specifically, first, XGBoost uses data parallel processing and algorithm optimization technology during training, making it more efficient than traditional gradient boosting methods; second, XGBoost provides L1 (Lasso) and L2 (Ridge) regularization to prevent model overfitting and improve the generalization ability of the model; third, XGBoost can automatically handle missing values; fourth, XGBoost supports multiple types of tasks such as regression, classification, sorting, and custom objective functions; fifth, XGBoost has a built-in cross-validation function that can adjust model parameters and select the best hyperparameter combination; sixth, XGBoost can provide feature importance scores, which helps understand the decision-making process of the model.
[0121] S53: Based on the initial model, perform vehicle driving trajectory analysis on each sample driving vector in the multiple sample driving vectors to obtain a prediction analysis label corresponding to each sample driving vector.
[0122] S54. Construct a loss function according to the prediction analysis label and the preset analysis label corresponding to each sample driving vector.
[0123] The core of Xgboost is to continuously iterate the weak learning machine (decision tree). The next decision tree generation fully considers the previous prediction results, so that the loss function between the model's predictive analysis label and the preset analysis label is minimized, that is, the loss function is used as the optimization target.
[0124] The loss function is determined as:
[0125]
[0126] in, Represents the original loss function. For whether there is an abnormality in vehicle driving, the logarithmic loss function is selected.
[0127] The logarithmic loss function is:
[0128]
[0129] Among them, y iis the preset analysis label of the ith item (0 indicates no abnormality, 1 indicates an abnormality). is a predictive analysis tag, Represents the regularization term, which is used to control the complexity of the model.
[0130] Pay attention to the need for certain predictive capabilities in judging driving abnormalities, that is, the model must have good generalization capabilities and select the L2 regularization term.
[0131] The L2 regularization term is:
[0132]
[0133] Where Ω(f k ) is the regularization term of the kth decision tree, λ is a parameter used to control the strength, T is the number of leaf nodes, and w j is the weight of the jth leaf node.
[0134] S55. Update the initial model according to the loss function.
[0135] In order to prevent the XGBoost model from overfitting and increase its generalization ability, parameter optimization was used in this case. Parameter debugging includes: the number of decision trees, tree depth, model learning rate, sampling ratio parameters, and time series difference window size for continuous iterative debugging.
[0136] Improve the tuning efficiency, use random search (RandomizedSearchCV) to search for the best transmission parameter combination and evaluate it, ensure the best model performance, and minimize the loss function as much as possible. Step S55 specifically includes:
[0137] Determine multiple hyperparameters of the initial model and the value range of each hyperparameter in the multiple hyperparameters; select the value of each hyperparameter from the value range of each hyperparameter through a random search function.
[0138] S56. When the loss function iterates to a minimum value, the initial model is determined as the vehicle analysis model.
[0139] An exemplary embodiment of the present disclosure provides a vehicle driving analysis system. Figure 6 , the system includes:
[0140] The acquisition module 61 is used to acquire the current driving data sequence.
[0141] The first determination module 62 is used to determine a current driving vector sequence corresponding to the current driving data sequence; the current driving vectors in the current driving vector sequence correspond one-to-one to the current driving data in the current driving data sequence.
[0142] The label analysis module 63 is used to input the current driving vector sequence into the trained vehicle driving analysis model to obtain a target analysis label set corresponding to the current driving vector sequence; the target analysis labels in the target analysis label set correspond one-to-one to the current driving vector in the current driving vector sequence.
[0143] The result analysis module 64 is used to determine the safety analysis result of the vehicle driving according to the target analysis tag set.
[0144] In one embodiment, the first determination module 62 includes:
[0145] An extraction unit, used for performing feature extraction on each current driving data in the current driving data sequence to obtain a current driving vector corresponding to each current driving data;
[0146] The first determining unit is used to determine a current driving vector sequence according to a current driving vector corresponding to each current driving data.
[0147] In one embodiment, the current driving data sequence includes current first driving data and current second driving data;
[0148] Before executing the acquisition module 61, the method further includes:
[0149] The collection module is used to collect initial driving data every first preset time period.
[0150] The updating module is used to update the initial driving data to the current first driving data; and to update the first driving data corresponding to the previous first preset duration to the current second driving data.
[0151] In one embodiment, the system further comprises a training module for training and obtaining a vehicle driving analysis model;
[0152] The training modules include:
[0153] An acquisition unit, used for acquiring a sample data set, the sample data set comprising a plurality of sample driving vectors and a preset analysis label corresponding to each of the plurality of sample driving vectors;
[0154] A first building unit, used to build an initial model;
[0155] An analysis unit, configured to perform a vehicle driving trajectory analysis on each sample driving vector among the plurality of sample driving vectors based on an initial model, and obtain a prediction analysis label corresponding to each sample driving vector;
[0156] A second construction unit is used to construct a loss function according to the prediction analysis label and the preset analysis label corresponding to each sample driving vector;
[0157] An updating unit, used to update the initial model according to the loss function;
[0158] The second determination unit is used to determine the initial model as the vehicle analysis model when the loss function iterates to a minimum value.
[0159] In one embodiment, the updating unit is further configured to:
[0160] Determine multiple hyperparameters of the initial model and the value range of each hyperparameter in the multiple hyperparameters; select the value of each hyperparameter from the value range of each hyperparameter through a random search function.
[0161] In one embodiment, the initial model includes an iterative weak learning machine model.
[0162] The system and method embodiments in the disclosed embodiments are based on the same disclosed concepts.
[0163] An exemplary embodiment of the present disclosure provides an electronic device, which may be the server 102 mentioned in the above embodiment, or may be other devices. Figure 7 As shown, the electronic device 700 may have relatively large differences due to different configurations or performances, and may include one or more central processing units (CPU) 710 (the processor 710 may include but is not limited to a processing system such as a microprocessor MCU or a programmable logic device FPGA), a memory 730 for storing data, and one or more storage media 720 (such as one or more mass storage devices) for storing application programs 723 or data 722. Among them, the memory 730 and the storage medium 720 can be short-term storage or permanent storage. The storage medium 720 stores at least one instruction, at least one program, code set or instruction set, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by the processor to implement the above method.
[0164] Furthermore, the CPU 710 may be configured to communicate with the storage medium 720 to execute a series of instruction operations in the storage medium 720 on the electronic device 700. The electronic device 700 may also include one or more power supplies 760, one or more wired or wireless network interfaces 70, one or more input and output interfaces 740, and / or one or more operating systems 721, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.
[0165] The input / output interface 740 may be used to receive or send data via a network. The specific example of the network may include a wireless network provided by a communication provider of the electronic device 700. In one example, the input / output interface 740 includes a network adapter (Network Interface Controller, NIC), which may be connected to other network devices via a base station so as to communicate with the Internet. In one example, the input / output interface 740 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.
[0166] It can be understood by those skilled in the art that Figure 7 The structure shown is only for illustration and does not limit the structure of the above electronic system. Figure 7 More or fewer components as shown, or with Figure 7 Different configurations shown.
[0167] An embodiment of the present disclosure also provides a computer-readable storage medium, which can be set in a server to store at least one instruction, at least one program, a code set or an instruction set related to a method in the method embodiment, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the above method.
[0168] Optionally, in this embodiment, the storage medium may be located in at least one of the multiple network servers of the computer network. Optionally, in this embodiment, the storage medium may include, but is not limited to, various media that can store program codes, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk, or an optical disk.
[0169] It should be noted that the above-mentioned sequence of the embodiments of the present disclosure is for description only and does not represent the advantages and disadvantages of the embodiments. The above-mentioned specific embodiments of this specification are described. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0170] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the xx method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the xx method embodiment.
[0171] A person skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware or by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a disk or an optical disk, etc.
[0172] The above description is only a preferred embodiment of the present disclosure and is not intended to limit the present disclosure. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present disclosure should be included in the protection scope of the present disclosure.
Claims
1. A vehicle driving analysis method, characterized in that: The method comprises: Get the current driving data sequence; Determining a current driving vector sequence corresponding to the current driving data sequence; Inputting the current driving vector sequence into a trained vehicle driving analysis model to obtain a target analysis label set corresponding to the current driving vector sequence; the target analysis labels in the target analysis label set correspond one-to-one to the current driving vector in the current driving vector sequence; A safety analysis result of the vehicle driving is determined according to the target analysis tag set.
2. The method according to claim 1, characterized in that The determining of a current driving vector sequence corresponding to the current driving data sequence comprises: Performing feature extraction on each current driving data in the current driving data sequence to obtain a current driving vector corresponding to each current driving data; The current driving vector sequence is determined according to the current driving vector corresponding to each current driving data in the current driving data sequence.
3. The method according to claim 1, characterized in that The current driving data sequence includes current first driving data and current second driving data; Before obtaining the current driving data sequence, the method includes: Collecting initial driving data at first preset time intervals; The initial driving data is updated to the current first driving data; and the first driving data corresponding to the previous first preset duration is updated to the current second driving data.
4. The method according to claim 1, characterized in that The method also includes the step of training and obtaining a vehicle driving analysis model; The vehicle analysis model obtained by training includes: Acquire a sample data set, wherein the sample data set includes a plurality of sample driving vectors and a preset analysis label corresponding to each of the plurality of sample driving vectors; Build an initial model; Based on the initial model, performing a vehicle driving trajectory analysis on each sample driving vector of the multiple sample driving vectors to obtain a prediction analysis label corresponding to each sample driving vector; Constructing a loss function according to the prediction analysis label and the preset analysis label corresponding to each of the sample driving vectors; Updating the initial model according to the loss function; When the loss function iterates to a minimum value, the initial model is determined as the vehicle analysis model.
5. The method according to claim 4, characterized in that The updating of the initial model according to the loss function comprises: Determining a plurality of hyperparameters of the initial model and a value range of each of the plurality of hyperparameters; The value of each hyperparameter is selected from the value range of each hyperparameter through a random search function.
6. The method according to claim 4, characterized in that The initial model includes a continuously iterative weak learning machine model.
7. A vehicle driving analysis system, characterized in that: The system comprises: An acquisition module, used to acquire the current driving data sequence; A first determining module is used to determine a current driving vector sequence corresponding to the current driving data sequence; the current driving vectors in the current driving vector sequence correspond one-to-one to the current driving data in the current driving data sequence; a label analysis module, used for inputting the current driving vector sequence into a trained vehicle driving analysis model to obtain a target analysis label set corresponding to the current driving vector sequence; the target analysis labels in the target analysis label set correspond one-to-one to the current driving vector in the current driving vector sequence; A result analysis module is used to determine the safety analysis result of the vehicle driving according to the target analysis tag set.
8. The system according to claim 7, characterized in that The first determining module includes: an extraction unit, configured to perform feature extraction on each current driving data in the current driving data sequence to obtain a current driving vector corresponding to each current driving data; The first determining unit is used to determine the current driving vector sequence according to the current driving vector corresponding to each current driving data.
9. An electronic device, characterized in that: The electronic device includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, at least one program, a code set or an instruction set is loaded and executed by the processor to implement the vehicle driving analysis method as described in any one of claims 1-6.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores at least one instruction or at least one program, and the at least one instruction or at least one program is loaded and executed by a processor to implement the vehicle driving analysis method as described in any one of claims 1-6.